Photo: Christian Ove Carlsson

CERE publication published in IFAC-PapersOnLine

Tuesday 18 Sep 18
|

Contact

John Bagterp Jørgensen
Professor
DTU Compute
+45 45 25 30 88

“A Thermodynamic Library for Simulation and Optimization of Dynamic Processes”

 

Tobias K.S. Ritschel, Jozsef Gaspar, Andrea Capolei, and John Bagterp Jørgensen
(IFAC PapersOnLine, 50 (2017) 3542-3547)
(Proceedings of the 20th World Congress of the International Federation of Automatic Control, July 9-14, 2017, Toulouse, France. DOI:https://doi.org/10.1016/j.ifacol.2017.08.951

Abstract: Process system tools, such as simulation and optimization of dynamic systems, are widely used in the process industries for development of operational strategies and control for process systems.

These tools rely on thermodynamic models and many thermodynamic models have been developed for different compounds and mixtures. However, rigorous thermodynamic models are generally computationally intensive and not available as open-source libraries for process simulation and optimization.

In this paper, we describe the application of a novel open- source rigorous thermodynamic library, ThermoLib, which is designed for dynamic simulation and optimization of vapor-liquid processes.

ThermoLib is implemented in Matlab and C and uses cubic equations of state to compute vapor and liquid phase thermodynamic properties. The novelty of ThermoLib is that it provides analytical first and second order derivatives. These derivatives are needed for efficient dynamic simulation and optimization. The analytical derivatives improve the computational performance by a factor between 12 and 35 as compared to finite difference approximations.

We present two examples that use ThermoLib routines in their implementations: (1) simulation of a vapor-compression cycle, and (2) optimal control of an isoenergetic-isochoric flash separation process. The ThermoLib software used in this paper is distributed as open-source software at www.psetools.org. 


News and filters

Get updated on news that match your filter.