In this paper we report a comprehensive analysis and tracking of the variabilities associated to multiscale simulations coupling a Coarse Gained Molecular Dynamics (CGMD) model of Lithium Ion Battery (LIB) electrodes fabrication with a 3D-resolved performance model at the cell scale. We quantify the impact on the final electrode mesostructure of the initial conditions in the CGMD simulations in terms of initial random distribution in the slurry of the active material and carbon binder domain particles (CBD). This study is carried out with a batch of simulations at different slurry active material/CBD compositions. Results show that for each electrode composition there is a statistical dispersion in the predicted cell performance curves, in agreement with in house experimental results. Furthermore, in house experiments revealed a different correlation regime depending on the C-rate at which the cathode is operating. The proposed methodology and findings allows us to pave the way towards the design of reliable approaches to control the variables correlations in the multiscale modeling of electrodes fabrication and its impact on LIB performance. (C) 2019 The Authors. Published by Elsevier Ltd.