In-process tool pullout monitoring during milling based on internal machine tool data

From the machining of difficult-to-cut materials, it is known that tools can be pulled out of the tool holder leading to a significant loss of workpiece quality and tool breakage. To prevent such costly failures, it is of high interest either to detect the occurrence of tool pullout before any critical threshold is exceeded or to avoid it. This paper investigates tool pullout during milling and introduces a model to quantify compensating tool length offsets using high-frequency machine data. For validating purposes, end milling experiments with varying cutting parameters were conducted where the actual tool pullout length was measured after each cutting cycle, and machine data were acquired with an industrial edge computer. The results show that the algorithm robustly detects the occurrence of tool pullout. Due to an undefined physical relationship between the axial tool pullout and the rotational twist, the tool pullout length cannot yet be determined exactly, so further investigations are necessary.