SOLAR HYDROGEN PRODUCTION BY THERMOCHEMICAL REACTION: DEVELOPMENT OF A PACKED-BED REACTOR

Solar water splitting is a promising process for the storage and conversion of sunlight power into cleanburning hydrogen gas, this paper presents a CFD analysis of hydrogen production via a proposed packed bed thermochemical receiver/reactor system. The idea of this study is the use of packed bed of spherical ceramic particles coated with active redox ferrite materials. The first step is an endothermic reaction, nickel ferrite (NiFe2O4) dissociate thermally into nickel oxide (NiO), ferrous oxide (FeO) and oxygen at 1473 K, this reaction take place under 2 KW concentrated solar energy. The second is the hydrolysis step at 1073 K to form hydrogen and NiFe2O4, the latter is recycled to the first step for a new production cycle. The CFD code solves the momentum, energy and species transport equations. The temperature fields of the reactor solid & the fluid phases were attained using the local thermal non-equilibrium model (LTNE). The LTNE model sources terms were computed through the user-defined functions to couple the energy equations of the fluid phase and solid phase. The complete model was used to analyse numerically the reaction through the packed bed in order to predict the thermal behavior under different conditions (inlet velocity, packing arrangement and solar concentrated flux).


INTRODUCTION
Hydrogen is actually considered as the most promising green fuel vector for the future as it has prodigious capability of reducing greenhouse gas and other environmental pollutants (C n H m ,NO x , heavy metals, radioactivity) emitted by fossil energy systems. It is considered that engines operating with hydrogen [1][2][3] can tremendously reduce dependence on nonrenewable fuel sources [4]. Thus, the use of hydrogen as an alternative fuel vector unlocks potential means of converting energy from solar radiation to storable chemical energy. It is envisaged that the hydrogen will replace smoothly fossil energy resources in automotive industry through fuel cell technology. It would be more efficient as fuel for internal combustion engines with comparable performance, emitting just water vapour [4,5].
Historically, it is observed that there have been tendency toward the decarbonization of energy sources from solid to liquid to gaseous fuels. From coal to oil back to natural gas again, the ratio of carbon to hydrogen in the molecule of each successive source of energy has decreased [6], which confirms the global tendency towards using hydrogen as a future source of energy. Currently a growth in demand for hydrogen is observed, with major applications in Ammonia production, petroleum processing, petrochemical production, oil and fat hydrogenation, fertilizer production, metallurgical applications and electronics industry [7], it is also anticipated that interest in using hydrogen as fuel for automotive industry will further enhance the demand of the green fuel.
Hence, it is necessary to expand the existing hydrogen production capacity by using renewable energy resources, in order to meet the worlthe d increasing demand. Consumption of hydrogen in different applications is shown in the following Figure [4]. There are basically three major pathways for Hydrogen production: electrochemical, thermochemical and biological methods. One of the most attractive methods for the acquisition of hydrogen could be water thermochemically splitting.

Figure 1. Hydrogen consumption by industry sector
Thermochemical hydrogen production process necessitates only water as a material input and heat as an energy source (rather than electricity for electrolysis process). The output of the process is H 2 and O 2 . The process itself comprises a set of chemical reactions, which sum the splitting of H 2 O molecule [8].
In the following section, the thermochemical Hydrogen production methods coupled with solar energy source are briefly described.
The exploitation of the massive amount of solar radiation and its conversion to chemical storable energy such as hydrogen is a subject of primary technological interest [9]. There are basically three methods and their combinations for producing hydrogen using solar radiations namely photochemical, electrochemical and thermochemical [10,11]. The latter is based on the use of concentrated solar radiations as the driving force that produce hydrogen from transformation of its feedstock [4]. There are different thermochemical hydrogen production pathways, namely steam reforming [12][13][14][15], gas cracking [16,17] and water splitting [9,[18][19][20]. There have been several papers in literature that have reviewed existing thermochemical cycles for hydrogen production [21][22][23][24]. The investigations for thermochemical cycles which could be made efficient hydrogen production from water was started worldwide in the late 1960s by James E. Funk and RM Reinstrom [8,25].
James E. Funk [21] was the first to evaluate the energy requirements for the production of hydrogen from water and the possibility of using two-step thermochemical processes. Thermocemical cycles evaluation in [22] by Sandia National Laboratories have been investigated last decades by several researchers . From over 350 multistep thermochemical cycles initially selected [22], 14 cycles are found suitable for solar application. The most thermochemical cycles promising candidates are based on Ferrite-based metal oxides, Zinc oxide, tin oxide as well as sulfates iodine. An attractive cycle for solar hydrogen production is the two-step solar thermochemical water-splitting using redox pair metal oxides, since it implicates less complex reactions, non-corrosive and nontoxic materials. Recently, enormous efforts has been made in developing concentration systems that exceed 5000 suns solar concentration ratios, capable of achieving temperatures above 1400°C, which are needed to release two-step thermochemical reactions using metal oxide. Ferrite-based thermochemical cycle has been the most explored cycle [23], it was demonstrated at 3;100;750 kW level within the project HYDROSOL, HYDROSOL 2, HYDROSOL-PLANT successively. The SiC-based monolith reactor model, was similar to the catalytic converter used for automobiles exhaust gas treatment catalytic converter. The reactor is constructed from ceramic multichanneled monoliths that absorbs the solar radiation flux [24]. The Ferrite-based cycle has been extensively studied in the solar furnace of Tokyo Institute of Technology [25] and Sandia National Laboratories [22,26] with locally developed solar reactor prototypes. Production step For this chemical process, several solar reactors were developed and tested at PSI's solar furnace, ZIRRUS and ROCA reactors were demonstrated at 10 kW solar radiation. Temperature reached a value >1900 K for both reactors.
The SI cycle was demonstrated within the USA DOE project, EU-FP6 HYTHEC and EU-FP7 HYCYCLES projects. In order to build a low cost feasible solar reactor, a packed bed based on ceramic is proposed. The body of this reactor/ receiver consists of spherical balls made of silicon carbide coated with the reactive ferrites material. The idea of studying the packed bed solar receiver/reactor is interesting because: -The spheres/ beads are inexpensive compared to the monolithic or foam structured media, and can fill the space of the cavity without issues of peripheral support/packing. -The wash-coating procedures are much easier and homogeneous than other porous structure previously used geometry.
Packed bed based on particles may be designed in many cases as a set of identical spheres occupying a determined space.

Figure 3. Reactor packing arrangement
When this space is limited to a predetermined space , such as cylindrical channels of a nanostructured porous structure filled by atoms or molecules of a substance, the geometric & thermophysical properties (Thermal conductivity , porosity ..) of the resulting porous matrix strongly depend on the arrangement (Porosity as example in [37][38][39]. Packed beds are used in a large industrial thermal-fluid systems applications, such as nuclear reactors [40,41], catalysts [42], and in cryogenics [43,44]. A number of exploratory studies investigated solar thermochemical reaction in packed bed reactor [45][46][47][48]. In Figure 4 a 5 kW solar reactor, subjected to solar radiation up to 2953 suns and operated at temperatures up to 1490 K was demonstrated at PSI [49]. The aim of the present paper is to study the feasibility of using this kind of porous structure as a receiver/reactor for direct thermal water splitting. It is important to mention that this specific design of directly irradiated reactor using ceramic balls, has not been studied extensively before, and could be very interesting technically.

REACTOR DESCRIPTION
The receiver/reactor body (Figure 6) consists of a cylindrical chamber (0.05 m internal diameter and 0.05 m long) that contains the ceramic spherical balls. The front of the receiver/reactor is directly irradiated by concentrated solar energy (maximal power of 2 kW constant solar flux density distribution).

Figure 7. Reactor design
The chamber of the reactor is surrounded by an insulator. A glass window is proposed to separate the system from the ambient to provide a controlled environment inside the chamber.

MATHEMATICAL AND NUMERICAL MODELLING
The study presents an approach to the computational modelling of a volumetric solar receiver/reactor consisting of an absorption cavity that is filled by a packed bed of ceramic spheres and irradiated from the flow inlet face. The variation of cavity thermal response is examined for a different sphere packing arrangements while a comparison against third party experimental results is shown. The model of the bed of packed spheres is based on a two-dimensional scalar transport in a homogeneous porous medium for which distinct solid and fluid temperature fields are retained.
The model approach used in this model is macroscopic, which is based on the averaging of fundamental flow equations such as continuity, momentum and energy as well as turbulence over a representative volume of the porous media [51]. The most research efforts in computation of turbulent flow in porous media [52][53][54] have been focused on developing macroscopic equations through the averaging volume techniques of the k-ε turbulence model. The governing equation assumptions are the following: -Uniform porosity through the porous structure.
-The flow field is steady and turbulent -Effect of Buoyancy, Hydrodynamic dispersion, viscous dissipation are negligible.
-Thermal expansion is negligible.

Continuity and momentum equation
According to the cited assumptions, the averaged continuity and momentum equation for the proposed model are: Where α is the permeability of the porous media, C_2 is the inertial coefficient. The first and the second term on the right hand side are the momentum source terms. The equation neglects the convective term as the flow is fully developed. The turbulence model are obtained by averaging the standard turbulence kinetic energy and the dissipation rate equation.

= ℎ (< > −< >)
Where h fs present the coefficient of the convective volumetric heat transfer and A fs is the exchange area between the solid matrix and the fluid.

Convective volumetric heat transfer coefficient
According to the following references [55] and [56] the volumetric heat transfer coefficient is constant over the length of the reactor. Table 4 gives an overview of some models existing in the literatue.
The model used to find the heat transfer correlation and the interfacial area density is a general correlation, which can be used for all Particle Reynolds Number [62] ℎ = (2 + 1,1 .
The area correlation for this model is given by:

Effective conductivity
The Effective thermal conductivity is the thermal conductivity used in porous media, it is calculated generally through correlations. The most accurate formula [63] is used in this study. (The default model used in ANSYS FLUENT) = ( )

Permeability and inertial loss coefficients
Ergun Equation is a semi empirical correlation applicable over a vast range of Reynold numbers and it is applicable for many type of packing: From Ergun correlation, the permeability and inertial loss coefficients are identified as:

Species transport equation
In this work, the two steps reactions were modeled separately (reduction and hydrolysis). The main objectives of these simulations are to study the kinetics of the reactions and the evaluation of the impact of these reactions on the temperature distribution. The species transport equation is expressed as:

Reduction reaction:
The thermal reduction is an endothermic reaction, it needs a temperatures >1400 K [6]. In this case the source term related to the chemical reaction (using Arrhenius-type temperature dependence model) is given by:

Hydrolysis reaction:
The second step of the cycle is an exothermic reaction. This reaction is carried out at temperatures around 1200 K. The volumetric heat source term in this case is given by:

Boundary condition
The walls were considered as static and adiabatic with an initial static temperature of 300 K, the inlet pressure of the fluid was defined to have a pressure of (1000 Pa above the atmospheric pressure). For the outlet boundary, the pressure was set to a value of 0 Pa.

Methodology of resolution
The physical model is a packed bed symmetric cylindrical , a two-dimensional mesh was generated and used in this study. This two dimensional heat transfer problem was solved using the CFD Computational Fluid Dynamics software ANSYS FLUENT . To couple the pressure and the velocities parameters with the second order upwind method, the SIMPLE algorithm was applied. The second order upwind method was used for the advection terms in the momentum equations and for the energy equation. The LTNE model sources terms were computed through the UDF to couple the energy equations of the fluid phase and solid phases. The convergence criterion was 10 −5 for all the equations of the model. Analysis were first conducted for N 2 flow in order to determine the thermal behaviour of the packed bed; then chemical reactions were included into the model to analyse the reactions kinetics and their impact on the reactor thermal behaviour. The data of the chemical reaction used in this simulation are listed below: Table 5. Reactions data

RESULTS & DISCUSSIONS
To verify the model developed, the model was validated with experimental and numerical results obtained by [68] using the same thermophysical properties of the cited model . Figure 8 presents the variation of temperature through the position. The results are showed good agreement with [68] experimental and numerical studies. In order to optimise the performances of the packed bed solar receiver, we conducted a series of sensitivity study of the inlet velocity, the balls arrangement and the inlet solar flux.    Another parametric study was related to the arrangement of the ceramic balls in the packed bed that affects the temperature distribution in the solid especially near the front surface of the receiver. The Figure 9. Temperature distribution, Left hand side: Solid phase (Triclinic balls arrangement) shows that the arrangement of the particles affect strongly the temperature behaviour of the receiver / reactor (comparison between Triclinic and cubic arrangements) As expected solid and fluid temperatures decrease when accounting for the reduction reaction. Figure 13 shows the temperature distributions with and without the reduction reaction. The temperature of the solid and fluid phases tend to homogenise after about 1 cm from the inlet of the packed bed. Results shows that the effect of the reaction on the temperatures is limited, however, the chemical reaction only consumes a low fraction of the solar energy entering the reactor Figure 11. 02 mole fraction Figure 14 shows the distribution of the oxygen in the reactor during the reduction step for the symmetric and wall lines. The results shows high oxygen concentrations close to the wall caused by the viscosity effect.

Conclusion
In the present study, CFD analysis of a water splitting solar reactor is given. The local thermal nonequilibrium model was used to investigate the temperature distributions. Numerical study was conducted to study the effect of the inlet velocity, solar radiation, ball arrangement and the reaction. The simulations were also compared with experimental data. The results shows that the temperature distribution depend strongly with the working conditions and the porous media properties. The results shows that the ideal temperature distribution with the maximum solid porous structure temperature located inside the absorber. It is remarkable that the packed balls arrangement has an interesting effect on the temperature distributions inside the reactor. Regarding reactive flow simulations, the impact of the reaction on the temperature distributions was analyzed, a temperature decrease reaching about 60 K was evidenced due to the endothermic reaction. In this study, pulsatile flow in an intracranial aneurysm is numerically investigated under the combined effect of the aneurysm size, the curvature of the artery and the flow rates. The aim of this study is to evaluate the flow field and the wall shear stress distribution along the saccular cavity wall during a cardiac cycle and use these findings to have a better understanding of the mechanism leading to the rupture of the vessel.

A fs
Exchange area between the solid matrix and the fluid , k-ε model Inertial coefficient