mul$%scale*simulaon*of*fluidized* bed*biomass*gasificaon* · mul$%scale*simulaon*of*fluidized*...
TRANSCRIPT
Mul$-‐scale simula$on of fluidized bed biomass gasifica$on
Ahmed F. Ghoniem Christos Altantzis, Addison K. Stark, Richard B. Bates
Akhilesh Bakshi, Rajesh Sridhar
Department of Mechanical Engineering MassachuseHs Ins$tute of Technology
NETL Workshop of Mul$phase Simula$ons
August 12-‐13, 2015
Project support by BP
Biomass to liquids pathways
• Gasifica$on offers improved feedstock u$liza$on versus exis$ng 1st genera$on routes • Gasifica$on allows produc$on of drop-‐in fuels (e.g. FT Diesel) as well as chemicals
Cellulosic biomass
Syngas
Pyrolysis Bio-oils
Acid Hydrolysis Aqueous products
FT Synthesis
Catalysis
Catalysis
Fermenta$on Sugar monomers
Alkanes
Liquid fuels
Fuels and chemicals
Ethanol
Thermochemical routes
Biochemical route
Fluidized bed gasi6ication of biomass
Raw biomass characteris/cs • High moisture content • Expensive par$cle size reduc$on
Entrained flow gasifier: Very dilute par$cle-‐gas mul$phase flow • High opera$ng temperatures • High levels of carbon conversion with low tar content • Requires very fine par$cles (<mm)
Not suitable for raw biomass, unless it is pre-‐torrefied, see Bates and Ghoniem, Bioresource Tech
3
Fluidized bed gasi6ication of biomass
Raw biomass characteris/cs • High moisture content • Expensive par$cle size reduc$on
Fluidized bed: Widely used paradigm for dense par$cle-‐gas mul$phase flows • High surface area contact between fluid and solid per unit bed volume • High levels of intermixing • Suitable for coarse par$cles with large residence $mes • Lower levels of carbon conversion with considerable tar content
Appropriate for biomass gasifica$on, poses opera$onal and modeling challenges
4
Multi-‐scale nature of biomass gasi6ication
Mul$-‐scale process: • Gas-‐phase chemistry • Surface chemistry • Single-‐par$cle modeling • Hydrodynamics • Mixing -‐ Segrega$on
Fluidized bed biomass gasi6ication Modeling
Challenges • Complex interplay between:
• Mul$phase fluid dynamics • Massive chemical reac$on networks
• Exis$ng chemical kine$c mechanisms are too computa$onally demanding
• Ranzi et al. [2] 460 species, 16000 reac$ons • Reliable chemistry reduc$on strategies are needed
• Simula$ng industrial scale reactors requires the development of advanced closure sub-‐models for the physics descrip$on
Typical structure of sodwood lignin (Faravelli et al., 2010 [2])
Multiscale simulations & technical challenges
2. Undesirable tar compounds
Technical challenges in fluidized bed gasifica/on
Demonstra/on scale d=2.5m
Commercial-‐scale d=5m
Lab scale d=0.25m
1. Expensive scale-‐up
3. Complex gas-‐solid mixing and hydrodynamics
Multiscale-‐multi physics simulations
2. Undesirable tar compounds
Technical challenges in fluidized bed gasifica/on
MIT’s modeling work 1. Bakshi et al. Efficient computa$onal schemes for
cylindrical FB reactors with CFD, Powder Tech. 2014. [3]
2. Altantzis et al., Sensi$vity of solids circula$on to walls, par$cle, flow characteris$cs using CFD. , Powder Tech., 2015 [4]
3. Bakshi, et al., Eulerian-‐Eulerian simula$on of dense solid-‐gas cylindrical fluidized beds; wall boundary condi$on and its impact on fluidiza$on. Powder Tech., 277 (2015) 47-‐62 [5]
4. Stark etal., Predic$on and valida$on of syngas and tar species from a reactor network model of fluidized bed biomass gasifica$on, Energy & Fuels, 2015 [6]
5. Bates et al, Char combus$on, gasifica$on and aHri$on in bubbling fluidized beds. Energy & Fuels 2015 [7]
6. Stark, A., Comprehensive par$cle and homogeneous chemistry modeling, PhD Thesis 2015 [8]
7. Sridhar, Reactor Network Modeling of biomass gasifica$on with detailed chemistry, Master Thesis 2015. 8
1. Expensive scale-‐up
3. Complex gas-‐solid mixing and hydrodynamics
CFD
Reactor network modeling Par/cle scale modeling
Computa/onal efficiency and scalability
Valida/on and metrics
Modeling length scales (m)
Internal + External Mass Transfer
Primary Pyrolysis Reaction zone
Gas Phase Secondary Pyrolysis Reactions, Tar Reformation and Oxidation Reactions
Heterogeneous Reactions
Heat Transfer (conduction, convection, radiation)
Abrasion causes attrition
Devola/liza/on
10-‐6-‐10-‐2 10-‐3-‐100 10-‐2-‐101
Network model Comprehensive
chemical kine/cs
Gasifica/on
Towards Multi-‐scale simulation of biomass gasi6ication
Approach Gas Phase Solid Phase Scale
Discrete Bubble Model
Lagrangian Eulerian Industrial (10 m)
Two Fluid Model
Eulerian Eulerian Engineering (1 m)
Discrete Element Model
Eulerian (unresolved)
Lagrangian Laboratory (0.1 m)
Discrete Particle Model
Eulerian (resolved)
Lagrangian Laboratory (0.1 m)
Molecular Dynamics
Lagrangian Lagrangian Mesoscopic (<0.001 m)
10
Two-Fluid Model (TFM) balances computational efficiency with
modeling fidelity making it suitable to industrial scale but challenges remain!
Accuracy
Compu
ta/o
nal efficien
cy
Computational approaches to 6luidization
Two Fluid Model
• Par$cle-‐Par$cle interac$ons – Kine$c theory of granular flows – Inter-‐par$cle drag law
• Segrega$on slope coefficient, Cs • Fric$on coefficient, Cf
• Par$cle-‐gas interac$ons – Drag laws
• Par$cle-‐wall interac$ons – Res$tu$on coefficient, e – Specularity coefficient, ϕ
Closure models involve parameters whose influence on hydrodynamics need inves$ga$on by means of: • Detailed experiments • Valida$on strategies • Use of detailed numerical methodologies in smaller
scales
• Ascending bubbles drive solids motion
• Bubbles act as
pathways for gas flow
Importance of bubble physics during 6luidization
Gas phase mo/on Solid phase mo/on
13
Bubble statistics Dense phase statistics
Time mean solids holdup
Size vs height Velocity vs Height
Original CFD data f(x,y,z,t)
Circulation flux
Metrics for 6luidization hydrodynamics
0.0
0.3
0.6
3D bubble statistics tool Bakshi at.al,
Bubble geometric characteristics • Surface area • Volume • Cord length (axial/radial) • Aspect ratios
Statistical analysis of bubble size and velocity distributions
3D bubble analysis for: • Accurate detection of the
number of bubbles in the bed (only a fraction is captured with 2D analysis)
• Azimuthal motion of bubbles
Dense phase mixing metric
• Circulation fluxes & circulation times are sensitive to the parameters in sub-models • Useful quantities for rigorous quantitative validation with novel experimental
measurements employing particle velocimetry • Strongly coupled with the bubbling behavior in the bed
16
100
cm
50 cm
Lab-Scale Model Delgado et al. 2013 [9]
• Lack of experimental data on φw => φw is a fitting parameter
φw tuned to 2D simulations is not appropriate for 3D simulations
• Thin rectangular beds extensively used in experimental studies employing non-intrusive measurements techniques
• Cylindrical beds more realistic geometries for scale-up and different hydrodynamics compared to pseudo-2D beds
• Variation of specularity coefficient to evaluate impact on simulation of pilot-scale model
100
cm
14.5 cm
Lab-Scale Model Rudisuli et al. , 2012 [11]
100
cm
10 cm
Lab-Scale Model Verma et al. , 2014 [10]
Parameter estimation and validation strategy Example: specularity coef6icient
100cm
15cm
100cm
30cm
Rudisuli et al. , 2012 (Op$cal probe measurements
for bubble dynamics)
Pilot scale reactor
100cm
50cm Delgado et al. 2013 [9] (Digital Image Analysis for solids mo$on)
Parametric analyses and valida$on • Parametric analyses with respect to the specularity coefficient • Valida$on based on experimental studies for sand-‐like par$cles in different bed geometries
• Thin rectangular beds are extensively used
in experimental studies employing non-‐intrusive measurements techniques
• Significant influence of the wall presence especially in the spanwise direc$on
• NOT suitable for comparison with 2D simula$ons
100cm
50cm Delgado et al. 2013
(Digital Image Analysis for solids mo$on)
Parametric analyses and valida$on • Parametric analyses with respect to the specularity coefficient • Valida$on based on experimental studies for sand-‐like par$cles in different bed geometries
19
Decreasing φ results in: • Larger bubble sizes
• Slugging fluidization
• Gas bypassing effects
• Higher bubble velocities
Uin = 2.5Umf φ = 0.0005
Negligible hindrance
Uin = 2.5Umf φ = 0.5
Significant hindrance
Simulation Results
Bubble hydrodynamics and solids motion significantly influenced by wall boundary condition
Thin rectangular bed
1×10-4 1×10-3 1×10-2 1×10-1 1×100
Specularity coefficient
0
2
4
6
8
Tim
e [s
]
tc+
tc-
tc
1×10-4 1×10-3 1×10-2 1×10-1 1×100
Specularity coefficient
0
2
4
6
8
Tim
e [s
]
tc+
tc-
tc
Case A Case D
Φ = 0.05 Φ = 0.4
Experiment Experiment
Circ
ulat
ion
time
[s]
Circ
ulat
ion
time
[s]
Uin=2.5Umf Uin=1.75Umf
20
• Circulation time ↑ when
• Uin ↓ - closer to Umf and less bubbling
• φ ↑ - wall resistance increases ⇒ less solid motion close to walls and smaller bubbles
• Appropriate φ ↓ as Uin ↑. For the range, 1.5Umf – 2.5Umf, Φ ∈ [0.05,0.5]
Choosing Φ
Thin rectangular bed
100cm
15cm
100cm
30cm
Rudisuli et al. , 2012 [11] (Op$cal probe measurements
for bubble dynamics)
Pilot scale reactor
100cm
50cm Delgado et al. 2013
(Digital Image Analysis for solids mo$on)
Parametric analyses and valida$on • Parametric analyses with respect to φ and the drag model • Valida$on based on experimental studies for sand-‐like par$cles in different bed geometries
• Realis$c geometries of gasifiers • Effect of boundary condi$ons for
different bed sizes
• Scaling-‐up simula$on towards industrially relevant sizes
22
φ = 0.01
Mechanism
0.2
0.4
0.6
0.8
Influence of specularity coefficient in a 3D cylindrical bed
solids distribu/on (circula/on flux / /me) bubble dynamics (Bubble dia/count)
Glass 2500 kg/m3, 1.1 mm Bed Dia = 10 cm U=2.0 Umf
Alumina 1040 kg/m3, 1.0 mm Bed Dia = 10 cm U=3.0 Umf
Suitable ϕ ?
• Range – 1.25-6.80 Umf, 860-2500 kg/m3, 0.289-1.1 mm • Bubble diameter comparisons show higher values of ϕ (0.01-0.3) more suitable for dense flows • Low sensitivity of metrics (solids / bubbles) for suitable ϕ • Variable ϕ model (Li&Benyahia [12]) matches the results of the suitable ϕ
Valida$on
2.3 Umf 6.8 Umf • Gidaspow model = pressure
drop data from packed + homogeneous fluidization experiments
• Syamlal-O’Brien model = terminal velocity correlation of particles from liquid-solid beds
Takeaways - 1. Drag model has a strong
impact on fluidization
2. Gidaspow model more suited for low velocities (2-4 Umf)
Effect on Pilot-‐Scale Bed
Influence of drag model
CFD
Reactor network modeling Par/cle scale modeling
Computa/onal efficiency and scalability
Valida/on and metrics
Modeling length scales (m)
Internal + External Mass Transfer
Primary Pyrolysis Reaction zone
Gas Phase Secondary Pyrolysis Reactions, Tar Reformation and Oxidation Reactions
Heterogeneous Reactions
Heat Transfer (conduction, convection, radiation)
Abrasion causes attrition
Devola/liza/on
10-‐6-‐10-‐2 10-‐3-‐100 10-‐2-‐101
Network model
Towards Multi-‐scale simulation of biomass gasi6ication
Comprehensive chemical kine/cs
Gasifica/on
Thermochemical Conversion of Biomass in FBBG
Gas-‐Phase Conversion Pathway
Solids Conversion Pathway
Solids -‐> Gas conversion: Complex interplay between Heat transfer and Chemistry
Gas-‐Phase Reac$ons: Mul$ple species and pathways
High char conversion crucial to op$mal efficiency and opera$on
Approach: Integrated par$cle/reactor network models to predict gasifier outputs
Reactor network model (Homogeneous reac=ons)
Char conversion par$cle model (Heterogeneous reac=ons)
Gasifica$on, aHri$on rates
Gas-‐phase concentra$ons Iterate un$l gas phase composi$on converges
Internal + External Mass Transfer
Primary Pyrolysis Reaction zone
Gas Phase Secondary Pyrolysis Reactions, Tar Reformation and Oxidation Reactions
Heterogeneous Reactions
Heat Transfer (conduction, convection, radiation)
Abrasion causes attrition
Devola$liza$on par$cle-‐scale model Inputs: Reactor temperature, pressure, feed par$cle size composi$on
C+H2O = CO+H2 C+CO2 = 2CO C+0.5O2= CO
Transport limita/ons during par/cle-‐Scale devola/liza/on
28
} Par$cle-‐phenomena directly influence overall conversion
} Interplay between } External convec$ve heat transfer
from the bed to the par$cle } Internal conduc$ve heat transfer } Primary pyrolysis reac$on kine$cs
Devolatilization Model Development
1-‐D par/cle model Devola/liza/on Mechanism
29
• Base of model is integration of heat transfer and chemical conversion:
– 1-D heat equation coupled with reactions
– Mechanism of Ranzi et al. [1]
• Boundary Conditions: – Convective and radiative heat transfer at
surface from particle trajectory history • Can come from CFD!
• System of equations are integrated with respect to time.
– Requires a stiff ODE integrator. – Simulation cost: 100 cpu-seconds
per second.
Ranzi et al. [1]
Single Particle Model Predictions
Resolves internal gradients and reac/on dynamics
Cellulose, Hemicellulose and Lignin devola$lize sequen$ally, yielding a triple pyrolysis wave
30
Single Particle Model Predictions
Predic/on of Devola/liza/on Species Rate of produc$on of tar precursors are of par$cular interest. Physical parameters have a strong effect. 20 gaseous species and char
31
Par/cle devola/liza/on /me versus par/cle size
0.005 0.01 0.015 0.02 0.025 0.030
50
100
150
200
250
300
Particle Diameter [m]
Con
vers
ion
Tim
e [s
]
500C
900C
“Tar+water” yields versus reactor temperature
Par$cle scale devola$liza$on model valida$on
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
500 600 700 800 900
Tar yield (kg/kg)
Reactor temperature ○C
Exp. 6mm
Lines: Best-‐fit by Gaston et al. 2011 [13] Points: Model predic$ons
32
Reactor network model (Homogeneous reac=ons)
Char conversion par$cle model (Heterogeneous reac=ons)
Gasifica$on, aHri$on rates
Gas-‐phase concentra$ons Iterate un$l gas phase composi$on converges
Internal + External Mass Transfer
Primary Pyrolysis Reaction zone
Gas Phase Secondary Pyrolysis Reactions, Tar Reformation and Oxidation Reactions
Heterogeneous Reactions
Heat Transfer (conduction, convection, radiation)
Abrasion causes attrition
Devola$liza$on par$cle-‐scale model Inputs: Reactor temperature, pressure, feed par$cle size composi$on
Approach: Integrated par$cle/reactor network models to predict gasifier outputs
C+H2O = CO+H2 C+CO2 = 2CO C+0.5O2= CO
Role of char conversion during FBG
• Gomez-‐Barea & Leckner 2013 [14]: • “The conversion of char is the most
decisive factor in FBG, because the main loss of efficiency is due to unconverted carbon in the ashes.”
• “...results show that value assigned for char conversion has a major effect on the temperature, gas hea=ng value, and therefore on other parameters like gas composi=on and cold gas efficiency”
• LHV ranges 5.4-‐7.75 MJ/Nm3 depending on assigned char conversion.
• Char conversion needs to be predicted accurately to reduce uncertainty in major output variables Calcula$ons for air-‐blown pilot scale wood pellet
FBG by Gomez-‐Barea & Leckner 2013 [14]:
Air/fuel equivalence ra$o
Char Particle Conversion Model
Devola$liza$on + primary fragmenta$on+
shrinkage
AHri$on pr
oduces
elutriable fi
nes
CO+H2
2CO
CO2
+H2O +CO2 +O2
Assump/on Jus/fica/on
Fines produced by aHri$on are assumed to be elutriated as soon as they are produced
Elutria$on $me scale is fast compared to aHri$on $me scale
Intrapar$cle diffusion limita$ons negligible during gasifica$on , dominant during combus$on
Thiele modulus <<1 for gasifica$on Thiele modulus >> 1 during combus$on
Isothermal par$cle Bi<1 35
Modi6ied attrition kinetics rate expression
Exis$ng models do not account for the reduc$on in density/hardness during conversion to affect the aHri$on rate
Current model aHempts to accounts for this
Fi\ed model results: Spruce wood pellet char gasifica$on in 60%vol CO2 800C
• Proposed structural model able to accurately fit results for aHri$on rate and overall conversion.
• ~25%wt of ini$al char is aHrited over the course of 180 minutes
0.E+00 1.E-06 2.E-06 3.E-06 4.E-06 5.E-06 6.E-06 7.E-06 8.E-06
0 50 100 150
Elut
riatio
n ra
te (k
g/m
in)
Time (minutes)
Expt. (Ammendola et al., 2013) Model
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
1
0 50 100 150
Car
bon
Con
vers
ion
(Xg)
Time (min)
Model Experiment (wood pellet)
“Shrinking core model” for char combustion; Internal and external diffusion limitations
Valida$on of combus$on/aHri$on par$cle model
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
0 5 10
Carbon
Con
version (X)
Time (min)
Ammendola wood chip Shrinking core model
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
0 10 20 30
Carbon
Con
version (X)
Time (min)
Ammendola 2013 spruce pellet Shrinking core model
Predic/ons are sensi/ve to: -‐Par$cle geometry/aspect ra$o -‐External mass transfer coefficient -‐Char par$cle density and diffusivity
Condi/ons 4.5% Oxygen at 800C 95.5% Nitrogen
Char gasifica$on/combus$on par$cle model
• Gasifica$on assisted aHri$on well represented by new model – Sensi$ve to assumed gasifica$on kine$cs
• Combus$on for different feedstocks is well represented by effec$veness factor model – Rate is sensi$ve to par$cle shape and size – Sensi$ve to external mass transfer coefficient model
• Can be incorporated in a RNM. • In CFD it will be part of a Lagrangian descrip$on
Reactor network model (Homogeneous reac=ons)
Char conversion par$cle model (Heterogeneous reac=ons)
Gasifica$on, aHri$on rates
Gas-‐phase concentra$ons Iterate un$l gas phase composi$on converges
Internal + External Mass Transfer
Primary Pyrolysis Reaction zone
Gas Phase Secondary Pyrolysis Reactions, Tar Reformation and Oxidation Reactions
Heterogeneous Reactions
Heat Transfer (conduction, convection, radiation)
Abrasion causes attrition
Devola$liza$on par$cle-‐scale model Inputs: Reactor temperature, pressure, feed par$cle size composi$on
Approach: Integrated par$cle/reactor network models to predict gasifier outputs
C+H2O = CO+H2 C+CO2 = 2CO C+0.5O2= CO
Importance of tar yields during biomass gasifica$on
Tar classifica$on based on, 1. Water solubility 2. Condensa$on temperatures
Downstream problems: 1. Condensa$on at low
temperatures might result in fouling or clogging of the gas pipelines
2. High solubility results in toxic wastewater which would require expensive disposal systems downstream
[16] van Paasen and Kiel, 2004
Reactor network model of fluidized biomass gasifier
• Fluidized bed is well mixed at rate faster than $mescale of devola$liza$on
• Devola$liza$on is uniform through bed
• Gas-‐phase reac$ons in emulsion can be modeled as a CSTR (con$nuously s$rred tank reactor)
• In freeboard few solids present thus liHle axial mixing
• Gas-‐phase reac$ons in freeboard can be modeled as a PFR (plug flow reactor)
43
Reactor network model of fluidized biomass gasifier
Mechanism: CRECK/Ranzi ~460 Species ~16000 Reac$ons
44
CRECK mechanism for primary pyrolysis and secondary gas phase
reac$ons
• 18 Reac$ons
• 33 Species • Diffusion Modeling
Biomass • 16,000 Reac$ons
• 460 Species • Homogeneous Gas-‐Phase
Primary Pyrolysis Products
Secondary Pyrolysis and Gasifica$on Products
45
Levoglucosan
Phenol
Coumaryl Alcohol
C=O
H2
Benzene
Toluene
Naphthalene
Experimental data available for valida$on: Air-‐blown bubbling fluidized bed gasifiers.
(Van Paasen and Kiel, 2004 [16]) • Biomass feed rate: 1kg/hr • Biomass feedstock: Beech • Air-‐Fuel Equivalence Ra$o: 0.25 • Fluidizing medium: Sand, 270 μm micron • Bed diameter= 7.4 cm
(Narvaez et al., 1996 [17]) • Biomass feed rate: 0.5kg/hr • Biomass feedstock: Pine sawdust • Air-‐Fuel Equivalence Ra$o: 0.25-‐0.4 • Fluidizing medium: Sand, 360 μm • Bed diameter=6 cm
(Kurkela & Stahlberg, 1992 [18]) • Biomass feed rate: 40kg/hr • Biomass feedstock: Pine sawdust • Air and Steam employed • Air-‐Fuel Equivalence Ra$o: 0.25-‐0.4 • Fluidizing medium: Sand, 600 μm • Bed diameter=15 cm • At pressure (4 bar) • Secondary Air injec$on.
46
Freeboard
Bubbling bed
Product Gas
Hk
Biomass
Air/Steam/O2
Hb
Db
Air/Steam/O2
RNM valida$on: Effects of temperature on major species, effects of CO and Water Gas
Shid Kine$cs.
Comparison of full mechanism with experimental data
• Full mechanism and reactor network model able to accurately represent the destruc$on of tars at higher temperatures
Tar Class predic$ons
49
We are under-‐predic$ng the
forma$on of PAH compounds… by orders
of magnitude
Well S$rred bed possibly over-‐predicts availability of oxygen everywhere, impeding
PAH forma$on… Transport likely playing
a role!
Well-‐s$rred Bed Zone is a strong assump$on
50
Increasing Gas velocity => Faster bubble growth and faster bubble flow Larger bubbles => Less gas exchange between emulsion and bubble
51
Well-‐s$rred Bed Zone is a strong assump$on
XN2 Xtari Rdevol Voidage • Bubbles carry majority
of excess fluidiza$on gas
• Devola$liza$on occurs in emulsion
• Full mixing occurs at splash zone
• Emulsion rela$vely rich…
RNM needs to capture this to predict PAH forma$on.
Improved RNM Formula$on
Improved RNM Formula$on
Devol. Gases
Air
Freebo
ard
(PFR)
Bubb
le
(PFR)
Emulsio
n (C
STR)
x 1-‐x
y 1-‐y
Devol. Gases
Air
Freebo
ard
(PFR)
Bubb
le
(PFR)
Emulsio
n (C
STR)
x 1-‐x
y 1-‐y
Impact of bubble-‐phase on gas-‐phase conversion
Impact of bubble-‐phase on gas-‐phase conversion
Devol. Gases
Air
Freebo
ard
(PFR)
Bubb
le
(PFR)
Emulsio
n (C
STR)
x 1-‐x
y 1-‐y
CFD
Reactor network modeling Par/cle scale modeling
Computa/onal efficiency and scalability
Valida/on and metrics
Modeling length scales (m)
Internal + External Mass Transfer
Primary Pyrolysis Reaction zone
Gas Phase Secondary Pyrolysis Reactions, Tar Reformation and Oxidation Reactions
Heterogeneous Reactions
Heat Transfer (conduction, convection, radiation)
Abrasion causes attrition
Devola/liza/on
10-‐6-‐10-‐2 10-‐3-‐100 10-‐2-‐101
Network model
Current work: Coupling CFD, particle thermochemistry and RNM
Comprehensive chemical kine/cs
Gasifica/on
Bubble phase and emulsion proper$es
Concluding remarks • Mul$-‐scale simula$on necessary to tackle the complexity of fluidized bed
biomass gasifica$on
• Eulerian Eulerian approach for fluidiza$on offers computa$onal scalability but requires closures and valida$on – Contribu$ons in appropriate metrics, robust parametriza$on, computa$onal
efficiency
• Par$cle scale modeling: – FBG dominated by par$cle scale processes of devola$liza$on, gasifica$on and
combus$on – Dominant physical/chemical processes must be iden$fied
• Impossible to model “all” processes
• Reactor network modeling – Flexible, incorporates comprehensive chemistry and par$cle models – Can benefit from CFD results (gas distribu$on, exchange … )
References [1] E. Ranzi, M. CorbeHa, F. Manen$, S. Pierucci, “Kine$c modeling of the thermal degrada$on and combus$on of biomass”, Chemical Engineering Science, 2014, 110:2-‐12 [2] T. Faravelli, A. Frassolda$, G. Migliavacca, E. Ranzi, “Detailed kine$c modeling of the thermal degrada$on of lignins” Biomass and Bioenergy, 2010,34(3):290–301 [3] A. Bakshi, C. Altantzis, A.F. Ghoniem, “Towards accurate three-‐dimensional simula$on of dense mul$-‐phase flows using cylindrical coordinates”, 2014, Powder Technology, 264:242-‐255 [4] C. Altantzis, R.B. Bates, A.F. Ghoniem, “3D Eulerian modeling of thin rectangular gas–solid fluidized beds: es$ma$on of the specularity coefficient and its effects on bubbling dynamics and circula$on $mes”, Powder Technology, 2015, 270(A):256-‐270 [5] A. Bakshi, C. Altantzis, R.B. Bates, A.F. Ghoniem, “Eulerian–Eulerian simula$on of dense solid–gas cylindrical fluidized beds: Impact of wall boundary condi$on and drag model on fluidiza$on”, Powder Technology, 2015, 277:47-‐62 [6] A.K. Stark, R.B. Bates, Z. Zhao, A.F. Ghoniem, “Predic$on and Valida$on of Major Gas and Tar Species from a Reactor Network Model of Air-‐Blown Fluidized Bed Biomass Gasifica$on”, Energy & Fuels, 2015, 29(4):2437-‐2452
References [7] R.B. Bates, C. Altantzis, A.F. Ghoniem, “Modeling of biomass char gasifica$on, combus$on and aHri$on kine$cs in fluidized beds”, Energy & Fuels, 2015, submiHed for publica$on [8] Addison Stark, “Mul$-‐Scale Chemistry Modeling of the Thermochemical Conversion of Biomass in a Fluidized Bed Gasifier”, MassachuseHs Ins$tute of Technology, PhD Thesis, 2015 [9] S. Sánchez-‐Delgado, C. Marugán-‐Cruz, A. Soria-‐Verdugo, D. Santana, “Es$ma$on ans experimental validaiton of the circula$on $me in a 2D gas-‐solid fluidized beds”, Powder Technology, 2013, 235:669-‐676 [10] V. Verma, J. T. Padding, N.G. Deen, J.A.M. (Hans) Kuipers, F. Barthel, M. Bieberle, M. Wagner, U. Hampel, “Bubble dynamics in a 3-‐d gas–solid fluidized bed using ultrafast electron beam X-‐ray tomography and two-‐fluid model”, AIChE Journal, 2014, 60(5):1632-‐1644 [11] M. Rüdisüli, T.J. Schildhauer, S.M.A. Biollaz, A. Wokaun, J.R. van Ommen, “Comparison of bubble growth obtained from pressure fluctua$on measurements to op$cal probing and literature correla$ons”, Chemical Engineering Science, 2012, 74(0):266-‐275 [12] T. Li, S. Benyahia, “Revisi$ng Johnson and Jackson boundary condi$ons for granular flows”, AIChE Journal, 2012, 58(7):2058-‐2068
References [13] K.R. Gaston, M.W. Jarvis, P. Pepiot , K.M. Smith, W.J. Jr. Frederick, M.R. Nimlos, “Biomass Pyrolysis and Gasifica$on of Varying Par$cle Sizes in a Fluidized-‐Bed Reactor”, Energy & Fuels, 2011, 25:3747–3757 [14] A. Gómez-‐Barea, B. Leckner, “Es$ma$on of gas composi$on and char conversion in a fluidized bed biomass gasifier”, Fuel, 2013,107:419–431 [15] P. Ammendola, R. Chirone, G. Ruoppolo, F. Scala, “The effect of pelle$za$on on the aHri$on of wood under fluidized bed combus$on and gasifica$on condi$ons”, 2013, Proceedings of the Combus=on Ins=tute, 34(2):2735-‐2740 [16] S.V.B. van Paasen, J.H.A. Kiel, “Tar forma$on in a fluidized-‐bed gasifier”, Report ECN-‐C-‐-‐04-‐013, Energy and Research Centre of the Netherlands (ECN), PeHen, The Netherlands, 2004 [17] I. Narváez, A. Orı́o, M.P. Aznar, J. Corella, “Biomass gasifica$on with air in an atmospheric bubbling fluidized bed. Effect of six opera$onal variables on the quality of produced raw gas”, Industrial and Engineering Chemistry Research, 1996, 35:2110–2120 [18] E. Kurkela, P. Ståhlberg, “Air Gasifica$on of peat, wood and brown coal in a pressurized fluidized bed reactor. I. Carbon conversion, gas yields and tar forma$on”, Fuel Processing Technology, 1992, 31:1–21