A1 2 JAFOCorreia

ICDS12
International Conference DURABLE STRUCTURES: from construction to rehabilitation

LNEC Lisbon Portugal 31 May - 1 June 2012
DURABLE STRUCTURES
12
LNEC Lisbon 31 May - 1June 2012
PROBABILISTICFATIGUEBEHAVIOUROFASTRUCTURAL
DETAILOFPUDDLEIRONFROMTHEEIFFELBRIDGE
J.A.F.O.Correia1,A.M.P.DeJesus1 andA.FernndezCanteli2
DS
1 UnitofDesignandExperimentalValidation,IDMEC,LAETA,
EngineeringDepartment,SchoolofSciencesandTechnology
UniversityofTrsosMontesandAltoDouro,
5001801VilaReal,Portugal.
Email:jcorreia@utad.pt;ajesus@utad.pt
2 DepartmentofConstructionandManufacturingEngineering,UniversityofOviedo,
33203CampusViesques,Gijn,Spain.
Email:afc@uniovi.es
IC
DURABLE STRUCTURES
LNE
C
1. INTRODUCTION
2. OVERVIEW OF THE DETERMINISTIC UNIGROW MODEL
3. PROBABILISTIC SWTN FIELDS
4. PROCEDURE TO GENERATE PSNR FIELDS

5. PROBABILISTIC FATIGUE DATA OF THE PUDDLE IRON FROM THE EIFFEL
(VIANA) BRIDGE
6. PREDICTION OF PROBABILISTIC SN FIELDS FOR A NOTCHED DETAIL
7. CONCLUSIONS
ACKNOWLEDGMENTS
ICDS12
DURABLE STRUCTURES
1. INTRODUCTION (1/2)
12
Probabilistic fatigue models are required to account conveniently for the

several sources of uncertainty arising in the prediction procedures, such as the
scatter in material behaviour.
DS
In this paper, a probabilistic approach is proposed to generate SN curves for

stress ratio R=0, applied to a structural detail of puddle iron from the Eiffel
Bridge, using local approaches based on probabilistic SWTN fields, in order to
model both fatigue crack initiation and fatigue crack propagation.
IC
DURABLE STRUCTURES
1. INTRODUCTION (2/2)
LNE
C
This study suggests a progress of the model proposed by Noroozi et al. to model
fatigue crack propagation, based on the local strain approach to fatigue, to
determine the crack propagation for a structural detail, and is applied to derive
probabilistic fatigue crack propagation fields (PSNpR fields) for a notched
detail, for distinct stress Rratios.
The probabilistic fatigue crack initiation fields (PSNiR fields), for notched plate,
are determined using an elastoplastic approach together with the PSWTN fields
to calculate the fatigue damage of the first elementary material block.
The global prediction (crack initiation and propagation) of the probabilistic SN
fields is presented as a unified approach. The predictions are compared with
available experimental SN fatigue data for the notched plate under
consideration.
ICDS12
DURABLE STRUCTURES
12
2. OVERVIEW OF THE DETERMINISTIC UNIGROW MODEL (1/3)

The UniGrow model was proposed by Noroozi et al. based on
the following assumptions:
The material is composed of simple particles of a finite

dimension * that represents the elementary material
block size, below which material cannot be regarded as a
continuum.
The fatigue crack tip is considered equivalent to a notch

with radius *.
The fatigue crack growth process is considered as being

successive crack increments due to crack reinitiations over
the distance *.
The fatigue crack growth rate can be determined as:
This equation may be used in an inverse way, to estimate

the * parameter, using constantamplitude fatigue crack
growth data for a material at various Rratios.
DS
IC
Noroozi,AH,Glinka,G,andLambertS.,
5
InternationalJournalofFatigue2005;
27:127796.
DURABLE STRUCTURES
LNE
C
The failure of the elementary material block may be computed using the Smith
WatsonTopper (SWT) relation:
where max and are the average values of the maximum elastoplastic stress and
strain range, through the first elementary material block ahead the crack tip.
This work suggests an extension of the model proposed by Noroozi et al. to model
fatigue crack propagation:
The stress intensity factors are determined for the detail under investigation
based on a finite element analysis, using the Jintegral method. Then, the elastic
stress fields are estimated ahead of the crack tip.
The actual elastoplastic stresses and strains, are computed using the Neubers
approach.
The residual stress distribution ahead the crack tip is computed using the actual
elastoplastic stresses computed at the end of the first load reversal and6
subsequent cyclic elastoplastic stress range, r=max.
ICDS12
DURABLE STRUCTURES
12
This work suggests an extension of the model proposed by Noroozi et al. to model fatigue
crack propagation (continuation):
The residual stress intensity factor, Kr, is computed using the weight function method
according to the following general expression:
DS
The weight function was found using the following expression:
The applied stress intensity factor (maximum and range values) is corrected using the
residual stress intensity value, resulting the total values, Kmax,tot and Ktot. For positive
applied stress ratios, Kmax,tot and Ktot may be computed as follows:
where Kr takes a negative value corresponding to the compressive stress field.
IC
Using the total values of the stress intensity factors, the first and second steps above are
repeated to determine the corrected values for the maximum actual stress and actual
strain range at the material representative elements.
7
DURABLE STRUCTURES
3. PROBABILISTIC SWTN FIELDS (1/2)
LNE
C
The UniGrow model needs a fatigue damage law to compute the number of cycles
to fail the elementary material block.
In this study, a probabilistic fatigue model is proposed rather than the
deterministic SWTN model.
The probabilistic SWTN model is developed as an extension of the PN model

proposed by Castillo and FernndezCanteli.
The SWT (=max.a) parameter was proposed by SmithWatsonTopper to take into

account the mean stress effects on fatigue life. Any combination of maximum
stress and strain amplitude that leads to the same SWT parameter predicts the
same fatigue life. The SWTN and aN fields exhibit similar characteristics.
ICDS12
DURABLE STRUCTURES
12
3. PROBABILISTIC SWTN FIELDS (2/2)
ThereforethePN fieldproposedbyCastilloandFernndezCanteli maybeextended

torepresentthePSWTN field:
DS
where p is the probability of failure, N0 and

SWT0 are normalizing values, and , and
are the nondimensional Weibull model
parameters. Their physical meanings are:
N0: threshold value of lifetime;
SWT0: fatigue limit of SWT;
: parameter defining the position of the
corresponding zeropercentile curve;
: scale parameter;
: shape parameter.
IC
The parameters log N0 and log SWT0 of the PSWTN model may be estimated using
the least squares method. The Weibull parameters are estimated by the maximum
9
likelihood method.
DURABLE STRUCTURES
4. PROCEDURE TO GENERATE PSNR FIELDS (1/3)
LNE
C
PREDICTIONOFPROBABILISTICSNi FIELDSFORASTRUCTURALDETAIL
10
ICDS12
DURABLE STRUCTURES
12
DS
PREDICTION OF PROBABILISTIC SNp

FIELDS FOR A STRUCTURAL DETAIL
IC
11
DURABLE STRUCTURES
LNE
C
GLOBALPREDICTIONOFPROBABILISTICSN FIELDSFORASTRUCTURALDETAIL
12
ICDS12
DURABLE STRUCTURES
12
5. PROBABILISTIC FATIGUE DATA OF THE PUDDLE IRON FROM

THE EIFFEL (VIANA) BRIDGE (1/2)
The elastic, static strength, cyclic and strainlife properties of the puddle iron from the
Eiffel Bridge are summarized in Tables 1 and 2.
Table2
DS
Table1
A notched detail consisting plate with a circular symmetric hole, as illustrated in Figure,
was considered in this investigation.
IC
13
DURABLE STRUCTURES
5. PROBABILISTIC FATIGUE DATA OF THE PUDDLE IRON FROM

THE EIFFEL (VIANA) BRIDGE (2/2)
The probabilistic SWTN model and the
respective parameters identified for this
material are applied to model both crack
initiation and crack propagation lives of a
notched detail.
LNE
C
This geometry was fatigue tested under

remote stress controlled conditions, for
stress ratio, R=0.0.
14
ICDS12
DURABLE STRUCTURES
12
6. PREDICTION OF PROBABILISTIC SN FIELDS FOR A NOTCHED DETAIL (1/10)

Modellingfatiguecrackinitiation
DS
The PSWTN model is used to model the fatigue crack initiation (failure of the first
elementary material block) at the notch root of the detail. Since the detail was tested
under elastoplastic conditions, elastoplastic stress analysis must be used to assess the real
stress conditions at the notch root.
An elastoplastic stress analysis is proposed based on SeegerHeulers model, which is a

generalization of the Neubers model. This model is used in conjunction with the Ramberg
Osgood description of the material. The initial loading is modelled using appropriate
monotonic properties of the material. The following set of equations was used to describe
the monotonic loading:
IC
15
DURABLE STRUCTURES

Modellingfatiguecrackinitiation(continuation)
LNE
C
Forcyclicloading,thefollowingequationswereusedtoassessthelocalstressrange:
SeegerHeulers model are used to compute the maximum stress and the stress range at
the notch root, respectively, and relate them to the nominal or remote stresses, applied
to the notch.
Using the probabilistic SWTN model for the puddle iron from the Eiffel Bridge, it is
possible to relate the nominal or remote stresses applied to the notched detail, with
the number of cycles to failure, for a given probability, p.
16
ICDS12
DURABLE STRUCTURES
12

DS
The notch detail shows an elastic stress concentration factor, kt, equal to 2.43, which is the
limit value of the fatigue notch reduction factor to be used in the elastoplastic fatigue
analysis.
IC
17
DURABLE STRUCTURES

LNE
C
The fatigue reduction factor, kf, was calibrated in order the resulting PSNi field to give a
satisfactory description of all available experimental fatigue data. The resulting fatigue
notch reduction factor is equal to 1.45. The fatigue reduction factor effects were only
accounted in modelling fatigue crack initiation. Figure shows the PSNi fields
corresponding to the fatigue crack initiation (dominating fatigue damage) for the notched
detail, using the PSWTN model for R=0.0.
18
ICDS12
DURABLE STRUCTURES
12
6.PredictionofprobabilisticSNfieldsforanotcheddetail(5/10)
Modellingfatiguecrackpropagation
The fatigue crack propagation modelling was supported by a bidimensional finite

element model of the notched detail, which was used to compute both stress intensity
factor and residual elastoplastic stress distribution.
DS
Taking into account the existing symmetry plane, only half of the geometry is modelled.
Plane stress quadratic triangular elements were used in the analysis due to the limited
specimen thickness.
The stress intensity factors were determined based on a linearelastic finite element
analysis using the Jintegral method.
Concerning the residual elastoplastic stress computation, a highly refined mesh at the
crack tip region was used in order to model the crack tip notch radius, *. The residual
stress intensity factor, Kr, was determined using the weight function method.
IC
A Von Mises yield model, with multilinear kinematic hardening, was used in simulations
using ANSYS 12.0 code. The plasticity model was fitted to the cyclic curve of the
19
material.
DURABLE STRUCTURES
LNE
C
Bidimensionalfiniteelementmodel residualelastoplasticstressdistribution
20
ICDS12
DURABLE STRUCTURES
12
DS
Bidimensionalfiniteelementmodel stressintensityfactorandweightfunction
IC
21
DURABLE STRUCTURES
LNE
C
The PSNp field was predicted using the PSWTN model of puddle iron from the Eiffel
Bridge together with the UniGrow model proposed by Noroozi et al. to structural detail,
for R=0, and using the * determined by Jos A.F.O. Correia et al (Correia JAFO, Jesus AMP,
FernndezCanteli A. A procedure to derive probabilistic fatigue crack propagation data. International Journal
Structural Integrity, ISSN 17579864, Issue 2, Vol. 3, 2012, p. 158183).
Weight functions used to determine the Kr for

notched detail, for the distinct crack lengths.
Stress intensity evolution with the crack length, for a unit

22
remote stress, which was used to determine the Kapplied.
ICDS12
DURABLE STRUCTURES
12
DS
EvolutionofKr withtheappliedstressintensityfactorrange
IC
23
DURABLE STRUCTURES
LNE
C
Global SN field predictions (crack initiation and crack propagation SN field)
Fatigue crack propagation had a small contribution to the total fatigue life for this detail.
24
ICDS12
DURABLE STRUCTURES
12
Conclusions
The global PSN field prediction for the notched detail (R=0.0), taking into account
the fatigue crack initiation and propagation, shows satisfactory results.
DS
The adaptation of the UniGrow model allows us to reproduce satisfactorily crack

propagation prediction using residual compressive stress estimation, based on
elastoplastic finite element analysis of the notched detail, and the PSWTN damage
model.
In this study, crack initiation is the dominating damaging process, while the fatigue
crack propagation represents a small influence on global predictions of the PSN
field. However its importance increases with the fatigue life, enhancing the slope of
the PSN field.
The procedure proposed to derive the probabilistic SN curves for structural details
proved to be quite efficient, since it can be used to reduce the need for extensive
testing and takes into account material variability.
IC
25
DURABLE STRUCTURES
ACKNOWLEDGMENTS
LNE
C
The authors acknowledge the Portuguese Science Foundation (FCT) for the
financial support through the doctoral grant SFRH/BD/66497/2009.
JosCorreia
Email:jcorreia@utad.pt
http://www.utad.pt/~jcorreia
http://www.utad.pt/~pontesfct
26

A1 2 JAFOCorreia

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

A1 2 JAFOCorreia

Uploaded by

Copyright:

Available Formats

ICDS12

International Conference DURABLE STRUCTURES: from construction to rehabilitation

LNEC Lisbon 31 May - 1June 2012

LNEC Lisbon 31 May - 1June 2012

4. PROCEDURE TO GENERATE PSNR FIELDS

LNEC Lisbon 31 May - 1June 2012

Probabilistic fatigue models are required to account conveniently for the

In this paper, a probabilistic approach is proposed to generate SN curves for

LNEC Lisbon 31 May - 1June 2012

LNEC Lisbon 31 May - 1June 2012

2. OVERVIEW OF THE DETERMINISTIC UNIGROW MODEL (1/3)

The material is composed of simple particles of a finite

The fatigue crack tip is considered equivalent to a notch

The fatigue crack growth process is considered as being

The fatigue crack growth rate can be determined as:

This equation may be used in an inverse way, to estimate

LNEC Lisbon 31 May - 1June 2012

2. OVERVIEW OF THE DETERMINISTIC UNIGROW MODEL (2/3)

LNEC Lisbon 31 May - 1June 2012

2. OVERVIEW OF THE DETERMINISTIC UNIGROW MODEL (3/3)

The weight function was found using the following expression:

where Kr takes a negative value corresponding to the compressive stress field.

LNEC Lisbon 31 May - 1June 2012

3. PROBABILISTIC SWTN FIELDS (1/2)

The probabilistic SWTN model is developed as an extension of the PN model

The SWT (=max.a) parameter was proposed by SmithWatsonTopper to take into

LNEC Lisbon 31 May - 1June 2012

3. PROBABILISTIC SWTN FIELDS (2/2)

ThereforethePN fieldproposedbyCastilloandFernndezCanteli maybeextended

where p is the probability of failure, N0 and

LNEC Lisbon 31 May - 1June 2012

4. PROCEDURE TO GENERATE PSNR FIELDS (1/3)

LNEC Lisbon 31 May - 1June 2012

4. PROCEDURE TO GENERATE PSNR FIELDS (2/3)

PREDICTION OF PROBABILISTIC SNp

LNEC Lisbon 31 May - 1June 2012

4. PROCEDURE TO GENERATE PSNR FIELDS (3/3)

LNEC Lisbon 31 May - 1June 2012

5. PROBABILISTIC FATIGUE DATA OF THE PUDDLE IRON FROM

LNEC Lisbon 31 May - 1June 2012

5. PROBABILISTIC FATIGUE DATA OF THE PUDDLE IRON FROM

This geometry was fatigue tested under

LNEC Lisbon 31 May - 1June 2012

6. PREDICTION OF PROBABILISTIC SN FIELDS FOR A NOTCHED DETAIL (1/10)

An elastoplastic stress analysis is proposed based on SeegerHeulers model, which is a

LNEC Lisbon 31 May - 1June 2012

6. PREDICTION OF PROBABILISTIC SN FIELDS FOR A NOTCHED DETAIL (2/10)

LNEC Lisbon 31 May - 1June 2012

6. PREDICTION OF PROBABILISTIC SN FIELDS FOR A NOTCHED DETAIL (3/10)

LNEC Lisbon 31 May - 1June 2012

6. PREDICTION OF PROBABILISTIC SN FIELDS FOR A NOTCHED DETAIL (4/10)

LNEC Lisbon 31 May - 1June 2012

The fatigue crack propagation modelling was supported by a bidimensional finite

LNEC Lisbon 31 May - 1June 2012

LNEC Lisbon 31 May - 1June 2012

LNEC Lisbon 31 May - 1June 2012

Weight functions used to determine the Kr for

Stress intensity evolution with the crack length, for a unit

LNEC Lisbon 31 May - 1June 2012

LNEC Lisbon 31 May - 1June 2012

Global SN field predictions (crack initiation and crack propagation SN field)

LNEC Lisbon 31 May - 1June 2012

The adaptation of the UniGrow model allows us to reproduce satisfactorily crack