We consider an industrial plant equipped with an on-site power generation system based on intermittent renewable energy. We investigate the problem of simultaneously planning the industrial production and the energy supply in this plant on a short-therm horizon. We propose to handle the resulting energy-oriented lot-sizing problem by a two-stage scenario-based stochastic programming approach. We discuss two alternative models differing with respect to when the decision on the production quantities is made. Preliminary computational results aimed at assessing the value of the stochastic solution are provided for both models.