Acoustic emission accuracy from a tensile test of a ceramic matrix composite

Abstract

Ceramic matrix composites (CMCs) on one level are excellent materials for acoustic emission (AE) analysis. They are excellent waveguides for AE waveform transmission due to the high modulus to density ratio. CMC inelastic behavior is due to micro- and macrocrack formation from matrix crack interaction with the fibers via a relatively weak fiber/matrix interface which create ideal stress waves. Because of this, AE is an excellent detector of microcracks in general, and most importantly in the case of CMCs, the initial or lowest stress crack formation. This property can be related to long time stressed-oxidation degradation of nonoxide composites, in particular. In addition, AE has been used to effectively determine the stress distribution for matrix cracks which cause the nonlinear stress–strain behavior. However, a key to quantitatively correlating AE with sources is first and foremost to locate where the AE originated. For a tensile test, most AE comes from the near-grip region and the radius region outside the gage area of interest. Outside the gage region AE would not be considered useful data pertaining to stress/strain behavior and must be sorted out from the AE dataset. Location is determined by the difference in time of arrivals (TOAs) of waveforms received on each sensor from a given AE source. Automated TOA techniques such as threshold voltage crossing or Akaike information criteria (AIC) have limitations in overall accuracy of differences in TOA (Δt) of two different sensors required for location analysis. This study has incorporated several signal filter and enhancement techniques and an approach toward increasing the accuracy of the classic TOA techniques. First TOA was determined for the two sensors of the AE tests “manually” based on first extensional peak of the waveform, this served as the “exact” difference in TOA. Δt’s were then determined for the various filter/TOA techniques and compared to those from the manual determined Δt. The best filter/TOA techniques resulted in more than two times better accuracy (defined as percentage of events within 0.1 µs of the exact Δt) than the conventional threshold crossing or AIC technique.