Limited Time Offer

Claim a 25% discount on all eLearning courses (including credentials) with code ELEARN25.

Offer is valid from March 10-31. Public courses excluded from promo. 

(55d) Product Tri-Section Based Crude Distillation Unit Model for Refinery Production Planning and Refinery Optimization

Li, F. - Presenter, East China University of Science and Technology
Qian, F., East China University of Science and Technology
Yang, M., East China Universtiy of Science and Technology
Du, W., East China University of Science and Technology
Mahalec, V., McMaster University
Crude distillation unit (CDU) is the start of the production process in a refinery. Accuracy of a CDU model has a very significant impact on the refinery production plan. High model accuracy is typically accomplished via nonlinear models, which leads to convergence difficulties when the entire refinery model is optimized. CDU model presented in this work is an MILP model with a modest number of binary variables, and it has accuracy on par with rigorous tray to tray CDU models. The model relies on the observation (Fu et al., 2016) that a line through the middle of the TBP curve of a CDU product (which is easily modeled via linear partial least squares) depends on the crude feed properties and the yields of the adjacent products. Novelty of the model presented in this work (product tri-section model) is that it does not require models of individual distillation towers comprising CDU, thereby leading to a much simpler model structure that is amenable to representation as a MILP. The model predicts the product true boiling point (TBP) temperatures and the bulk properties (e.g., specific gravity, sulfur content). High accuracy and mixed-integer linear structure make this model eminently suitable for the plant-wide refinery production planning optimization. Significant reduction in the computational effort required for the optimization of nonlinear refinery models is illustrated by comparison with previous work, while its high accuracy is illustrated by comparison with a rigorous tray to tray model in AspenPlus.