Designing a Distributed Control System (DCS) for power plants involves complex allocation decisions at two levels: functional and physical. At the functional level, operations are assigned to Automation Units to balance processing loads, while at the physical level, cables are assigned to electrical cabinets to minimize separation. We model this problem as a 0-1 Integer Program, which is computationally challenging even for commercial solvers. To address this, we propose a Generalized Benders Decomposition (GBD) framework that exploits the two-level structure, combined with symmetry-breaking constraints to eliminate equivalent solutions and accelerate convergence. Computational experiments on representative instances show that GBD outperforms direct resolution methods, and symmetry-breaking further improves efficiency, making it a promising approach for DCS design.