<p>Laser-based powder bed fusion (PBF-LB) enables high-resolution fabrication of complex metal structures. However, improving their quality requires sophisticated process development, including the development of processing strategies and validation of process parameter sets. A key aspect of this development is the mesoscale virtualization of heat transport, which addresses the limited in-situ observability of the physical manufacturing process. To close experimental gaps, e.g. to be able to consider physical interactions with temperature fields, process virtualization with numerical models is usually employed. To obtain computationally optimized models, the explicit implementation of fluid flows is often avoided. However, inaccurate virtual representations of the actual process can limit the validity of the results. A possible compensation measure is the integration of effective properties, which allows an implicit implementation of the effects that are not explicitly modeled. In this context, the present research develops a method to increase the accuracy of virtual simulations by adjusting a three-dimensional energy source formulation. Central challenges, such as the implicit representation of fluid dynamic effects that strongly depend on the laser power and scan speed, are solved by adapting the effective penetration depth and absorptivity. This is achieved by evaluating simulation data for melt track depth and width using literature data from physical experiments, aggregated in a generalistic regression model. The presented approach provides transferability to other materials and parameter spaces, significantly improving the virtual modeling accuracy of heat transport in simplified mesoscale models. This supports process development and contributes to the optimization of process results in the PBF-LB.</p>