Risk-based design optimization for powder bed fusion metal additive manufacturing

Abstract

Powder bed fusion is a widely used additive manufacturing (AM) process for producing complex, small-batch parts that are impractical to manufacture using conventional methods. However, its broader adoption is hindered by process-induced defects. The challenge in AM stems from inherent material and process uncertainties. Therefore, it is critical to account for these uncertainties in the design optimization and control of powder bed fusion AM processes. In this work, we formulate and solve a design optimization under uncertainty problem for a powder bed fusion metal AM process. Our objective is to minimize energy consumption while enforcing a risk-based constraint formulated with a buffered probability of failure on residual stress, along with a constraint on melting temperature to ensure a successful build. We use surrogate models for the residual stress and temperature snapshots to accelerate optimization; we train these models using data from high-fidelity finite element simulations. We validate the optimization results through additional high-fidelity simulations. The validated results demonstrate that the proposed optimization reduces energy consumption, enhances process reliability, and contributes to more robust and sustainable additive manufacturing.