Test operation of a 100 kW pilot plant for solar hydrogen

production from water on a solar tower

M. Roeb
a,
*
, J.-P. Sack
a
, P. Rietbrock
a
, C. Prahl
b
, H. Schreiber
a
, M. Neises
a
, L. de Oliveira
a
,
D. Graf
a
, M. Ebert
b
, W. Reinalter
b
, M. Meyer-Gru nefeldt
b
, C. Sattler
a
, A. Lopez
c
,
A. Vidal
c
, A. Elsberg
d
, P. Stobbe
d
, D. Jones
e
, A. Steele
e
, S. Lorentzou
f
, C. Pagkoura
f
,
A. Zygogianni
f
, C. Agraotis
f
, A.G. Konstandopoulos
f
a
Deutsches Zentrum fu r Luft- und Raumfahrt e.V. (DLR), Institute of Technical Thermodynamics, Solar Research, Linder Ho he, 51147 Ko ln, Germany
b
Deutsches Zentrum fu r Luft- und Raumfahrt e.V. (DLR), Institute of Technical Thermodynamics, Solar Research, Plataforma Solar de Almera,
Carretera de Sene s s/n, km 5, E-04200 Tabernas, Spain
c
Centro de Investigaciones Energe ticas, Medioambientales y Tecnolo gicas (CIEMAT), Plataforma Solar de Almera, Carretera de Sene s s/n,
km 5, E-04200 Tabernas, Spain
d
Stobbe Tech Ceramics, Malmmosevej 19c, 2840 Holte, Denmark
e
Johnson-Matthey Technology Centre, Blounts Court, Sonning Common, Reading RG4 9NH, UK
f
Aerosol and Particle Technology Lab./Centre for Research and Technology Hellas, 6th km Harillaou-Thermi Road, 57001 Thermi-Thessaloniki, Greece
Received 30 November 2009; received in revised form 5 March 2010; accepted 19 April 2010
Available online 11 May 2010
Communicated by: Associate Editor Tatsuya Kodama
In memoriam: Peter Rietbrock
Abstract
The present work describes the realisation and successful test operation of a 100 kW pilot plant for two-step solar thermo-chemical
water splitting on a solar tower at the Plataforma Solar de Almer a, which aims at the demonstration of the feasibility of the process on a
solar tower platform under real conditions. The process applies multi-valent iron based mixed metal oxides as reactive species which are
coated on honeycomb absorbers inside a receiverreactor. By the introduction of a two-chamber reactor it is possible to run both process
concepts in parallel and thus, the hydrogen production process in a quasi-continuous mode. In summer 2008 an exhaustive thermal qual-
ication of the pilot plant took place, using uncoated ceramic honeycombs as absorbers. Some main aspects of these tests were the devel-
opment and validation of operational and measurement strategy, the gaining of knowledge on the dynamics of the system, in particular
during thermal cycling, the determination of the controllability of the whole system, and the validation of practicability of the control
concept. The thermal tests enabled to improve, to rene and nally to prove the process strategy and showed the feasibility of the control
concept implemented. It could be shown that rapid changeover between the modules is a central benet for the performance of the
process.
In November of 2008 the absorber was replaced and honeycombs coated with redox material were inserted. This allowed carrying out
tests of hydrogen production by water splitting. Several hydrogen production cycles and metal oxide reduction cycles could be run with-
out problems. Signicant concentrations of hydrogen were produced with a conversion of steam of up to 30%.
2010 Elsevier Ltd. All rights reserved.
Keywords: Thermochemical cycle; Hydrogen; Ferrites; Receiverreactor; Solar tower
0038-092X/$ - see front matter 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.solener.2010.04.014
*
Corresponding author. Tel.: +49 2203 601 2673.
E-mail address: martin.roeb@dlr.de (M. Roeb).
www.elsevier.com/locate/solener
Available online at www.sciencedirect.com
Solar Energy 85 (2011) 634644
1. Introduction
Emission-free hydrogen production from water is possi-
ble by solar-powered thermo-chemical cycles. A two-step
cycle can be based on a metal oxide redox pair system,
which can split water molecules by abstracting oxygen
atoms and reversibly incorporating them into its lattice.
During the rst step of this cycle (the water splitting) the
reduced and therefore activated material usually the
lower-valence oxide of a metal exhibiting multiple oxida-
tion states is oxidised by taking oxygen from water and
producing hydrogen according to the reaction:
MO
reduced
H
2
O !MO
oxidised
H
2
1
In the next step (the regeneration step) the material is
reduced again (regenerated), setting some of its lattice
oxygen free according to the reaction:
MO
oxidised
!MO
reduced
1=2O
2
2
The redox pairs used for this cycle are ferrites and mixed
metal oxides with iron as the main component (Charvin
et al., 2007; Ishihara et al., 2008; Kodama et al., 2006;
Allendorf et al., 2006; Han et al., 2007; Steinfeld et al.,
1999, 2002; Sturzenegger and Nu esch, 1999; Gokon
et al., 2008).
The HYDROSOL research group has introduced the
concept of monolithic honeycomb solar reactors for per-
forming these redox pair cycles for the production of
hydrogen from the splitting of steam using solar energy
(Agraotis et al., 2005) inspired from the well-known auto-
mobile catalytic converters. The basic idea proposed, devel-
oped and demonstrated within the HYDROSOL and
HYDROSOL-II projects was to combine a monolithic
honeycomb support structure capable of achieving high
temperatures when heated by concentrated solar radiation,
with a redox pair system suitable for the performance of
water dissociation and for regeneration at these tempera-
tures. With this conguration, the complete operation of
the whole process (water splitting and regeneration of the
metal oxide) can be achieved by a single solar energy con-
verter. In addition, by using a two-chamber reactor the two
steps of the cycle can be performed in parallel and therefore
a hydrogen production process in a quasi-continuous mode
can be achieved (Roeb et al., 2006a). After the HYDRO-
SOL reactor, several other dierent approaches on solar
receiver concepts for using mixed ferrites have been pro-
posed in the scientic literature (Kaneko et al., 2006; Kod-
ama and Gokon, 2007; Diver et al., 2006, 2008). However
the essential advantage of the HYDROSOL reactor over
the other reactor concepts above is that, in contrast to
those which in order to meet the technical requirements
of the thermochemical cycle made up of two process steps
performed at dierent temperature levels with dierent heat
demands either use two dierent reactors or employ reactor
parts that are continuously rotating between a higher- and
a lower-temperature zone the HYDROSOL reactor con-
tains neither moving parts nor moving solid particles. The
dierent heat demands of each step are realised in the
HYDROSOL process not by moving the reactors but by
adjusting the ux density on each reactor module when
the status of the cycle is switched from regeneration to
splitting and vice versa by re-alignment of a part of the
solar concentrators. The proof-of-principle of this concept
was validated with iron-based oxides coated on ceramic
monolithic structures, which were placed inside a solar
receiverreactor. After successful experimental demonstra-
tion of several cycles of alternating hydrogen and oxygen
production and elaboration of process strategies, presented
in previous contributions (Roeb et al., 2006b, 2008), the
present work describes the realisation and successful test
operation of a 100 kW pilot plant on a solar tower, which
aimed at the demonstration of the feasibility of the process
under the real operation conditions of a solar tower
platform.
2. Pilot plant
2.1. Layout and operation strategy
The successful testing and the suitable behaviour of the
laboratory-scale reactor in the earlier stages of the project
have provided a basis for the design of the pilot reactor.
Whereas the general reactor concept was kept, the scale-
up of the reactor from 10 kW
th
to 100 kW
th
was basically
realised by increasing the absorber surface. The reactor is
set up of two reactor modules to run both steps of the cycle
in parallel (photograph in Fig. 1a). Fig. 1b represents a
schematic explosion view of the reactor to demonstrate
its main components. Three times three (a total of 9) indi-
vidual pieces of square-shaped monolithic honeycomb
absorbers made of siliconized silicon carbide (siSiC) (Fend
et al., 2004; Agraotis et al., 2007), each with a dimension
of 146 146 mm were assembled as one absorber module.
The honeycomb absorbers exhibit a length of 60 mm and
are mounted to form a square with slightly shaped concave
surface. They are kept in place by springs applying force
from the rear side. Into the center channel of each honey-
comb monolith a thermocouple has been inserted from
the back to monitor the absorber temperature near the irra-
diated front face. The end of the thermocouple is located
5 mm behind the front face and is shielded from direct
solar radiation. To additionally monitor the temperature
distribution on the receiver surface a thermocamera is
installed in a distance of about 10 m on an arm in front
of the receiver. With dierent lters it is possible to mea-
sure the quartz glass window temperature and the temper-
ature on the surface of the absorber inside the reactor
behind the quartz glass window.
The feed gas is fed in through channels in the outer part
of the reactor surrounding the hot core. Product and
sweep gases are collected in one central exhaust pipe
attached to the rear part of the housing. Each module of
the reactor is equipped with a quartz window xed by a
water-cooled window frame. The distance between the cen-
M. Roeb et al. / Solar Energy 85 (2011) 634644 635
ters of the two modules is 1.3 m so that the dierent and
homogeneous ux distributions required for each process
step can be realised separately on each module without
aecting the other one. The pilot reactor has been installed
on the tower of the so-called Small-Solar-Power-System
(SSPS) solar tower plant (Fig. 2) of the Plataforma Solar
de Almer a (PSA), Spain. The heliostat eld of this solar
facility is able to provide about 1.5 MW
th
(Geyer and
Schiel, 1987). Thus, for the pilot plant only a part of the
heliostat eld is needed.
For powering the two modules of the reactor with dier-
ent solar ux it is necessary to divide the heliostat eld into
dierent parts and to actuate those separately (Roeb et al.,
2008, Fig. 3). The heliostats (of the so-called Martin Mar-
ietta type), consist of 12 facets each with a total surface of
39.3 m
2
. The power supply of each heliostat is realised
(a)
(b)
housing
Quartz
pane
sealing
Inner housing
absorber
Product gas duct
Gas
inlet/outlet
Fig. 1. (a) 100 kW
th
pilot reactor and (b) explosion view of reactor design.
E Ex xp pe er ri im me en nt ta al l P Pl la at tf fo or rm m
Fig. 2. SSPS heliostat eld and tower.
636 M. Roeb et al. / Solar Energy 85 (2011) 634644
independently by an own photovoltaic cell. All in all 93
individual-tracking heliostats are available with an average
clean reectivity of 87% concentrating the power at the
aperture. During the splitting process, steam is fed in from
a steam generator into the so-called east reactor chamber
operating at 800 C. At the same time, nitrogen as ushing
gas is fed into the west reactor chamber operating at
1200 C, in order to release the oxygen from the metal
oxide redox system. After a half-cycle of 2030 min, a part
of the heliostats focus is moved from the west to the east
chamber to realise the necessary temperature increase up
to 1200 C to perform the regeneration of the redox system
in the east chamber, whereas the west chamber is cooled
down to 800 C to proceed with the splitting of water at
that temperature.
2.2. Hardware
Besides the solar part and all peripheral units for gas
and steam supply and feeding, pre-heating, and product
gas piping and treatment, the system additionally includes
several devices for analytics. Those are besides thermocou-
ples at various positions of the absorber/reactor, a moving
bar ux measurement system, an infrared camera to mon-
itor alternatively the surface temperature of the absorber or
of the quartz window, a mass spectrometer and a gas chro-
matograph including pumping and gas cooler to enable the
monitoring of the hydrogen produced. Fig. 4 gives a sche-
matic overview on the peripheral hardware of the pilot
plant.
The feed streams of steam and nitrogen are adjusted by
mass ow controllers. Up- and downstream the controllers
a temperature and pressure measurement probe is installed.
The steam generator is equipped with a reservoir including
charging level sensors. Three gas pre-heaters are installed,
two for heating the nitrogen feed stream and one for hold-
ing the temperature of the nitrogen/water steam mixture
above 110 C in order to avoid condensation. For the
switch between hydrogen production and metal oxide
regeneration and vice versa magnetic valves are used to
alternate the ow between the two modules.
Temperature measurements are performed at the inlet
and at the outlet of the reactor. In the exhaust line probes
for the gas chromatograph (GC) and for the mass spec-
trometer (MS) measurements are integrated. There, a part
of the product gas is sucked from the reactor o-gas stream
distance from tower
Focal length
of heliostats
Focus 1 (east): Groups 1, 2, 3, 4
Focus 2 (west): Groups 5, 6, 7, 8
x
Switch-over groups (east and west): 0, 9, 10, 11
y
Reserve group (east and west): R
Tower
11
12
13
14
15
22
23
24
25
26
21 27
33
34
35
36
37
32 38
31 39
43
44
45
46
47
42 48
41 49
63
64
65
66
67
62
68
61
69
83
84
85
86
87
82
81
51
52
53
54 55
56
57
58
71
72
73
74 75
76
77
78
91
92
93
94 95
96
97
98
B1
B2
B3 B4
B5
B6
A1
A2
A3
A4
A5
A6
A7
C2
C3
C4
C5 C6
C7
C8
C9
C1 CA
59 m
67 m
77 m
89 m
97 m
104 m
113 m
122 m
133 m
143 m
156 m
169 m
67 m
136 m
115 m
162 m
97 m
5
6
7
8
1
3
2
4
9
11
10
0
R
Fig. 3. Partitioning of the SSPS-CRS heliostat eld at the PSA.
M. Roeb et al. / Solar Energy 85 (2011) 634644 637
into those systems for analysis. All these components are
housed beside the reactor in a cabinet on the tower. Out-
side the tower, a thermo camera is located in a distance
of about 10 m on an arm in front of the receiver, which
measures the temperature distribution at the absorber.
Using several lters, measurements of the temperature on
the quartz glass window and on the surface of the absorber
behind the window can be carried out.
3. Thermal qualication
First, a series of thermal tests with uncoated monoliths,
i.e. without a redox system, were carried out. The objective
was to gain knowledge about the thermal behaviour of the
pilot plant in particular of the receiverreactor and the
possibility and practicability of controlling it by dierent
means. All tests were carried out with air instead of nitro-
gen and without water vapour. A specic intention was to
determine a practicable operational strategy concerning the
optimum of temperatures and mass ows. These tests were
also aimed at the denition of standard operational param-
eters as reference for parametric tests on thermal perfor-
mance and hydrogen production. The experimental focus
of attention was on the one hand the temperature at the
central honeycomb of both modules which in most cases
exhibited also the maximum temperature and on the
other hand the average temperature of all monoliths. To
control the temperature, the volume ow and pre-heater
temperature for the working medium air as well as
the number of heliostats focused on the two modules of
the receiver were systematically varied or used to compen-
sate for the uctuations of the receiver temperature caused
by other (outer) parameters like Direct Normal Irradiance
(DNI), ambient temperature, wind speed and direction.
During the thermal test campaign, the two monoliths
were operated individually with dierent solar power, dif-
ferent operating temperatures and dierent mass ow of
the feed gas to most authentically simulate the operation
when producing hydrogen. Thus, dierences and inuences
of both modules can be observed by parallel operation
(Fig. 5). The operational conditions were switched after a
typical time of about 30 min for a half-cycle by moving
the focus of some heliostats from one module to the other
and by switching the valves for the gas supply.
The inuence of defocusing and focusing of heliostats
was examined for both temperature levels (Fig. 6). It was
observed that the absorbers heat up faster than they cool
down. Thus, after the switch, the level of 1200 C targeted
for the regeneration is reached earlier than the level of
800 C targeted for water splitting. The time for switching
the temperature levels is dened as the period until both
reach 95% of their desired level. The speed by which helio-
stats are shifted determines the time needed to mutually
transfer the status of the modules. So far, more than ten
minutes are needed to cool down the hotter modules to
800 C. This is in particular true for the rear part of the
Fig. 4. Flow sheet of the process hardware.
638 M. Roeb et al. / Solar Energy 85 (2011) 634644
monoliths and the outlet temperature of the gas: a sharp
reaction after changing solar ux is observed at the absor-
ber front, where the thermocouples are located, whereas
the outlet temperature needs more time to react on a
change of solar power like the one described. In fact, the
eective time for a switch is much shorter (<5 min), since
only the time needed to heat up the cold module is essen-
tial for the process, since a sucient high temperature level
for the regeneration is necessary, whereas water is already
split at temperatures near 1200 C. Even if the conditions
are not optimal with respect to thermodynamics and get
better when approaching 800 C the kinetics of water split-
ting become slower with decreasing temperature. In previ-
ous laboratory and solar furnace experiments a trade-o of
these two dierent eects was found in this temperature
range between 800 and 1200 C with some dominance of
the kinetic eect (Roeb et al., 2008). Nevertheless the split-
ting predominantly has to be carried out at temperatures
below 1000 C to conserve the stability of the redox mate-
rial and support when coming into contact with steam
above these temperatures. Much more critical for the eec-
tiveness of the process is the regeneration temperature.
Bench scale experiments showed that sucient high reac-
tion rates and yields for this regeneration reaction can only
be achieved at temperatures higher than 1150 C (Fig. 7).
At the same time it has to be ensured that the temperature
Fig. 5. Cyclic procedure with two full cycles performed with non-preheated air (mass ow: 25 m
3
/h).
12:00 13:00 14:00 15:00 16:00
0
200
400
600
800
1000
1200
DNI normalized - west module
DNI normalized - east module
Focus 1
Focus 2
18-07-2008
Time [hh:mm:ss]
A
v
e
r
a
g
e

t
e
m
p
e
r
a
t
u
r
e

[

C
]
0
5
10
15
20
25
30
35
40
H
e
l
i
o
s
t
a
t
s

f
o
c
u
s
e
d

[
N
H
]
Fig. 6. Progression of absorber temperatures and number of heliostats needed.
M. Roeb et al. / Solar Energy 85 (2011) 634644 639
of the redox material should not signicantly exceed
1250 C to avoid the degradation of the material.
To achieve a fast switching of the temperature levels, the
number of heliostats moved should be as high as possible
until the target temperatures are reached. Fig. 6 gives an
impression on the strategy of realising such a fast change-
over and on the number of heliostats needed to keep the
two temperature levels. The number of heliostats focused
on the modules was adjusted according to the require-
ments; during the switchover of temperature levels between
the modules, the heliostats were not simply refocused from
one module to the other. Additionally to the refocused
heliostats, some heliostats were moved to standby to
accelerate the cool down, and on the other hand on the
module supposed to heat up, additional heliostats were
focused to facilitate the switchover. The number of helio-
stats was further adjusted in order to achieve the exact level
of required temperatures. Fig. 8 shows the solar power pro-
vided by the heliostats to each of the two chambers, as
measured by a moving bar system in front of the receiver
apertures. To keep a temperature level of 800 C roughly
17 kW were necessary, whereas about 45 kW were needed
to keep a temperature level of 1200 C.
Parametric studies were carried out to nd out parame-
ters suitable to control the process, especially to adjust and
maintain the temperature levels within the half-cycles.
Those parameters were rstly the mass ow of air (used
as the working medium), secondly the pre-heating temper-
ature of the feed (here air), and thirdly the number of helio-
stats focussed on the two apertures.
Concerning the mass ow of air it can be said in general
that a variation of the mass ow of the feed could be a use-
ful instrument to simultaneously inuence the temperature
prole of the whole receiver. This is at least true for the
ne-tuning of the operational temperature, which means
that temperature uctuations caused by solar ux uctua-
tions of up to about 80 K could be compensated by mass
ow adjustments. For such compensation the mass ow
has to be changed signicantly (5080% mass ow reduc-
tion of the initial value for temperature increase, or by a
factor of 45 for a comparable temperature decrease).
Since this inuences the conditions for the chemical process
drastically in particular the reaction kinetics the mass
ow is not regarded as the rst choice of control
parameters.
Feed pre-heating had only a very small eect on absor-
ber temperatures if the temperature of the working medium
air is raised from ambient temperature to 250 C at the
inlet of the reactor. Therefore they are not suitable as a
means to inuence the operational temperature signi-
cantly and to control the process.
The preferred way of controlling the process tempera-
ture is by the heliostats themselves. Fig. 9 represents the
results of an experiment involving control by heliostats
only. The individual heliostats of the SSPS eld provide
dierent foci which dier in size and solar ux distribution.
Some provide very sharp-edged foci, others provide very
big images with low solar ux. This fact was used to set
up a control strategy by using so-called high-ux mirrors
for the coarse adjustment, i.e. temperature steps of about
100 K, and low-ux heliostats for the ne-tuning of the
temperature control, i.e. temperature steps of about 10
20 K. An accurate temperature control is in particular nec-
essary for the high temperature reaction, the regeneration,
to avoid on the one hand overheating and on the other
hand to ensure sucient reaction rates (see above). Fig. 9
demonstrates the eect of varying the number of those dif-
ferent heliostats focused on the two apertures and the fea-
0 10 20 30 40 50 60
0.00
0.05
0.10
0.15
0.20
T = 1080 C
T = 1135 C
T = 1150 C
T = 1180 C
n
H
2

i
n

m
m
o
l
/
g
f
e
r
regeneration time in min
Fig. 7. Specic amount of additional hydrogen versus regeneration time for dierent regeneration temperatures (ironzinc mixed oxide).
640 M. Roeb et al. / Solar Energy 85 (2011) 634644
sibility of the described control concepts by just using the
number of heliostats for temperature control. For both
temperatures, 800 C and 1200 C, the control by heliostats
can be applied to suciently ensure steady states.
4. Hydrogen production experiments
When preparing the hydrogen production experiments
preliminary tests with water vapour in the reactor were car-
ried out. A mass ow of 3.5 kg/h steam and a mass ow of
15 Nm
3
/h nitrogen turned out as the most suitable opera-
tion conditions to start with. After that the blank siSiC
monolithic absorber structures were replaced by a set of
18 monoliths coated with ironzinc mixed oxide. In the
beginning of the hydrogen production tests, both chambers
were initially heated up to 800 C to achieve steady-state as
far as possible. Afterwards, two full cycles (twice heating
up to 1200 C and twice cooling down to 800 C) were car-
ried out in both reactor modules (Fig. 10).
The composition of the o-gas stream was detected by
the GC for all cycles for both modules and by the MS just
for the western module. The hydrogen concentrations
detected in one of the rst experiments are displayed in
Fig. 11. There, the dots symbolize the signal given by the
GC and the line in between was calculated as linear inter-
polation between two sampling points. This curve only
Fig. 8. Power on the aperture of each module.
Fig. 9. Eect of varying number of heliostats on temperature.
M. Roeb et al. / Solar Energy 85 (2011) 634644 641
gives a rough idea about the progress of hydrogen concen-
tration during the test, since only every 100 s a measure-
ment was performed and recorded. Nevertheless, some
qualitative information can be drawn out. The rst broad
peak at t = 5000 s was obviously caused by splitting resid-
ual water in the apparatus. But apparently the highest out-
put of hydrogen was produced in the eastern module
during the rst cycle at about t = 7000 s. The measured
concentration corresponds to a conversion of 30% of the
steam fed in. After that a reduction of hydrogen concentra-
tion and therefore of also the yield by a factor of about two
occurred. This eect is similar to what has been observed
earlier in smaller reactors in the lab and in the solar furnace
and is attributed mainly to deactivation of the particular
redox system and to a minor extent to inhomogeneous tem-
perature distribution of the absorber. There is evidence
that some of the zinc material in the mixed oxide, volatiliz-
es during cyclic operation resulting in a reduction of the
activity of the redox material from its initial value. The
strongly diminished hydrogen production indicated by
the last peak was caused by the occurrence of a leakage
and therefore by inltrated air in the reactor. Nevertheless,
about 35 g of hydrogen were produced in the very rst
three half-cycles, which would mean that about 500 g of
hydrogen could have been produced, if the experiment
was extended all over the day. The target is to increase
the daily production to more than 3 kg per day by mainly
improving the hydrogen production rate of the redox mate-
rial, by decreasing the heat rejection of the receiver and by
improving the recovery of sensible heat in the process.
More recently another experimental campaign using the
same coated monoliths has been carried out. Despite sev-
eral weeks of interruption the redox material exhibited
the same activity with that during the cycles shown in
Fig. 11. After several cycles the hydrogen yield settled at
a more or less constant level, which is about a factor of ve
less than the rst cycle. Currently the redox material has
been renewed by replacing the coated absorber monoliths
Fig. 10. Temperature and steam mass ow rate of rst water splitting tests.
Fig. 11. Concentration of hydrogen detected by the GC.
642 M. Roeb et al. / Solar Energy 85 (2011) 634644
to enable to start another experimental campaign with
optimised process conditions.
5. Conclusions
A 100 kW pilot plant for two-step solar thermo-chemi-
cal water splitting via monolithic honeycomb solar reactors
has been developed, installed and test operated at the SSPS
solar tower plant at PSA in Spain. The reactor concept is
strongly based on the HYDROSOL two-chamber solar
receiverreactor developed and tested for quasi-continuous
hydrogen production in the solar furnace of DLR in
Cologne. The feasibility of the process has been demon-
strated under real conditions at PSA. An exhaustive ther-
mal qualication of the pilot plant has been carried out
using uncoated ceramic honeycombs as absorbers. Those
tests helped to develop and validate operational and mea-
surement strategy as well as to create essential knowledge
on the dynamics of the system, in particular during thermal
cycling. One main aspect was the necessity of a rapid
changeover between the modules as a central benet for
the performance of the process. Such quick changeover
after completion of a half- cycle could be realised by mov-
ing as many as possible heliostats until the target tempera-
tures are reached.
Potential control parameters have been analysed which
are capable of ensuring sucient constant temperature lev-
els. It was concluded that there is only little usefulness of
employing feed gas pre-heating and mass ow of the feed
gas as control parameters, whereas the preferred way of
controlling the process temperature is by the heliostats
themselves. Solar ux uctuations are compensated by
adding or removing individual heliostats to the two foci.
A control strategy was set up by using high-ux mirrors
for the coarse adjustment and low-ux heliostats for the
ne tuning of the temperature control. For both tested
cycle temperatures, 800 C and 1200 C, the control by
heliostats can be applied to ensure suciently steady states.
In the end the practicability of the implemented control
concept could be shown.
Some test series applying honeycombs coated with redox
material have been also carried out. This allowed practical
testing of solar hydrogen production by water splitting in a
100 kW-scale. Several hydrogen production and metal oxide
reduction cycles could be run without problems. Signicant
concentrations of hydrogen were produced with a conver-
sionof steamof upto30%. Like inlab-scale experiments deg-
radation of the redox material is an issue in the pilot plant as
well. This problem is being addressed in on-going material
development studies. The pilot plant tests are being contin-
ued by using absorber monoliths with dierent coating and
by stepwise optimising the process conditions.
Acknowledgements
The authors would like to thank the European Commis-
sion for co-funding of this work within the Project
HYDROSOL-II Solar Hydrogen via Water Splitting in
Advanced Monolithic Reactors for Future Solar Power
Plants (SES6-CT-2005-020030), under the Sixth Frame-
work Programme of the European Union (20022006).
References
Agraotis, C., Roeb, M., Konstandopoulos, A.G., Nalbandian, L.,
Zaspalis, V.T., Sattler, C., Stobbe, P., Steele, A.M., 2005. Solar water
splitting for hydrogen production with monolithic reactors. Solar
Energy 79, 409421.
Agraotis, C., Mavroidis, I., Konstandopoulos, A.G., Hoschmidt, B.,
Stobbe, P., Romero, M., Fernandez-Quero, V., 2007. Evaluation of
porous silicon carbide monolithic honeycombs as volumetric receivers/
collectors of concentrated solar radiation. Journal of Solar Energy
Materials and Solar Cells 91, 474488.
Allendorf, M.D., Diver Jr., R.B., Miller, J.E., Siegel, N.P., 2006.
Thermodynamic analysis of mixed-metal ferrites for hydrogen pro-
duction by two-step water splitting. In: Proceedings of ISEC 2006
ASME International Solar Energy Conference, Denver, Colorado,
July 813.
Charvin, P., Abanades, S., Flamant, G., Lemort, F., 2007. Two-step water
splitting thermochemical cycle based on iron oxide redox pair for solar
hydrogen production. Energy 32, 11241133.
Diver Jr., R.B., Siegel, N.P., Miller, J.E., Moss, T.A., Stuecker, J.N., 2008.
Development of a CR5 solar thermochemical heat engine prototype.
In: 14th Solar PACES International Symposium, SolarPaces, Las,
Vegas, USA, March 47.
Diver Jr., R.B., Miller, J.E., Allendorf, M.D., Siegel, N.P., Hogan, R.E.,
2006. Solar thermochemical water-splitting ferrite-cycle heat engines.
In: Proceedings of ISEC 2006 ASME International Solar Energy
Conference, Denver, Colorado, July 813.
Fend, T., Hoschmidt, B., Pitz-Paal, R., Reutter, O., Rietbrock, P., 2004.
Porous materials as open volumetric solar receivers: experimental
determination of thermophysical and heat transfer properties. Energy
29 (56), 823833.
Geyer, M., Schiel, W., 1987. High ux experiment test program. In: Schiel,
W., Geyer, M., Carmona, R. (Eds.), The IEA/SSPS High Flux
Experiment. Springer-Verlag, Berlin.
Gokon, N., Mizuno, T., Nakamuro, Y., Kodama, T., 2008. Iron-
containing YSZ (yttrium-stabilized zirconia) system for a two-step
thermochemical. Journal of Solar Energy Engineering 130, 011018-1
011018-6.
Han, S.B., Kang, T.B., Joo, O.S., Jung, K.D., 2007. Water splitting for
hydrogen production with ferrites. Solar Energy 81, 623628.
Ishihara, H., Kaneko, H., Hasegawa, N., Tamaura, Y., 2008. Two-step
water splitting process with solid solution of YSZ and Ni-ferrite for
solar hydrogen production. Journal of Solar Energy Engineering 130,
044501-1044501-3.
Kaneko, H., Fuse, A., Miura, T., Ishihara, H., Tamaura, Y., 2006. Two-
step water splitting with concentrated solar heat using rotary-type
solar furnace. In: 13th Solar PACES International Symposium,
SolarPaces, Seville, Spain, June 2023.
Kodama, T., Gokon, N., 2007. Thermochemical cycles for high-temper-
ature solar hydrogen production. Chemical Reviews 107, 40484077.
Kodama, T., Nakamuro, Y., Mizuno, T., 2006. A two-step thermochem-
ical water splitting by iron-oxide on stabilized Zirconia. Journal of
Solar Energy Engineering 128, 37.
Roeb, M., Monnerie, N., Schmitz, M., Sattler, C., Konstandopoulos, A.,
Agraotis, C., Zaspalis, V.T., Nalbandian, L., Steele, A., Stobbe, P.,
2006a. Thermo-chemical production of hydrogen from water by metal
oxides xed on ceramic substrates. In: Proceedings of the 16th World
Hydrogen Energy Conference, Lyon, France, June 1316.
Roeb, M., Sattler, C., Klu ser, R., Monnerie, N., de Oliveira, L.,
Konstandopoulos, A.G., Agraotis, C., Zaspalis, V.T., Nalbandian,
L., Stobbe, P., Steele, A., 2006b. Solar hydrogen production by a two-
M. Roeb et al. / Solar Energy 85 (2011) 634644 643
step cycle based on mixed iron oxides. Journal of Solar Energy
Engineering Transactions of the ASME 128, 125133.
Roeb, M., Sack, J.-P., Rietbrock, P., Neises, M., Ebert, M., Reinalter, W.,
Schmitz, M., Sattler, C., Lorentzou, S., Pagkoura, C., Zygogianni, A.,
Agraotis, C., Konstandopoulos, A.G., Stobbe, P., Jones, D., Steele, A.,
Lopez, A., Romero, M., 2008. Development and vericationof a two-step
thermochemical process for solar hydrogen production from water. In:
Mancini, Thomas (Ed.), 14th SolarPACES Biannual Symposium, 14th
Biannual SolarPACES Symposium, Las Vegas, NV (USA), March 47.
Steinfeld, A., 2002. Solar hydrogen production via a two-step water
splitting thermochemical cycle based on Zn/ZnO redox reactions.
International Journal of Hydrogen Energy 27, 611619.
Steinfeld, A., Sanders, S., Palumbo, R., 1999. Design aspects of solar
thermochemical engineering a case study: two-step water-splitting
cycle using the Fe
3
O
4
/FeO redox system. Solar Energy 65 (1), 43
53.
Sturzenegger, M., Nu esch, P., 1999. Eciency analysis for manganese-
oxide-based thermochemical cycle. Energy 24, 959970.
644 M. Roeb et al. / Solar Energy 85 (2011) 634644