Cryogenic air separation unit (ASU) is considered as the currently available commercial oxygen production method for oxy-fuel combustion power plants; however, this method leads to significant energy penalty and economic cost. Real-time optimizing system operations during dynamic processes (such as flow rate change, oxygen product purity change, and flexible operation) are expected to achieve remarkable energy savings. Dynamic exergy provides a powerful indicator for real-time evaluating the system thermodynamic performance and quantifying the impact of a control strategy. In this work, some important transient exergy parameters of ASU systems under typical dynamic operating scenarios were first obtained through combining steady-state and dynamic process simulations. Next, control penalty and cost for internal control structures (layers and loops) were determined for the optimizations of control strategy and operation. Feedforward-feedback control structure and ASU-following control strategy are more suitable for ASU regulation and flexible operation, respectively, because more efficient thermodynamic performance is achieved during the investigated operating scenarios. The control structure, layer and loop play different roles in terms of energy behavior and require reasonable regulation to optimize energy behavior. This study provides an important insight into using control optimization aided by the dynamic exergy method to implement energy-efficient operations for industrial plants. (C) 2018 Elsevier Ltd. All rights reserved.