A novel scale‑resolving strategy aimed at mitigating the grey‑area phenomenon in hybrid RANS–LES simulations is proposed. The approach relies on a stochastic energy‑injection mechanism—in the form of a random source term added to the momentum equations—which models the backscatter of energy from the subgrid to the resolved scales. The methodology is based on a system of stochastic differential equations whose underlying random variables are constrained to satisfy prescribed analytical autocorrelation functions. The strategy is specifically designed for unstructured flow solvers, addressing a major limitation of existing Stochastic BackScatter (SBS) formulations originally conceived for block‑structured grids. A rigorous mathematical derivation is developed to identify the constraints required to ensure a physically consistent correlation behaviour of the stochastic variables within a general unstructured framework. The resulting model is integrated into the open‑source CFD solver SU2 and coupled with the classic Detached Eddy Simulation (DES) approach based on the Spalart–Allmaras turbulence model. Particular attention is given to preventing spurious energy accumulation near the grid cut‑off. To verify this and to calibrate the correlation level of the stochastic variables, simulations of decaying homogeneous isotropic turbulence are performed. These tests confirm that the proposed strategy reproduces an energy‑backscatter behaviour consistent with the Eddy‑Damped Quasi‑Normal Markovian (EDQNM) closure. Finally, the effectiveness of the approach in reducing grey‑area artefacts is evaluated through the simulation of two canonical test cases: the backward‑facing step and the plane free shear layer. In both scenarios, the proposed methodology enhances the growth of resolved turbulence in regions where traditional DES formulations remain overly dissipative, leading to an improved prediction of shear‑layer dynamics.