Professional Documents
Culture Documents
1 s2.0 S0038092X15006970 Main PDF
1 s2.0 S0038092X15006970 Main PDF
com
ScienceDirect
Solar Energy 125 (2016) 339–359
www.elsevier.com/locate/solener
Received 7 September 2015; received in revised form 4 December 2015; accepted 14 December 2015
Available online 8 January 2016
Abstract
As concern over the CO2 emissions associated with power production from traditional fossil fuels grows, there has been more interest
in using renewable energy sources such as solar energy. However, there are many issues with using solar energy on a large scale including
dispatchability and economically viable storage methods. A potential solution to these problems is using a hybrid solar-fossil fuel power
generation system. Within such systems, solar reforming has been shown to be a promising integration method with steam redox reform-
ing as an effective reforming process. In this article, a receiver reactor concept for a solar steam redox reformer is presented. The receiver-
reactor consists of a dumbbell shape absorber system that has two distinct absorbers. This absorber system setup allows for the switching
between reduction and oxidation steps without having to constantly change inlet streams to the reactor and is designed such that the inlet
connections do not interfere with the solar window. In addition, at any point in time only one solar absorber is irradiated by the solar
energy (during the reduction step). A computational model is developed to assess its performance. Simulations show that the receiver-
reactor strongly absorbs the solar radiation and most of the radiative heat transfer occurs in the front half of the reactor. Moreover,
results show that higher conductivity absorber materials yield smaller temperature variations not only within the reactor but also with
time, and therefore are more suitable for long term reactor operation.
Ó 2015 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.solener.2015.12.024
0038-092X/Ó 2015 Elsevier Ltd. All rights reserved.
340 E.J. Sheu, A.F. Ghoniem / Solar Energy 125 (2016) 339–359
Nomenclature
The solar reforming integration method involves carriers can be used, redox reforming does not require
reforming the fuel (natural gas) into syngas (which has a the use of expensive noble metal catalysts (as is usually
higher heating value) using solar energy. The syngas is then the case with solar dry and steam reformers). Steam redox
used as a fuel for the gas turbine to produce power in a reforming can also yield product compositions with more
combined cycle. Solar reformers can be categorized into desirable H2/CO ratios than direct steam reforming. In this
two main categories: directly irradiated and indirectly irra- article, a design for a receiver-reactor that can be used for
diated. In this work, the focus will be on directly irradiated the steam redox process is discussed. In addition, a compu-
solar reformers as higher temperatures can be reached in tational model that can aid in the design of the solar reac-
these reformers which leads to higher methane conversion. tor is presented. Finally, a set of base case results from the
However, it should be noted that indirectly irradiated reactor model is shown and the validity of the results
reformers can also be of great interest due to cost consider- discussed.
ations. For a directly irradiated solar reformer, the solar
receiver and reactor are one integrated unit. In traditional 2. Receiver–reactor design
solar receivers used in solar only power production, a solar
absorber (usually a porous ceramic disk or wire mesh) is As mentioned previously, the solar reformer studied
placed inside a solar receiver and irradiated by the solar herein uses steam redox reforming which consists of two
energy. A heat transfer fluid is then sent through the solar main reactions: a (metal) reduction reaction and an oxida-
absorber and heated by convection. To apply this concept tion reaction. In previous work, iron oxide has been shown
to solar reforming, the solar absorber merely becomes the to be a promising option as an oxygen carrier for a steam
reaction site. Most often in these solar reformers, the solar redox reformer and hence is utilized in this analysis (Sheu
absorber is a porous ceramic disk or ceramic honeycomb and Ghoniem, 2014). For this reformer, the two states
with any catalyst needed for the reforming process coated for the oxygen carrier are magnetite (Fe3O4) and wustite
onto the porous surface or channels. The most commonly (FeO). Thus, the two main reactions for the reformer stud-
used solar reforming processes are steam reforming, dry ied herein are
reforming, and redox reforming. There has been much
Reduction : CH4 þ Fe3 O4 ! 3FeO þ CO þ 2H2
experimental work with steam and dry solar reformers
(Anikeev et al., 1998; Berman et al., 2007; Buck et al., DH ¼ 266:60 kJ=mol
o
1991; Maria et al., 1986; Dahl et al., 2004; Levy et al., Oxidation : 3FeO þ H2 O ! Fe3 O4 þ H2
1992; Wörner and Tamme, 1998), and to a less extent on
DH ¼ 60:44 kJ=mol
o
redox solar reformers as well (Kodama et al., 2000;
Steinfeld et al., 1998; Steinfeld et al., 1993). Meanwhile Note that CO2 can also be used as the oxidizing agent.
there has been much less computational work, with only However, steam is used in this analysis due to the faster
a few numerical studies on steam and dry solar reformers oxidation rates (Stehle et al., 2011).
(Ben-Zvi and Karni, 2007; Petrasch and Steinfeld, 2007; In chemical looping applications, often times two sepa-
Skocypec et al., 1994). Development of computational rate reactors are used (one for oxidation and one for reduc-
models for solar reformers in conjunction with experimen- tion) with a fluidized bed circulating the oxygen carrier
tal studies can aid in the design and operation of the solar between the two reactors (Adanez et al., 2012). However,
reformer for use in hybrid solar fossil fuel power the energy required for the circulating bed as well as the
production. associated pressure drop can be detrimental to reformer per-
Previous system analysis has shown that steam redox formance (Adanez et al., 2012). Moreover, there are other
reforming has some advantages for use in hybrid power complexities associated with maintaining the fluidized bed,
cycles (Sheu and Ghoniem, 2014) and will be the focus in separating the particles, and particle agglomeration at high
this work. Moreover, since a number of different oxygen temperatures (Adanez et al., 2012). Therefore, the reactor
E.J. Sheu, A.F. Ghoniem / Solar Energy 125 (2016) 339–359 341
design proposed here is one where the reaction site is the switch between being irradiated (during reduction) to not
solar absorber which has the oxygen carrier coated on to being irradiated (during oxidation). The rotor in between
it (similar to what is commonly used in direct dry and steam the two absorbers is insulated to ensure that the rotor is
solar reformers). The solar absorber used in this reactor is a not damaged by the high operating temperatures and no
honeycomb ‘‘disk” with micro-channels, as this type has heat transfer between the two absorber. The inlet ports
been shown to be advantageous in terms of fuel conversion are also designed such that the reducing/oxidizing agent
and more even temperature distribution (Levy et al., 1992). enters the dumbbell absorber system from the top and bot-
A schematic of how the reactor is incorporated with the tom. This top entrance ensures that there is no interference
solar collector is shown in Fig. 1. Note that in the schematic of the solar irradiating the absorber during the reduction
the reactor is shown with a dish solar collector/receiver sys- from any inlet connections. Moreover, the dumbbell sys-
tem; however, the reactor could also be implemented with a tem and the outer housing of the receiver reactor needs
solar tower type receiver as well. to be manufactured such that the dumbbell system just fits
Unlike more traditional steam and dry reformers, the inside the housing to minimize any spillage of the inlet
steam redox process is a two step process and therefore streams. The reactor would also have to carefully assem-
requires some sort of switching mechanism in order to go bled to ensure that the inlet port of the dumbbell system
between the reduction and oxidation processes. One way aligns with the inlet port of the housing. Note that due
of achieving this is to switch the inlet streams going into to high temperature operation and the moving parts within
the reactor between the reducing and oxidizing agent. How- the reactor, the sealing of the inlet and outlet ports may not
ever, the control and execution of this inlet stream switching be perfect and there can be spillage of gases (i.e., unre-
can be quite difficult in terms of implementation. Thus, for formed gases pass through the space in between the absor-
this reactor design, the solar absorber itself is the compo- bers). However, as discussed previously, this reactor is
nent that is moving and switching between the two reform- designed for use in hybrid solar-fossil fuel power genera-
ing steps. Essentially, this reactor has two solar absorbers tion rather than production of pure syngas. Thus, all the
which switch between the two inlet ports that have constant products (whether reacted or not) are sent to a gas turbine
flow of the reducing and oxidizing agent. Moreover, in this for power production. Therefore, the unreacted gases pass-
reactor design, the solar energy is only used during the ing through the reactor is not necessarily a detriment for
reduction step as this step is highly endothermic while the the reactor. Since the two absorbers switch between the
oxidation step is slightly exothermic. Thus at any point in two steps, they must be identical and the reduction and oxi-
time, only one of the solar absorbers will be irradiated by dation time must be equal. Also note that during the
the solar radiation. Schematics of the reactor design for this switching, the inlet gases would just flow through the reac-
switching reactor are shown in Fig. 2. tor to the outlet without undergoing any reforming. The
The two absorber system in the reactor is essentially a switching time is much smaller than the reduction/oxida-
dumbbell shape. For a certain cycle time scyc , the dumbbell tion time so therefore the amount of gases that do not pass
system will rotate 180 degrees every scyc divided by 2. This through the actual reforming site should be minimal.
rotation allows for the switching of the absorber between As discussed previously, the absorber is a honeycomb
the reduction and oxidation steps without having to switch disk due to potential advantages in temperature distribu-
inlet streams. The rotation also allows the absorber to tion and fuel conversion. The absorber can be manufac-
tured from a number of different materials. The most
commonly used materials include Alumina, SiC, and Cor-
dierite. The selection of material is important as the absor-
ber material will also function as the support for the
oxygen carrier which is coated onto the channel surfaces.
For the base case results presented later, Alumina and
SiC are used for the absorber material. The length of the
absorber is chosen to be 50 mm based previous experimen-
tal studies for solar dry and steam reformers (Buck et al.,
1991; Skocypec et al., 1994). For this type of solar receiver
absorber, the absorber is often relatively short. The reason
is that the radiative flux decays quite rapidly in the axial
direction. This decay will be discussed in more detail when
the base case results are presented. The length is mainly
determined by the reaction kinetics. The absorber channel
width (d) is 4 mm based on previous experiments with hon-
eycomb absorbers (Levy et al., 1992). The length and chan-
nel size can have great effect on reactor performance.
Further sensitivity and optimization studies could be used
Fig. 1. Schematic of receiver reactor with solar collector. to select optimal channel design parameters.
342 E.J. Sheu, A.F. Ghoniem / Solar Energy 125 (2016) 339–359
(a)
(b)
(c)
Fig. 2. Conceptual schematic of receiver reactor: (a) simplified outer view of reactor, (b) simplified transparent view of reactor from front (left) and back
(right), (c) detailed schematic of reactor (side view).
During the reduction step, methane and a carrier gas available. The CO2 also has the potential to further reform
flows into the reactor. The carrier gas chosen here is CO2 the methane through the process of dry reforming; how-
to prevent coking that may occur during the methane ever, any dry reforming is neglected in this analysis due
partial oxidation process. Note that steam can also be used to reasons that will be discussed later in Section 3.5. The
as the carrier gas in situations where CO2 is not readily methane flow rate is chosen such that chemical energy of
the methane flowed into the reactor is 100 kW. The cycle
time is 6 min. This cycle time was chosen to achieve signif-
icant conversion (reduction) of the oxygen carrier during
the reduction step. The flow rate during the oxidation step
is also chosen such that the oxygen carrier is completely re-
oxidized (back to original state) during the time period for
oxidation (cycle time divided by 2). All parameters used for
the base case design of the reactor are summarized in
Table 3 (Section 4).
Now that the overall design of the reactor-receiver has
been presented. The computational model developed to
study this reactor will be discussed.
3. Computational model
where h is the heat transfer coefficient, ds is the solid thick- For the model initialization, the solid temperature is set
ness, and k s is the thermal conductivity of the solid to the steady state solid temperature calculated assuming
(Table 3). Therefore the solid phase can be thought of as no gas flow. To achieve a steady state, surface radiation
a transient fin with temperature variation only along the losses are taken into account at the front surface of the
axial direction of the channel. The flow inside the channel channel. Without the gas flow, the solid temperature can
is laminar (Re < 100). Moreover, the ratio of channel reach temperatures as high as 2000 K, which is what is
length to size (d 2ds ) is large (25); therefore the flow expected if the solid is merely being heated by the solar
is assumed to be fully developed for a large part of the radiation with only minimal surface radiation losses. For
channel with an entrance region that is a relatively small every subsequent cycle, the initial condition for the appro-
portion of the channel. The heat and mass transfer coeffi- priate variables (i.e., temperature) are set to the values
cient correlations do include entrance region effects. The obtained at the end of the previous cycle. The model is con-
input solar flux is assumed to be already concentrated. verged when the reformer has reached a cyclic steady state.
There is also no pressure drop taken into account as the In this analysis the reformer is defined to have reached a
pressure drop calculated for laminar fully developed flow cyclic steady state when the temperature profile of the solid
is relatively small (<1 kPa) compared to the reactor operat- is the same (<1 K difference) at the end of the cycle as it is
ing pressure (20 bar). at the beginning of the cycle. Note that as the solid temper-
The channels have a square cross section while the ature profile reaches a cyclic steady state, so does the gas
absorber is a circular cross section (Fig. 2). Therefore, there temperature profile.
will be some non fully formed channels (i.e., not a square). The model is solved using COMSOL. The different
In the computational model, these channels are neglected. equations solved in the computational model will now be
As shown in Fig. 4, the reactor operation is modeled as discussed in more detail.
cyclic with two steps as follows: (1) Methane (along with a
carrier gas – CO2) is fed into the channel and metal reduc- 3.1. Species conservation
tion occurs, and (2) the input stream is switched from
methane to steam and metal oxidation occurs. During the There are three different sets of species conservation
oxidation step, no solar energy is provided to the reactor. equations used in this model: one set for the species in
The absorber switching time is not considered as that time the gas phase, one set for the gas–solid interactions, and
(on the order of seconds) is much smaller than the actual a conservation equation for the oxygen carrier. The equa-
reduction/oxidation time (on the order of minutes). The tions for the species conservation are as follows
model takes into account both the gas phase and the solid
phase (porous oxygen carrier layer and dense support) with @C i;g @ uC i;g
Gas : Ac þ Ac ¼ P c J gs;i ð1Þ
convective heat transfer and mass transfer between the @t @x
solid and gas phases (Fig. 4). Radiative transport from Gas–solid interactions : 0 ¼ P c J gs;i þ P c Ri;s ð2Þ
the solar irradiation (when appropriate) and conductive @ð1 s ÞC oc
heat transfer are also taken into account in the solid phase. Oxygen carrier only : Aoc ¼ P c Roc ð3Þ
@t
The mass of the dense layer (ðds doc Þqs ) is much
larger than that of the porous oxygen carrier layer where C i;g is the molar concentration of species i in the gas
(ð1 s Þdoc qoc ) in order to minimize temperature fluctua- flow (mol/m3), Ac is the cross sectional area of the channel
tion and ensure thermal stability. Therefore, all heat (m2), u is the velocity of the flow (m/s), P c is the perimeter
transfer properties for the solid are taken to be that of of the channel (m), Aoc is the cross sectional area of the oxy-
the dense support. As mentioned previously, the absorber gen carrier layer (m2), s is the porosity of the solid, C oc is
material and is chosen to be alumina or SiC for the base the molar concentration of the oxygen carrier within the
case results. porous layer (mol/m3), Ri;s is the surface reaction rate for
gaseous species i (mol/m2/s), Roc is the surface reaction rate
for the solid oxygen carrier (mol/m2/s). Both Ri;s and Roc
are dependent on the thickness of the porous layer (in addi-
tion to various kinetic parameters) and are discussed in
more detail later in Section 3.5. J gs;i is the molar flux of spe-
cies i between the solid and gas flow (mol/m2/s) and is
defined as
J gs;i ¼ hm ðC i;g C i;s Þ ð4Þ
where hm is the mass transfer coefficient (m/s). The mass
transfer coefficient is calculated based on a simplified Sher-
wood number (Sh) correlation for laminar flow in catalytic
Fig. 4. Schematic of reactor operation model in channel during reduction monoliths with a square cross section (Gundlapally and
(top) and oxidation (bottom). Balakotaiah, 2011; West et al., 2003):
344 E.J. Sheu, A.F. Ghoniem / Solar Energy 125 (2016) 339–359
3.3. Radiation transport model where I c represents the collimated part of the radiative
intensity and the I d represents the diffuse part of the radia-
In this model, the solid is considered to be an absorbing, tive intensity.
emitting, and isotropically scattering material. The solid is The collimated part the radiative intensity can be calcu-
also assumed to have homogeneous and gray radiative lated as (Modest, 2013)
properties. In order to determine the radiative flux, the @I c
radiative intensity I must be determined. The one dimen- ¼ K ext I c ð12Þ
@z
sional transport equation based on the assumptions men-
tioned previously is as follows Substituting Eqs. (11) and (12) into Eq. (8) yields the fol-
Z 4p lowing equation for the diffuse part of the radiative
@I rs intensity
¼ ra I b ðT s Þ K ext Ið^sÞ þ Ið^sÞdX0 ð8Þ
@z 4p 0 Z 4p
@I d rs
¼ ra I b ðT s Þ K ext I d þ ðI d þ I c ÞdX0 ð13Þ
where I is the radiative intensity which is a function of @z 4p 0
direction vector ^s; ra is the absorption coefficient, rs is
To solve for these different radiative intensities a modi-
the scattering coefficient, K ext is the extinction coefficient
fied P1 approximation is used (Modest, 2013). The P1
(ra þ rs ), I b is the blackbody emission, and X is the angle
approximation is valid when the material is optically thick.
of the radiation.
Optical thickness can be calculated as (Modest, 2013)
The radiative properties (scattering, absorption, and
extinction coefficients) are determined from models using sL ¼ K ext L
geometrical optics. Geometrical optics are valid when
where sL is the optical thickness and L is the channel
(Petrasch et al., 2007)
length. For the materials used in the simulations, the opti-
pL cal thickness are approximately 15 and 10. Therefore all
a5
k optical thicknesses are greater than 1 and the material solid
can be considered to be optically thick.
where L is the channel (solid) length and k is the wave-
For solving using the P1 approximation, the incident
length of the radiation. For the typical wavelengths of solar
radiation function G will be used instead of the radiative
radiation and the base case parameters used herein
intensity I. The incident radiation function is defined as
pL
k
105 . Thus since geometric optics are used the radiative Z 4p
properties are only dependent on the porosity of the solid. GðzÞ ¼ Ið^sÞdX ð14Þ
In this reformer, the solid properties are based on the dense 0
support layer, thus the radiative properties are taken for By using G the dependence on direction is removed which
materials that have very low porosity (<0.05%). makes for easier solving of equations. Similar to I; G can
The radiative flux (qr ) is the integration of radiative also be divided in collimated and diffuse parts yielding
intensity overall all angles in a specific direction
Z 4p G ¼ Gc þ Gd ð15Þ
qr ¼ Ið^sÞ^sdX ð9Þ Applying the P1 approximation, the diffuse radiative inten-
0
sity can be written as
For the the energy conservation equation, the gradient of 1
the radiative flux is needed and can be calculated as Id ¼ ðGd þ 3qd Þ ð16Þ
4p
Z 4p Z 4p
@qr @I Substituting Eq. (16) into Eq. (13) and integrating over all
¼ ^sdX ¼ 4pra I b ðT s Þ K ext Ið^sÞdX
@z 0 @z 0 solid angles yields
Z 4p Z 4p
rs @qd
þ Ið^sÞdX0 dX ð10Þ ¼ ra ð4pI b ðT s Þ Gd Þ þ rs Gc ð17Þ
0 4p 0 @z
The radiative intensity within the solid has two distinct where qd is the radiative flux due to the diffuse radiative
parts: (1) the collimated portion which consists of the intensity.
remaining collimated beam from the solar radiation after To obtain an additional equation, substituting Eq. (16)
partial extinction, absorption and scattering and (2) the dif- into Eq. (13), integrating over all solid angles and multiply-
fuse portion which results from emission from boundaries ing by a specific direction vector yields
and within the medium and the scattered parts of the radi- 1 @Gd
ation from the collimated solar irradiation. For this model, qd ¼ ð18Þ
3K ext @z
it is assumed that the collimated solar radiation penetrates
directly into the solid without changing direction. Thus the Eqs. (17) and (18) are the governing equations for the mod-
radiative intensity can be rewritten as ified P1 approximation method (Modest, 2013).
Combining Eqs. (17) and (18) yields an equation for Gd
I ¼ Ic þ Id ð11Þ as
346 E.J. Sheu, A.F. Ghoniem / Solar Energy 125 (2016) 339–359
@ 1 @Gd This boundary condition is chosen because previous exper-
¼ ra 4pI b ðT s Þ Gd þ rs Gc ð19Þ
@z 3K ext @z imental work has shown in these types of absorbers, the
solid temperature variation decreases towards the back half
Gc can be easily solved for from Eq. (12) and is equal to of the absorber (Muir et al., 1994).
Gc ¼ I o expðK ext zÞ ð20Þ Note that during the oxidation step (i.e., no radiation),
the front face boundary condition is still the same as the
After solving for Gd and Gc from Eqs. (19) and (20) explic- convection and surface radiation losses are again assumed
itly, the radiative flux term in Eq. (6) can be calculated to be minimal.
using Eq. (10) (integrated over all solid angles to use G For the radiative transfer equation, the radiation flux at
instead of I) as follows the front surface is set to the solar flux as discussed previ-
ously and defined as
@qr @ 1 @G
¼
@z @z 3K ext @z 1 @G
qr jz¼0 ¼ ¼ Io
3K ext @z z¼0
where G is the sum of Gd and Gc .
For the back end, Marshak’s boundary condition is used
3.4. Boundary conditions (Modest, 2013) and defined as
1 @Gd 1
For the species conservation, the boundary conditions ¼ rT s jz¼L Gd jz¼L
4
are simply based on the inlet flow rate of each of the 3K ext @z z¼L 4
species:
C i;g ðz ¼ 0Þ ¼ C i;in 3.5. Kinetics
For the gas flow energy equation, the boundary condition To model the chemical reaction kinetics, kinetics param-
is the inlet temperature of the gas flow. eters from two different studies were used. The different
T g ðz ¼ 0Þ ¼ T in kinetic parameters for both the reduction and oxidation
step are summarized in Table 1. How these parameters
For the solid energy equation, an energy balance for the are used in the rate expressions as well as the experimental
front face (i.e., the side of the absorber being irradiated) conditions used to obtain these parameters will be dis-
yields cussed next.
qcond jz¼0 þ qr jz¼0 ¼ qconv jz¼0 þ qrad;s jz¼0 þ I o
3.5.1. Reduction
where qcond is the conductive flux, qr is the radiation flux As discussed previously, the reduction reaction for the
transported within the solid, qconv is the convection heat reformer studied here is
loss, qrad;s is the surface radiation loss, and I o is the input
CH4 þ Fe3 O4 ! 3FeO þ CO þ 2H2 DH o ¼ 266:60 kJ=mol
solar flux. The solar absorbers used in these types of solar
receivers are often coated such that they completely absorb The kinetics parameters are taken from (Moghtaderi and
the solar radiation (Ávila Marı́n, 2011). Thus, the surface Song, 2010). This kinetics study was done for the reduction
radiation loss is assumed to be minimal. Moreover, the of hematite to wustite with methane (Fe2 O3 ! FeO). The
convective heat loss is also assumed to be minimal as the reduction of iron oxide occurs in steps – hematite
solar receiver is usually constructed such that there is insu- (Fe2O3) ? magnetite (Fe3O4) ? wustite (FeO) ? iron
lation to minimize the convection loss. Finally, at the irra- (Fe) (Adanez et al., 2012). The first reduction stage
diated surface the radiation flux dominates over the (Fe2O3 ? Fe3O4) is what is most often used in chemical
conduction loss. Therefore, the radiation flux at the surface looping combustion and is a comparatively fast process
is assumed to be equal to the input solar flux. This assump- (Adanez et al., 2012). Therefore, despite the study not
tion also leads to the following boundary condition for the being done for the exact same reaction utilized herein,
temperature gradient
@T s
¼0
@z z¼0
Table 1
Kinetic parameters used in computation model
(Moghtaderi and Song, 2010; Stehle et al., 2015).
where essentially the front surface is non-conducting.
For the other side of the solid (i.e., the exit), the bound- Kinetic parameter Value
ary condition is such that the surface is adiabatic or k 0;red 4e2 [mol0.2 m0.6 s1]
mathematically, Ea;red 25 [kJ/mol]
a 0.89
@T s k 0;oxd 4e3 [mol0.5 m0.5 s1]
¼0
@z z¼L Ea;oxd 47.3 [kJ/mol]
E.J. Sheu, A.F. Ghoniem / Solar Energy 125 (2016) 339–359 347
step that is actually being used can be seen as the rate- et al., 2013). For the particle size used in the experiment,
limiting step. Furthermore, it is assumed that only FeO is the oxygen carrier layer thickness should be between 26
formed during reduction and is not further reduced to Fe and 31 lm. The chosen oxygen carrier layer thickness
since there is always Fe3O4 available for reduction (oxygen (30 lm) falls within this range. In addition to porosity of
carrier not fully reduced during reactor operation) and the layer is chosen to be the same as the porosity used in
therefore, the further reduction of FeO to Fe should be the kinetics experiment (Moghtaderi and Song, 2010).
minimal due to reduction stages of the oxygen carrier. As this study was done at atmospheric pressure, there is
The kinetics from (Moghtaderi and Song, 2010) are one- no pressure dependence for the conversion rate. Based on
step extrinsic kinetics. The oxygen carrier is prepared by chemical looping combustion studies the pressure depen-
direct mixing of Fe2O3 and alumina powders. A paste dence can be written as (Abad et al., 2007)
was then created from this mixture before being dried
and calcined. The calcined paste was then pulverized in a k0
k 0;p ¼ ð23Þ
ball mill with particle sizes in the range of 90–106 lm Pa
(Moghtaderi and Song, 2010). The oxygen carrier has a where k 0;p is the pre-exponential factor at pressure P ; k 0 is
58% wt. active metal oxide content. The kinetic data was the pre-exponential factor at atmospheric pressure, and a
obtained using thermogravimetric analysis in the tempera- is the pressure coefficient determined for experiment. The
ture range of 800–950 °C at atmospheric pressure. Methane value of a used here in (Table 1) is taken from the study
is used as the reducing agent with Argon as the carrier gas done with CO/H2 as the reducing agent as there was no
(Moghtaderi and Song, 2010). data found for methane (Abad et al., 2007).
The kinetic parameters derived from this study are Including this pressure dependence yields the following
based on the shrinking core model for spherical grain expression for the conversion rate:
geometry. Therefore, the conversion rate is written as a
dX P 2
(Moghtaderi and Song, 2010) ¼3 ð1 X Þ3 k red C 0:2
g ð24Þ
dt Po
dX 2
¼ 3ð1 X Þ3 k red C 0:2
g ð21Þ where P is the pressure of the reactor and P o is the atmo-
dt
spheric pressure.
where X is the conversion of Fe3O4 to FeO (X ¼ 0 at start To give an idea of what times are needed for significant
of reduction), C g is the concentration of reducing gas oxygen carrier conversion, the oxygen carrier conversion
(methane), and k red is the Arrhenius rate constant defined versus time is shown in Fig. 5. Note that the plot shows
as the conversion for typical reformer operating conditions
Ea;red (i.e., pressurized condition).
k red ¼ k 0;red exp ð22Þ As discussed previously, the internal diffusion is mini-
RT
mal. Therefore, the conversion rate shown in Eq. (24) can
where R is the universal gas constant and k 0;red and Ea;red be adapted for this study using parameters pertaining to
are pre-exponential factor and activation energy, respec- the oxygen carrier and the stoichiometric coefficients for
tively, determined from experiments (Table 1). species i as follows
Since these kinetics are based on spherical particles and
mi dX
in the reformer the oxygen carrier is basically used with a Ri;s ¼ q voc doc ð25Þ
plate-like geometry (porous layer atop a support layer), moc oc dt
special attention must be paid as to how the geometry where mi is the stoichiometric coefficient of species i; mo c is
can affect the reaction rate. The two main ways that the the stoichiometric coefficient of the oxygen carrier, voc is
geometry can affect the reaction rate are external mass the volume fraction of the oxygen carrier, and doc is the
transfer and internal diffusion. The external mass transfer thickness of the oxygen carrier layer.
is already characterized by Eq. (2). The internal diffusion, As mentioned previously, CO2 is used as a carrier gas
however, is inherently part of the kinetics as the kinetics during the reactor operation. CO2 has the potential to react
are extrinsic. Moreover, the experiments were done assum- at the surface with the methane to produce more syngas
ing minimal internal diffusion effects. Therefore, in order to through dry reforming. Experimental work with dry
apply these kinetics to the solar steam redox reformer, the reforming and FeO as the catalyst has been done before
thickness of the porous layer (Table 3) must be chosen (Gokon et al., 2002). This study found that the presence
carefully to ensure that the internal diffusion is indeed min- of Fe3O4 had no effect on the dry reforming reaction sug-
imal. In order to ensure that the internal diffusion is similar gesting that the magnetite stage (the fully oxidized stage
between the spherical geometry utilized in the experiment in this reactor) is not a catalyst for the dry reforming reac-
and the plate like geometry used in the reformer, the con- tion (Gokon et al., 2002). However, there is still the poten-
version times for internal diffusion should be matched tial for the FeO that is produced during the reduction step
(Zhao et al., 2013). In order to match the two conversion to catalyze the dry reforming reaction. From the study
times, the thickness of the oxygen carrier layer needs to done in (Gokon et al., 2002), it was concluded that the
348 E.J. Sheu, A.F. Ghoniem / Solar Energy 125 (2016) 339–359
0.8 (Stehle et al., 2011) will be used. Note that the model can
T = 1000 K
always be improved with more accurate kinetics.
0.7
T = 1100 K As mentioned previously the kinetic study is done with
0.6
T = 1200 K an iron rod. The experiments were done in a temperature
T = 1300 K
range of 973–1473 K which matches well with the typical
Conversion X (−)
For the oxidation kinetics, the parameters from (Stehle The first order dependence for the conversion on the reac-
et al., 2015) are used. The kinetics study was done for the tion rate is chosen because the conversion of the oxygen
oxidation of an iron rod to magnetite with steam or CO2. carrier with time calculated from this expression (Fig. 6)
During the oxidation of pure iron with steam, wustite is of the same order of magnitude of other experiments
(FeO) is an intermediate formed during oxidation but then done with steam oxidation in the literature (Stamatiou
is readily oxidized to the final magnetite stage at the end of et al., 2010).
oxidation (Stehle et al., 2011). Therefore, even though the From the modified Hydrogen production rate, the gen-
current reformer oxidizes wustite to magnetite rather than eral reaction rate for species i to be used in the species con-
pure iron, these kinetics parameters can be seen as a con- servation modeling can be expressed as
servative estimate and still valid for use. However, it should Ri;s ¼ mi Xk oxd C 0:5 ð30Þ
g
be noted that this kinetic study is done with an iron rod
rather than a porous oxygen carrier layer on a support
which is what is modeled herein. This difference can be 3.6. Numerical validation
important in terms of reaction rates. There have been pre-
vious kinetic studies for doped ferrites or iron oxide based As there is no experimental data with this exact type of
porous structures (Mehdizadeh et al., 2012); however in solar reformer, the model is validated as follows: first, the
order to match the reactor design, which uses pure iron heat transfer modeling is validated by comparing model
oxide as the oxygen carrier, the kinetics parameters from results with literature results for a solar receiver used to
E.J. Sheu, A.F. Ghoniem / Solar Energy 125 (2016) 339–359 349
0.6
the temperature results are not necessarily representative
0.5 of what the reformer temperature results will be.
0.4 For the species conservation modeling, as mentioned
previously, the model results are compared to an
0.3
experimental study for dry reforming of methane in a
0.2 Ni–Cr–Al alloy foam absorber (Gokon et al., 2009). The
0.1
absorber is coated with c-Alumina support particles by
the washcoat method. Then Ruthenium is loaded on the
0 reformer via a drop wise application of RuCl3 and ethanol
0 1 2 3 4 5 6 7 8
Time (min) solution. The Ru loading is such that it is 30 wt% of the
mass of the support. The absorber is placed in a tubular
Fig. 6. Conversion of oxygen carrier versus time using modified rate quartz reactor which is within an electric furnace. The reac-
expression for oxidation (for typical reformer operating conditions).
tant gases are sent in from one end of the reactor through
the absorber and then the methane conversion is measured
heat up air (Wu et al., 2011) then second, the species con- at the other end of the reactor. The methane conversion is
servation modeling is validated by comparing to an exper- measured at four different temperatures: 600 °C, 650 °C,
imental study of a solar dry reformer at isothermal 700 °C, and 750 °C. The ratio of Methane to Carbon
conditions (Gokon et al., 2009). For both steps of the val- Dioxide is set to 1 and Argon is used as a carrier gas.
idation, model parameters are set to those used in the liter- The comparison of the methane conversion measured
ature studies. Thus, the parameters do not match with what experimentally and the methane conversion calculated
are used in the base case results that will be presented later. from the model is shown in Fig. 9. Note that channel
However, this validation does illustrate whether the model dimensions are estimated from pore properties in experi-
equations that are used are valid and whether they are ment. Also note that for this comparison, since the exper-
being numerically solved correctly. When presenting the imental study is for dry reforming and not redox
base case results, the reasonableness of those specific results reforming, the kinetics of the computational model are
will be discussed. changed to match those of the experimental study. For
For the heat transfer modeling validation, the model the experimental study, the kinetics are based on the Lang-
results are compared to the study done by Wu et al. muir–Hinshelwood model with the reaction rate dependent
(2011) (Fig. 7). Since this part is just for the heat transfer on partial pressures and various parameters determined
modeling, the species conservation modeling is not taken experimentally including rate constant and adsorption
into account. The Wu et al. study is done for turbulent equilibrium constants (Gokon et al., 2009). The experimen-
flow. Thus, to validate the model with the literature results, tal study is also done at isothermal conditions so the heat
the fluid temperature is taken as an input and the solid tem- transfer modeling is not taken into account during this part
perature is compared. Model parameters such as absorber of the validation.
length and heat transfer coefficient are taken from (Wu From Fig. 9 it can be seen that overall the methane con-
et al., 2011). The results are compared for several different version calculated from the model is consistent with the
inlet velocities. For completeness, the radiative flux calcu- experimental results from literature. There is a larger dis-
lated is also shown in Fig. 8. The radiation flux monoton- crepancy at the higher temperature likely because the kinet-
ically decreases along the channel and decays very quickly. ics parameters best matched experimental results in the
For the most part, the model results match reasonably well 600–700 °C range (Gokon et al., 2009), and thus a larger
with the literature results (Fig. 7). There are some discrep- discrepancy at the higher temperature would be expected.
ancies at the front face of the absorber. This discrepancy Thus, since the results match reasonably well, the species
can be due to the channel width/solid thicknesses values conservation model can be considered valid.
used. As mentioned previously, the literature study is done Overall, the literature studies used from comparison are
for a porous media; therefore, the parameters for channel not necessarily representative of the solar steam redox
size used in the model are estimated from the pore size reformer operation; however, they do allow for validation
given for the literature study. of the model equations themselves. With the differing
The difference in these parameters can affect not only the parameters used in the base case results, the reasonableness
front surface temperature (due to differing solid surface of those results will also be discussed later.
area) but also the other heat transfer processes (i.e., con- Now that the model have been validated, the base case
vection). Overall, the model results match reasonably well simulation results and their validity will be discussed.
350 E.J. Sheu, A.F. Ghoniem / Solar Energy 125 (2016) 339–359
550 550
Temperature (K)
Temperature (K)
500 500
450 450
400 400
350 350
300 300
0 0.01 0.02 0.03 0.04 0.05 0 0.01 0.02 0.03 0.04 0.05
Position (m) Position (m)
Inlet Velocity = 1.30m/s Inlet Velocity = 1.08m/s
700 700
Temperature (K)
Temperature (K)
600 600
500 500
T - Lit
f
T - Lit
s
400 400 T - Model
s
300 300
0 0.01 0.02 0.03 0.04 0.05 0 0.01 0.02 0.03 0.04 0.05
Position (m) Position (m)
Fig. 7. Comparison between model and literature results for validation of energy conservation numerics (Wu et al., 2011).
600 24
22
500 Literature
20 Model
Methane Conversion (%)
Energy Flux (kW/m 2 )
400 18
16
300
14
200 12
10
100
8
0
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0.045 0.05 6
600 650 700 750
Position (m)
Temperature ( oC)
Fig. 8. Radiative flux along channel for heat transfer modeling validation.
Fig. 9. Comparison between model and literature results for validation of
species conservation numerics (Gokon et al., 2009).
4. Base case results
taken for low porosity alumina/SiC solids or honeycomb
For the base case, both an alumina absorber and SiC (Skocypec et al., 1994; Pitz-Paal et al., 1997). Note that
absorber are simulated. The alumina absorber is simulated as discussed previously, the absorber front face is assumed
to compare model results to previous experimental work in to be coated such that the emissivity of the absorber is very
order to examine whether the model results agree with small which means the surface radiation losses are minimal.
trends observed previously. The SiC absorber is simulated The reactor design and operating parameters used for
to determine if a material with a higher thermal conductiv- the base case results are shown in Table 3. The concen-
ity is beneficial to overall reactor performance. The proper- trated solar irradiation value (400 kW/m2) was chosen
ties of the oxygen carrier and absorber/support materials based on values obtained for an actual reformer integrated
are summarized in Table 2. The optical properties are with a parabolic dish (Buck et al., 1991). As discussed
E.J. Sheu, A.F. Ghoniem / Solar Energy 125 (2016) 339–359 351
Table 2 0.18
Oxygen carrier and support material properties.
Parameter Alumina SiC 0.16
u @ 0.1*L
Conductivity of solid (k s ) 30 [W/m/K] 100 [W/m/K] u @ 0.5*L
Density of solid (qs ) 3900 [kg/m3] 3200 [kg/m3] 0.14 u @ outlet
Velocity (m/s)
Volume fraction of oxygen carrier (voc ) 0.45 0.45 0.12
Absorption coefficient (ra ) 140 [m1] 50 [m1]
Scattering coefficient (rs ) 160 [m1] 151 [m1]
0.1
Reduction (With Solar) Oxidation (No Solar)
0.08
Table 3
Design and operating parameters used for base case results.
0.06
Parameter Value
Absorber diameter (D) 40 [cm] 0.04
Channel size (d) 4 [mm] 0 50 100 150 200 250 300 350
Channel length (L) 50 [mm] Time (s)
Thickness of solid (ds ) 1 [mm]
Thickness of oxygen carrier layer (doc ) 30 [lm] Fig. 10. Gas velocity profile throughout cycle at different points along the
Cycle time (scyc ) 6 [min] channel for Alumina absorber (reduction: t = 0–180 s, oxidation: t = 180–
Volume fraction of fuel at inlet 0.5 360 s).
Fuel flow velocity (ured ) 0.1 [m/s]
Steam flow velocity (uoxd ) 0.04 [m/s] where gCH 4 is the cycle average methane conversion, m_ CH 4 is
Pressure (P) 20 [bar]
Inlet gas temperature (T in ) 800 [K]
methane flow rate at the inlet and outlet of the channel,
Solar irradiation (I o ) 400 [kW/m2] and the integration over time is for only the reduction step.
The oxygen carrier conversion is defined as
previously the flow rate during reduction is determined C oc;FeO
X ¼ ð32Þ
such that the chemical energy of the inlet fuel flow is C oc;FeO þ C oc;Fe3 O4
100 kW. The fuel flow rate is such that the total solar input where X is the oxygen carrier conversion, C oc;FeO and
is approximately 10% of the total fuel and solar input. The C oc;Fe3 O4 are the molar concentrations of the two states of
flow rate during oxidation is chosen such that all the oxy- the oxygen carrier (wustite and magnetite). X is one when
gen carrier can be fully oxidized. The cycle time is chosen the oxygen carrier is fully reduced and is zero when it is
such that there is a base case oxygen carrier conversion fully oxidized.
of approximately 30% (Fig. 17). Note that the cycle time The solar utilization efficiency is defined as
in essence determines the oxygen carrier conversion, and
for this base case, the oxygen carrier conversion is on the DH p jT p
gsu ¼ ð33Þ
lower end since that is not the objective of this reactor. Fur- DH r jT r þ Qsolar
thermore, very high or complete oxygen carrier can lead to
material degradation over time. The inlet gas temperature
(800 K) is chosen based on operation of reformer within 1500
t = 0s t = 60s
400 400
-200 0
0 0.01 0.02 0.03 0.04 0.05 0 0.01 0.02 0.03 0.04 0.05
Position (m) Position (m)
t = 120s t = 180s
400 400
Energy Flux (kW/m 2 )
200 200
100 100
0 0
0 0.01 0.02 0.03 0.04 0.05 0 0.01 0.02 0.03 0.04 0.05
Position (m) Position (m)
t = 240s t = 300s
10
Energy Flux (kW/m 2 )
0
0
-10 -20
-20
-40 Radiation Flux
-30
Conduction Flux
-40 -60
0 0.01 0.02 0.03 0.04 0.05 0 0.01 0.02 0.03 0.04 0.05
Position (m) Position (m)
@G
Fig. 12. Radiation (qr ¼ K1ext @z
) and conduction flux (qcond ¼ k s @T
@z
) along channel throughout cycle for Alumina absorber.
1330 1550
1500
1320 300 kW/m 2
2
1450 400 kW/m
2
500 kW/m
Temperature (K)
Temperature (K)
1310
1400
1300 1350
1300
1290
1250
1280
1200
1270 1150
0 20 40 60 80 100 120 140 160 180 0 20 40 60 80 100 120 140 160 180
Time (s) Time (s)
Fig. 13. Reactor gas outlet temperature throughout reduction step (with Fig. 14. Average reactor temperature throughout the reduction step for
solar radiation) for Alumina absorber. different concentrated solar fluxes.
where DH p jT p and DH r jT r is the enthalpy of the reformer actually used within the reactor whether through heating
or reforming.
products and reactants at product temperature T p and
The chemical efficiency is defined as
reactant temperature T r , respectively, and Qsolar is the
amount of solar energy entering reactor. The solar utiliza- DH op
gchem ¼ ð34Þ
tion efficiency represents how much of the solar energy is DH or þ Qsolar
E.J. Sheu, A.F. Ghoniem / Solar Energy 125 (2016) 339–359 353
t = 0s t = 60s
1400 1600
Temperature (K)
Temperature (K)
1400
1200
1200
1000
1000
800 800
0 0.01 0.02 0.03 0.04 0.05 0 0.01 0.02 0.03 0.04 0.05
Position (m) Position (m)
t = 120s t = 180s
1600 1600
Temperature (K)
Temperature (K)
1400 1400
1200 1200
1000 1000
800 800
0 0.01 0.02 0.03 0.04 0.05 0 0.01 0.02 0.03 0.04 0.05
Position (m) Position (m)
t = 240s t = 300s
1400 1400
Temperature (K)
Temperature (K)
1200 1200
T gas
1000 1000 T
solid
800 800
0 0.01 0.02 0.03 0.04 0.05 0 0.01 0.02 0.03 0.04 0.05
Position (m) Position (m)
Fig. 15. Comparison of solid and gas temperature along the channel for different times throughout cycle for Alumina absorber (reduction: t = 0–180 s,
oxidation: t = 180–360 s).
58 0.45
t = 0s
0.4 t = 60s
57 t = 120s
Oxygen Carrier Conversion (-)
t = 180s
0.35
Methane Conversion (%)
56
0.3
55
0.25
54
0.2
53
0.15
52 0.1
51 0.05
0
50 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0.045 0.05
0 20 40 60 80 100 120 140 160 180
Time (s) Position (m)
Fig. 16. Instantaneous methane conversion during reduction step for Fig. 17. Oxygen carrier conversion along channel during reduction step
Alumina absorber. for Alumina absorber.
0.12 variation with time than the outlet of the channel. During
reduction, the radiation/conductive heat transfer (which
0.1 Reduction (With Solar) Oxidation (No Solar) raises the temperature) dominates over the chemical reac-
tion energy transfer (which lowers the temperature due to
0.08 the endothermic nature of the reaction) near the inlet of
the channel. Moving along the channel, the chemical reac-
0.06
tion energy transfer (the thermal energy required for the
endothermic reduction reaction) starts to dominate. Also
0.04
0 50 100 150 200 250 300 350 note that the spatial temperature variation decreases along
Time (s) the channel (i.e., temperature variation within first half of
channel is much larger than temperature variation within
Fig. 18. Gas velocity profile throughout cycle at different points along the
second half). This matches what has been observed in solar
channel for SiC absorber (reduction: t = 0–180 s, oxidation: t = 180–
360 s). reformer experimental studies where most of the tempera-
ture variation occurred in the front half of the absorber
(Buck et al., 1991; Skocypec et al., 1994). The decrease in
1420 variation also shows that the adiabatic boundary condition
T solid @ inlet at the back surface is reasonable. During oxidation, there is
1400 T solid @ 0.5*L
no longer any solar irradiation. The reactor is mainly
T @ outlet
1380
solid cooled by convection because the oxidation reaction is only
slightly exothermic. Moreover, the front face temperature
Temperature (K)
t = 0s t = 60s
400 400
-200 0
0 0.01 0.02 0.03 0.04 0.05 0 0.01 0.02 0.03 0.04 0.05
Position (m) Position (m)
t = 120s t = 180s
400 400
Energy Flux (kW/m 2 )
200 200
100 100
0 0
0 0.01 0.02 0.03 0.04 0.05 0 0.01 0.02 0.03 0.04 0.05
Position (m) Position (m)
t = 240s t = 300s
10 10
Energy Flux (kW/m 2 )
Energy Flux (kW/m 2)
Radiation Flux
5 5 Conduction Flux
0 0
-5 -5
-10 -10
0 0.01 0.02 0.03 0.04 0.05 0 0.01 0.02 0.03 0.04 0.05
Position (m) Position (m)
@G
Fig. 20. Radiation (qr ¼ K1ext @z
) and conduction flux (qcond ¼ k s @T
@z
) along channel throughout cycle for SiC absorber.
et al., 1994; Muir et al., 1994). The concentrated solar flux operation. This case will be discussed later when discussing
used in the base case (400 kW/m2) falls within the range the SiC absorber.
used in the experimental study, and the temperature profile A comparison of solid and gas temperature along the
in Fig. 11 shows that the maximum temperature variation channel length throughout the cycle is also shown in
within the solid during the reduction step is around 150 Fig. 15. At the channel outlet, the gas temperature is essen-
degrees which falls within this range. In addition, for the tially equal to the solid temperature. Fig. 15 also shows
same range of solar fluxes the outlet reactor temperature that during reduction the maximum solid temperature is
is between 800 and 1100 °C (Skocypec et al., 1994; Muir at the front face (where the solar radiation is) and during
et al., 1994) which also matches well with the results oxidation the maximum solid temperature is within the
(Fig. 13). channel.
As the amount of concentrated solar flux can greatly Moving onto conversion, the instantaneous methane
affect the temperature results, the average reactor tempera- conversion during the entire reduction step is shown in
ture calculated from the model for the range of solar fluxes Fig. 16. Similar to the cycle average methane conversion,
used in previous experimental studies is shown in Fig. 14. the instantaneous methane conversion is defined as
As expected, as the amount of solar flux increases, the reac- R
m_ CH 4 ;out dt
tor temperature increases. Again these reactor tempera- gCH 4 ¼ 1 R1s ð35Þ
tures are of the same order of magnitude as previous m_
1s CH 4 ;in
dt
experimental results (Skocypec et al., 1994; Muir et al., where gCH 4 is the instantaneous methane conversion, m_ CH 4
1994). is methane flow rate at the inlet and outlet of the channel.
Since the temperature variation along the channel solid Essentially the instantaneous methane conversion is the
is quite high, absorber material degradation during solar conversion corresponding to one second of time through-
reforming tests has been observed (Buck et al., 1991; out the reduction step rather than a cycle average.
Muir et al., 1994). Therefore, a higher thermal conductivity The instantaneous methane conversion decreases during
absorber material might be beneficial for long term reactor the reduction step as the reaction rate decreases due to the
356 E.J. Sheu, A.F. Ghoniem / Solar Energy 125 (2016) 339–359
t = 0s t = 60s
1300 1400
Temperature (K)
Temperature (K)
1200
1200
1100
1000
1000
900
800 800
0 0.01 0.02 0.03 0.04 0.05 0 0.01 0.02 0.03 0.04 0.05
Position (m) Position (m)
t = 120s t = 180s
1400 1600
Temperature (K)
Temperature (K)
1400
1200
1200
1000
1000
800 800
0 0.01 0.02 0.03 0.04 0.05 0 0.01 0.02 0.03 0.04 0.05
Position (m) Position (m)
t = 240s t = 300s
1400 1400
Temperature (K)
Temperature (K)
1200 1200
T
gas
1000 1000 T solid
800 800
0 0.01 0.02 0.03 0.04 0.05 0 0.01 0.02 0.03 0.04 0.05
Position (m) Position (m)
Fig. 21. Comparison of solid and gas temperature along the channel for different times throughout cycle for SiC absorber (reduction: t = 0–180 s,
oxidation: t = 180–360 s).
57 0.45
t = 0s
t = 60s
0.4
56 t = 120s
Oxygen Carrier Conversion (-)
t = 180s
0.35
Methane Conversion (%)
55
0.3
54 0.25
0.2
53
0.15
52
0.1
51 0.05
0
50 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0.045 0.05
0 20 40 60 80 100 120 140 160 180
Position (m)
Time (s)
Fig. 23. Oxygen carrier conversion along channel during reduction step
Fig. 22. Instantaneous methane conversion during reduction step for SiC
for SiC absorber.
absorber.
5. Conclusion References
Overall, a design for a solar receiver-reactor for steam Abad, A., Adánez, J., Garcı́a-Labiano, F., de Diego, L.F., Gayán, P.,
redox reforming has been presented. The receiver-reactor Celaya, J., 2007. Mapping of the range of operational conditions for
consists of a dumbbell shape absorber system that has Cu-, Fe-, and Ni-based oxygen carriers in chemical-looping combus-
tion. Chem. Eng. Sci. 62 (1–2), 533–549, Fluidized Bed Applications.
two distinct absorbers. This absorber system setup allows
Adanez, J., Abad, A., Garcia-Labiano, F., Gayan, P., de Diego, L.F.,
for the switching between reduction and oxidation steps 2012. Progress in chemical-looping combustion and reforming tech-
without having to constantly change inlet streams to the nologies. Prog. Energy Combust. Sci. 38 (2), 215–282.
reactor. In addition, at any point in time only one solar Anikeev, V., Bobrin, A., Ortner, J., Schmidt, S., Funken, K.H., Kuzin, N.,
absorber is irradiated by the solar energy (during the reduc- 1998. Catalytic thermochemical reactor/receiver for solar reforming of
natural gas: design and performance. Solar Energy 63 (2), 97–104.
tion step). The receiver reactor is designed such that the
Ávila Marı́n, A.L., 2011. Volumetric receivers in solar thermal power
inlet connections do not interfere with the solar window. plants with central receiver system technology: a review. Solar Energy
A computational model that can help aid in design and 85 (5), 891–910.
evaluation of reformer performance has also been devel- Ben-Zvi, R., Karni, J., 2007. Simulation of a volumetric solar reformer. J.
oped. The model accounts for mass and heat transfer Solar Energy Eng. 129 (2), 197–204.
between gas and solid phase as well as radiation transport Berman, A., Karn, R.K., Epstein, M., 2007. Steam reforming of methane
on a Ru/Al2O3 catalyst promoted with Mn oxides for solar hydrogen
and conduction through the solid. The model is considered production. Green Chem. 9, 626–631.
to be solved when the reformer has reached a cyclic steady Buck, R., Muir, J.F., Hogan, R.E., 1991. Carbon dioxide reforming of
state operation. Using this developed model, two base case methane in a solar volumetric receiver/reactor: the CAESAR project.
designs were simulated – one with an Alumina absorber Solar Energy Mater. 24 (1–4), 449–463.
Dahl, J.K., Weimer, A.W., Lewandowski, A., Bingham, C., Bruetsch, F.,
and one with a SiC absorber. The base case results yielded
Steinfeld, A., 2004. Dry reforming of methane using a solar-thermal
reasonable results as compared to previous experimental aerosol flow reactor. Indust. Eng. Chem. Res. 43 (18), 5489–5495.
solar reformer studies. The results show that both absor- Gokon, N., Oku, Y., Kaneko, H., Tamaura, Y., 2002. Methane reforming
bers strongly absorb the solar radiation and most of the with CO2 in molten salt using FeO catalyst. Solar Energy 72 (3), 243–250.
radiative heat transfer occurs in the front half of the absor- Gokon, N., Osawa, Y., Nakazawa, D., Kodama, T., 2009. Kinetics of CO2
ber. Therefore, increasing the absorber (channel) length reforming of methane by catalytically activated metallic foam absorber
for solar receiver-reactors. Int. J. Hydrogen Energy 34 (4), 1787–1800.
only contributes to increasing the methane conversion. Gundlapally, S.R., Balakotaiah, V., 2011. Heat and mass transfer
The alumina absorber experiences slightly higher reactor correlations and bifurcation analysis of catalytic monoliths with
temperatures (especially front face temperatures) than the developing flows. Chem. Eng. Sci. 66 (9), 1879–1892.
SiC absorber; however, average reactor temperature for International Energy Outlook, 2013. Tech. Rep., U.S. Energy Information
both absorbers are approximately the same and thus both Administration.
Kodama, T., Ohtake, H., Matsumoto, S., Aoki, A., Shimizu, T.,
achieve very similar methane conversions. The solar utiliza- Kitayama, Y., 2000. Thermochemical methane reforming using a
tion efficiency is also essentially the same for both absor- reactive WO3/W redox system. Energy 25 (5), 411–425.
bers as the average reactor temperatures are similar and Levy, M., Rubin, R., Rosin, H., Levitan, R., 1992. Methane reforming by
therefore the losses associated with heating the carrier gas direct solar irradiation of the catalyst. Energy 17 (8), 749–756.
Maria, G.D., Tiberio, C., D’Alessio, L., Piccirilli, M., Coffari, E.,
are comparable. Furthermore, the results show that a
Paolucci, M., 1986. Thermochemical conversion of solar energy by
absorber material with a higher thermal conductivity might steam reforming of methane. Energy 11 (8), 805–810.
be better suited for this type of solar reformer in terms of Mehdizadeh, A.M., Klausner, J.F., Barde, A., Rahmatian, N., Mei, R.,
long term operation. 2012. Investigation of hydrogen production reaction kinetics for an
As discussed previously, many different parameters can iron-silica magnetically stabilized porous structure. Int. J. Hydrogen
effect reformer performance. For instance, while increasing Energy 37 (18), 13263–13271.
Modest, M.F., 2013. Radiative Heat Transfer. Academic Press.
the channel length can lead to higher methane conversion Moghtaderi, B., Song, H., 2010. Reduction properties of physically mixed
due to longer residence times, it can increase the tempera- metallic oxide oxygen carriers in chemical looping combustion. Energy
ture variation which is detrimental for long term operation. Fuels 24 (10), 5359–5368.
Therefore, future work will include performing sensitivity Muir Jr., J.F., Hogan, R.E., Skocypec, R.D., Buck, R., 1994. Solar
reforming of methane in a direct absorption catalytic reactor on a
analysis to determine which design/operating parameters
parabolic dish: I – Test and analysis. Solar Energy 52 (6), 467–477.
have the largest impact on reactor performance. By identi- Petrasch, J., Steinfeld, A., 2007. Dynamics of a solar thermochemical
fying which parameters are most important, the design of reactor for steam-reforming of methane. Chem. Eng. Sci. 62 (16),
the reactor can be further improved. 4214–4228.
E.J. Sheu, A.F. Ghoniem / Solar Energy 125 (2016) 339–359 359
Petrasch, J., Wyss, P., Steinfeld, A., 2007. Tomography-based monte carlo Stehle, R., Bobek, M., Hooper, R., Hahn, D., 2011. Oxidation reaction
determination of radiative properties of reticulate porous ceramics. J. kinetics for the steam-iron process in support of hydrogen production.
Quant. Spectrosc. Radiative Transfer 105 (2), 180–197. Int. J. Hydrogen Energy 36 (22), 15125(11).
Pitz-Paal, R., Hoffschmidt, B., Bhmer, M., Becker, M., 1997. Experimen- Stehle, R., Bobek, M., Hahn, D., 2015. Iron oxidation kinetics for {H2}
tal and numerical evaluation of the performance and flow stability of and {CO} production via chemical looping. Int. J. Hydrogen Energy
different types of open volumetric absorbers under non-homogeneous 40 (4), 1675–1689.
irradiation. Solar Energy 60 (3–4), 135–150. Steinfeld, A., Kuhn, P., Karni, J., 1993. High-temperature solar thermo-
Sheu, E.J., Ghoniem, A.F., 2014. Redox reforming based, integrated chemistry: production of iron and synthesis gas by Fe3O4-reduction
solar-natural gas plants: reforming and thermodynamic cycle effi- with methane. Energy 18 (3), 239–249.
ciency. Int. J. Hydrogen Energy 39 (27), 14817–14833. Steinfeld, A., Brack, M., Meier, A., Weidenkaff, A., Wuillemin, D., 1998.
Sheu, E.J., Mitsos, A., 2013. Optimization of a hybrid solar-fossil fuel A solar chemical reactor for co-production of zinc and synthesis gas.
plant: solar steam reforming of methane in a combined cycle. Energy Energy 23 (10), 803–814.
51, 193–202. West, D.H., Balakotaiah, V., Jovanovic, Z., 2003. Experimental and
Sheu, E.J., Mitsos, A., Eter, A.A., Mokheimer, E.M.A., Habib, M.A., Al- theoretical investigation of the mass transfer controlled regime in
Qutub, A., 2012. A review of hybrid solar-fossil fuel power generation catalytic monoliths. Catal. Today 88 (12), 3–16, Environmental
systems and performance metrics. ASME J. Solar Energy Eng. 134 (4), Catalysis and Reaction Engineering.
041006:1–041006:17. Wörner, A., Tamme, R., 1998. CO2 reforming of methane in a solar driven
Skocypec, R.D., Hogan Jr., R.E., Muir, J.F., 1994. Solar reforming of volumetric receiver reactor. Catal. Today 46 (2–3), 165–174.
methane in a direct absorption catalytic reactor on a parabolic dish: II Wu, Z., Caliot, C., Flamant, G., Wang, Z., 2011. Coupled radiation and
– Modeling and analysis. Solar Energy 52 (6), 479–490. flow modeling in ceramic foam volumetric solar air receivers. Solar
Stamatiou, A., Loutzenhiser, P., Steinfeld, A., 2010. Solar syngas Energy 85 (9), 2374–2385.
production from H2O and CO2 via two-step thermochemical cycles Zhao, Z., Chen, T., Ghoniem, A.F., 2013. Rotary bed reactor for
based on Zn/ZnO and FeO/Fe3O4 redox reactions: kinetic analysis. chemical-looping combustion with carbon capture. Part 1: Reactor
Energy Fuels 24 (4), 2716–2722. design and model development. Energy Fuels 27 (1), 327–343.