469862 Optimisation-Based Design of a Heat-Integrated Crude Oil Distillation System Using Rigorous Simulation and Surrogate Models

Tuesday, November 15, 2016: 2:43 PM
Monterey II (Hotel Nikko San Francisco)
Dauda Ibrahim, Chemical Engineering, The University of Manchester, Manchester, United Kingdom, Jie Li, School of Chemical Engineering and Analytical Science, The Univesirity of Manchester, Manchester, United Kingdom, Gonzalo Guillén-Gosálbez, Centre for Process Systems Engineering, Imperial College of Science, Technology and Medicine, London, United Kingdom and Megan Jobson, School of Chemical Engineering and Analytical Science, The University of Manchester, Manchester, United Kingdom

In petroleum refining industry, the crude oil distillation system plays a key role in the overall process. The system comprises a complex heat-integrated distillation column and a heat recovery system where the crude oil feed is partially vaporised. The presence of large number of degrees of freedom (column structure and operating conditions) and the complex interactions existing between individual units makes the design of crude oil distillation unit one of the most challenging tasks within process system engineering. Due to the complex nature of the process, traditional design methods consider designing the complex column and the heat recovery system in separate steps. In this approach, however, interactions between the two subsystems are not taken into account. Some researchers have applied standard optimisation techniques to simultaneously design the crude oil distillation unit and the heat recovery system, but these approaches typically rely on shortcut column models, to avoid the convergence problems associated with rigorous column models. The use of these shortcut models can lead to large errors, as they might fail to predict the behaviour of the complex crude oil distillation unit accurately.

This paper introduce a new framework for the design of crude oil distillation system that integrates a rigorous tray-by-tray crude oil distillation system model implemented in a process simulator (Aspen HYSYS) with an optimiser coded in MatLab. The approach takes advantage of the physical and thermodynamic property models, crude oil characterisation models and column sizing models available in the process simulator.

The proposed approach is implemented in three steps. First, the column superstructure of [1] for optimal feed location, total number of trays and optimal operating conditions is adapted to build the column superstructure using a rigorous tray-by-tray model in Aspen HYSYS. The superstructure consists of conditional and unconditional trays in all column sections, which are represented using Murphree tray efficiencies. In the second step, the problem is formulated as an MINLP with the number of trays in each column section defined as integer variable and the operating conditions as continuous variables. The heat recovery system is represented using Pinch analysis in order to determine the minimum utility requirements that will be used in the calculations. In the final step, an optimisation algorithm (genetic algorithm) is applied to optimise the system, thereby determining the optimal column configuration and operating conditions that minimise a given objective (e.g. total annualised cost).

To expedite the solution procedure, the proposed approach also explored the use of surrogate models based on neural networks that are constructed from a set of samples generated using the original simulation model.

The proposed approach has been applied to the design of a crude oil distillation system that separates crude oil into five products. Two design objectives of interest are considered namely, minimum total annualised cost and minimum energy cost. The results show that significant savings in total annualised cost and energy cost can be achieved using our approach without compromising the solution accuracy.


[1] Yeomans, H. & Grossmann, I. E. Optimal design of complex distillation columns using rigorous tray-by-tray disjunctive programming models. Ind. Eng. Chem. Res. 2000, 39 (11), 4326−4335

Extended Abstract: File Not Uploaded
See more of this Session: Process Design I
See more of this Group/Topical: Computing and Systems Technology Division