In the additive manufacturing of components using powder-based laser metal deposition (LMD) / directed energy deposition (DED), controlling the heating of the volume during build-up is crucial for ensuring process stability and contour accuracy. When process parameters remain constant, inherent heating leads to variations in deposited layer thickness due to changes in the melt pool volume. These variations result in contour deviations and may even halt the process if parameters deviate from their optimal range. Especially with complex geometries, this often necessitates labor-intensive process development to adjust parameters and build-up strategies accordingly.


The melt pool volume serves as a key indicator of process stability in LMD. Measurement of the melt pool area can be achieved using a coaxial camera in the beam path. By adjusting the laser power during the process, the goal is to minimize melt pool variance. To accomplish this, an AI model is trained using data from LMD processes with consistent parameters during geometry build-up. The AI model learns the correlations between laser power, geometry, and other component-specific factors, and the size of the melt pool area. Subsequently, the trained AI model can predict the local laser power required for a stable process. This eliminates the need for extensive experimentation and evaluation to determine suitable process parameters, particularly when component geometry changes.