Comparison of different approaches for predicting material removal power in milling process

Accurately characterizing the energy consumption of machining processes is a starting point to increase manufacturing energy efficiency and reduce their associated environmental impacts. As a significant contributor of machining power consumption, the material removal power can be predicted by three different approaches: multiplying specific cutting energy by material removal rate (approach I), multiplying cutting forces by the cutting speed (approach II), and modeling the power consumption as exponential functions of cutting parameters (approach III). However, there is no general agreement about the accuracy of different modeling approaches. Therefore, this paper aims to test and compare different modeling approaches with respect to their suitability to predict the material removal power in the milling process. In order to obtain the coefficients in the models, experiments were carried out on a machine center. Three types of workpiece materials (carbon steel, aluminum, and ductile iron) are selected for cutting tests. Four-factor (cutting speed, feed, depth of cut, and width of cut) four-level orthogonal experiments are employed based on Taguchi’s method. Dynamometer and power acquisition devices were used to measure the cutting force and machine power. Then a set of models were established using coefficients obtained from literatures or regression analysis of experimental data. Values of material removal power predicted by the three approaches are compared with those from confirmation experiments. When using coefficients from literatures, the prediction accuracy varies from 51.2 to 87.7% for steel, 49.3 to 64.6% for aluminum, and 57.2 to 90.9% for ductile iron, depending on the sources of coefficients. When the coefficients are obtained experimentally, the prediction accuracy of all approaches is over 83.9%. In this case, approach III achieves the highest prediction accuracy, followed by approach II and approach I for steel. Approaches I and III give the highest prediction accuracy for aluminum and ductile iron, respectively. Approach III is recommended to be used in industry due to its high prediction accuracy and moderate implementation difficulty.