The Discrete Element Method (DEM) is a numerical computational technique that simulates the macroscopic material behaviour by solving the equations of motion of its constituents. For a correct prediction of this behaviour, are set as input data the mechanical characteristics of the elements, the so-called microparameters. However, there is no recipe for determining these microparameters based solely on the macroscopic responses of the simulated material. It is required an additional step known as Calibration. The method widely used in this calibration is trial and error, although is an inefficient method due its unfavorable scale factor. This work proposes a new approach using the Generalized Simulated Annealing global optimization method, minimizing the normalized quadratic area between the experimental and calculated curves of the axial stress-strain and volumetric-axial deformations curves simultaneously. Comparison is done using triaxial tests for both synthetic and real data whose results demonstrate the usefulness and applicability of the proposed approach.