Plant operation support system with prediction functions that realizes safe and optimal operation

The number of plant accidents caused by misrecognition or misoperation is increasing in Japan. We have developed an operation assist system, Mirror Plant, to prevent such accidents. Mirror Plant is an online simulator based on a first principles model, which can provide wide-ranging support in situation recognition or decision making to plant operators through prediction and visualization. This paper focuses on new prediction functions incorporated into this system, i.e., steady state prediction and transient state prediction. As these prediction functions are based on real-time process data, they can take into account the continual changes in plant conditions. Using these prediction functions to investigate the consequences of future changes in operation enables safe plant operation, even if operation conditions are close to being off specification or in the explosive territory.