The selected store consist of a vertical tank made of plexiglass of approximately 200 liters. The tank length is 1.0 m with an internal diameter of 0.5 m. The Plexiglass material is 8 mm thick. The tank has two inlet/oulet ports located at the top and bottom walls. The ports have an inner diameter of 0.04 m. Two plates with a diameter of 0.30 m and located 0.04 m away from the ports are use to reduce the inertia of the entering mass flow rates.

The PCM modules consist of rectangular section rings (0.01×0.02 m) located at different positions of the tank. The PCM used is a commercial hydrate salt (Sodium Acetate) with a latent heat of 240 kJ/kg and with a melting temperature of 56 C.

Three devices have been tested: a) a sensible storage device without PCM modules; b) an hybrid device with a 6% of its volume occupied by PCM modules; c) an hybrid device with a 10% of its volume occupied by PCM modules. These devices will be hereafter referred as designs S, H6 and H10 respectively (see Figure 1).

The stores have been tested considering the assay C in ENV 12977-3. The test has been performed as follows: i) Charging the store with a mass flow rate of 50 l/h and with an inlet temperature of 60zC until the outlet temperature at the bottom of the store exceeds 55 C; ii) Discharging the complete store with a mass flow rate of 100 l/h and with an inlet temperature of 20 C. Ambient temperature is fixed to 20 C.

CFD results

Considering the geometry of the stores and the working conditions explained above, the following hypothesis have been assumed: the heat transfer and fluid flow phenomena is axialsymmetric, convection is negelected in the phase change phenomena involved in the PCM modules, the physical properties are constant in accordance with the Boussinesq ap­proximation (density variations are only relevant in the buoyancy terms of the momentum
equations), the fluid behaviour is Newtonian, the viscous dissipation and the influence of pressure in temperature is negligible, and the radiating medium is non-participant. The ther­mal loss of the tank have been modelled considering heat transfer coefficients of 3 W/m2K.

The computational domain has been discretized using cylindrical coordinates. The size of the control volumes (CVs) has been maintained almost constant throughout the domain, except for the zones near the inlet/outlet ports. The computational grid was composed of nTx CVs. The simulated time was discretized using a constant time increment.

In order to verify the numerical solutions obtained, and to analyse the sensitivity of numer­ical parameters that account for the discretization (mesh spacing and time increment), the store without PCM modules has been simulated considering different grid densities (16×53, 32×106 64×212 and 128×424 CVs) and time-steps (0.25 s, 1 s and 4 s). After this analy­sis, a discretization of 32×106 CVs and a time-step of 1s has been selected to perform the numerical simulations.

In Figure 1 illustrative temperature countours for the charging sequence and for the tested devices are plotted. As can be observed, the thermal behaviour of all tanks are very similiar. Thermal stratification is clearly showed. However, the thermocline thickness is slightly higher in the case of H10 design.

In Figure 2 temperature profiles are given for the discharging test sequence. As can be

seen, after 2 hours the store without PCM is practically discharged while in the hybrid stores PCM modules of the top part of the tank, still have a high temperatures.