Towards a Muti-scale Modeling Framework for Fluidized Bed Reactor Simulation Addison Killean Stark 1,3, Christos Altantzis 2,3, Ahmed F Ghoniem 3 November 16, 2016 November 16, 2016 1 ARPA-E, 2 NETL, 3 MIT.
Thermochemical Biomass Conversion Pathways Fluidized Bed Gasification Chemical Modeling Challenges INTRODUCTION AND BACKGROUND
Biomass reactor technology: Fluidized bed Particle-scale processes O 2 H 2 O CO 2 Char Char conversion Attrition CO H 2 Combustion+ Gasification Product gases (CO 2, H 2, H 2 O, CO) Freeboard Char Reactor-scale processes Gas-solid mixing/segregation Gas/solids Volatile Solids circulation segregation segregation Advantages High levels of thermal inertia, mixing heat transfer Feedstock flexibility Drying and Devolatilization Biomass Volatiles (H 2 O, CO2, C x H y,ch 4 ) Char Fuel Fluidization agent Bubbles Bubble hydrodynamics Bubble Emulsion -Rise -Growth -Coalescence -Mass transfer Challenges Design Scale-up Optimization 2
Fluidized Bed Biomass Gasifiers (FBBG) and Pyrolyzers (FBBP) Parameter FBG FBP Temperature Range [C] Equivalence Ratio (n air /n fuel ) Fuel Particle Diameter Solids Residence Times Gas Residence Times Fluidization Regime 700-1000 500-800 0.2-0.4 0 1-50 mm <1mm >10 2 s <10 1 s 10 0 s(+) 10 0 s(-) Bubbling - Slugging Turbulent - Entrained 3
Multi-Scale Reactive Modeling APPROACH 4
Multi-scale fluidized bed modeling approaches Lattice Boltzmann Methods (LBM) Lagrangian Eulerian Models (LEM) Eulerian Eulerian model (EEM) Discrete Bubble Model (DBM) Two Phase Theory (TPT) Fluid-particle interaction Particle-particle interaction Particle-particle interaction; bubble behavior Industrial scale, solids motion; bubble behavior Industrial-scale reactive simulations 10-5 m 10-2 m 10-1 m 10-0 m 10 1 m 5
Multi-scale Modeling of FBBG Reactive CFD Particle Scale Modeling Reactor Network Modeling 10-6 -10-2 10-3 -10 0 10-2 -10 1 Modeling Length Scales (m)
Multi-Scale chemistry simulations tackle technical challenges Technical challenges in fluidized bed gasification Current Model Development 1. Predicting major gas species and tar components over range of operating conditions 2. Influence of thermally large particles (>>1mm) on chemical conversion 3. Complex gas-solid mixing and hydrodynamics 1. Development of Reactor Network Model with detailed kinetic model 2. Detailed Particle-Scale Model coupled coupling heat transfer with detailed kinetic model 3. Reactive 3D Cylindrical CFD Simulations of Air-, Steam/Air- and Steam- Blown Experimental FBBGs. 7
Multi-scale Modeling Contributions and Findings Reactive CFD Particle diameter shown to be positively correlated with large PAH formation. Particle Scale Modeling Non-uniformity of gases in bed-zone identified as major driver of tar growth and PAH formation Increased particle Reactor diameter Network shown Modeling to: Improve lateral devolatilization zone uniformity Increase axial devolatilization zone segregation 10-6 -10-2 10-3 -10 0 10-2 -10 1 Modeling Length Scales (m)
Outline of the remainder. Model Development Reactor Network Model Detailed Particle Model Reactive CFD Influence of Particle Diameter on Tar Chemistry Coupling of detailed particle with RNM Influence of Superficial Gas Velocity on Tar Chemistry Coupling Reactive CFD simulation to improve RNM 9
Coupled particle and RNM Mechanistic predictions Influence of Bed Material Catalysis REACTOR MODELING (RNM)
Multi-scale Modeling of FBBG Reactive CFD Particle Scale Modeling Reactor Network Modeling 10-6 -10-2 10-3 -10 0 10-2 -10 1 Modeling Length Scales (m)
Chemical Conversion of Biomass in FBBG Gas-Phase Conversion Pathway Solids Conversion Pathway Solids Gas-Phase -> Gas Reactions: conversion: Complex Multiple species interplay and between pathways Heat Very complex transfer and chemistry Chemistry
Simple Reduced Network Model of FBBG Fluidized bed is well mixed at rate faster than timescale of devolatilization Devolatilization is uniform through bed Gas-phase reactions in emulsion can be modeled as a CSTR (continuously stirred tank reactor) Mechanism: CRECK/Ranzi In freeboard few solids ~460 Species present thus little ~16000 axial mixing Reactions Gas-phase reactions in freeboard can be modeled as a PFR (plug flow reactor) 13
PARTICLE SCALE PYROLYSIS MODEL
Multi-scale Modeling of FBBG Reactive CFD Particle Scale Modeling Reactor Network Modeling 10-6 -10-2 10-3 -10 0 10-2 -10 1 Modeling Length Scales (m)
Devolatilization Model Development 1-D particle model Devolatilization Mechanism Base of model is integration of heat transfer and chemical conversion: 1-D heat equation coupled with reactions Mechanism of CRECK at Poli. Milano Boundary Conditions: Convective and radiative heat transfer at surface from particle trajectory history Can come from CFD! Ranzi et al. 16
REACTOR-SCALE REACTIVE CFD SIMULATIONS
Model, Tools & Validation Two Fluid Model Solid and gas phases described using fully interpenetrating continua Particles are not individually tracked => Computationally efficient Constitutive relations required for particle-scale interactions Conservation of Mass ε g v s v g Apparent density = Volume fraction x real density Conservation of Momentum Solids Stress Tensor particle-particle interactions Drag Model particle-gas interactions 18
Model, Tools & Validation Most validation studies based on comparison of global parameters like pressure drop and bed height Need to develop suitable metrics and compute them efficiently to accurately describe hydrodynamics Original CFD data Bubble Statistics Solids mixing Velocity Distribution Circulation flux Time mean solids holdup 19
Model, Tools & Validation Bubble Statistics CS 2 Simulation 2D Vertical Slice or Cross Section using simulation data Digital Image Analysis Bubble detection using ImageJ Threshold void fraction = 0.7 Bubble statistics using MATLAB Lagrangian Velocimetry Bubble numbering based on lateral and axial positions Velocity using identical numbered bubbles in consecutive frames CS 1 CS 1 For large-scale beds, 2D bubble statistics cannot detect smaller bubbles and their azimuthal trajectories!
INFLUENCE OF DEVOLATILIZATION ON CONVERSION Particle diameter effects Internal and external heat transfer
Multi-scale Modeling of FBBG Reactive CFD Particle Scale Modeling Reactor Network Modeling 10-6 -10-2 10-3 -10 0 10-2 -10 1 Modeling Length Scales (m)
Batch Fluidized Bed Pyrolysis Conditions (NREL 4 FB Reactor) Product Gas Freeboar d H b Biomass Bubbling bed D b Gaston et al (2011) Nitrogen Doctoral Thesis Defense, December, 2014
Gas species diameter dependence vs Gaston 2011 Model Expt. Little particle diameter influence is observed for major gas species in both model and experiment: Fast Gas Phase Reactions approach pseudo-equilibrium Stark et al. 2016 Under Review. 24
Influence of Particle Diameter on PAH growth Stark et al. 2016 Under Review. 25
Sinapoyl Aldehyde Yields are the (potential) link. Stark et al. 2016 Under Review. 26
Implications and Summary of Coupling of Detailed Particle Model with RNM Grinding smaller is effective in decreasing primary tar precursors. Particle model + grinding model + tar clean-up model: will elucidate optimum grinding size for different biomasses. Decreases PAH growth potential. Faster fluidization will decrease larger tar precursors (Synapol Aldehyde, Coumaryl, etc) favoring smaller compounds such as phenol. Increased heat transfer rate is achieved at the particle scale. CFD will inform upper-bounds of superficial gas velocity before increased small-particle entrainment leading to efficiency losses. 27
QUANTIFICATION OF THE INFLUENCE OF BED MIXING ON PAH FORMATION AND GROWTH 28
Multi-scale Modeling of FBBG Reactive CFD Particle Scale Modeling Reactor Network Modeling 10-6 -10-2 10-3 -10 0 10-2 -10 1 Modeling Length Scales (m)
Tar Class predictions We are underpredicting the formation of PAH compounds by orders of magnitude Well Stirred bed possibly overpredicts availability of oxygen everywhere, impeding PAH formation Transport likely playing a role! Stark et al. Energy & Fuels 2015. 30
Driver Tar Conversion pathways Primary tars: Class 2 tars (phenol, cresol etc) stand at a precipice. If oxygen is available, quickly consumed/cracked to smaller gases and target species (CO & H2). With decreased oxygen concentrations, PAH growth is favored. Well-mixed bed => good distribution of oxygen. Poorly-mixed bed => relatively rich zones. Stark et al. Chem. Eng. J. 2016.
Well-stirred Bed Zone is a strong assumption Increasing Gas velocity => Faster bubble growth and faster bubble flow Larger bubbles => Less gas exchange between emulsion and bubble 32
Well-stirred Bed Zone is a strong assumption X N2 X tari R devol Voidage Bubbles carry majority of excess fluidization gas Devolatilization occurs in emulsion Full mixing occurs at splash zone Emulsion relatively rich RNM needs to capture this to predict PAH formation. Stark et al. Chem. Eng. J. 2016. 33
Emulsion (CSTR) Bubble (PFR) Freeboard (PFR) Improved RNM Formulation 1-x x Devol. Gases y Air 1-y Stark et al. Chem. Eng. J. 2016. 34
Gas velocity plays an important role in mixing 2xUmf 7.25xUmf 14.5xUmf 35
Devolatilization Zone dependence Stark et al. Chem. Eng. J. 2016. 36
Tar distribution 37
Emulsion (CSTR) Bubble (PFR) Freeboard (PFR) Calculating parameters from CFD 1-x x Devol. Gases y Air 1-y Time averaged data is used to analyze pseudosteady state operation. Average bed height Average voidage Reaction zones Average voidage is used to calculate zone volumes Gas mass flux weighted voidage is used to calculate mass flows through zones N2 mass flux weighted voidage is used to calculate air flow through each zone. Stark et al. Chem. Eng. J. 2016. 38
Emulsion (CSTR) Bubble (PFR) Freeboard (PFR) Calculating parameters from CFD Global Air:Fuel Ratio 1-x x Devol. Gases y Air 1-y 39
Impact of bubble-phase on gas-phase conversion. Emulsion (CSTR) Freeboard (PFR) Bubble (PFR) 1-x x Devol. Gases y Air 1-y 40 Stark et al. Chem. Eng. J. 2016.
Improving Gas Flow Distribution Total Gas Flow = dense flow + visible bubble flow+ through-flow Bubble swarms offer low-resistance pathway for shortcut of gas => minimal contact with dense phase V g [m/s] 3.00 0.40 D = 50 cm, H 0 = 50 cm 0.2x real time U = 0.72 m/s LLDPE (1.15 mm, 800 kg/m 3 ) Bakshi et al. in Prep. 41
SUMMARY OUTCOMES FUTURE DIRECTIONS 42
Implications and Summary Devolatilization Effects Grinding smaller is effective in decreasing primary tar precursors. Particle model + grinding model + tar cleanup model: will elucidate optimum grinding size for different biomasses. Decreases PAH growth potential. Grinding too small can localize devolatilization In large beds Dbed:Dparticle ratio may be too high resulting in poor devolatilization distribution Gas By-passing Gas-Solids mixing is non-linear with superficial gas velocity At slow flow rates solids mixing may be insufficient With increasing flow-rates increasing bypassing occurs and rich emulsion phase is observed Rich emulsion leads to increased PAH formation O2:HC ratio is crucial. 43
Future Directions Phenomena Influence of long-lived char in bed on chemistry. Intra-particle heat and mass transfer influence on devolatilizaton chemistry Large-scale reactor hydrodynamics and solid mixing on gas-phase mixing. Model Development Incorporation of particle population model into RNM Development of shrinkingcore representations for eulerian CFD, hybrid DEM simulations. Extend advanced RNM to include multiple bubble paths, repeating structures. 44
Contributions https://www.researchgate.net/project/multiscale-modeling-of-biomass-gasification-inbubbling-fluidized-beds Coupling Particle and RNM: Stark, AK & Ghoniem, AF. Biomass Devolatilization at the Particle Scale: The Influence of Particle Diameter on Polycyclic Aromatic Hydrocarbon (PAH) formation in Fluidized Bed Gasification and Pyrolysis Reactors. Under Review, Fuel. RNM development and mechanistic study of conversion pathways: Stark, AK, Bates, RB, Zhao, Z & Ghoniem, AF. Prediction and Validation of Major Gas and Tar Species from a Reactor Network Model of Air-Blown Fluidized Bed Biomass Gasification. Energy and Fuels, 29(4), 2015. Bates, RB, Altantzis, C, Ghoniem, AF. Modeling of Biomass Char Gasification, Combustion and Attrition Kinetics in Fluidized Beds. Energy and Fuels 30(1), 2016. Coupling CFD and RNM: Stark, AK, Altantzis, C, Bates, RB, & Ghoniem, AF. Towards an Advanced Reactor Network Modeling Framework for Fluidized Bed Biomass Gasification: Incorporating Information from Detailed CFD Simulations. Chemical Engineering Journal, 303, 2016. Fundamental CFD Simulations: Bakshi, A, Altantzis, C, Bates, RB, Ghoniem, AF. Study of the Effect of Reactor Scale on Fluidization Hydrodynamics Using Fine-Grid CFD Simulations Based on the Two-Fluid Model. Powder Technology 299, 2016. Bakshi, A, Altantzis, C, Bates, RB, Ghoniem, AF. Multiphase-Flow Statistics using 3D Detection and Tracking Algorithm (MS3DATA): Methodology and Application to Large-Scale Fluidized Beds. Chemical Engineering Journal 293, 2016. Bakshi, A, Altantzis, C, Bates, RB, Ghoniem, AF. Eulerian-Eulerian Simulation of Dense Solid-Gas Cylindrical Fluidized Beds: Impact of Wall Boundary Condition and Drag Model on Fluidization. Powder Technology 277, 2015. Altantzis, C, Bates, RB, Ghoniem, AF. 3D Eulerian Modeling of Thin Rectangular Gas-Solid Fluidized Beds: Estimation of the Specularity Coefficient and Its Effects on Bubbling Dynamics and Circulation Times. Powder Technology 270, 2015. 45