An integrated framework for enhanced machining: Machinability analysis, thermal modelling, and sensor-based monitoring

Licentiate thesis, 2024

To advance the development and optimisation of sustainable machining processes, it is imperative to integrate sensor-based tool wear monitoring and robust wear predictive models that incorporate batch-to-batch material variations. These resources play a pivotal role in streamlining production lines and managing variations that hinder productivity or compromise product quality.

The first area of focus of this research is the understanding of batch-to-batch variations in the medium carbon micro-alloyed steels and their influence on machinability. This is accomplished by the characterisation of two batches of a modified pearlitic-ferritic C38 steel in terms of microstructural characteristics and non-metallic inclusion properties. After each batch is machined to a constant spiral cutting length at different cutting conditions, the cutting tools are examined and reveal significantly different tool wear responses for the different batches. In fact, one batch exhibits considerably reduced machinability compared to the other for all cutting conditions. A comparative investigation of tool wear is then presented and discussed in light of the identified microstructural differences between batches. It is determined that the less machinable batch possesses a lower ferritic volume fraction, higher hardness, and more abrasive nitrides than the other batch. Additionally, the thermal loads are inferred to be higher during the cutting of the less machinable batch due to the difference in precipitation hardening between both batches.

The second focus of this research is the enhancement of semi-analytical thermal models that predict the temperature profiles during machining by incorporating the effect of a variable heat flux in the secondary shear zone due to the existence of sticking and sliding zones along the tool-chip interface. An accurate prediction of temperature distribution on the cutting tool is pivotal in enhancing the accuracy of tool wear estimations, especially for thermally induced wear mechanisms such as diffusion. While the modified models perform similarly well to the original models in predicting the value of the maximum temperature, they perform significantly better in predicting its location along the tool-chip interface, reducing the error percentage by an order of magnitude. The calculation of a variable versus constant tool-chip heat partition, and the inclusion of a round versus ideal cutting edge, are determined to have a significant effect on the temperature profiles.

The third focus of this research is the assessment of the 3-axis dynamometer, 3-axis accelerometer, and microphone as viable solutions in the identification of tool wear. For this purpose, longitudinal turning tests are performed with cutting tools that possess different levels of flank wear. The sensor signals acquired from the cutting tests are post-processed using the Discrete Wavelet or Hilbert-Huang transforms. The accelerometer exhibits a success rate of up to 89% in identifying the worn tool using the second and third IMF functions of the Hilbert-Huang transform. The microphone exhibits a lower success rate (78%) using the same algorithm. The Discrete Wavelet transform performs significantly worse than the Hilbert-Huang transform for all investigated cutting conditions and sensors.