Thermochemical Conversion of Biomass to Fuels
Scientists in the Computational Science Center at the National Renewable Energy Laboratory (NREL) used direct numerical simulations (DNS) in two biomass conversion projects to thoroughly investigate:
Biomass thermochemical conversion processes, such as gasification and pyrolysis, hold great promise for the production of second- and third-generation biofuels and will play a determining role in meeting the U.S. renewable fuels targets set for the next 20 years. However, current technologies based on solid-gas fluidized beds face significant challenges, noticeably increasing the risks associated with the development of industrial-scale facilities.
The major challenge is improving control of the product gas composition, which determines the extent of expensive post-treatments such as scrubbing and conditioning before liquid transportation fuels can be obtained. This lack of control is due in part to the very complex bed dynamics. Bubbles, upset conditions, and gas streaming greatly impact the residence time of the biomass decomposition products inside the reactor, which in turn alters the composition of the raw output gas.
Empirical techniques alone cannot provide enough detail data to develop optimization strategies in a reasonable timeframe and, therefore, accurate computational fluid dynamics (CFD) simulations of this kind of systems have the potential to accelerate considerably the development and deployment of commercial-scale conversion units. However, the wide range of time and lengths scales encountered in the reactor, along with the lack of detailed understanding of the biomass chemistry, renders predictive simulations of biomass conversion in fluidized beds extremely challenging.
To increase the fundamental understanding of these systems and develop accurate models for large-scale simulations, a multi-level approach takes place, from particle scale to laboratory scale.
Flow past freely falling spheres in a periodic box using an immersed boundary technique.
Principal investigator: Perrine Pepiot, National Renewable Energy Laboratory
Collaborator: Olivier Desjardins, Assistant Professor, University of Colorado at Boulder
Gas-Solid Coupling in Fluidized Bed Reactors
Predicting the dynamics of fluidized beds requires a careful description of gas-particle and particle-particle interactions. This project uses resolved-particle DNS of dense particulate flows with moving and colliding particles based on a conservative immersed-boundary approach to thoroughly investigate gas-solid coupling in fluidized bed reactors. Database construction is underway of flows past freely moving particles at various Reynolds numbers and levels of packing, covering a large range of scales, thanks to the excellent scaling properties of the numerical tools. The objectives are to provide more appropriate closures for gas-particle interactions in both Lagrangian and Eulerian descriptions of the particulate phase, enable large eddy simulation (LES) of dense solid-gas flows, and characterize the nature of the gas flow in terms of turbulence and energy dissipation laws.
Tar Formation during Biomass Gasification
The objective is to perform DNS of the 4-inch fluidized bed reactor used at NREL to investigate tar formation during biomass gasification. The solid phase is described using a point-particle assumption, and the coupling between particle transport and gas phase flow is integrated within the DNS/LES code NGA [Desjardins et al., J. Comp. Phys., 227(18):8395-8416, 2008), which has shown excellent scaling properties up to 12K cores. The results will help characterize the interactions between bed dynamics and chemical processes, with a focus on the reaction pathways leading to undesirable products such as aromatic species.
Point-particle direct numerical simulation of a fluidized bed reactor containing 12 million particles.
Principal investigator: Perrine Pepiot, National Renewable Energy Laboratory
Data visualization: Kenny Gruchalla, National Renewable Energy Laboratory
Collaborators: Olivier Desjardins, Assistant professor, University of Colorado at Boulder, and Jesse Capecelatro, Ph.D. candidate, University of Colorado at Boulder
