Fatigue Crack Paths 2003

That type of reasoning leads to a damage accumulation different on the two critical

plane relative to the two loads. The deduced fatigue life can then be understood as an

upper bound estimation. Indeed, no interaction mechanisms have been accounted for.

The two loads are viewed as acting alone on the damage accumulation.

Another way to deal with damage accumulation is to carry out a linear summation of

damage accumulated in each blocks. More exactly, each individual damage will be

)i(

D

)i(D is the damage accumulated in the

characterised by means of the ratio

, where

)i(R D

block i as for a constant amplitude loading and

represents the damage to failure for

)i(RD

this block.

Whenapplied to the previous example, the initiation condition is given by :

)i(R )i(

)i(R )i(

DD

DD

=

⎜ ⎛ ⎛ ∑ ∑ ⎟ ⎞ ⎜

⎟ ⎠ ⎟ ⎞

1

⎜ ⎝

⎟ ⎠

+

⎜ ⎝

i

tension

i

torsion

The two different types of damage rule (with or without interaction) has been

applied together with the aforementioned model. The predictions are compared with the

experimental data on Fig. 3 for the two mild steels and for the different loading modes.

It appears that the two damage accumulation methods lead (as expected) to two

bounds of life prediction. Whenno interaction is considered the predictions lie on the

non conservative side of the graph. Andthe predictions are almost all conservative with

a total interaction between the microcracks. Let us notice that the degree of non

interaction seems also to depend on the length of each individual block.

To get the best life prediction, one should precisely define and estimate an

interaction factor which could reflect the ability of a misoriented crack to find the path

imposed by a new stress state. Unfortunately the estimation of this coefficient would

require some unusual and time consuming fatigue tests. The previous definition of the

two bounds is then a good compromise between prediction accuracy and simple model

identification.

C O N C L U S I O N

Using any critical plane damage models, the damage accumulation under different

loading modes is uneasy to deal with. Indeed, the orientation of the critical plane is most

of the time related to the loading condition. Whenthis condition changes, the critical

plane orientation changes but howdoes the damage accumulates from one loading mode

to the other ?

This study tried to understand if some interaction appears between non similar

oriented damages. Someexperimental results of two mild steels submitted to sequential

loading blocks of different types (tension, bending, torsion, combined bending and