The air separation unit (ASU) is one of the core elements of integrated gasification combined cycle (IGCC) power plants. The ASU separates air into pure oxygen and nitrogen, to be sent to the gasifier and the gas turbine, respectively. This system consumes about 10% of the gross power output generated in IGCC, so its economical operation is important for lowering the overall power generation cost. The use of an elevated-pressure air separation unit (EP ASU), in which the operating pressure is higher than in a conventional ASU, is known to lead to significant energy savings. In this research, controlled variable selection for an EP ASU was studied, considering both the controllability and economics, that is, with the objective of maintaining economically near-optimal operations in the presence of anticipated load changes. The main tool used for this was the so-called "minimum singular value rule" within the overall framework of self-optimizing control (SOC). For the purpose of selecting and testing self-optimizing control structures, equation-based modeling of EP ASU was carried out and implemented on the commercial software platform gPROMS. Then, the minimum singular value rule was applied using steady-state gain matrices (obtained from the simulator) to select a small number of candidate sets for controlled variables, to which rigorous analyses based on nonlinear simulation and optimization could be applied to pick the top choice. Before the minimum singular value rule was applied, however, certain process insights and heuristics were used to reduce the number of candidate sets down to a manageable level. The economic losses as a result of adopting a fixed control structure were assessed by comparing the hourly operating costs achieved under SOC with the equivalent values obtained by performing full nonlinear optimizations for the given scenarios. In addition, for the suggested control structure, proportional plus integral (PI) control loops were designed, and their dynamic performance was tested in order to make sure that it is attractive in terms of not only economics but also controllability. The finally selected control structure is compared with those presented in previous works.