In the last years, energy costs have gained great attention, because of their constant rise and environmental implications. More researchers have directed their attention towards energy-efficient production planning systems, as their adoption is usually not tied to large investments. This study presents a scheduling model that minimizes the total energy production costs. In particular, the industrial system considered is a two-machines flow-shop without intermediate buffers, so that a no-wait or a blocking condition can arise. Moreover, batches of different sizes are possible. We develop a Genetic Algorithm (GA) that minimizes total system energy costs based on energy states of the two machines. A numerical analysis is carried out in order to show the effectiveness of the proposed model. Finally, managerial insights are proposed, supported by a scenario analysis.