Data-driven Analysis and Optimization of Externally Heat-Integrated Distillation Columns (Ehidic)

Peng Qiu,Bo Huang,Zhenghua Dai,Fuchen Wang
DOI: https://doi.org/10.1016/j.energy.2019.116177
IF: 9
2019-01-01
Energy
Abstract:To present a data-driven framework for analyzing and optimizing a highly coupled complex system with multivariable in large search space, externally heat-integrated distillation columns (EHIDiC) for propylene/propane separation is studied. By analyzing a large dataset from Aspen Plus-MATLAB distributed computing, we not only make a wide range of global optimization but also achieve a further understanding of EHIDiC. First, each independent variable has a different degree of influence on the total annualized cost (TAC), in which the location of heat-integrated stages in the rectifier has the greatest impact. Second, the capital and operation costs of the compressor are the dominant parts of the TAC, which need to be considered carefully in design and optimization. Third, the EHIDiC scheme in lowest TAC performance prefers heat integrated between the bottom of the stripper and the top of the rectifier, while heat rejects from the bottom of the rectifier causes the highest TAC performance, regardless of heat distribution on the stripper. Fourth, compared with Genetic Algorithm optimization result, the data-driven analysis can discern more underlying scenarios with less computing time in much bigger search zones. At last, a more TAC saving EHIDiC scheme is constructed by shifting some separation work from the stripper to the rectifier.
What problem does this paper attempt to address?