Present study addresses optimum design of MEK manufacturing plant based on the flowsheet simulation and two-step cost optimization. The plant consists of a reactor that converts 2-butanol to MEK by dehydrogenation and several separation units, which are interconnected by the recycle streams. Due to the existence of recycle streams, the flowsheet simulation exhibited convergence difficulty. In the present study, the difficulty has been alleviated by introducing following techniques: first, In-line Fortran is used to make-up material loss problem; second, tearing and making a convergence component group; third, reordering calculating sequence in the flowsheet sequence. Aspen Plus steady-state simulator is used for the flowsheet simulation and two-step cost optimization strategy has been employed to determine the design and operating variables. In the optimization part, SQP-based infeasible path approach is applied in order to avoid convergence problems due to the existence of several recycle streams. Two-step cost optimization strategy was adopted as an alternative of simultaneous optimization of design and operating variables. First step is to determine the design variables via sensitivity analysis of capital and operating cost in the distillation column. In the second step, operating variables were optimized for the given design specification by SQP-based infeasible path approach.