(163b) Model Reduction for Complex Systems Analysis | AIChE

(163b) Model Reduction for Complex Systems Analysis


Schmal, P. - Presenter, Process Systems Enterprise Inc.
Kilinc, M., Carnegie Mellon University
Sahinidis, N., Carnegie Mellon University
Lawal, A., Process Systems Enterprise
Chowdhury, A., Process Systems Enterprise
We present results from the application of the ALAMO methodology to the development of reduced-order models (surrogate models) for several case studies of interest to the design, analysis, control, and optimization of chemical processes. We begin by considering several simple systems, including flash calculations and steam tables. We then assess the capabilities of ALAMO to build simple and accurate models for an air separation unit. Air separation is an energy intensive process that involves heat integration and material recycles between columns. As a result, this system is often difficult to model. A high-fidelity model for an air separation unit was developed in gPROMS. ALAMO was utilized in order to reduce the complex simulation model to a reduced-order algebraic surrogate model. Extensive computational results are presented to compare the two models in terms of run time and relative accuracy.