Laser Powder Bed Fusion (L-PBF) has immense potential for the production of complex, lightweight, and high-performance components. Optimizing process parameters both for existing and new materials is crucial for maximizing the capabilities of this technique. Traditional process parameter optimization involves a large amount of experimentation, which makes it a costly and time-consuming endeavor. An elegant improvement upon this is leveraging numerical simulations to find optimal process parameters. Current research mostly focuses on modeling single-line scans, most likely to maintain computational efficiency. Unfortunately, optimization of process parameters using this single-line geometry will lead to process parameter sets optimized for thin walls rather than bulk material. A novel approach to optimizing bulk process parameters is introduced in this work. The approach utilizes a discrete element simulation, with a ray tracing-modelled laser heat source[1, 2]. The approach significantly reduces the cost and time consumption in contrast to conventional optimization methods. The simulation is sped up by using GPU acceleration, enabling efficient simulation of multiple fully scanned layers. This will result in process parameters optimized for the bulk material rather than thin walls. The practical use of the optimization method is demonstrated in a case study where optimal process parameters are developed for AlSi10Mg. The optimization entails just 5 days of computation time, which would have been at least 8 months without the GPU acceleration. Through experimental validation, it was found that the numerically optimized process parameters were indeed the highest quality, with an optical density of 99.91%.