Among the many changes in industrial production brought by Industry 4.0, product shorter lifetime and wider variety are two of the hardest challenges that manufacturing systems must face to stay competitive. To this end, Reconfigurable Manufacturing Systems (RMS) are among the most valuable tools, mainly due to being highly scalable, i.e. production tasks and/or resources can be reorganized in very reasonable time to promptly adapt the throughput to quick demand changes. The use of the Scalability property of RMSs can be generalized to achieve efficiency and responsiveness in many situations displaying variability in some form. Such is the case of a variable energy cost profile: e.g. with Time-Of-Use pricings, an RMS allows to use lower-consumption, lower-throughput configurations in costlier periods, and higher-consumption, higher-throughput configurations in off-peak periods, hence to fulfill a given demand with minimum energy-related costs. However, designing an RMS to achieve this goal is difficult, as two decision problems must be faced together that are hierarchically related, usually tackled in distinct moments, and yet strongly intertwined: line balancing and production planning. Moreover, several criteria can be considered to assess the suitability of a design, giving rise to a Multi-Objective Bilevel Optimization Problem, balancing and planning being the upper-and lower-level. We consider here three criteria, namely the number of workstations and-to take into account demand uncertainty-the expected values of service level and energy cost w.r.t. a set of demand scenarios. We propose a three-phase matheuristic and discuss its economical impact on instances derived from the literature.