Aluminum electrolysis is a challenging multiphysics problem: electric currents generate electromagnetic forces that induce motion in both the electrolytic bath and the molten aluminum, thereby displacing the interface between the fluids [1]. This work focuses on the consumption of carbon anodes due to electrochemical reactions. These anodes are consumed at approximately 2 cm/day, which requires an anode replacement every 24 days. The objective of this study is to predict the anodes shape and their impact on the process. A model involving the electric potential is coupled with a Level Set equation to predict the interface between the anode and the electrolytic bath. The consumption rate of the anode is proportional to the local current density. Since the consumed portion of the anode is small compared to its size, an anisotropic adaptive mesh refinement algorithm is advocated, as in [2]. Additionally, elliptic re-initialization is added following [3]. Numerical results are presented in an industrial framework. This work is funded by the Rio Tinto Aluminium Péchiney company. [1] Steiner G., Simulation numérique de phénoménes MHD: Application `a l’´electrolyse de l’aluminium, Thése No 4469 EPFL, 2009. [2] Passelli P., Picasso M. Anisotropic adaptive finite elements for aluminium electrolysis, Advances in Applied Mechanics 59, 2024, 1-17. [3] Basting C., Kuzmin D. A minimization-based finite element formulation for interface-preserving level set reinitialization, Computing, 95, 2012.