UCLA

Process safety is a crucial issue in the area of process systems engineering as accident prevention is a top priority in process operations. Operational safety needs to be directly incorporated into control system and safety system to handle disturbances and device failures in the chemical processes. Motivated by the above considerations, this dissertation provides various methods and case studies to demonstrate the integration of safety considerations into controller design. First, we present the dynamic interactions between feedback control and safety systems, and elaborate on the effectiveness by showing applications to a continuous stirred tank reactor (CSTR) and a high-pressure flash drum separator with both classical and model-based controllers. Then, a Safeness Index function is developed to be utilized as a constraint in model predictive control (MPC) design to provide coordination between control and safety systems. The proposed Safeness-Index based MPC is applied to a flash drum and an ammonia production process to enhance process operational safety. Moreover, a large-scale ammonia process network is studied with respect to process operational safety with multiple model predictive controllers to avoid extremely high temperature in the presence of significant disturbances. Additionally, the 2015 explosion accident at the refinery operated by ExxonMobil in Torrance, California is analyzed. A control-based approach is presented for how the accident could have been potentially avoided by simulating accident conditions in the fluid catalytic cracking unit (FCC). Lastly, a method for combining neural network models with first-principles models is presented. The improved performance of this hybrid model is demonstrated in both real-time optimization (RTO) and MPC of a CSTR and a distillation column. Aspen Plus Dynamics, a commercial process simulation software, is integrated with Matlab to carry out the above simulations for large-scale chemical processes to demonstrate the applicability and effectiveness of the proposed control methods.