Utilizing Cyber-Physical System Based on Plant Models, Proactive and Safe Control, Toward Full Automating Operation

Due to the rapid development of computing, sensing and network technologies, advanced Internet of Things (IoT) technologies enable the integration of the physical and the cyber world into a cyber-physical system. Although IoT is just a gateway into and out of the real world and cyber world, data and information from the physical world are analyzed and learned in the cyber world, and in which a model that describes behaviors in the physical world is created. Utilizing the cyber-physical system is expected to create new value to the physical world. In the cyber-physical system, the one of the major goals must be to realize automating plant operation system.

We have developed “Mirror Plant” which is an online dynamic plant simulator. The rigorous models based on physical and chemical laws are used in Mirror Plant. Mirror Plant receives sensor-measured data directly from the plant for the model calibration. The model parameters inside Mirror Plant are automatically adjusted in the real time. As the result, Mirror Plant can perfectly simulate the dynamic behavior of the actual plant. By making computer run faster than reality, Mirror Plant can accurately predict near future plant behavior. When Mirror Plant tells an alarm will occur in 30 minutes, the operator can make action to prevent predicted accident before something happened. Mirror Plant realizes a proactive plant operation yet to be seen in the conventional plant operation and contributes to safer plant operation. Mirror Plant is already used in Japanese commercial petrochemical plants and LNG processing facility by operators who are directly involved in production and technical staffs who support production operation.

The development of IoT technology is accelerating the digital transformation in the manufacturing site, where the physical world and the cyber world will interact closely with each other. It is expected that big data utilization and AI technology will lead to more sophisticated plant model like Mirror Plant in the cyber world. The plant model which can exactly simulate the target various applications. By using the high-fidelity plant model, the unattended operation at night and fully automated operation of plants will be realized in future. As the new automating plant control technology, we introduce Economic Model Predictive Control (EMPC), which uses nonlinear plant models created in the cyber world and delivers optimal and safe control system design. We applied EMPC to a chemical batch process with strong nonlinearity. EMPC is the one-layer type controller in which both control and optimization function are involved. We introduce the optimal operation result by EMPC shows high productivity with high yield and energy saving compared to the conventional PID control. If a plant model is available, anyone can use EMPC and specify any production requirements, such as maximum productivity ensuring environment operation.