A hierarchical approach to production planning for complex manufacturing systems is presented. A single facility comprising of a number of work-centers that produce multiple part types is considered. The planning horizon includes a sequence of time periods, and the demand for all part types is assumed to be known. The production planning problem consists of minimizing the holding costs for all part types as well as the work-in- process, and the backlogging cost for the end items. We present a two- level hierarchy that is based on aggregating parts to part families, work-centers to manufacturing cells and time periods to aggregate time periods. The solution at the aggregate level is imposed as a constraint to the detailed level problem which employs a decomposition based on manufacturing cells. This architecture uses a rolling horizon strategy to perform the production management function. We have employed perturbation analysis techniques to adjust certain parameters of the optimization problems at the detailed level to reach a near- optimal detailed production plan.