Methodological proposal for advanced control for operational point change in interactive systems with intensified units: purification of lactic acid

• Autores: Angel Santiago Cabezas López
• Directores de la Tesis: Jorge Mario Gómez Ramírez (dir. tes.), Federico Lozano Santamaria (codir. tes.), Nicolás Rios Ratkovich (tut. tes.)
• Lectura: En la Universidad de los Andes (Colombia) ( Colombia ) en 2026
• Idioma: inglés
• Tribunal Calificador de la Tesis: Galo Antonio Carrillo Le Roux (presid.), David Alejandro Liñán Romero (secret.), Diego Camilo Pradilla Raguá (voc.)
• Enlaces
  • Tesis en acceso abierto en: hdl.handle.net
• Resumen

  • This thesis develops a methodological proposal to perform operating-point changes in multigrade chemical processes using digital-twin environments and advanced control strategies. The study focuses on the catalytic-distillation purification of lactic acid, a representative intensified system with strong interactions, nonlinear dynamics, and tight thermal and hydraulic constraints. The work demonstrates how a sequential digital-twin architecture can articulate steady-state optimization, dynamic modelling, regulatory control, and model-based predictive control to achieve economically consistent transitions between product purities of 80, 88, and 99 mol %.

    The steady-state digital twin reproduces the thermodynamic and kinetic behaviour of the process and supports the formulation and solution of a real-time optimization problem that delivers economically feasible operating points. These operating points define the reference conditions required for the subsequent control layers. The dynamic digital twin incorporates hydraulic dynamics and a complete regulatory layer configured with proportional–integral–derivative controllers, which stabilise levels, pressures, and temperatures within each unit and provide the operating environment in which supervisory control strategies can act.

    A linear state-space model is identified from step-test data generated on the dynamic digital twin, capturing the closed-loop behaviour of the system under regulatory control. This model supports the design of a Model Predictive Control layer responsible for executing operating-point changes in closed loop. The controller achieves stable transitions with low overshoot, zero offset, and settling times consistent with the dynamics of the column train.

    The thesis then extends the formulation to an Economic MPC structure to assess how explicitly incorporating process economics affects transition trajectories and steady-state solutions. The comparison highlights how economic terms modify the controller’s decisions and offers insights into the trade-offs between economic performance and tracking accuracy.

    Overall, the thesis demonstrates that sequential digital-twin development provides a coherent and transferable route to analyse, design, and evaluate advanced control systems for intensified processes requiring flexible and economically aware operating-point changes.