Sei sulla pagina 1di 27

February 9, 2013

Lecture 9

14.5 Release

Fatigue Analysis

1

Release 14.5

Introduction

- This lecture provides only the basic ideas of fatigue. Some discussion of

the background to the checking methods is included as well.

- The main purpose of this lecture is to demonstrate the implementation of ANSYS technology in studying fatigue in fixed marine structures.

- Joint fatigue assessment (FATJACK) is included in Design Assessment and

can be connected to Static Structural, Transient Structural and Harmonic

Response analyses.

- In order to use FATJACK in the most efficient way, global models are often studied first. Although such models contain approximations, they have the benefit of giving a good insight into the structural behaviour. At a later stage local models may need to be solved (see submodeling).

February 9, 2013

2

Release 14.5

Nomenclature

- Attribute Groups define the input data to FATJACK

- DA result objects define the output. They can be requested either before or after the analysis

- Inspection Points are the positions to check for fatigue around the brace where it connects to the chord

- As WB is geometry-based the user needs to select bodies rather than individual elements

- Elements that do not have results will be semi transparent in the graphics window

- Stress Concentration Factors can be returned for the brace or the chord

side

- As always the results can be parameterised!

February 9, 2013

3

Release 14.5

Stress Concentration Factors

- The distribution of stresses across the section of a member may be abrupt

due to stress raiser in the region (bolt holes, notches, welds). When the

variation is abrupt so that within a very short distance the stress intensity increases greatly, the condition is described as stress concentration.

- The term is vague and it implies some sort of irregularity not inherent in the member.

- In short, SCF is the ratio σact/σ, i.e. actual stress divided by nominal stress.

stress

The

nominal

stress

is

based

on

the

net

section

ignoring

any

redistribution caused be the irregularity. Consider the following example:

February 9, 2013

Sample

moment, the geometry becomes

subjected

nom

I

My

to

, which for the current

uniform

bending

nom

6 M

b

(

D h

)

2

But the actual stress is much higher because of the stress concentration that occurs at the root of the notch!

4

Release 14.5

Stress Concentration Factors (cont.)

- SCFs relate hot spot stresses to the nominal stresses in a member

computed from the axial forces and in-plane and out-of-plane bending

moments.

- FATJACK provides facilities for both defining explicit SCF values and for automatically generating the SCF values using one or more established empirical formulations. Explicit and generated SCF values may be mixed as necessary to achieve the desired result.

- Some default values exist in FATJACK, e.g. for tubular section:

brace side SCF for in-plane bending=3.5

brace side SCF for out-of-plane bending=3.5 chord side SCF for axial force=3.0 chord side SCF for in-plane bending=3.5 chord side SCF for out-of-plane bending=3.5

February 9, 2013

5

Release 14.5

Fatigue Assessment

- When a steel member is subject to a sufficiently large fluctuating tensile

stress, small crack-like defects will grow in size and eventually reach a big

enough size to cause the member to fail.

- Most marine structures are subject to chaotic stress fluctuations during their design life making any approach based on endurance limit stress impractical. Random operation loads (wind, wave, earthquake, VIV) make the use of S-N curves necessary in this case.

- The use of S-N curves is an alternative approach for the assessment of in- service cracks. Based on specimens subject to fluctuating loading the

number of cycles to failure N is inversely proportional to the stress range S

to the power m of 3 to 4. This is expressed by the following relationship:

N AS

m where A and m are obtained experimentally

- The method used within FATJACK to carry out the fatigue assessment is based upon the use of S-N curves

February 9, 2013

6

Release 14.5

Fatigue Assessment (cont.)

- The environment is also important, fatigue life is reduced in freely

corroding conditions (Bradshaw et al. 1984).

- Tubular joints are used throughout fixed marine structures and a number of different S-N curves already exist in the public domain.

- Crack propagation is mostly driven by tensile stresses but high tensile

residual stresses in the material exist creating tensile stresses even when

the applied stress is compressive. This compression-tension cycle requires the use of the full stress range in fatigue calculation. A typical S-N curve is

shown below.

7

February 9, 2013

Release 14.5

Miner’s Rule

- Miner’s rule is based on the concept of fatigue damage. For variable

amplitude environmental load, Miner’s rule allows for the number of

different amplitude stress cycles to be considered.

- The fatigue damage for a joint that is subject to n number of cycles/year

N AS

m

simply n/N. If the same joint is subject to a variable amplitude loading the load cycles can be divided into groups of approximately an equal stress

range. Thus, if there are G groups with stress range Sg and number of

cycles ng in each group, the fatigue damage, per group, is

D

g

n / N

g

g

where

N

g

AS

m

g

- Miner’s rule states that failure, under variable amplitude loading, is to occur when

G

g 1

D

g

1

February 9, 2013

8

Release 14.5

Miner’s Rule (cont.)

n
n
n
n
n
n
Miner’s Damage Rule:
i 
1
2
3
4
5
N
N
N
N
N
N
i
1
2
3
4
5
 amp
N
1
Damage =1.0
n 1
N
2
n 2
N
3
n
3
N
3
n
4
N
4
n 5
N
5

log N

Since the calculation is based on the number of cycles/yr for the constituent waves, the resulting damage is that associated with one year of operation. The fatigue life is simply the reciprocal of damage, that is life=1/D

February 9, 2013

9

Release 14.5

When dealing with marine structures, two important additions to the

normal structural mass have to be considered:

- The added (entrained) mass of water, which moves with submerged or partially submerged structural members.

- The increase in mass due to marine growth (which varies over the life of

the structure).

It must be remembered that the added water is external to the structure and is over and above any water contained within a hollow structure. For example a hollow pipe has an effective mass equal to:

- The mass of the member

- The mass of the water contained within the pipe

February 9, 2013

10

Release 14.5

In theory added mass (mass/unit length) is determined by potential flow

theory but added mass values for common sections are widely

documented.

Closely with added mass are the drag and inertia coefficients (OCTABLE)

Added mass is defined in OCDATA as, the ratio of added mass of the

external fluid over added mass for a circular cross section

Obviously the added mass can have a profound effect on modal and dynamic analyses and should be calculated as accurately as possible.

February 9, 2013

11

Release 14.5

Rainflow Counting

- Consider the stress-time history recorded at a probe as shown in Fig. A. Questions like “what stresses are significant, what counts as a cycle, and what is the measure of damage incurred spring into mind!

- Consider the first fully reversed cycle and the second fully reversed cycle.

- It is clear that to impose the stress-time history on a part, it is necessary that the time trace look like the solid line plus the dashed line.

- Fig. B begins and ends with the max stress-time history.

- Acknowledging the existence of a single stress-time history is to discover a “hidden” cycle as shown in the dashed line.

- To ensure that the hidden cycle is not lost, start the curve with the largest (or smallest) stress and add previous history to the right side as shown in Fig. B.

February 9, 2013
12

Release 14.5

Rainflow Counting (cont.)

- Imagine we flood the curve with water and we gradually drain it to monitor the valleys that are formed.

- Valley #1 has σmax=80Pa, σmin=-60Pa, σav=70Pa, σm=10Pa

- Valley #2 has σmax=60Pa, σmin=40Pa, σav=10Pa, σm=50Pa

- Valley #3 has σmax=-20Pa, σmin=-40Pa, σav=10Pa, σm=-30Pa

February 9, 2013

13

Release 14.5

Rainflow Counting (cont.)

Based on ASTM E1049-85 (2005) Standard Practices for Cycle Counting in Fatigue Analysis

Reduces spectrum of varying stress into simple stress reversals

Allows the application of Miner’s rule to assess fatigue life of structure subject to complex loading

It is possible to use results from up to 1000 different transient dynamic analyses

Results can be combined using a probabilistic approach, output includes:

14
Release 14.5

fatigue life (based on Miner´s rule)

usage factors

damage per wave (history)

February 9, 2013

ANSYS - Code Checks (FATJACK/BEAMST)

Joint Code & Member Code checks including:

AISC 10th edition working stress and 2nd edition LRFD

API RP2a-WSD 21st edition working stress

RP2A-LRFD 1st edition
BS5950 part 1 1992
NORSOK 2000
NORSOK NS3472 1984
NPD 1992

DS449 1984 (with 1994 amendments)

DS412 1984 (with 1994 amendments)

ISO 19902 implementation started

Easy-to-use code check facilities including:

Code checks on time histories

Code checks on combined load cases

Visualization of code checks

Ability to use them in combination with ANSYS calculations

February 9, 2013

15

Release 14.5

ANSYS - Fatigue Assessments

FATJACK module offers both deterministic and spectral fatigue capabilities

for tubular frame structures subjected to waves and current or wind including wind gusts

can be used in frequency and time domain

sea states: JONSWAP, Pierson-Moskovitz, Occhi-Hubble, Scot-Weigel and Shell New Wave, or user-defined wave spectra

FATJACK includes explicit SCF definitions

SCFJ if crown & saddle SCF is known e.g. from empirical formulae

SCFA if SCF is known at specific locations e.g. from FE

SCFB if SCF is constant across a section

SCFP if SCF values vary with location

Automatic (empirically derived) SCF definitions based on

Efthymiou, Wordsworth, Kuang or DS449

February 9, 2013

16

Release 14.5

Analysis Type Selection

At v14 support for spectral fatigue was added:

February 9, 2013

17

Release 14.5

Analysis Type Selection (cont.)

The following types of fatigue analysis are supported in Workbench:

- Time History: enables the selection of joints to be included, along with the definition of the cycles for rainflow counting and target life. Upstream systems are usually transient analyses with random ocean loading coming from different wave directions

- Stress History: enables the selection of joints to be included, along with the definition of the target

life. Wave conditions (heights, periods, directions) are automatically determined with OCEAN commands in the upstream system(s). Wave occurrence data can be provided with attribute

groups. Upstream systems can be either static or transient structural

- Spectral (static/dynamic): Enables the selection of joints, along with definition of the peak stress, wave spreading and target life of the analysis. Wave transfer function, spectrum, and additional frequency data should be provided in a text file containing the FATJACK commands. Wave load cases are automatically determined using the HROCEAN command provided in upstream system(s)

in the order that they are defined. Upstream systems are usually of the Harmonic Response type; note that both the static and harmonic options of the HROCEAN command can be used when performing a spectral analysis

- Deterministic: Enables the selection of joints, along with definition of the target life. Wave load

cases are automatically determined using the harmonic ocean wave procedure provided in upstream system(s). Upstream systems should be of harmonic response type; only the static option of the HROCEAN is appropriate here (HROCEAN,STATIC)

February 9, 2013

18

Release 14.5

Deterministic Methods

- Deterministic methods apply Miner’s rule directly

- The ocean environment is idealised by representative wave cases, with a defined number of loading cycles (occurrences/year). MAPDL commands will be used to produce a series of wave cases. The

structure is then analysed to determine the stress state and then the

fatigue life

- In deep waters (D>0.5λ) where the structure responds sinusoidally to the

analysed. The pair of loadcases represents the real and imaginary parts of a solution which retains both magnitude and phase information

- The above is achieved by undertaking a harmonic response analysis

- FATJACK

uses this data to generate the stress ranges directly from the

incoming member forces for the loadcase pairs

- In shallow waters linear theory is invalid and a more detailed analysis needs to be adopted, i.e. stress-history approach (not covered here)

February 9, 2013

19

Release 14.5

Harmonic Ocean Wave Procedure (HOWP)

mass of the water outside the pipe. A damping matrix must be added

separately if one is needed. The load vector is computed based on the loads at a given time, but the standard analysis method usually misses some important effects, as all peak loads rarely occur at the same time

So that all relevant ocean wave loading effects are accounted for, a

specialized variation of the harmonic analysis is available. The harmonic

ocean wave procedure (HOWP) applies to regular waves only (Airy and

Wheeler one-component waves, as well as Stokes and Dean’s Stream Function waves), and works only with the full-solution harmonic analysis

method (HROPT,FULL)

February 9, 2013

20

Release 14.5

Harmonic Ocean Wave Procedure (HOWP) (cont.)

- HOWP is implemented with the HROCEAN command. The frequency is

obtained by the wave period defined via OCDATA & OCTABLE when

HROCEAN, HARMONIC.

- Ocean loads are calculated with the assumption that the structure is stationary

- ANSYS calculates the forces on each load component of each element at

NPHASE solutions, spread evenly over one wave cycle. Then, the minimum and maximum, and the phase between them, are calculated. The

command uses the resulting information to generate the real and

- For every component of every element, special calculations are performed to obtain real and imaginary loads. All loads are sinusoidal at the given frequency, but the magnitude and phase angle of every component of every element needs to be determined, as follows:

February 9, 2013

21

Release 14.5

Harmonic Ocean Wave Procedure (HOWP) (cont.)

- First, assume that the wave profile is represented as a simple cosine wave,

as shown below:

February 9, 2013

22

Release 14.5

Harmonic Ocean Wave Procedure (HOWP) (cont.)

- A series of static analyses is performed ranging from 0 to 360 degrees. The

number of these analyses is controlled via the HROCEAN command. The

result is a roughly sinusoidal force pattern, as shown below:

February 9, 2013

23

Release 14.5

Harmonic Ocean Wave Procedure (HOWP) (cont.)

- From this force history, the maximum and minimum forces are calculated

by fitting the highest and lowest points. Then:

A

(

Y

max

Y

min

)

2

(

X

max

X

min

180)

(

X

max

2

X

min

180)

2

if

if X

X X X

max

max

min

min

Y A cos( X ) A coscos X A sinsin X

- Acosφ is the coefficient on the real load vector and Asinφ is the coefficient on the imaginary load vector

February 9, 2013

24

Release 14.5

HROCEAN, Type, NPHASE

Type Specifies how to include ocean wave information in a harmonic analysis:

HARMONIC (default) Performs a harmonic analysis using both real and

imaginary load vectors (calculated via HOWP). This behaviour is the

default. This option works by performing a harmonic analysis running at a frequency determined by the wave period (specified via OCTABLE).

STATIC Performs a static analysis using both real and imaginary load

vectors (calculated via HOWP). This option works by performing a harmonic analysis running at a frequency of 0.0.

OFF Deactivates a previously activated HOWP and performs a standard

harmonic analysis.

NPHASE Positive number specifying the number of phases to calculate forces. This value must be at least 8 (defaults to 20)

February 9, 2013

25

Release 14.5

Salmon C.G. & Johnson J.E. (1980) Steel structures design and behaviour,

Harper and Row, New York

Gurney T.R. (1979) Fatigue of welded structures, Cambridge Uni Press

Knott J.F. (1973) Fundamentals of fracture mechanics, Butterworths UK

Lalani, M. & Teddett, I.E. (1985) Design of tubular joints for offshore structures, UEG, London

Barltrop, N. & Adams, A.J. (1991) Dynamics of Fixed Marine Structures,

Butterworth-Heinemann

Bishop, N & Sherratt, F. (2000) Finite Element Based Fatigue Calculations, NAFEMS

February 9, 2013

26

Release 14.5

Workshop 7 Fatigue Analysis

Goal:

Learn how to set up a Design Assessment system with FATJACK

Learn about Solution Selection and analysis factorization

Define input in DA

Request output from DA

Study the results