A distributed predictive control framework based on a state-space model with constraints on the output and control-input is proposed. By a benchmark process, the performance of this framework is analyzed and compared with centralized control strategies in a regulation problem of a distillation column.