The physical modeling of turbulence for the partial differential equations closure used to numerically solve large-scale leading-edge vortex flows maintains a significant grade of complexity and interest in the research community. This plays a major factor for the discrepancies with experimental or high fidelity data. A baseline common turbulence model is enhanced by means of a series of additional destruction terms which are formulated with a correlated physical feature of vortical flows as fundament and calibrated by means of an automatic gradient descent optimization procedure towards experimental data. The numerical accuracy is augmented for a certain cluster of cases around the calibration case in the parametric space which describes the classification of vortex flows over highly swept aerodynamic planforms.