Dynamics Driven Cost Optimization for Hybrid Manufacturing

While additive manufacturing (AM) offers many advantages, parts produced by AM may not provide sufficient surface finish and accuracy for all engineering applications. Therefore, machining operations are often required to qualify the final part geometry. The combination of AM and machining is known as hybrid manufacturing (HM). HM uses an AM process, such as laser powder bed fusion (LPBF), to create a preform that is then machined on a computer numerically controlled (CNC) machine tool. The combination of user-selected machining parameters, including radial immersion, axial depth, and spindle speed, determines the effectiveness of the machining operation. Improper combinations may result in chatter, an unstable machining condition caused by self-excited vibrations, or high amplitude vibrations that result in dimensional inaccuracies, known as surface location error (SLE). The effectiveness of the machining process is directly dependent on the flexibility, or compliance, of both the workpiece and the tool-holder-spindle-machine assembly. The dynamics of the workpiece are initially set by the AM preform geometry and continuously change as material is removed during machining.

This work describes a novel framework that simultaneously optimizes the preform geometry and machining parameters. The optimization objective is to minimize HM cost for industrially relevant geometries while avoiding chatter and mitigating SLE. A parametric representation of the preform geometry is coupled with a finite element analysis (FEA) solver to predict evolving preform dynamics as material is removed by machining. A structural shape optimization algorithm is introduced to minimize static compliance for a given amount of preform volume. A numerical, time-domain simulation is then used to select optimal machining parameters. Finally, Bayesian optimization is used to find the optimal preform volume that minimizes HM cost