Advanced composite manufacturing processes use polymers and fibers pre-impregnated with thermoplastic or thermoset resins as raw-materials. These highly sensitive materials are considerably perishable, affecting the production process in several ways and requiring special inventory cares. This perishable condition of raw materials in composite manufacturing constitutes the main motivation for this paper. We study a multi-level lot-sizing problem with raw-material perishability and batch-ordering. Several types of perishable raw-material items are procured in batches from suppliers. Once received, two different inventory sub levels are considered: a special storage location where the material can be stored avoiding deterioration, and a secondary location at the shop floor where the material becomes available for production and starts deteriorating. We proposed a mixed-integer programming formulation for the problem and perform computational experiments with sensitivity analyses, demonstrating its potentials for practical applications in planning composite production.