Production scheduling in dynamic real-time optimization with closed-loop prediction
Introduction
Decision making in large chemical and petrochemical plants typically follows a hierarchical structure, with basic PI/PID controllers at the lowest level, an MPC system that provides set-points to the lower-level controllers, a real-time optimization (RTO) layer that provides economics-based set-points to the MPC system and possibly also inputs directly to the plant, and a scheduling and planning layer that provides production targets [12], [31].
Increased globalization and deregulation of energy markets have created a dynamic plant operating environment, requiring frequent transitions in order to meet varying product demands and respond to fluctuating energy prices. Backx et al. [2] recognized the impact on process plants of an increasingly transient and competitive marketplace, and advocated a strategy of “intentionally dynamic operation” for enhanced profitably within this setting. Traditional plant decision making systems are more suited to steady-state production of commodity products, rather than the current highly dynamic plant environment. Integration across the various levels of decision making has been recognized as a key challenge in enterprise-wide optimization [22]. Such considerations have led to efforts toward integration of the decision-making layers, or elements thereof, including economic MPC [1], [16], dynamic RTO (DRTO) [25], [27], [42], and integration of scheduling and control [5].
This paper extends a recently proposed DRTO scheme based on closed-loop prediction of the plant response under constrained MPC [25], [28] to include production scheduling decisions. The production scheduling includes the sequencing of products at different specification levels, with the overall objective of maximizing the economics of production over a prediction horizon, where the economics of the transition dynamics are accounted for. The optimal decisions are communicated to the MPC system through time-varying set-point trajectories, thereby allowing the standard regulatory MPC formulation to be retained.
The remainder of this article is organized as follows. Section 2 presents a brief review of literature on DRTO and different approaches for combining scheduling and control. Section 3 presents the detailed formulation of the integrated DRTO and scheduling framework implemented in this work, and the solution approach followed. Section 4 demonstrates the performance of the method through application to SISO and MIMO case studies. Section 5 concludes the article with a summary of findings and identifies future research directions.
Section snippets
Dynamic real-time optimization
RTO traditionally utilizes a steady-state process model for calculating economically optimal operating conditions that are provided to the plant and associated control system [12], [31]. Steady-state RTO systems are limited in their execution frequency by the requirement for (near) steady-state plant conditions, and also do not directly account for plant dynamics. These factors pose limitations in the application of steady-state RTO to plants that exhibit slow dynamics or that undergo frequent
Integrated scheduling and closed-loop DRTO (CL-DRTO) formulation
The conceptual design of the two-layer architecture for integrating scheduling and control within a DRTO framework is shown in Fig. 1. The upper layer incorporates a linearized plant model as a surrogate to generate the predicted plant response. The MPC optimization subproblems are embedded along the DRTO prediction horizon to generate the control actions for the DRTO plant model whose outputs provide disturbance estimates for subsequent MPC calculations. This sequential closed-loop prediction
Case study 1: single-input single-output (SISO) linear dynamic system
The proposed formulation was first applied to a SISO linear dynamic system as a proof of concept to assess whether the algorithm behaved as expected. The dynamic model is a SISO transfer function model, and to give some physical meaning, the input is assumed to represent a volumetric flow rate, and the output a molar concentration. The system is illustrated in Fig. 5 which depicts a CSTR. The flow rate u enters with an inlet concentration of component A, a reaction occurs, and the output is the
Conclusion
Process plants are operating in a significantly more dynamic environment than before, leading to the development of systematic approaches for integration of scheduling and control. To this end, a DRTO framework based on closed-loop prediction of a plant’s response under constrained MPC is extended in this work to include scheduling decisions. The integrated DRTO system determines an economics-based production sequence and provides appropriate set-point trajectories to an underlying MPC system
CRediT authorship contribution statement
Jerome E.J. Remigio: Conceptualization, Methodology, Software, Investigation, Writing - original draft, Visualization. Christopher L.E. Swartz: Supervision, Conceptualization, Methodology, Writing - review & editing.
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgements
Support from the McMaster Advanced Control Consortium (MACC) is gratefully acknowledged. The authors also thank Hao Li for his assistance with the in-house software code that was utilized.
References (46)
- et al.
Economic optimization using model predictive control with a terminal cost
Ann. Rev. Control
(2011) - et al.
Integration of model predictive control and optimization of processes
IFAC Symposium: Advanced Control of Chemical Processes
(2000) - et al.
Integrated production scheduling and process control: a systematic review
Comput. Chem. Eng.
(2014) - et al.
The explicit linear quadratic regulator for constrained systems
Automatica
(2002) - et al.
Optimal grade transition campaign scheduling in a gas-phase polyolefin FBR using mixed integer dynamic optimizaton
- et al.
Integration of scheduling and control with online closed-loop implementation:fast computational strategy and large-scale global optimization algorithm
Comput. Chem. Eng.
(2012) - et al.
RTO: an overview and assessment of current practice
J. Process Control
(2011) - et al.
Integration of scheduling and control under uncertainties: review and challenges
Chem. Eng. Res. Des.
(2016) - et al.
A simulation-based optimization framework for integrating scheduling and model predictive control, and its application to air separation units
Comput. Chem. Eng.
(2018) - et al.
A time scale-bridging approach for integrating production scheduling and process control
Comput. Chem. Eng.
(2015)
A tutorial review of economic model predictive control methods
J. Process Control
Optimal operation: scheduling, advanced control and their integration
Comput. Chem. Eng.
Closed-loop formulation for nonlinear dynamic real-time optimization
IFAC-PapersOnLine
Approximation of closed-loop prediction for dynamic real-time optimization calculations
Comput. Chem. Eng.
A two-level strategy of integrated dynamic optimization and control of industrial processes - a case study
Dynamic real-time optimization of distributed MPC systems using rigorous closed-loop prediction
Comput. Chem. Eng.
Process scheduling under uncertainty: review and challenges
Comput. Chem. Eng.
State-of-the-art review of optimization methods for short-term scheduling of batch process
Comput. Chem. Eng.
Integrated scheduling and dynamic optimization of grade transitions for a continuous polymerization reactor
Comput. Chem. Eng.
A survey of industrial model predictive control technology
Control Eng. Pract.
Production scheduling and linear MPC: complete integration via complementarity conditions
Comput. Chem. Eng.
A mixed-integer formulation for back-off under constrained predictive control
Comput. Chem. Eng.
A state-space model for chemical production scheduling
Comput. Chem. Eng.
Cited by (24)
Integration of chemical process operation with energy, global market, and plant systems infrastructure
2024, Computers and Chemical EngineeringThe quest towards the integration of process control, process operations, and process operability –Industrial need or academic curiosity?
2024, Computers and Chemical EngineeringA scenario-based framework for the integration of scheduling and control under multiple uncertainties
2023, Journal of Process ControlProduction scheduling under demand uncertainty in the presence of feedback: Model comparisons, insights, and paradoxes
2022, Computers and Chemical EngineeringA stochastic optimization framework for integrated scheduling and control under demand uncertainty
2022, Computers and Chemical EngineeringCitation Excerpt :The CL-DRTO formulation used in this study builds on that first proposed by Jamaludin and Swartz (2017b) for online economic optimization of plants with centralized model predictive control, and later extended to processes with distributed MPCs (Li and Swartz, 2018). More recently, Remigio and Swartz (2020) integrated scheduling within the CL-DRTO formulation, leading to an integrated scheduling and control problem. The DRTO is deemed closed-loop because it models the plant response under the action of the model predictive control (MPC).