The reactive bed is compacted in the reactor between the gas diffuser which is, in that case, fixed by rigid wedges, and the heat exchanger wall. A controlled pressure is applied to the coolant and therefore under the heat exchanger wall. During reaction, the coolant level and the displacement sensor give information on the relative position of the wall.

The reactions can occur after compacting. Thermodynamic constraints were fixed, i. e. the heat exchanger and evaporator/condenser temperatures were chosen according to thermodynamic equilibrium of solid/gas reaction and liquid/gas reaction [1].

The overall reaction advancement X is the number of mole of hexahydrated salt (which is deduced from the level in the liquid reacting water tank) divided by the total number of salt moles in the block. X is time dependant, and it is include between 0 and 1.

To simplify the identification of heat and mass transfer parameters, some constraints were applied:

• Temperature and pressure are fixed in the stability domain of the reactive salt i. e. at the boundaries of the reaction (X = 0 and X = 1).

• Transfers are assumed 1D because of the small ratio thickness / diameter of the sample, from 10 to 50 mm / 300 mm. Moreover, the external boundary is thermally insulated and airtight.

The measurements from the heating wire, thermocouples in the reactive bed and fluxmeter were used to identify the effective thermal conductivity. This identification uses the Fourier’s law in steady state.

Owing to the range of particle size and working pressure, the diffusion through the porous media is controlled by Darcy’s law. Nevertheless, the Knudsen diffusion could be envisaged. There are 3 categories of flow according to the mean free path of the fluid through the pores of the porous media.

In our case, the mean free path is close to the mean diameter of pores. The convective flow and the diffusive flow are not negligible because of Darcy’s law and Knudsen’s law respectively, equation 2.

к dp Dk dp p dz p dz

The Darcy’s law is modified by the introduction of the Klinkenberg’s coefficient b which depends on the Knudsen diffusivity Dk, the dynamic viscosity g, and the permeability k. The pressure evolution through the reactive porous media is given by this equation combined with the mass balance equation, equation 3. A numerical solution has been implemented using the commercial software: COMSOL®.

The determination of the permeability and the Klinkenberg’s coefficient was carried out in transient state. First, the gas tank volume and the gas volume above the reactive block are disconnected. Each of them is filled by steam up to the pressure pg and pd, respectively for the gas storage tank and the dead volume in the reactive bed. Then, the two volumes are connected together. The experimental evolution of the two pressures versus time is compared with the simulated one. The permeability and the Klinkenberg’s coefficients are determined by these pressure evolutions. Each volume pressure is simulated according to equation 3. The typical simulations are presented in figure 3. The first equalization of pressure Peg characterizes the gas tank volume and the dead volume above the reactive bed. As this pressure equalization occurs in less than 1 ms, we assume that it does not involve the porous volume of the reactive bed. After this first step, the gas diffuses through the porous volume of the reactive bed, and the pressure evolves from (peg, teg) to (pec, tec). The identification of k and b is carried out between these two points.

Pg

teg< 1 ms

Figure 3 : Typical evolution of pressure versus time in transient state