Powder quality is an important aspect in metal laser powder bed fusion (LPBF) additive manufacturing processes. The quality of powder, described by the particle size distribution (PSD), the shape of the particles and the amount of satellites, must be closely controlled to guarantee a stable process that delivers high quality parts. The need to precisely meet the powder specification makes its manufacturing expensive and increases the risk of a bottleneck. In a first step a particle-resolved simulation of the LPBF process [1] is used to study the impact of changes in the PSD of powder on the melt pool form and the probability of defects, such as key-holing and lack of fusion. Several powders made from 316L steel with different PSD and layer thickness are simulated, and the effect on the melt pool form and defect generation are analyzed. The simulation results are validated by comparison to test printings on an Aconity Mini. In a second step the results of the particle-resolved simulation were used to calibrate a fast layer-scale simulation of the LPBF-process. The fast calculation treats the powder as continuous medium and adjust the material properties of the metal to mimic the powder. The laser absorption is described by a 3D volumetric heat source model. Material and heat source model parameter are adjusted to gain a melt pool form, which is comparable to the results of the particle resolved simulation and experiments. The fast calculation of the melt pool shape is then used to come to a fast defect prediction for LPBF processes. The aim is to be able to adapt the metal LPBF process parameters to changing powder properties with limited computational effort in order to keep the part quality stable.