The production of liquefied natural gas (LNG) is a highly energy intensive process, as required liquefaction temperature is approximately -160 degrees C at atmospheric pressure. In this study, we propose a novel bi-level optimizing operation system for an LNG process, which consists of a real-time steady-state optimizer (RTSSO) and a decentralized control system. The RTSSO computes the optimal operating conditions such that the compressor power is minimized, while the decentralized control system performs real-time feedback actions to attain the target operating points against various disturbances. Special attention was given to the decentralized control system so that i) the process operation can be rapidly stabilized, and ii) the developed system can be seamlessly applied to an actual process. The performance of the proposed operation system was validated in a numerical LNG plant that precisely replicates an actual plant that produces 100 t of LNG per day. (C) 2016 Elsevier Ltd. All rights reserved.