Acoustic emission assessment of measurement errors caused by gaps in chemical composition analyzes carried out using a portable spark spectrometer

During in situ tests, it was observed that the chemical composition analysis results obtained using a portable spark spectrometer significantly varied as the distance between the tested surface and the spectrometer probe nozzle increased. It was also observed that the nature of the sound emitted during the burning process was altered with the variation of this distance. Based on the Huygens-Fresnel principle, it was supposed that frequencies of the sound emitted during the burning process, as well as its magnitudes may be directly related to the quality of the chemical composition results. This paper investigated the relation between the errors associated with the results of chemical composition analyzes and the pressure of the leaked sound, since both depend on the gap thickness. The study was based on a full factorial design 3 x 8, where the test surface shape was investigated on three levels (flat; 220 mm diameter cylindrical and 325 mm diameter cylindrical) and the vertical distance factor was evaluated on eight levels (0 mm; 0.1 mm; 0.2 mm; 0.3 mm; 0.4 mm; 0.5 mm; 0.6 mm; 0.7 mm). During the tests, frequencies of the emitted sound signals were monitored using a precision microphone and a data acquisition module. The Analysis of Variance showed with a 95% confidence level that the investigated factors (vertical distance and test surface shape) and the interaction between them produced statistically significant effects in the values of the response variable (wt%) of the X65 steel. The Kruskal-Wallis and MannWhitney U tests showed that at least one experiment exhibited a mean value different from others and a significant number of pairs presented differences between the means. Through the analysis of acoustic sound signals, the validity of the measurement results was evaluated. An acceptance criterion based on relative error in wt% of chemical composition was defined. (C) 2019 Elsevier Ltd. All rights reserved.