Fast additive manufacturing distortion prediction using backward interpolation

Powder Bed Fusion Additive Manufacturing (PBFAM) processes have emerged as important industrial processes that are capable of manufacturing complicated part features, such as hollow designs, lattice structure and other unique design structures. The current state of computing technology requires significant time to predict the part distortion formed during the PBFAM build process. In this paper, a novel Backward Interpolation (BI) model is presented for fast estimation of part distortion based on build process parameters. In the Backward Interpolation approach, we propose the concept of distortion factor that represents the distortion contribution of unit forces on the node to all other nodes that have equal or lower height in the FEM mesh. To start with, the CAD model is meshed in a layer by layer manner and the FEM model is voxelized and sliced for calculating the distortion factor. Next, the as-built distortion before cutoff from the substrate is calculated based on distortion factors and the internal forces generated by the inherent strain of the voxel. Subsequently, the spring back distortion is calculated, and the total part distortion is then finally determined as the summation of part distortion and the spring back distortion. An experimental validation of the methodology was conducted, and the predicted distortion results appeared to agree well with the distortion obtained from the built part. The comparison of distortion obtained from Backward Interpolation and distortion data from published work was further used to validate this algorithm. The computational efficiency of the proposed Backward Interpolation method was evaluated using an example part and demonstrated that the part distortion values were obtained within a few minutes using a desktop computer, confirming the significant decrease in computing time using this approach.