A mixed-integer (MILP) formulation is presented for the synthesis and operational planning of utility systems for multiperiod operation with varying demands. The design of the utility system optimizes the configuration choice and equipment sizes from a superstructure of alternatives. The optimization problem takes into account the investment costs, operating costs of units for all periods, and the planning for scheduled maintenance of equipments. The multiperiod formulation includes binary variables for the choice of units for design, as well as binary variables for operational status of existing equipment during each period, producing a model that is computationally expensive to solve.A bilevel decomposition method is proposed that consists of solving a higher level design problem for choosing the design configuration in a reduced space that excludes the operational binary variables in each time period. For the choice of units obtained from the first level, the lower level planning problem consists of optimizing their operation and operational status (on or off) for each period. A new concept of design cuts is introduced to tighten the gap between solutions obtained from both levels. Numerical results are presented for several example problems.