(26f) Reduced-Order Model for a Large-Scale Real-Time Optimization of a Residue Fluidized Catalytic Cracker

Masuku, C. M., Carnegie Mellon University
Bai, A., Carnegie Mellon University
Wu, K., Carnegie Mellon University
Liporace, F. S., Petrobras
Biegler, L., Carnegie Mellon University
Niederberger, J., Petrobas
Efficient nonlinear programming algorithms and modeling platforms have led to powerful process optimization strategies [1]. However, these algorithms are challenged by recent evolution and deployment of large-scale equation-oriented (EO) real-time optimization (RTO) models in industrial Oil & Gas facilities. Petrobras’ RECAP unit in Mauá, Brazil, currently has an RTO system that was implemented in Aspen Plus for a Distillation, RFCC and Solvents units, all integrated in one large-scale model. This optimization model tracks the changes in the plant’s input, calculate an objective function based on profit maximization considering the constraints of operating conditions and then solved using Aspen Plus Optimizer (incorporating: simulation, optimization, parameter estimation, and data reconciliation modes). Since the model of the whole plant is large, including more plant sections requires accurate and efficient reduced-order models (ROMs) to save running time and costs.

The residue fluidized catalytic cracking unit is chosen as a starting point for building the ROM in PYOMO using a trust-region framework for glass box/black box optimization [2]. Since ROMs are sufficiently accurate only in a restricted zone around the point in decision variable space where they are constructed [3], they are evaluated and compared with the detailed EO model in the following aspects: 1. Ease to generate the model; 2. Model running time; 3. Optimal result accuracy (objective function and variables); 4. Sensitivity to the offset (ε); and 5. Stability of extrapolation.

[1] L.T. Biegler, Y.-D. Lang, W. Lin, Multi-scale optimization for process systems engineering, Comput. Chem. Eng. 60 (2014) 17–30.

[2] J.P. Eason, L.T. Biegler, A trust region filter method for glass box/black box optimization, AIChE J. 62 (2016) 3124–3136.

[3] A. Agarwal, L.T. Biegler, A trust-region framework for constrained optimization using reduced order modeling, Optim. Eng. 14 (2013) 3–35.

Keywords: Real Time Optimization, Reduced Order Models, Crude Distillation, Equation-Oriented Optimization, Trust Region Method.