The Direct Numerical Simulation (DNS) of the equations governing fluid dynamics phenomena computes the evolution of all the significant flow structures by resolving them with a properly refined computational grid. However, when the convection effects are dominant with respect the diffusive ones, this requires very fine meshes, making DNS computationally unaffordable for an operative viewpoint. Therefore, for many realistic problems the simulation of turbulent flows is performed by introducing different models. The equations can be properly averaged (in different ways, quite often in time), or filtered (usually in space). In this work, we focus on the latter approach leading to Large Eddy Simulation (LES) techniques. In this context, we consider an alpha model with a nonlinear differential low-pass filter [1, 2, 3] for the simulation of a wide variety of fluid flows at large Reynolds number with under-refined meshes. We consider several test cases coming from geophysical fluid dynamics [2, 3] and hemodynamics [1]. Moreover we will show some preliminary outcomes related to the extension of this pipeline to the study of magnetohydrodynamic turbulence. [1] Girfoglio, M., Quaini, A., and Rozza, G. A Finite Volume approximation of the NavierStokes equations with nonlinear filtering stabilization. Computers and Fluids (2019) 187: 27-45. https://doi.org/10.1016/j.compfluid.2019.05.001 [2] Girfoglio, M., Quaini, A., and Rozza, G. A novel Large Eddy Simulation model for the QuasiGeostrophic equations in a Finite Volume setting. Journal of Computational and Applied Mathematics (2023) 418: 114656. https://doi.org/10.1016/j.cam.2022.114656 [3] Clinco, N., Girfoglio, M., Quaini, A., and Rozza, G. Filter stabilization for the mildly compressible Euler equations with application to atmosphere dynamics simulations. Computers and Fluids (2023) 266: 106057. https://doi.org/10.1016/j.compfluid.2023.106057