Additive Manufacturing process simulation for metal laser powder bed fusion has achieved a degree of mainstream status in recent years. It is however, still with significant limitations. One major limitation is the inability to deliver linked, multi-scale results over a full part volume, in a reasonable calculation time. It has been estimated that running a complete part at the individual scan vector level would require millennia to solve. The present work describes a novel multi-scale, analytical model-based process simulation approach that enables thermal history and defect prediction at the individual scan vector and powder layer level in minutes. The approach has been validated and is in production use today. The model is able to predict risk of local defects including keyholing, lack of fusion, balling up, and surface roughness. Further it can determine the process parameter window locally for any location within the part, based on the local geometry and temperature of the substrate before the laser is applied locally. The analysis also includes probabilistic effects and supports uncertainty quantification. Once fully incorporated into a commercial software platform, this solution will enable engineers to evaluate build processes and risk of defects given the proposed process parameters, or even to define how process parameters and scan vectors need to be modified locally to mitigate risk. It will also enable identification of high-risk areas where inspection and testing focus would be the most valuable to support model-guided qualification and certification plans.