Fatigue failure model

6.4. Fatigue failure model#

Go to handbook

6.4.1. Model description#

Fatigue is a failure mechanism incurred by cyclic loading, leading to the initiation and extension of cracks, which degrade the strength of materials and structures. We consider here case of high-cycle fatigue failure, i.e., failures that occurs after a modelled component is exposed to large numbers of load cycles. The limit state function for this type of failure can be written as:

\[ g\left( D_{cr}, A, \text{SSF}, \Theta \right) = D_{cr} - \Theta \cdot 10^{- A} {\text{SSF}}^{B} \sum_{j = 1}^{k}{N_{j} {S_{eq,j}}^{B}}, \]

This expression contains a set of variables that we consider uncertain (\(D_{cr}, A, \text{SSF}, \Theta\)) and parameters that we consider to be known with a sufficiently high accuracy (\(B, \{S_{eq,j}, N_j\}_{j=1,\cdots,N}\)). A summary of these variables and their meaning is given in Table 6.4.1.

Table 6.4.1 Input variables for reliability analysis#

Name

Description

Unit

Type

\(D_{cr}\)/D_cr

Threshold for accumulated damage

\(-\)

uncertain

\(A\)/A

S/N curve slope

\(\log(N/m^2)^{-1}\)

uncertain

\(B\)/B

S/N curve intercept

\(-\)

deterministic

SSF

Global stress scaling factor

\(-\)

uncertain

\(S\)/S

Load collective distribution

\(N/m^2\)

deterministic

\(N\)/N

Number of load cycles

\(-\)

deterministic

\(\Theta\)/Theta

Model uncertainty

\(-\)

uncertain

6.4.1.1. Load collective#

The load collective is the set of load events that the component was subjected to during its lifetime. The collective is denoted here by \(\{S_{eq,j}, N_j\}_{j=1,\cdots,N}\). To simplify usage of this interactive tool, the user can specify a distribution from which the load collective is sampled and a number of total load cycles \(N\). The generated distribution is then shown in a plot after the analysis.

6.4.2. Interactive reliability prediction#

This page offers an interactive reliability prediction that lets the user specify the properties of all variables listed in Table 6.4.1. The value of deterministic variables can be selected with a slider. Uncertain variables are characterized by:

  • Distribution denoted by “Dist” and can be choosen from a set of parametric probability distributions;

  • Mean value denoted by “E” and can be selected with a slider;

  • Coefficient of variation denoted by “C.o.V.” and can be selected with a slider.

Note

To run the interactive reliability prediction on this page, click the –> Live Code button on the top of the page. Wait a few seconds until the Kernel has loaded and run the cell below with Run.

from nrpmint.booktools import fatigue_failure

# start the web user-interface
fatigue_failure.web_ui()
Contents