A model predictive controller (MPC) was developed in order to optimise energy management of an ice-cream warehouse refrigeration system coupled to a phase-change material (PCM) tank. The controller's internal model was based on a steady-state refrigeration cycle model and energy balances on the PCM tank and warehouse. Considering crystal growth in ice-cream, the quality evolution was based on Ostwald ripening equation. The product temperature field was determined by a 1-D finite difference equation. An original PCM model was developed for better performance of Levenberg–Marquardt's optimisation algorithm which was modified in ordepreviso*r to take into account control variables' bounds. Simulation of the ice-cream storage for a period of 90 days lead to good results on the optimised control sequence with efficient energy management thanks to the PCM tank. Ice crystals size remained below the defined target: 26 microm. This study showed the great potential of MPC to reduce energy consumption and guarantee food quality.