The increasing scale and complexity of modern process systems, along with the growing demand for resource efficiency, profitability, and sustainability, underscore the need for advanced control strategies. As systems become larger and more interconnected, optimizing their performance is essential to minimize resource consumption, reduce emissions, and ensure operational profitability. To meet these objectives, effective control strategies must be capable of adapting to varying process conditions while addressing scalability and robustness challenges.

Model Predictive Control (MPC) is a well-established method known for its adaptability and flexibility. However, when applied to large-scale systems, MPC faces several challenges, including computational complexity, communication constraints, and the need for reliable coordination among subsystems. To address these issues, decentralized and distributed control strategies have been developed, distributing control tasks among subsystems. Despite this, these strategies often struggle with reliability due to communication delays and coordination difficulties.

Hierarchical control structures have emerged as a promising solution, providing modularity and robustness to handle the complexities of large-scale systems. However, the design and implementation of these structures remain a significant challenge, especially in chemical engineering applications, where process units exhibit non-uniform dynamics and require controllers of varying sizes. As a result, innovative approaches are necessary to manage these complexities while ensuring optimal system performance.

This study presents a unified hierarchical control optimization framework that simultaneously determines the optimal control architecture and operational control policy through a single mixed-integer nonlinear programming (MINLP) formulation. The proposed Multi-Rolling Horizon (MRH) strategy determinesoptimal control architectures while maintaining structural feasibility and computational tractability, embedding all possible controller-subsystem assignments and communication topologies into one integrated optimization problem. The methodology systematically balances regulation accuracy, control effort, communication overhead, and computational complexity through a normalised composite function, ensuring that the resulting control topology is directly deployable for distributed MPC execution. Preliminary test cases on small-to-medium scale systems demonstrate the framework's potential for optimizing control structures, showing how the unified approach can balance multiple performance objectives, including regulation accuracy, control effort, communication overhead, and computational complexity. This research contributes a systematic framework for control architecture optimization that offers a step toward more integrated approaches in process control design, with future work needed to validate its applicability across diverse industrial systems.