(497e) Distributed Safeness Index-Based Predictive Control for Enhanced Process Operational Safety

Though industry has traditionally decoupled the designs of the process control and safety systems, recent work [1] has demonstrated that coordinating the control and safety systems through a common metric termed the Safeness Index can enhance process operational safety. The Safeness Index is a process-specific metric that is a function of the closed-loop state and therefore mathematically formalizes a systems approach to process safety [2]. It is developed from information on past process accidents, first-principles models, industrial safety studies, and process operating data. Thresholds on this index are used to set constraints in model predictive control and to trigger actions of the safety system. An economic model predictive control (EMPC) design with Lyapunov-based stability constraints and Safeness Index-based constraints (termed Safeness Index-based LEMPC) was developed in [1] with a centralized control architecture. Distributed control designs (e.g., sequential and iterative [3], [4]) can be beneficial for reducing the computation time of EMPC and are thus an attractive method for implementing Safeness Index-based LEMPC that may enhance its acceptance by industry.

In this work, we investigate the development of sequential and iterative distributed Safeness Index-based LEMPC designs. Due to the hard Safeness Index-based constraints that require the EMPC to maintain the process state within a region where the Safeness Index is below a given threshold, the distributed Safeness Index-based LEMPC design poses unique feasibility challenges that suggest that inputs should be partitioned such that those inputs that most affect the value of the Safeness Index are computed by one distributed controller. Feasibility considerations also affect the order in which the distributed controllers are recommended to be evaluated in a sequential distributed Safeness Index-based LEMPC design. Implementation strategies for both sequential and iterative distributed Safeness Index-based LEMPC are developed that are proven to maintain closed-loop stability in a bounded region of state-space and to always drive the closed-loop state into the region in state-space where the Safeness Index is below a threshold whenever it exits this region. A chemical process example is used to demonstrate the proposed control designs.

[1] Albalawi A, Durand H, Christofides PD. Process operational safety using model predictive control based on a process safeness index. Computers & Chemical Engineering. in press.

[2] Leveson NG, Stephanopoulos G. A system-theoretic, control-inspired view and approach to process safety. AIChE Journal. 2014;60:2-14.

[3] Liu J, Chen X, Muñoz de la Peña D, Christofides PD. Sequential and iterative architectures for distributed model predictive control of nonlinear process systems. AIChE Journal. 2010;56:2137-2149.

[4] Scattolini R. Architectures for distributed and hierarchical model predictive control - A review. Journal of Process Control. 2009;19:723-731.