Prediction of Surface Roughness in Turning of EN19 Steel Using Acoustic Emission

Abstract

Finding surface roughness of a turned product involves taking the product from the machine and measuring the surface roughness separately, which involves downtime during the machining process. This paper covers a new method of predicting the surface roughness, which involves recording the sound generated during machining and analyzing the sound level versus frequency graph for patterns specific to a particular condition, i.e., if the surface roughness produced during machining is high, it produces distinctive graph when compared to graph generated during machining of the product which got less surface roughness values. If a correlation is done for specific patterns in the graphs generated and surface roughness values, the process can be automated such that the sound generated during the machining is analyzed and checked for predetermined correlated conditions and surface roughness can be estimated during the machining process itself without removing the workpiece. EN19 (AISI 4140) steel round stock is taken and is turned at different speeds, feeds and depths of cut parameters combination. The surface roughness values of the turned workpieces were measured separately. Sound generated during the machining processes is recorded by using a sound recorder. The recorded sound is processed in audio-editing software to remove any ambient noises and to eliminate the sound generated due to chips from the machining sound. Frequencies versus sound level graphs are generated and peak amplitude values are noted. Prediction conditions were framed by analyzing graphs generated between experiment number—peak amplitude and experiment number—surface roughness. Confirmation experiments were performed and the sound recorded was analyzed and surface roughness was predicted within a close range and the surface roughness was later measured by a profilometer showed the predicted surface roughness values were true.