(29a) Robust Operational Optimization of Crude Oil Distillation Systems

Operational optimization of crude oil distillation systems can bring significant economic and environmental benefits considering their massive throughput and extensive energy use. However, crude feed compositions, characterized by true boiling point (TBP) curves, are usually not available due to complex crude oil movement and mixing operations.

Instead of employing expensive online crude composition analyzers, this work develops a low-cost method without exact feed TBP data based on the so-called robust optimization technique.The method can return the optimal operating conditions satisfying process constraints for a range of predefined crude scenarios. The robust operational optimization workflow can be divided into two stages, an offline stage and an online stage. In the offline stage, possible crude scenarios are analyzed and a linear optimization model for each crude scenario is generated. Based on the generated linear models, an uncertainty set for model parameters is then identified for robust optimization. In the online stage, current states of the objective function and process constraints are adapted from real-time plant data. Combining the uncertainty set and the current states, the robust operational optimization model is generated and solved.

The effectiveness of the proposed method is demonstrated by a case study. The case study also shows that robust optimal solutions lose about 2% of optimization potentials compared with the situation that perfect online feed TBP data is available.