Dissertation Defense - Final

University of Maryland, College Park
Modeling of Solar Particle Receivers for
Hydrogen Production and Thermochemical
Energy Storage
Andrew S. Oles
December 11th, 2014
Committee: Professor Greg Jackson, Chair
Professor Ken Kiger
Professor Amir Riaz
Professor Peter Sunderland
Professor Michael Zachariah

How Concentrating Solar Works
Electricity
Heliostat
Field
Solar
Receiver
Storage Generation
Hot Storage
Cold
Storage
• Central receiver designs
− High outlet temperatures for efficient
power cycles or chemical processes
− Amenable to high solar concentrations
for cost effective
2

• Concentrating solar power require new receiver and storage
technologies to meet DOE targets for cost of solar-thermal electricity
(SunShot Initiative)
• Solid particle receivers have potential as next-generation design
– Outlet temperatures > 600 °C for higher-efficiency power-cycles or high-
temperature chemistry (like H2O splitting for renewable H2)
Motivation
3

Falling Particle Receivers
• Low-stress on solid materials for high
temperature solar absorption
– Low-cost construction
– Extended material life
• Potential for effective energy storage
– High heat capacity
– Stable materials for high-T storage
• Potential as a reactor
– High temperature redox chemistry
– Potential for fuel, chemicals, or even
metals production
Conc. Solar
Radiation
Cold Particle
Flow In
Hot Particle
Flow Out 4

Thermochemical Fuel Production
• Oxide reduction can be used for thermochemical energy storage or
fuel productions
• Ceria is a common material studied for solar fuel production (Kodama
et al., Haile et al., Steinfeld et al., Abanades et al., Davidson et al.)
Particle
Receiver
5

• Background
• Inert Particle Receiver Simulations
– Model description
– Prototype-scale results
– Commercial-scale results
• Reactive Particle Receiver Simulations
– Reactive particle modeling
– Ceria particle results
– Perovskite particle results
• Reactive Particle Receiver CFD Simulations
– Comparison of simplified and CFD model
Outline
• Background
6

• Inert particle receivers can achieve most SunShot performance
requirements with proper design
– Integrated storage with high-Cp particles
– Low-cost materials stable in air over large temperature range
– Work with next-gen (supercritical Rankine) power cycles with firing
temperatures above 650 ºC
• Challenges in inert particle receiver design
– Difficult to design with complex interaction of radiation-driven heat
transfer and multi-phase particulate flow
– Tradeoffs between receiver “solar-absorption” efficiency ηsolar and
particle outlet temperatures Tp,out needed for high-efficiency power
cycles or high-temperature chemistry.
Inert Particle Receivers
7

Gas and Particle Dynamics Model
Side View of
Receiver
Sheath Gas
Particle
Curtain
Non-
participant
gas
• Particle momentum solved in Lagrangian
frame
• Solid-gas mass and momentum coupling
• Air entrainment adapted from semi-
empirical approach of Liu[1]
– Gaussian gas-phase velocity profile, uy,g
– Entrainment proportional to mean uy,g
• Empirical particle spreading of curtain
thickness (Δzcurt) based on Kim et al.[2]
[1]: Liu, Z. (2003). University of Wollongong Thesis Collections.
[2] Kim, K., et al. (2009). Sol. Energy. 83, 1784-1793.
   g
ρ
ρρ
d
uu
CC
ρ
ρ
dt
du
p
gp
p
2
gy,py,
SD
p
gpy,
4
3 



uz,g,entrained
=auy,g
8

Heat-Transfer Model
• Particle curtain transport adapted from the
approach of Röger et al.[3]
– Particle temperatures and energy balance
solved on Eulerian grid
– Gas-particle heat transfer modeled with
Ranz-Marshall correlation:
– Improved internal curtain heat-exchange
derived between 2 semi-transparent surfaces
[3] Röger, M. et al. (2011). J. of Sol. Energy Eng., 133.
ṁp
hp(Tin,f)
ṁp
hp(Tin,b)
ṁp,f
hp(Tf)
ṁp,b
hp(Tb)
frad,Q
fconv,Q
fsol,Q
bsol,Q
curtQ
brad,Q
bconv,Q
  iiiλiλ
M
m iλiλ
iλiλ
curt yxfTfTσ
ρρ
εε
Q ΔΔ∑
-1
,
4
i',
4
i'
1 ',,
',,
mm
mm
mm



  curtconvsolradp,inoutp, QQQQhhmp
 
3/12/1
PrRe6.02 gp
p
pp
k
dh
Nu 


9

Radiation Transport Model
• Radiation balance solved via surface-to-surface radiation method
– Hottel’s zonal method[3] is employed for semi-transparent cells with view
factors calculated from Gaussian Integration
– Curtain transmittance τrad depends on particle
diameter dp and volume fraction fv: 







 curt
p
v
rad z
d
f
τ Δ
2
3
exp
𝜹 𝒌𝒋 𝒒 𝒐𝒖𝒕,𝝀 𝒎,𝒊
′′
= 𝝆 𝝀 𝒎,𝒊 𝒒𝒊𝒏𝒄,𝝀 𝒎,𝒊
′′
+ 𝝉 𝝀 𝒎,𝒊 𝒒𝒊𝒏𝒄,𝝀 𝒎,𝒊′
′′
+ 𝜺 𝝀 𝒎,𝒊 𝒇 𝝀 𝒎,𝒊 𝝈𝑻𝒊
𝟒
+ 𝒒 𝒔𝒐𝒍𝑹𝒆𝒇𝒍,𝝀 𝒎,𝒊
′′
𝑸 𝒓𝒂𝒅,𝒊 = 𝑨 𝒇
𝒎=𝟏
𝑴
𝜺 𝝀 𝒎,𝒊 𝒒𝒊𝒏𝒄,𝝀 𝒎,𝒊
′′
− 𝜺 𝝀 𝒎,𝒊 𝒇 𝝀 𝒎,𝒊 𝝈𝑻𝒊
𝟒
10

Ly,r
Ly,a
x
y
z
Prototype-Scale Receiver Model Parameters
Geometry Lx (m) Ly (m) Lz (m)
Receiver – r 1.85 5.00 1.50
Aperture – a 1.00 3.00 -
Curtain – c 1.00 5.00 Δzcurt
Property Units Baseline Range
dp μm 280 [100, 700]
ṁ’p kg s-1m-1 2.0 [1.0, 4.0]
εp
[4] - 0.85 [0.1-1.0]
Tp,in K 600 [300, 1100]
𝒒 𝑺𝒐𝒍𝒂𝒓
′′ kW m-2
1000 [100, 1500]
ρp
[4] kg m-3 3560 -
Cp,p
[4] J kg-1K-1 264+2.07T-1.12e-3T2
[4] Siegel, N., et al. (2010). J. of Sol. Energy Eng., 132.
λ range (μm) εwall,λ
[4]
0-4.5 0.20
4.5-∞ 0.80
11

ṁ'p = 4.0 kg s-1m-1, Tp,in = 600 K, 𝒒 𝑺𝒐𝒍𝒂𝒓
′′
= 1000 kW m-2, dpart = 600 μm
Left Wall Front Wall Right Wall
Bottom WallTop Wall Rear Wall
Curtain Front Curtain Rear
′′
= 1000 kW m-2, dpart = 100 μm
′′
= 1000 kW m-2, dpart = 280 μm
Prototype Receiver Wall and Curtain Temperatures
Wall Temperatures Particle Temperatures
12

• Smaller dp decreases curtain τrad
– Lower velocity due to greater
drag increases fv
• For smaller dp where τrad < 0.25,
ηsolar remains constant at ~84%
Impact of dp on Receiver Performance
0.65
0.70
0.75
0.80
0.85
440
460
480
500
520
540
100 200 300 400 500 600 700
ηsolar
ΔTp(K)
dp (μm)
dp (μm)
13

Directly Irradiated
Zone
Directly Irradiated
Zone
ṁ'p
(kg s-1m-1)
0.0
0.2
0.4
0.6
0.8
1.0
500
700
900
1100
1300
1500
0 10 20 30 40
ηSolar
OutletTp(K)
ṁ'p (kg s-1m-1)
Particle, rear
Particle, front
Efficiency
ṁ'p
(kg s-1m-1)
• Increasing ṁ'p transmit reduces light to rear of the curtain and to
back walls.
• This increases ηsolar to maximum of ~ 88% at the expense of lower
Tp,out and higher T-gradients between front and rear of the curtain.
• Optimal flow-rate between 8 and 10 kg s-1m-1 achieve near maximum
ηsolar at higher Tp,out.
Impact of ṁ'p on Performance
14

• Prototype-scale results demonstrate need for high flow rates and
longer falls to achieve higher Tp,out while maintaining high ηsolar.
• Sandia National Labs[5] have been studying large, commercial-scale
receivers at their solar field facility.
• It is important to assess performance trade-offs at these larger
commercial scales before large-scale investments can be made for
plants using particle receivers.
• Commercial-scale receiver design requires evaluation of important
operating parameters for further development
– Impact of εp on performance
– Advantages of selective absorption, with εp in solar spectra and low εp at
longer wavelength
Commercial-scale Particle Receiver Simulations
[5] Ho,C. (2014). Personal Communication.
15

Ly,r
Ly,a
x
y
z
Commercial-Scale Receiver Model Parameters
Receiver – r 12 21 15
Aperture – a 11 20 -
Curtain – c 11 21 tcurt
Property Units Baseline
dp μm 280
ṁ’p kg s-1m-1 40
ρp
[4] kg m-3 3560
Cp,p
[4] J kg-1K-1 264+2.07T-1.12e-3T2
Tp,in K 600
′′ kW m-2
1000
16
λ range (μm) εp,λ εwall,λ
[4]
0-2.5 0.1-0.9 0.2
2.5-4.5 0.1-0.9 0.2
4.5-∞ 0.1-0.9 0.8

• ηsolar and Tp,out both increase
monotonically with εp
• Due to high ṁ'p, minimal
solar irradiation reaches rear
of curtain.
Impact of Grey Particle Emissivity on Performance
0.0
0.2
0.4
0.6
0.8
1.0
500
700
900
1100
1300
1500
0.1 0.3 0.5 0.7 0.9
ηSolar
Tp(K)
εp (-)
Front Temperature
Rear Temperature
Efficiency
17

Solar Irradiance and Particle Emittance
0
300
600
900
1,200
1,500
1,800
250 750 1250 1750 2250 2750 3250 3750
SpectralIrradiance/Emittance(Wm-2nm-1)
Wavelength (nm)
Solar Source 5600 K Source
1000 K Blackbody 1300 K Blackbody
1600 K Blackbody 1900 K Blackbody
0
100
200
300
1750 2250 2750 3250 3750
18

ṁ'p = 40 kg s-1m-1, Tp,in = 600 K, 𝒒 𝑺𝒐𝒍𝒂𝒓
′′
= 1000 kW m-2, dpart = 280 μm
εp,λ<2.5μm=0.90, εp,λ>2.5μm=0.50
′′
= 1000 kW m-2, dpart = 280 μm
εp,λ<2.5μm=0.90, εp,λ>2.5μm=0.90
′′
= 1000 kW m-2, dpart = 280 μm
εp,λ<2.5μm=0.90, εp,λ>2.5μm=0.10
Impact of Particle IR Emissivity on Temperatures
19

Performance
Measure
Units
IR emissivity (λ > 2.5 μm)
ελ = 0.1 ελ = 0.3 ελ = 0.5 ελ = 0.7 ελ = 0.9
ηSolar (-) 0.892 0.888 0.884 0.881 0.878
ηgas (-) 0.005 0.005 0.005 0.005 0.005
ηrad,lost (-) 0.090 0.094 0.098 0.101 0.104
ηconv,lost (-) 0.012 0.012 0.012 0.013 0.013
Tp,out (front) K 1307 1303 1300 1296 1294
Tp,out (rear) K 648 648 648 648 648
Impact of Particle IR Emissivity on Receiver Performance
Results for Inlet Tp,in = 600 K with ελ<2.5 = 0.9
20

Outline
• Background
21

• Undoped and doped ceria has been proposed by many authors[6-10]
for solar thermochemical fuel production because it:
– Preserves its (flourite) crystal structure under large degrees of
reduction, δ
– Maintains thermal stability with melting temperature >2800 K
– Exhibits high catalytic activity for H2O and CO2 reduction
• Lab-scale tests have demonstrated the capability to reliably yields H2
or CO, but have had trouble identifying practical receiver geometries
Ceria as a Solar Material
Parameter Value
ρpart (kg/m3)
7215 (Ce2O4)
6200 (Ce2O3)
cp,part (J/kg-K) ~460[11]
kpart (W/m-K) 12.0[11]
λ range
(μm)
frad (%)
Solar
εp,λ
[10]
Solar
frad (%)
1600 K
εp,λ
[10]
1600K
0-0.6 31 0.57 0 0.36
0.6-1.25 54 0.26 7 0.17
1.25-3.5 15 0.09 64 0.08
3.5-∞ 0 0.51 29 0.34
[6] Chueh, W, & Haile, S. (2010) Phil. Trans. Roy. Soc A, 368.
[7] Scheffe, J., Steinfeld, A. (2012) Energy & Fuels, 26.
[8] Lapp et al. (2012) Energy, 37.
[9] Le Gal et al. (2011). Energy & Fuels, 25.
[10] Marabelli & Wachter. (1987) Phys. Rev. B., 36.
[11] Mogensen et al. (2000). Sol. State. Ion., 129.
22

Ceria Modeling
δb
δsb
δs
Diff. R1 R2
• Species fractions related to δ:
Diffusion
Ce2O4(b) + Ce2O3(sb) ↔ Ce2O3(b) +Ce2O4(sb)
D∞ = 1.0 e-4 m2/s [12] Ea,diff = 333.4 kJ/kmol [12]
     
     δρ
δρ
2VOCe
21OOCe
0
O32
0
O42

 -
dr
μd
TR
ρD
j OOO
diff
0

23
Reverse Incorporation
Ce2O4(sb) + VO(s) ↔ OO(s) + Ce2O3(sb)
kfwd,R1 = 3e6 kmol/s[13] βR1 = 0.5[13]
Surface Exchange
2 OO(s) ↔ O2(g)+2 VO(s)
σO2 = 0.75[14] βR2 = 0.5[13]
 







 







 



TR
Xk
TR
Xkn
ex
ssbred
ex
ssbred
,R1
(sb)OCeO(s)R1rev,
,R1
(sb)OCe(s)VR1fwd,R1
exp+
1
exp
32
42O




 











 




















 

TRTRW
P
TR
kn
ex
sred,R22
(s)V
O
O
O
ex
sred,R22
O(s)R2fwd,R2
exp
2
1
exp2
O
2
2
2






[12] Giordano et al. (2011). Energy & Fuels, 25. [13] DeCaluwe et al. (2010). J. Phys. Chem, 114.
[14] Leistner et al. (2012) Appl. Cat. B, 415.

0.0001
0.001
0.01
0.1
1
1.E-321.E-281.E-241.E-201.E-161.E-121.E-081.E-041.E+00
δinCeO2-δ
1773
1673
1573
1473
1373
1273
1173
1073
973
873
• Zinkevich et al. (2010) model incorrectly accounted for δ dependence
– Corrected Zinkevich model correctly models T >1000 K
– Corrected Zinkevich model has reasonable low-T performance
• Surface thermodynamics fit ∆𝒉 𝒓𝒆𝒅,𝒔
𝟎
− ∆𝒉 𝒓𝒆𝒅,𝒃
𝟎
and ∆𝒔 𝒓𝒆𝒅,𝒔
𝟎
− ∆𝒔 𝒓𝒆𝒅,𝒃
𝟎
by using in-situ XPS data of DeCaluwe et al. (2011)
Thermodynamic Model
Equilibrium PO2 (atm) compared to experimental values[12]
24

• Thermodynamics must capture temperature dependence of ideal-
state and excess properties under partially reduced conditions.
– Ideal-state temperature dependence captured with SGTE polynomial
– Ideal entropy of mixing by dilute solution (thermodynamically consistent)
– Non-ideal bulk excess free energy calculated with Redlich-Kister terms
• Chemistry based on reversible mass action kinetics with rates and
excess free energy term modeled as in DeCaluwe et al. (2011)
Ceria Thermochemistry for Reactive Particle Model
25
  








4
0,
2
32
ln,Δ
OCe
OCeex
red
X
X
RTδTμ







 

TR
kk
0
red
R1fwd,R1rev, exp
  







 









TR
μμμ
TRWπ
P
σk
0
O(s)
0
O
0
(s)V
O
0
OR2fwd,
2O
2
2
5.0
exp
2
       δTμδTμTμδTμ ex
red
ex
redredred ,Δ,ΔΔ,Δ 0,0


Ly,r
Ly,a
x
y
z
Prototype-Scale Receiver Model Parameters
Receiver – r 1.85 5.00 1.50
Aperture – a 1.00 3.00 -
Curtain – c 1.00 5.00 tcurt
dp μm 300 [200, 700]
ṁ’p kg s-1m-1 2.0 [1.0, 4.0]
Tp,in K 1100 [1000, 1400]
′′ kW m-2
1000 -
PO2,in atm 1·(10-5) -
26
λ range (μm) εwind,λ
[15] ρwind, λ
[15] εwall,λ
[4]
0-0.6 0.00 0.073 0.20
0.6-1.25 0.00 0.071 0.20
1.25-3.5 0.046 0.068 0.20
3.5-∞ 0.91 0.011 0.80
[15] Heraeus. (2007).

Reactive particle wall temperatures
′′
= 1000 kW m-2, dpart = 300 μm, σstick=0.75
27

Directly Irradiated Zone
′′
= 1000 kW m-2, dpart = 300 μm, σstick=0.75ṁ'p = 1 kg s-1m-1, Tp,in = 1300 K, 𝒒 𝑺𝒐𝒍𝒂𝒓
′′
= 1000 kW m-2, dpart = 300 μm, σstick=0.10
Directly Irradiated Zone
• Ceria is rate-controlled by surface reaction
• Cooling outside directly irradiated zone by radiation loss and reaction
• Lower σstick cases do not reach equilibrium by exit
Impact of varying ceria kinetics
28

0
0.02
0.04
0.06
XCe2O3(2δ)
Particle Flow Rate (kg s-1m-1)
1 2 3 4
0
0.1
0.2
0.3
ηtot
1300
1500
1700
1900
1000 1100 1200 1300 1400
Tp,out(K)
Tp,in (K)
Impact of varying Inlet Tp
29

• Smaller particles capture more energy chemically
– Greater surface area and Tp
• Reactive particles can achieve higher ηSolar than inert particles
• Particles much lower than 300 μm can have stability problems[4]
Impact of dp and reaction on performance
 
Chem
Solar
k
k
Sensible η
Q
hmhm
η
tot

 
 


1
kout,kout,kin,kin,
 
Solar
n
i
ipreac
O
ireactg
Chem
Q
Th
W
m
η
cells




1
,
,,
Δ
2
0
0.05
0.1
0.15
0.2
0.25
100 200 300 400 500 600 700
Efficiency
dP (μm)
ηSensible ηChem ηInert
30

• Receiver design is not optimized for ceria production
– To achieve high Tp at this scale requires low ṁ'p
• Design requires evaluation in context of a full-system
– Strategies for power production or heat recovery
• Undoped ceria performance is low due to very high Tp and low εp
– Doping strategies being explored, but face challenges due to cycling
and slow oxidation kinetics. [6,8-10]
• Lower-Tp cycles with better optical properties can achieve higher
performance
• Perovskites are a class of materials with similar solid-structures and
high εp
– Favorable reduction thermodynamics at much lower temperatures
– Cannot be used for fuel production
Ceria conclusions and perovskite motivation
31

Surface Exchange
2 OO(s) ↔ O2(g)+2 VO(s)
ksurf,∞ = 0.109 m/s [18] Ea,surf =74.30 kJ/mol [18]
La0.1Sr0.9Co0.8Fe0.2O3-δ Particle Model
δb
δs
Diffusion
Surf
Exch.
• Species fractions related to δ:
Diffusion
LSCFO3(b) + VO(s) ↔ LSCFO2(b) + OO(s)
D∞ = 1.01e-4 m2/s [18] Ea,diff = 55.96 kJ/mol [18]
dr
μd
TR
ρD
an OOO
partdiff
0
= ( )sseqsurfsurf kn δδρ -,
0
=
Parameter Value
ρpart (kg/m3)
6580[16] (LSCFO3)
6051[16] (LSCFO2)
cp,part (J/kg-K) 145[16]
ε (-) 0.90[17]
[16]: Beale, S. et al. (2011). ECS Transaction, 35: 935-943.
[17]: Guar, A. et al. (2013) Euro. Fuel Cell Conf.
[18]: Choi, M. et al. (2011). Sol. State Ionics, 11: 269-274.
[ ] [ ] ( )
[ ] [ ] δρ
δρ
0
O20.20.80.90.1
0
O30.20.80.90.1
VOFeCoSrLa
1OOFeCoSrLa
==
== -
32

• Assume ΔHO(δ) and ΔSO(δ) are constant with temperature[19]
• Ideal thermodynamics fit to NASA Polynomial (Ref. state: δ0 =0.45)
LSCF Thermodynamics
ΔHO = -433.27δ - 55.835
ΔSO = -76.414δ - 168.78
1000 °C
950 °C
900 °C
800 °C
1000 °C
950 °C
900 °C
800 °C
          OO
eqO
OeqOOLSCFOLSCFO μTμ
P
P
RTTμPTμδTμδTμ Δ
2
1
ln
2
1
,
2
1
,-, 0
0
,0
, 2
2
22223


















[19]: Choi, M. et al. (2012). Sol. State Ionics, 12: 22-27.
33

′′
= 1000 kW m-2, dpart = 300 μm, k = ksurf
Directly irradiated zone
′′
= 1000 kW m-2, dpart = 300 μm, k = 10* ksurf
Directly irradiated zone
Influence of Reaction Rate
• Process is kinetically limited by surface rates.
• Reduction is driven strongly by Tp, even at high PO2.
• Faster reduction decreases ΔTp and improves efficiency.
34

Influence of Reaction Rate
1100
1130
1160
1190
1220
1250
200 300 400 500 600
Tp,out(K)
dP (μm)
k x 10 k x 1
• ηSolar is relatively constant at both kinetic rates
• ηChem increases while ηSensible decreases with faster kinetics
• Smaller dp particle curtains have lower τ, greater surface area, and
slower up
• With faster kinetics, ηSensible actually decreases with dp
0
0.2
0.4
0.6
0.8
1
200 300 400 500 600
StorageEfficiency
dP (μm)
k x 10 - ηSolar k x 10 - ηChem
k x 1 - ηSolar k x 1 - ηChem
35

• LSCF transmits more than inert particles at higher ṁ'p due to higher ρ
• Transition in temperature curve ~1000 K due to reaction
• Tradeoff between higher ṁ'p and higher Tp shows inflection in O2
production around 20 kg s-1m-1
Commercial-scale, influence of ṁ'p
36
ṁ'p
(kg s-1m-1)
ṁ'p
(kg s-1m-1)ṁ'p
(kg s-1m-1)

• LSCF tests demonstrate significant potential to improve storage
density significantly via chemical reduction
• LSCF equilibrium show strong Tp dependence  large ΔSO desirable
• Reactive particles require careful consideration of storage conditions
• Ideal operation requires evaluation in full-cycle context
Perovskite conclusions
0
400
800
1200
1600
0.0
0.2
0.4
0.6
0.8
1.0
20 30 40 50 60 70 80
OutletTp(K)
Efficiency
ṁ'p (kg s-1m-1)
37

Outline
• Background
38

• Validate simplified model assumptions
– Gas entrainment model developed for non-reactive, isothermal flow
– Curtain stability untested with simplified model
• Evaluate impact of gas-flow on performance
– Internal gas-flow impacts wall and particle temperatures through
recirculation
• Test alternative gas-flow conditions for improvements
– Opportunity to improve O2 injection in vicinity of reaction
– Improved thermal impacts of gas
CFD Model Motivations
39

• Lagrangian-frame particle tracking
• Particle temperatures and reaction integrated along the fall
• Gas-phase coupling
• Stochastic particle tracking to account for turbulent dispersions
Particle Model
   44
1
,, Δ pRppreacreacpgpp
N
m
p
mpmp TTσεahnTTha
dt
dT
cm
m










cell
mm
drops
pY
m
V
t
Wn
n
N
dS
Δ


  














cell
ipmmpipig
pD
ppdrops
pM
i
V
t
uWnmuu
C
dρ
μ
n
N
dS
Δ
24
Re18
,,,2


        






cell
refmpmmgppp
drops
pT
V
t
ThThnTTha
n
N
dS
Δ00


   
p
gpi
ipig
pD
pp
ip
ρ
ρρg
uu
C
dρ
μ
dt
du 
 ,,2
,
24
Re18
40

• To implement in CFD framework, kinetic mechanism modified to
depend on degree of surface reduction (δs)
• Simplified model shows δs stays in equilibrium with δb.
• Optimization method calculates δs in equilibrium with δb.
Modified Ceria Reaction Mechanism
   spsredbpbredsb δTμδTμμ ,Δ,ΔΔ ,, 
Profiles of Tp and δ for bulk and surface along fall
41

• Solves for radiation intensity, Iλ, at every location and in specified
directions, θ and φ
• Directions determined by splitting Cartesian grid into Nθ x Nφ
discretizations in each octant
• Particle source terms determined by collecting contributions from
each injection
Discrete Ordinates (DO) Radiation Model
           ')'(',,, , ΩΦ
4
4
0
2
dsssrI
π
σ
SrInasrIσaassrI
π
λ
pI
pλλbλλppλλ















cell
pp
drops
p
p
V
t
εd
π
n
N
da
Δ
4
2

   












cell
pσp
drops
p
p
V
t
εfd
π
n
N
σd
Δ
11
4
2








cell
pppλ
drops
pI
pλ
V
t
Tεaf
n
N
dS m
Δ4
,

Property Value
′′
(kW m-2) 917.8
Beam Direction [0, 0, -1]
Beam Width
Δθ x Δφ (deg)
0.001 x 0.001
Diffuse Fraction 0.0
Property Value
Nθ x Nφ 9 x 5
𝑵 𝜽 𝒑
x 𝑵 𝝋 𝒑 7 x 7
Δλ1 (μm) [0, 4.5]
Δλ2 (μm) [4.5, 100]
42

Prototype-Scale Run Parameters
Lx (m) Ly (m) Lz (m)
Receiver – r 1.85 5.00 1.50
Aperture – w 1.00 3.00 -
Curtain – c 1.00 - 0.01
Gas Inlet - i 1.00 - 0.10
dpart μm 300 [200, 500]
ṁ'part kg s-1m-1 2.0 [2.0, 4.0]
Tin K 1100 -
εp - 0.3347 -
PO2,in atm 1·(10-5) -
ug,in m/s 1.0 -
′′ kW m-2
917 -
λ range (μm) εWall,λ
[4]
0-4.5 0.20
4.5-∞ 0.80 Lc,z
Li,z
Lc,x
Lagrangian Particle Injection
Locations
43

Wall and Curtain Temperatures
44

• Gas recirculation cells form due to particle-entrainment and buoyancy
• Tg greater than Tp in early fall, less than Tp in later half
• Minimal backflow occurs around edges of curtain
Gas Profiles
45

• Higher ṁ'p reduces temperatures
– Lowers thermodynamic forcing
– Slows kinetics
• δeq falls due to lower Tp and
increasing PO2
• Higher ṁ'p releases more O2
despite lower δ
Impact of varying ṁ'p (kg s-1m-1)
0.025
0.020
0.015
0.010
0.005
0.000
0 1 2 3 4 5
Meanδp(-)
Distance from inlet (m)
0.04
0.03
0.02
0.01
0.000 1 2 3 4 5
PO2(atm)
1900
0 1 2 3 4 5
1100
1300
1500
1700
MeanTp(K)
46

• Gas recirculation
– Pre-heats particles along first half
of fall
– O2 from exit recirculates to inlet
• Higher gas flow-rate around
particles in CFD
– Dampens PO2 rise from reaction
• Higher max Tp with CFD model
– δb reaches equilibrium before exit
– Cooling damped by reoxidiation
outside directly irradiated zone
Comparison of Simplified and CFD Models
CFD Model
Simplified Model
47

• Higher ṁg impact gas absorption and wall temperatures
• Isotropic radiation reduces reflection out of window
• Higher Tp increases chemical storage
Comparison of Simplified and CFD Models
0.0
0.2
0.4
0.6
0.8
1.0
2.0 3.0 4.0
ṁ'p (kg s-1m-1)
0.0
0.2
0.4
0.6
0.8
1.0
2.0 3.0 4.0
FractionQSolar
ṁ'p (kg s-1m-1)
CFD Model Simplified Model
48

• Gas injected at the bottom of the receiver near the front and rear wall
– Promotes curtain stability
– Pre-heats entrained gas
Impact of Alternative Gas Injection Strategies
49

Impact of Alternative Gas Injection Strategies
2100
2000
1900
1800
1700
1600
1500
1400
1300
1200
1100
Temperature(K)
Top
Injection
Bottom
Injection
.020
.018
.016
.014
.012
.010
.008
.004
.002
.000
δb(-)
Top
Injection
Bottom
Injection
Curtain Temperatures Curtain Reduction
50

• Smaller particles are more efficient
– Below dp ~200 μm, minimal improvement in performance
• Trade-off between ηSolar, mean Tp, and Tp,front-Tp,rear with increasing ṁ'p
– Ideal ṁ'p of 8-10 kg s-1m-1 to balance ηSolar and mean Tp
• Ideal flow values relate to curtain τ, with optimal τ <25% to achieve
higher Tp with minimal changes in ηSolar
• Important to maximize εp, but ideal selectivity improves ηSolar < 2% for
Tp below 1300 K
– At temperature above 1600 K, selectivity can improve ηSolar ~ 5%
Conclusions – Inert Particles
0
0.2
0.4
0.6
0.8
1
600 800 1000 1200 1400 1600
ηSolar
Tp,out (K)
dp
Tp,in
𝑞 𝑆𝑜𝑙𝑎𝑟
′′
εp
51

• General reactive-particle conclusions:
– Particle size is even more important due to lower τ and higher surface
area
– Best performance occurs when reaction cycle is properly scaled to
particle reaction-rate
– Reactors require analysis in context of a full-cycle to optimize
• Ceria operating Tp is too high and εp too low: Max ηChem ~ 7%
– At maximum ceria ηSolar ~ 35%, the ηChem < 1%
• Perovskite particles show promise due to low reduction Tp , high εp,
and ability to work above atmospheric PO2
• CFD simulations demonstrate the importance of capturing gas-flow
effects
Conclusions – Reactive Particles
52

• Test new materials
– Perovskites and other dark materials with fast kinetics, low reduction Tp,
and lower cost
• Receiver architectural changes
– Shorter particle drops
– Layered curtains
– Investigate more alternative gas-injection strategies
• Improvements to simplified model
– Improved gas-treatment to include influence of gas over larger range
– Semi-empirical gas flow along walls to capture recirculation
• Improvements to CFD model
– Improve particle-radiation coupling to allow for multi-bin particle
properties and anisotropic scattering
Future Work
53

• Presentations
– Concentrated Solar Thermal Energy for H2O and CO2 Splitting. Oles, Jackson, Thamire, Gibbons.
ASME-ES2012
– Simulation of High-Temperature Receivers Using Ceria Particles. Oles, Jackson, Gibbons. ASME-
ES2013
– Simulation of High-Temperature Receivers Using LSCF Particles. Oles, Jackson. ASME-ES2014
– Impacts of Spectral Selectivity in Directly Irradiated Particle Receivers. Oles, Jackson, Ho. ASME-
ES2014
• Publications
– Parametric design modeling of concentrated-solar falling-particle receivers. Oles, Jackson. WIP.
– Investigation of absorption selectivity on concentrated-solar falling-particle receiver performance. Oles,
Jackson, Ho. WIP.
– Modeling of a concentrated-solar falling-particle receiver for ceria reduction. Oles, Jackson. Solar
Energy.
– Modeling of storage enhancement in a falling-particle solar receiver utilizing reactive perovskite
particles. Oles, Jackson. WIP
– Modeling reactive ceria particles in a falling-particle solar receiver using CFD. Oles, Jackson. WIP.
Presentations and Publications
54

• Thank you to Dr. Gregory Jackson for his help and direction as my
Ph.D. advisor.
• Thank you to Warren Citrin for financial support through the Warren
Citrin Fellowship for Entrepreneurial Engineering Students
• Thank you to my dissertation committee – Dr. Kiger, Dr. Riaz, Dr.
Sunderland, and Dr. Zachariah – for your time and scrutiny of this
work.
• Thank you to Dr. Cliff Ho at Sandia Natl. Labs for his collaboration
and direction.
• Thank you to my lab-mates including Will Gibbons, Lei Wang, Josh
Pearlman, Babak Eslami, Danica Gordon, and Esteban Echeverria.
• Thanks to Amanda and my family for their support and
encouragement to make this possible.
Acknowledgements
55

Dissertation Defense - Final

More Related Content

What's hot

Viewers also liked

Similar to Dissertation Defense - Final

Dissertation Defense - Final