Case depth evaluation of induction-hardened camshaft by using magnetic Barkhausen noise (MBN) method

Magnetic Barkhausen noise (MBN) method is one of the non-destructive\ua0evaluating (NDE) techniques used in industry to monitor the quality of ferromagnetic products during manufacture. In this article, case depth evaluation of the camshaft lobes by this means after induction hardening is described. A\ua0routine industrial monitoring practice is found to have limitation to evaluate the thickness of this process-hardened\ua0layer. With the aid of metallography on selected samples, this uppermost layer is found to have one, or more than one microconstituents. This infers that each type possesses different physical properties in response to the MBN measurement. Consequently, the interpretation of the MBN signal/data for case depth evaluation is not straight-forward.\ua0From metallography, a\ua0qualified component should have a\ua0uniform layer of martensite with grains ≤ 50\ua0\ub5m and the thickness around\ua03.0–5.0\ua0mm. This gives the magnetoelastic parameter (i.e. mp) in a\ua0range of\ua020–70 in industrial MBN measurement. The mp outside this range corresponds to either a\ua0non-martensitic\ua0type or a\ua0martensitic type with grains > 50\ua0\ub5m. In fact, the characteristic features of a\ua0Barkhausen burst like peak intensity, width and position can be used to categorise different microstructural conditions. Then, the case depth of the qualified components, or the thickness of the qualified martensite, can be estimated. Statistical regression decision tree model helps to divide this qualified group into three sub-groups\ua0between 3.0 and\ua06.0\ua0mm, and each can be identified by the decision criteria based on the specific ranges of the mp reading, the RMS of peak intensity and the peak position. In the end, a\ua0physical model is used to show how the difference of microstructures is influencing the magnetic flux, and thus the mp. Nevertheless, more information is needed to improve the model for this application.