Discrete element method modeling of biomass fast pyrolysis granular flows

Qi, Fenglei
Major Professor
Mark Mba-Wright
Committee Member
Journal Title
Journal ISSN
Volume Title
Research Projects
Organizational Units
Mechanical Engineering
Organizational Unit
Journal Issue
Mechanical Engineering

Organic biomass is an abundant renewable resource on the earth and properly utilizing biomass resources could provide an alternative energy to traditional fossil fuels and help to mitigate the impacts of energy consumption on environment and climate change. Fast pyrolysis is one way to achieve thermochemical conversion of biomass organic materials into bio-oil at mild temperature (500 oC) in the absence of oxygen. Due to high heating rate requirement and low thermal conductivities of biomass materials, physical processes such as particulate flows, mixing and heat transfer have complicated effects on biomass fast pyrolysis at both reactor scale and particle scale. Besides the intensive research of the chemistry of biomass fast pyrolysis, study of the underlying physics is also necessary for gaining more knowledge of biomass fast pyrolysis processes in practical reactors.

In this research, the biomass pyrolysis reactive granular flow in a double screw reactor is numerically investigated and the underlying physics such as particle mixing and heat transfer in the reactor are studied. A new Discrete Element Method (DEM) model was proposed with extended capability of modeling particle-particle and particle-wall heat transfer and integrating biomass devolatilization reaction models for simulating reactive granular flows. In the DEM model, the particle hydrodynamics is modeled by adopting Hertz-Mindlin nonlinear soft sphere model. The particle-scale heat transfer model considers both conductive and radiative heat transfer between particle and particle/wall. The biomass devolatilization model involves coupling with energy equation in an adaptive time step manner and considers the variation of solid particle thermal properties with temperature and conversion process.

Particle flow and mixing have a great impact on biomass fast pyrolysis process by affecting the heat transfer dynamics in the granular flow. The DEM was first employed to investigate the granular flow and particle mixing in a double screw reactor. Visual observations suggest the simulation captures the particle mixing trends observed in the experiments. Results indicate that the mixing index profile in the axial direction shows a mixing-demixing-mixing oscillation pattern. Increasing screw pitch length is detrimental to mixing performance; decreasing the solid particle feed rate reduces the mixing degree; and increasing the biomass to glass bead size ratio decreases mixing performance. A comparison of a binary, single-sized biomass and glass particle mixture to a multicomponent mixture indicates that the binary system has similar mixing pattern as a multicomponent system.

The developed particle-scale heat transfer model was validated by modeling heat transfer in packed beds and comparing simulation predictions with experimental measurements. The simulation results of the heat transfer in the double screw reactor indicate an existence of both spatial and temporal temperature oscillations in the granular flow. The effects of the operating conditions on the average temperature profile, biomass particle temperature probability distribution, heat flux and heat transfer coefficient are analyzed. The results show that the particle-fluid-particle conductive heat transfer pathways are the dominant contributors to the total heat flux, which accounts for approximately 70%-80% in the total heat flux. Radiative heat transfer contributes 14%-26% to the total heat flux. The heat transfer coefficient in the double screw reactor varies in a range of 70 to 110 W/(m2K) depending on the operating conditions.

The proposed approach was applied to simulating biomass fast pyrolysis process in the double screw reactors. Results show that the heat of pyrolysis needs to be considered for accurate prediction of biomass pyrolysis process in the reactor. The hemicellulose and cellulose decompositions are predicted to start around 480 K and 600 K, separately, and the predictions are in agreement with experimental studies. The product yield predictions also have a good agreement with experimental studies. Results indicate that both decreasing particle size and reducing feedstock volumetric fill level in the reactor are favorable to the biomass pyrolysis process.

A multi-objective kinetic parameter regression model was proposed for estimating parameters in kinetic models in the last part of this research. The proposed regression model integrated a multi-objective particle swarm optimization algorithm with ODE solver from CVODE. A case study indicates that this regression model has a better performance comparing to traditional deterministic optimization solvers.