Development of a 5 kW
th
Windowless Packed-Bed Reactor
for High-Temperature Solar Thermochemical Processing
Christian Wieckert
1,a), Nikolaos Tzouganatos
1,2,b)and Aldo Steinfeld
2,c)1Solar Technology Laboratory, Paul Scherrer Institute, 5232 Villigen PSI, Switzerland, 2 Department of Mechanical and Process Engineering, ETH Zurich, 8092 Zurich, Switzerland.
a)Corresponding author: [email protected] b)[email protected]
Abstract. Solar thermochemical reactors operating at high temperatures and requiring sealing of the reaction chamber
usually make use of a transparent quartz window for the access of incoming concentrated solar radiation, but windows become critical components when scaling up. Indirect irradiation by an intermediate solar absorber eliminates the need for windows at the expense of a less-efficient conduction heat transfer to the reaction site. Here, we present the design, fabrication, and testing of a 5 kWth windowless packed-bed reactor. The designed was aided by a coupled Monte Carlo
ray-tracing and finite element thermal/structural numerical model. A series of C/SiC ceramic matrix composites and monolithic solar absorbers were experimentally tested for their thermal behavior, structural stability, and oxidation resistance in air. The solar reactor was further demonstrated for performing the carbothermal reduction of ZnO at temperatures above 1200°C, yielding peak Zn yield rates of 0.05 mol/min and a solar-to-chemical energy conversion efficiency of 15%.
INTRODUCTION
The two-cavity packed-bed solar reactor concept has been applied for high-temperature thermochemical processing of solid feedstocks, namely: the carbothermal reduction of ZnO1-3,the purification of Waelz oxide4, and
the steam-based gasification of carbonaceous feedstock5-6, and it has been experimentally demonstrated on
lab-scale1,2,4-5 and pilot-scale setups up to 300 kW
th3,6. It consists of two cavities in series, the upper one functioning as the
solar absorber and the lower one functioning as the reaction chamber that can accommodate bulky feedstock of various sizes, shapes, and morphologies without prior processing. Previous designs used a quartz window to seal the upper cavity and protect the absorber plate separating the two cavities. However, the quartz window remains a critical component for scaling up. This article presents the design, fabrication, and testing of a 5 kWth windowless
packed-bed solar reactor. The main challenge for the realization of the windowless reactor is the development of a gas-tight absorber assembly separating the two cavities that is chemically and mechanically stable towards air and product gases at temperatures of up to 1400°C and subjected to severe thermal shocks. Various geometries and materials are experimentally tested for their structural stability, chemical resistance, and heat transfer efficiency when exposed to high-flux irradiation. The design of the reactor configuration is aided by thermal/structural numerical modelling, and a lab-scale prototype is demonstrated by performing the carbothermal reduction of ZnO.
EXPERIMENTAL SETUP
A schematic representation of the windowless solar reactor configuration and experimental setup is shown in Fig. 1. Experimentation was carried out at PSI’s High-Flux Solar Simulator (HFSS)7: an array of ten 10 high-pressure
Xe-arc lamps that provides a continuous beam of concentrated thermal radiation under similar radiative heat transfer characteristics of solar concentrating dishes and towers. The radiative flux at the focal plane can be adjusted by the number of lamps used and a Venetian blind-shutter. The solar reactor is designed for a beam-down type concentrating
system8, which is realized in the lab by redirecting the HFSS’s horizontal beam with a 50x50 cm2 water-cooled,
Al-polished 45°-deflection mirror into a vertical beam. The solar reactor consists of two cavities in series separated by a solar absorber plate. The upper cavity (UC) serves as an open cavity-receiver with a 60 mm-diameter circular aperture for the access of concentrated solar irradiation. The lower cavity (LC) consists of a 135 mm-diameter, 230 mm-height octagonal enclosure made of 8 mm-thick recrystallized silicon carbide (ReSiC) walls and contains a packed bed of solid reactants. Both cavities are lined by Al2O3/SiO2 thermal insulating material and the LC is contained in a stainless
steel vessel. Concentrated solar radiation is absorbed in the UC, transferred by conduction across the separating absorber plate, emitted towards the LC, and absorbed by the top surface of the packed bed, thus providing the process heat to drive the highly endothermic chemical reactions. During the reaction, the solid feedstock is converted predominantly into gaseous species that exit through a 50 mm-diameter outlet port located on the lateral walls of the LC.
To gas-seal the solar reactor against atmospheric air, the solar absorber body is sandwiched between a holding ring (160 mm inner diameter, 290 mm outer diameter) and a flange ring (160 mm inner diameter, 173.5 mm outer diameter), both made of Inconel 617 and sealed by the gasket washer, as depicted in the enlargement in Fig. 1. The entire assembly is then integrated to the upper face of the stainless steel vessel. In contrast to the previous windowed reactor design, the use of highly oxidation-resistant Inconel 617 parts permit operation without active cooling, thereby also diminishing the temperature gradients that are detrimental for the structural stability of the solar absorber.
FIGURE 1. Schematic of the 5 kWth windowless packed-bed solar reactor configuration, peripheral components, and main
sensors. On top right: Top and cross-sectional views of the solar absorber assembly.
Temperatures are measured using 10 K-type thermocouples embedded at critical reactor positions, as indicated in Fig. 1. A representative value of the upper cavity temperature is obtained by thermocouple T1 mounted close to the
surface of the insulation monoliths. The temperature at the region of maximum incoming solar radiative flux is measured by thermocouple T4 mounted at the center point of the lower surface of the solar absorber body.
monitoring of the temperature gradient across the holding ring, while the temperature difference T4 –T5 is indicative
of the thermal stresses developed throughout the solar absorber specimens during operation. Thermocouples T6 and
T8 embedded behind the ReSiC walls monitor the lower cavity temperature at different cavity heights, while T10 and
T9 measure at the top and the bottom of the packed bed, respectively. The pressure in the lower cavity of the reactor
(P1) and at the upstream side of the filter unit (P2) is measured online to detect potential overpressure. Pressure relief
valves are set to open at an overpressure of ~ 150 mbar in case of pressure build-up. The gas composition of the non-condensable gases is analyzed online by a gas chromatograph (GC).
Four solar absorbers (A-D), shown in Fig. 2 (a), were exposed to concentrated solar radiation and investigated for their thermal and structural behavior, as well as their chemical resistance to air and the gaseous products evolving during the carbothermal reduction of ZnO. Figure 2 (a) shows top-view photographs of a circular plate (A) and a (quasi-) hemispherical cavity (B), respectively, both made of Si-lined carbon fiber-reinforced SiC ceramic matrix composite (C/SiC CMC). This novel class of structural materials aims at combining the properties of SiC with enhanced fracture toughness to avoid the brittle fracture behavior exhibited by monolithic ceramic materials. Besides these ceramic matrix composites, pure monolithic pressureless sintered alpha-SiC (SSiC) was tested as it exhibits a superior oxidation resistance in air at temperatures up to 1600°C. Key physical, thermal and mechanical properties 9-10 of the two material types investigated are listed in Table 1. In presence of the largely non-uniform distribution of
the solar radiative flux, the use of a circular plate made of SSiC is not favored as high thermal stresses are expected to develop throughout the specimen during reactor operation. Indeed, SSiC circular plates tested under similar experimental conditions in a previous study2 experienced brittle fracture during heat up. Therefore, a hemispherical
SSiC cavity (C) of similar dimensions to the C/SiC CMC specimen and a (quasi-) cylindrical SSiC cavity (D) were selected, as depicted in Fig. 2 (a).
TABLE 1. Key material properties of the solar absorber specimens9-10.
Si-lined C-fiber reinforced SiC CMC (C/SiC matrix composite)
Pressureless sintered alpha-SiC (SSiC)
Maximum operating temperature in air (°C) 1350 1600
Density (g/cm3) 2.4 3.15
Thermal conductivity (W m-1 K-1) 20 - 40 125 (@ 20oC)
Expansion coefficient (10-6/K) 1.8 - 3.0 3.0 - 5.0
Young’s modulus (GPa) 20 - 30 400
Flexular strength (3-point) (MPa) 50 - 95 400
Thermal shock parameter (W mm-1) 46 26
Fracture toughness KIc (MPa m1/2) 23-30 4
The size and shape of the UC insulation monoliths varied for each solar absorber configuration as these affected the radial distribution of the solar flux incident on the solar absorber. In order to achieve a flux distribution that would induce relatively small temperature gradients and low thermal stresses throughout each solar absorber, the design of these components was based on a numerical thermal and structural modeling.
NUMERICAL THERMAL AND STRUCTURAL MODEL
The design of the UC insulation monoliths as well as decisions on the operation strategy of the HFSS for each reactor configuration were aided by three coupled numerical models, as schematically shown in Fig. 2 (b): (i) a 3D Monte Carlo (MC) ray-tracing simulation to acquire the solar radiative flux absorbed by the surfaces composing the upper cavity qabs″,(ii) a 2D-axissymmetric FE thermal simulation to compute the temperature distribution throughout
the solar reactor, and (iii) a 2D-axissymmetric FE structural simulation using the temperature profiles throughout the solar absorber specimens to compute the corresponding stress distributions and predict potential material failure based on the Mohr-Coulomb failure theory11, which is applicable for materials with compressive strengths that far exceed
their tensile strength such as ceramics. Recursive simulations were conducted until determining the final reactor design and a suitable operation strategy for each solar absorber. The UC insulation shape and size, the position of the focal
plane relative to the reactor aperture, the solar power input, and the reactor heat-up rate were varied until reaching an acceptable result in terms of structural stability of the solar absorber.
(a) (b)
FIGURE 2. (a): Photographic representation of the four solar absorber configurations: A) top-view of a 170 mm diameter, 3
mm-thick circular C/SiC CMC plate, B) top-view of a 170 mm diameter, 3 mm thick, 40 mm high C/SiC CMC hemispherical cavity, C) top- and front-views of a 170 mm diameter, 3 mm thick, 40 mm high SSiC hemispherical cavity, and D) top- and
front-views of a 170 mm-outer-diameter, 50 mm inner-diameter, 3 mm thick, 90 mm high SSiC cylindrical enclosure. (b): Schematic of the thermal/structural numerical simulation. The solar flux absorbed by the upper cavity components qabs″ is used as
an input to the thermal simulations providing the temperature distribution and maximum temperature difference ΔTmax
throughout the solar absorber configurations. The maximum (σ1), medium (σ2), and minimum (σ3) principal stresses, and safety
factors (SFstress) according to the Mohr-Coulomb stress criterion obtained from the structural simulations are used to assess
potential fracture of the specimen during operation.
Monte Carlo ray-tracing model: An in-house MC ray-tracing code12 was used to compute the radiative flux
absorbed by the upper cavity components of the reactor. The 45° angle, Al-polished mirror was modeled as a specularly reflecting surface with no optical losses. The upper cavity surfaces were handled as diffusely reflecting, and were radially and circumferentially segmented in order to obtain a high resolution radiative flux map. Appropriate surface segmentation was achieved by recursive mesh refinement until reaching a well-resolved solution. A total of 107 rays were used to simulate the radiation emitted by the eight Xe arc lamps of the HFSS utilized during the
experiments.
Thermal FE Model: A transient 2D-axisymmetric FE thermal model was developed for solving the energy
conservation equation in the solid portion of the reactor. The contribution of the solar radiative flux absorbed by the UC surfaces is accounted for by applying the circumferentially averaged 3D radiative flux maps obtained by the MC simulations as source terms. Diffusely emitting and reflecting surfaces, uniform radiative fluxes over each surface element and non-participating media were assumed within the enclosure. All simulations were performed in absence of solid reactants in the lower cavity. A total hemispherical absorptivity of 0.6, 0.88 and 0.75 was assumed for the insulation monoliths, the absorber specimens and the LC ReSiC wall, respectively13,14. Surface-to-ambient thermal
radiation emitted from the exterior surfaces of the UC insulation monoliths and the reactor steel shell toward the surroundings (Tenv = 291.15 K) is accounted for by source terms. Since the governing energy conservation equation
is solved only in the solid portion of the reactor, convective heat transfer affecting the UC and LC surfaces, as well as the exterior surfaces of the UC insulation monoliths and the reactor steel shell, is modeled using empirical Nusselt correlations for forced and natural convection and is introduced as a source term into the energy conservation equations of the respective grid elements. The governing equations were solved using ANSYS Mechanical15 with ~ 44,000, ~
45,000, and ~ 55,000 grid elements for the circular plate, hemispherical and cylindrical cavity configurations, respectively. Grid independent results were yielded for an element size of 3·10-3 m and a time step of 10 s after
recursive refinement of the spatial and temporal grids. To adequately resolve the steep temperature gradients in the regions close to the directly irradiated surfaces of the UC, a locally refined grid, ranging from 1·10-3 m at the region
of the insulation monolith surfaces to 1·10-4 m throughout the solar absorber specimens, was applied.
Structural FE Model: To compute the stress distribution throughout the solar absorber specimens and predict a
possible material failure according to the Mohr-Coulomb failure criterion11, the numerically computed temperature
distributions throughout the specimens were introduced into a 2D-axissymmetric FE structural simulation performed with ANSYS Mechanical15. An uniformly distributed pressure load of 10 MPa was imposed on the upper outer edge
of the solar absorber specimens to represent the gasket stress applied by the flange ring, whereas the vertical displacement of the lower outer edge of the specimen was set equal to zero. The stress-free temperature was set equal to 295.15 K. The governing equations were solved using ~ 26,000, ~ 30,000, and ~ 43,000 grid elements for the circular plate, hemispherical and cylindrical cavity configurations, respectively.
NUMERICAL RESULTS
The final UC insulation shape and the relative focal plane-reactor aperture position for each configuration are depicted in Fig. 3. The reactor aperture was aligned to the secondary focal plane for all the solar absorber configurations except for the cylindrical SSiC cavity, where it was positioned 20 mm below the focal plane in order to achieve a more uniform flux distribution and further reduce the thermal stresses developed throughout the solar absorber.
FIGURE 3. Cross section of the upper cavity for the a) circular plate, b) hemispherical, and c) cylindrical configurations,
comprising the solar absorber specimens and the insulation monoliths. The position of thermocouples T1 and T4, as well as the
Stepwise increase of the radiative power input at about 300 Wth increments every 6 min until reaching the
steady-state solar power input, Qsolar, was identified as a reasonable operation strategy for the HFSS. Similar reactor heat-up
conditions were applied to all the solar absorber specimens to enable comparison of their structural behavior under comparable thermal stress loading. The values of Qsolar of 4.75 kW for the circular plate and hemispherical
configurations and of 4.35 kW for the cylindrical configuration were obtained by recursive MC-thermal simulations aiming to achieve a LC steady-steady state temperature of about 1200°C. Lower power input is required for the latter owing to the smaller size and, thus, thermal mass of the UC insulation for this configuration, as shown in Fig. 3.
Figure 4 depicts the circumferentially averaged radiative fluxes absorbed by the specimens at the end of the reactor heat-up period. For the circular plate, the peak radiative flux of 435 suns (1 sun = 1 kW/m2) is obtained at the center
of the Si-lined C/SiC CMC. The flux distribution map acquired for both the C/SiC CMC and SSiC hemispherical specimens is identical due the similarity in their optical properties. Higher peak fluxes in the order of 500 suns are obtained at the center of the specimens despite their larger surface area since somewhat steeper radiative flux gradients were preferred over exposure of the critical concave-to-convex transition region to high levels of incoming solar radiation. For the cylindrical SSiC cavity, a 150 mm outer-diameter, 84 mm inner-diameter, 8 mm thick insulation part (α = 0.6, ρ = 0.413) extending over the upper 2.5 cm of the SSiC specimen was designed to protect the curved
portion between the horizontal and the vertical part of the specimen from exposure to incident solar radiation, as illustrated in Fig. 3 (bottom). Lower fluxes compared to the previous absorber configurations are attributed to the larger surface area over which the incoming radiation is distributed.
FIGURE 4.Distribution of the circumferentially averaged radiative flux absorbed by the solar absorber specimens, normalized to the radiative power input Qsolar entering through the reactor aperture at the end of the reactor heat-up stage (4.75
kW for geometries A, B and C and 4.35 kW for geometry D in Fig.2 (a)).
Figure 5 illustrates the maximum principal stress distribution throughout the solar absorber specimens at the maximum thermal stress loading level (a) for the C/SiC CMC and (b) for the SSiC specimens including the safety factors according to the Mohr Coulomb stress criterion. For the hemispherical cavity made of SSiC low safety factors of 0.8 and 0.9 are obtained at the critical gasket and inflection point regions, respectively, as shown in Fig. 5 (b, top) indicating potential material failure during operation. Figure 5 (b, bottom) illustrates the better performance of the cylindrical SSiC cavity due to its larger heat transfer area and the higher uniformity of the incident solar flux distribution, both leading to lower temperature gradients and thermal stresses throughout it.
(a) (b)
FIGURE 5. (a) Principal stress distributions throughout the circular plate (top) and hemispherical C/SiC CMC (bottom) at the
maximum thermal stress loading level. (b) Principal stress distributions throughout the hemispherical (top) and cylindrical cavity made of SSiC (bottom) at the maximum thermal stress loading level. Indicated in grey are the safety factors according to the
Mohr Coulomb stress criterion at structurally critical regions.
EXPERIMENTAL RESULTS
Two experimental campaigns were performed. In the first one, experimental runs without solid reactants in the LC and a conservative reactor heat-up rate as guided by the numerical simulations were applied to investigate the structural behavior of the solar absorber specimens and the Inconel parts as well as the performance of the high-temperature sealing. In the second campaign, the heat-up period was halved and solid feedstocks were introduced into the LC prior to each experimental run. Mixtures of commercially available ZnO powder (Alfa Aesar 011558, purity: 99.0 % min, mean particle size 44 μm) and beech charcoal gravel (proFagus GmbH, Bodenfelde, Germany, particle size: 0.5 – 1.0 mm)2 were used as feedstock for the solar-driven carbothermal production of Zn, according to the overall net reaction:
ZnO + C = Zn(g) + CO ∆ 239.9 (1)
For both experimental campaigns, the solar reactor was purged with N2 before starting an experimental run. All
experimental runs were performed at a slight overpressure of ~ 10 mbar. The heat-up period of the reactor commenced upon igniting eight Xe-arc lamps of the HFSS. The solar power input was increased to ~ 4.4 – 4.75 kWth in steps of
about 300 Wth over a time period of 96 min and 48 min for the first and second type of experiments, respectively, by
adjusting the orientation of the Venetian blind-shutter. For the first type of experiments T10 was maintained constant
for ~ 30 – 45 min before switching off the HFSS. For the carbothermal reduction experimental runs T10 was kept
constant until reaching reaction completion, as indicated by the small temperature difference across the packed bed and the low CO and CO2 concentrations detected by the GC. Stabilization of the temperature at a predetermined value
was achieved by reducing the solar power stepwise to 75 – 80 % of the nominal power. The molar production rate of Zn was calculated based on the oxygen balance in CO and CO2 in the absence of any other source of oxygen apart
from the ZnO powder. Measurement of the variation of Qsolar was performed for each geometrical configuration using
a water-cooled copper-coiled calorimeter.
Experimental runs without feedstock were carried out to investigate the performance of all four solar absorber
specimens with respect to their structural stability, chemical inertness, and thermal behavior, while the leak tightness of the solar reactor over the whole experimental duration was monitored by measuring the reactor overpressure online. The temperature difference across the specimen T4 – T5 was monitored as a decisive parameter for the structural
stability. The tests demonstrated successfully the technical feasibility of the high-temperature sealing solution. Under the described operation conditions the specimens A, B, and D, depicted in Fig. 2, performed well. Maximal temperature differences T4 – T5 of about 400°C were measured for specimens A and B and of 590°C for specimen D.
The hemispherical specimen made from SSiC (C) failed during the reactor heat-up period, in line with the predictions from the numerical model, where specimen C obtained a safety factor below unity. While withstanding the mechanical stresses the C/SiC CMC specimens showed a slight mass loss of 0.9% (A) and 0.4% (B) which is attributed mainly to evaporation of free Si contained in the ceramic matrix and/or the specimen lining at hot spots with temperatures exceeding the maximum operating temperature of the material, as well as to oxidation of free carbon in air. The cylindrical SSiC cavity remained intact over the whole experimental run, as no morphological change after visual inspection or mass loss could be detected.
Carbothermal reduction experiments were carried out with specimens A, B, and D to investigate their chemical
resistance to the product gases evolving during the carbothermal reduction of ZnO. ZnO powder and beech charcoal gravel were hand-mixed at a ZnO:C molar ratio of 1:0.8 into batches of ~ 325 g and loaded into the reactor, forming a 35 mm-high packed bed. The structural stability of the solar absorber specimens was tested under more severe thermal and stress loading conditions by reducing the reactor heat-up time by half compared to the preliminary thermal tests without feedstock. Temperatures in the range of 1202 – 1219°C were obtained under quasi steady-state conditions at the top of the ZnO:C packed bed (T10).
The variation of solar reactor temperatures, gas product molar flow rates and solar power input as a function of time for a representative experimental run using the circular C/SiC CMC plate as absorber body is shown in Fig. 6. Peak CO and Zn(g) production rates were reached after 105 min and 111 min at 0.042 mol/min and 0.053 mol/min, respectively, while the LC temperature reached a steady-state value of T10 = 1202°C. Peak gas product molar flow
rates are of particular importance for the two-cavity solar reactor since they are of similar order of magnitude as the steady-state values obtained during prolonged reactor operation when introducing larger amounts of solid feedstock into the LC. The solar-to-chemical energy conversion efficiency of the process defined as
, (2)
where and LHVi represent the molar flow rate and the lower heating value of species i = Zn(g), CO, and C,
respectively, and , acquired a peak value of ~ 12.8% over the course of the experimental run. The temperature difference T4 – T5 obtained a considerably higher peak value of 516 °C after the first ~ 48 min compared
to the first type of experiments as a result of the shorter heat-up period.
FIGURE 6. Temporal variation of reactor temperatures, molar flow rates of Zn(g), CO, CO2, and H2, and solar radiative power
input over the course of a typical experimental run for the solar-driven carbothermal reduction of commercial ZnO powder, using the circular Si lined carbon fiber-reinforced SiC CMC plate as the intermediate solar absorber specimen.
After the end of the experimental run, small amounts of ash, unreacted ZnO powder, and charcoal gravel, corresponding in total to 3.6% of the initial feedstock mass, were detected in the reaction chamber, indicating almost complete conversion of the reactants. Condensed zinc vapor was predominantly recovered from the water-cooled condenser, while a lesser fraction was detected in the outlet pipes and the filter unit. The solar absorber was proven structurally stable and the solar reactor remained leak tight. The absence of any morphological change on the lower side of the plate indicated no chemical interaction between the absorber material and the gaseous products evolving during the ZnO reduction.
Carbothermal reduction tests using the hemispherical C/SiC CMC cavity and the cylindrical SSiC cavity as solar absorbers were carried out at quasi steady-state LC temperatures of T10 = 1212 °C and T10 = 1219 °C, respectively.
Both configurations remained undamaged and no evidence of chemical reaction between the solar absorber bodies and the product gases could be obtained after visual inspection of the samples. Peak Zn(g) production rates of 0.05 mol/min and 0.048 mol/min were reached, leading to peak solar-to-chemical conversion efficiencies of 12.85% and 15%, respectively. Despite the introduction of higher natural convection losses from the open UC due to the elimination of the window, the larger heat transfer area and the higher thermal mass due to the larger size of the windowless reactor, as well as the higher temperature differences between the reactor shell and the surroundings vis-à-vis the preceding windowed two-cavity solar reactor design, solar-to-chemical conversion efficiencies of similar order of magnitude16 were obtained for both reactors. This is attributed to the elimination of water-cooled parts in the
windowless design that serve as large heat sinks in solar thermochemical reactors. Since the reactor design aimed at the experimental demonstration of the windowless concept rather than achieving high efficiency, increased conversion efficiency is expected after optimizing the reactor design. Considerably higher solar-to-chemical conversion efficiencies are to be expected at a pilot scale plant17.
SUMMARY AND CONCLUSIONS
A 5 kWth windowless packed-bed solar reactor for thermochemical processes was designed, fabricated, and
experimentally demonstrated using various solar absorber configurations and materials. The reactor design and the operation strategy of the HFSS were aided by a coupled 3D Monte Carlo, 2D thermal and 2D structural numerical model. Thermal tests without solid reactants were carried out at LC’s temperatures of about 1200°C using a conservative reactor heat-up rate in order to reduce the thermal stresses throughout the solar absorber bodies. The circular C/SiC CMC plate and the cylindrical SSiC cavity remained intact. Leak tightness of the reactor was demonstrated over the course of the experimental runs. Under similar solar flux distributions, the hemispherical Si-lined C/SiC CMC cavity exhibited superior structural behavior vis-à-vis the SSiC cavity: the latter experienced brittle failure during reactor heat-up owing to its lower resistance to fracture. Despite its structural superiority, the CMC material demonstrated lower chemical stability. Optimization of the CMC material e.g. by identification of a carbon fiber content that allows for maximization of its structural stability while minimizing the adverse effects on its chemical resistance in air would be of interest. An alternative material worth to be investigated is SiC-fiber reinforced SiC. Carbothermal reduction experiments proved satisfactory chemical resistance to the gaseous products of the reaction. Peak Zn(g) production rates of ~ 0.05 mol/min were obtained for a LC temperature of 1219°C, leading to solar-to-chemical conversion efficiencies in the range 12.8 – 15%. In sum, the technical feasibility of the packed-bed windowless solar reactor concept for performing thermochemical processes was successfully demonstrated.
ACKNOWLEDGMENTS
Funding by the Swiss State Secretariat for Education, Research and Innovation (Grant Nr. 16.0183), EU’s Horizon 2020 Program (Project INSHIP – Grant No. 731287), and EU’s 7th Framework Program (Project STAGE-STE – Grant
No. 609837; Project SFERA-II – Grant No. 312643) is gratefully acknowledged. The authors also thank D. Wuillemin and V. Schnetzler for their contribution to the reactor design and fabrication, as well as Y. Bäuerle and M. Locher for the operation of PSI’s high-flux solar simulator.
REFERENCES
1. T. Osinga, U. Frommherz, A. Steinfeld and C. Wieckert, ASME J. Sol. Energy Eng. 126, 633–637 (2004). 2. S. Kräupl, U. Frommherz and C. Wieckert, ASME J. Sol. Energy Eng. 128, 8–15 (2006).
3. Wieckert, U. Frommherz, S. Kräupl, E. Guillot, G. Olalde, M. Epstein, S. Santén, T. Osinga and A. Steinfeld, ASME J. Sol. Energy Eng. 129, 190-196 (2007).
4. N. Tzouganatos, R. Matter, C. Wieckert, J. Antrekowitsch, M. Gamroth and A. Steinfeld, TMS-Journal of Metals 65, 1733-1743 (2013).
5. N. Piatkowski, C. Wieckert and A. Steinfeld, Fuel Process. Technol. 90, 360–366 (2009).
6. C. Wieckert, A. Obrist, P. von Zedtwitz, G. Maag and A. Steinfeld, Energy & Fuels 27, 4770-4776 (2013).
7. J. Petrasch, P. Coray. A. Meier, M. Brack, P. Haeberling, D. Wuillemin and A. Steinfeld, ASME J. Sol. Energy Eng. 129, 405-411 (2007).
8. A. Yogev, A. Kribus, M. Epstein and A. Kogan, Int. J. Hydrogen Energy 23, 239-245 (1998).
9. SGL Carbon Group, SIGRASIC 6010 GNJ - Short Fibre Reinforced Ceramics, Data sheet, Meitingen, Germany, 2005.
10. Ceramdis GmbH, Properties of ceramic materials, Data sheet, Elsau, Switzerland, 2014. 11. W.-F. Chen and E.M. Lui, Handbook of structural engineering (CRC press, New York, 2005).
12. J. Petrasch, “A free and open source Monte Carlo ray tracing program for concentrating solar energy research”,
ASME 4th Int. Conf. on Energy Sustainability (American Society of Mechanical Engineers 2010), pp. 125–132.
13. M.F. Modest, Radiative heat transfer (Academic press, Oxford, UK 2013). 14. Insultech AG, Properties of ReSiC, Data sheet, Trimbach, Switzerland, 2002. 15. ANSYS® Academic Teaching Advanced, Release 15.0.
16. N. Tzouganatos, C. Wieckert and A. Steinfeld, AIChE J. 62, 4586-4594 (2016). 17. N. Piatkowski and A. Steinfeld, AIChE J. 57, 3522–3533 (2011).
