50% of the energy consumed in food cold chain are for chilling, freezing and storage steps. For cold storage facilities, refrigeration is the main source of energy consumption with an average of 50 kWh/m3/year. A possible strategy to reduce their energy consumption is to take advantage of the variation of performance of the refrigeration systems during a day due to external and internal temperature variations. Electricity cost can also widely change during a day, depending on the local production. This strategy can be based on additional storage capacities to delay the refrigeration running period, but also on the use of the thermal inertia provided by the storage warehouse itself and the products stored. An experimental test has been performed in a frozen warehouse storage facility to evaluate the potential energy savings of such a strategy. Using coupled models of the building, products and refrigeration system, using weather forecasts and products flux statistics, a predictive control algorithm was developed and applied to determine periodically the optimal set-points of the storage warehouse. Technical implementation of the control system is presented. Energy savings potential and impact on products of the temperature fluctuations are also discussed.