In this paper we formulate a computational framework for the characterization and performance optimization of gravity tables. Segregation of a multi-dispersed population of grains is assessed using a comprehensive numerical framework integrating Computational Fluid Dynamics (CFD) simulations with a Discrete Element Method (DEM) setup in the open-source software OpenFOAM. We have carried out combined qualitative and quantitative assessments of process conditions, such as deck shape, deck inclination (longitudinal and transverse), vibration speed and fluidizing air velocities at the surface. We assess the performance of our framework by employing a quantitative metric rho(stat), which represents the efficiency of separation in terms of density differences between the two separated product fractions. Additionally, qualitative flow behaviour is evaluated by examining relevant velocity and material distribution trends. We show that these separation profiles are highly sensitive to changes in the process conditions, with the presence of discernible optimal operating points. For the chosen gravity table, we have identified a longitudinal tilt of 6 degrees, an eccentric speed between 300 and 350 RPM, deck air velocities of up to 1.5 m/s and a transverse tilt of 0.75 degrees as a set of optimal operating conditions. (C) 2017 Elsevier B.V. All rights reserved.