In milling processes involving roughing and finishing operations, machining conditions selected during process planning not only affect machining efficiency but also influence the surface quality of subsequent finishing stages. However, the selection of machining parameters is still largely based on empirical knowledge, and the quantitative relationship between process efficiency, machining load, and surface roughness remains insufficiently clarified. To address this issue, this study proposes a digital twin–based decision-support framework to systematically analyze machining performance and surface quality in milling processes.
In the proposed approach, CAM-generated roughing and finishing G-code programs are analyzed using two complementary digital twin tools: Siemens Run My Virtual Machine (RMVM) for machining time estimation and MACHPRO for cutting force simulation. A three-level full factorial design is adopted to configure axial (AP) and radial (AE) machining parameters, enabling systematic evaluation of their effects on machining time and cutting force without extensive trial machining.
Representative machining conditions are then validated through physical finishing experiments, with surface roughness (Ra) measured as the primary quality indicator. Experimental results show that machining time and cutting force exhibit opposing trends with respect to AP and AE variations, revealing an inherent trade-off between process efficiency and machining load. Cutting force is further shown to correlate strongly with surface roughness trends, indicating its effectiveness as a process indicator for surface quality variation.
To support dual-objective decision-making, Gaussian Process Regression (GPR) is employed as a trend-fitting tool to construct continuous surrogate representations of machining time and cutting force, from which a Pareto front is derived. This Pareto-based representation enables flexible selection of cutting conditions according to different production requirements. The results demonstrate that the proposed digital twin–based framework can significantly reduce the reliance on extensive trial-and-error experiments during process planning.
By integrating simulation-driven evaluation, experimental validation, and Pareto-based decision analysis, suitable cutting parameter combinations can be identified in a virtual environment and directly transferred to actual machining, achieving a desirable balance between machining efficiency and surface quality in milling processes involving roughing and finishing operations.
Fig. 7 Summary of the integrated digital twin–based experimental and decision-support workflow, including machining time and cutting force simulation, experimental validation, and Pareto-based dual-objective analysis. (online color)
Fullsize Image
View full abstract