Crack Paths 2012

which were caused by various variable amplitude loadings. Application of the reference

concept is limited to loadings satisfying the condition of "stationarity over short

distance". It means that all significant events, especially overloads, occur sufficiently

regularly and frequently. A proposal of physical explanation of reference features is

described in the second part of this paper.

T E X T U R AFLR A C T O G R A BPAHSYE DO N3D R E C O N S T R U C T I OFN S

T H EM O R P H O L O FGFYR A C T U RSEU R F A C[3E]

An image and a 3D reconstruction of a crack morphology present formally the same

type of R2 R transformation. Instead of image brightness, the local height of fracture

surface may be used. Therefore, methods of textural image analysis may be applied also

on 3D reconstructions. Relations in the data matrix, e.g. similarity, regularity and orde

ring are the source of information, which is expressed by means of numerical features.

Experiment was accomplished on six C T specimens made from aluminum alloy,

loaded in pairs by a constant cycle, a sequence of constant cycles with regular overload,

and a block of cycles with random characteristics.

3D reconstructions were acquired by four methods: as an output of confocal

microscope Olympus Lext, as a calculation by M E Xsoftware from stereo-pairs made

by SEM,as a calculation by DFFsoftware from a series of equidistantly focused images

made by metallographic microscope, and as an output of interferometrical Zygo

microscope. The image documentation of crack surfaces was realized by SEM.

Reconstructed or imaged areas are situated along central axis of crack surface, their

spacing and format size being in tenths of millimeters. 3D reconstructions were pre

processed with the aim to remove declination and large-scale waving of crack surface.

Images were normalized.

0.1 0.2 0.5 1 10S.125pecimenA7, Metal. microsc., Wavelet Cra k gr. rate, xper. vexp [Pm/cycle] C

10.5 SpecimenA7, S E M2D, Wavelet

,m o d e l

,m o d e l

0.12

r a c k g r . r a t e

gr . r a t e

U = 0.9813

1 U = 0.9974

0.1 0.2

0.5

c k

C r a

Crack gr. rate, exper. vexp [Pm/cycle]

a

b

Fig. 1: Examples of results. Comparison of input (known) and output (estimated by m o

del) crack growth rates. Constant cycle loading, analysis based a) on 3D recon

structions from images made by a metallographic microscope, b) on S E Mimages.

Markers represent single 3Dreconstructions or images of the fracture surface.

808