427201 First Principles Based Design of 3D Pyrolysis Reactors Using LES

Wednesday, November 11, 2015: 3:49 PM
250B (Salt Palace Convention Center)
David J. Van Cauwenberge1, Pieter A. Reyniers1, Jens Floré2, Kevin M. Van Geem1 and Guy B. Marin1, (1)Laboratory for Chemical Technology, Ghent University, Ghent, Belgium, (2)BASF Antwerpen N.V., Antwerp, Belgium

Pyrolysis is the main process for the production of many valuable organic building block chemicals such as ethylene, propylene and vinyl chloride. The free-radical gas phase reactions are accompanied by secondary reactions leading to the formation of a carbonaceous coke layer on the inner walls of the reactor tubes. This layer leads to an increased pressure drop over the reactor causing a loss in selectivity. Additionally, the insulating effect of the layer forces a higher furnace firing rate and higher tube metal temperatures to maintain the same conversion. In industrial plants regular decoking procedures are hence inevitable.

Because of this detrimental effect on the overall economics, the design of novel reactor geometries exhibiting improved heat transfer and reduced metal temperatures with the aim of increasing run lengths has received quite some attention [1]. In order to assess the effect of these geometries however, one must account for the chemical kinetics, as well as the complex flow phenomena in the reactor. While reliable microkinetic models with thousands of reactions and hundreds of intermediates are widely applied, implementing these efficiently in a Computational Fluid Dynamics (CFD) code remains a challenge [2]. Commercial packages such as ANSYS Fluent for example do not even allow kinetic models with more than 50 species.

Description: Description: C:\Users\djvcauwe\Pictures\MERT\Filmpje\T_tube\06 (4).png

Figure 1: Streamwise periodic temperature field for a spirally corrugated pipe.

In the present contribution, detailed kinetics were incorporated in the open-source CFD code OpenFOAM by reduction of a butane pyrolysis reaction mechanism, automatically generated with Genesys [3]. The quasi-steady-state approximation is applied in order to reduce the stiffness of the system and the computational time. A large eddy simulation (LES) methodology is employed in order to capture the turbulent-chemistry interaction on a smaller scale than the classical Reynolds-averaged Navier-Stokes (RANS) turbulence models. The length of the computational domain is limited by extending on the periodic methodology of Patankar et al. [4] for fully developed velocity and temperature fields. This novel approach is validated by comparison with full-scale simulations for a variety of geometries and Reynolds numbers. Additionally, a pilot plant setup with a standard tubular reactor and a spirally corrugated tube is simulated and compared with the experimental results. The advantages and disadvantages of using three-dimensional reactor configurations instead of conventional tubular reactors are discussed.


[1] C.M. Schietekat, D.J. Van Cauwenberge, K.M. Van Geem, G.B. Marin, Computational fluid dynamics-based design of finned steam cracking reactors, AlChE J., 60 (2013).

[2] K. He, M.G. Ierapetritou, I.P. Androulakis, Integration of on-the-fly kinetic reduction with multidimensional CFD, AlChE J., 56 (2010) 1305-1314.

[3] N.M. Vandewiele, K.M. Van Geem, M.-F. Reyniers, G.B. Marin, Genesys: Kinetic model construction using chemo-informatics, Chem. Eng. J., 207–208 (2012) 526-538.

[4] S.V. Patankar, C.H. Liu, E.M. Sparrow, Fully Developed Flow and Heat Transfer in Ducts Having Streamwise-Periodic Variations of Cross-Sectional Area, J. Heat Transfer, 99 (1977) 180-186.



DVC gratefully acknowledges financial support from the combined IWT and BASF Antwerpen N.V. Baekeland program. PAR acknowledges financial support from a doctoral fellowship from the Fund for Scientific Research Flanders (FWO). The authors also acknowledge the financial support from the Long Term Structural Methusalem Funding by the Flemish Government – grant number BOF09/01M00409. The computational work was carried out using the STEVIN Supercomputer Infrastructure at Ghent University, funded by Ghent University, the Flemish Supercomputer Center (VSC), the Hercules Foundation and the Flemish Government – department EWI.




Extended Abstract: File Not Uploaded