We have designed a simple multi-scale method that identifies turbulent motions in hydrodynamical grid simulations. The method does not assume any a priori coherence scale to distinguish laminar and turbulent flows. Instead, the local mean velocity field around each cell is reconstructed with a multi-scale filtering technique, yielding the maximum scale of turbulent eddies by means of iterations. The method is robust, fast, and easily applicable to any grid simulation. We present here the application of this technique to the study of spatial and spectral properties of turbulence in the intra-cluster medium, measuring turbulent diffusion and anisotropy of the turbulent velocity field for a variety of driving mechanisms: a) accretion of matter in galaxy clusters (simulated with ENZO); b) sloshing motions around cool-cores (simulated with FLASH); c) jet outflows from active galactic nuclei (AGN, simulated with FLASH). The turbulent velocities driven by matter accretion in galaxy clusters are mostly tangential in the inner regions (inside the cluster virial radius) and isotropic in regions close to the virial radius. The same is found for turbulence excited by cool-core sloshing, while the jet outflowing from AGN drives mostly radial turbulence motions near its sonic point and beyond. Turbulence leads to a diffusivity in the range D turb ∼ 10 29 −10 30 cm 2 s −1 in the intra-cluster medium. On average, the energetically dominant mechanism of turbulence driving in the intra cluster medium is represented by accretion of matter and major mergers during cluster evolution.