Preparation of fatigue specimens with controlled surface characteristics

Masatoshi Kuroda, James Marrow

    Research output: Contribution to journalArticlepeer-review

    Abstract

    The response surface methodology (RSM) coupled with central composite design (CCD) have been employed to design fatigue specimens of austenitic stainless steels with controlled surface characteristics, for the purpose of investigating the effects of machining-induced residual stresses and roughness on fatigue behaviour. Simple cylindrical specimens having various surface characteristics were first produced by changing the final cutting conditions (spindle speed, feed rate and cutting depth) of the lathe. The fitting of the response surface model for the data was studied by analysis of variance (ANOVA). The response surface model employed in the analysis adequately represented the largest peak to valley height (roughness Ry) and the axial residual stress, although a good fit was not always achieved on the responses of the mean spacing of adjacent local peaks (roughness S) and the microhardness. Fatigue specimens were then prepared for several surface conditions, selected by interpolation of the response surface model, and the predictions were successfully compared with experiments. This approach to prepare fatigue specimens of austenitic stainless steels can form the basis of a systematic study of the effect of surface preparation parameters on fatigue behaviour. © 2007 Elsevier B.V. All rights reserved.
    Original languageEnglish
    Pages (from-to)396-403
    Number of pages7
    JournalJournal of Materials Processing Technology
    Volume203
    Issue number1-3
    DOIs
    Publication statusPublished - 18 Jul 2008

    Keywords

    • Analysis of variance
    • Cutting depth
    • Feed rate
    • Residual stress
    • Response surface methodology
    • Surface roughness

    Fingerprint

    Dive into the research topics of 'Preparation of fatigue specimens with controlled surface characteristics'. Together they form a unique fingerprint.

    Cite this