The tests were disturbed by the clouds passing in the sky and by the drift of the heliostats during the solar tracking. The second effect was sensibly reduced after the change of the tracking algorithm in the control software of the heliostats field. For this reason, the last days of tests were less affected by instantaneous defocus to correct the aiming point, and the system behaved more stably. It yields more quasy-steady states to compare the results. Thus, only the results of the square metallic cups tests are presented, in reason of clarity. Moreover the position of the inner thermocouples is better fixed, because they were installed by Emitec during he manufacturing and they could not move, what leads to a higher data reliability of these tests.

For a roughly constant value of the solar radiation flux, the axial temperature distribution around the central channel is shown in fig.3 for different mass flow rates per cup. It is assumed that the mass flow rates per cup are all in the same range and then, the mass flow rate per cup is estimated by dividing the overall mass flow rate per the number of cups (35 ones for the square cups tests, since once cup of the Solair receiver was removed and closed). In this chart, the wall temperatures are plotted versus the parameter x/L, being x the distance from the irradiated entrance along the channel and L the total channel length. Inside each cup, a similar temperature profile is maintained for the three flow rates, but the mean temperature increases as the mass flow rate decreases, as it was expected.

The influence of the cell size on the axial distribution is also obtained from data depicted in the figure: for a smaller cell or channel diameter, the maximum wall temperature is reached earlier, and then the temperatures decrease with a stronger slope. On the other hand, for absorbers with a larger cell size, the maximum temperature is reached deeper in
the channel. With only 4 data we cannot assure, as it seems in the chart, that the maximum for the 500 cpsi absorber is located at a x/L value of 0.43, which corresponds to a distance of 30 mm from the entrance. It should be somewhere between 10 and 30 mm, but the positions chosen for the thermocouples do not allow to determine the location of the maximum temperature. Similarly, for the absorber with 600 cpsi, the maximum temperature should be near the thermocouple located at 10 mm on the channel axis.

These temperature profiles give information about the penetrability of the radiation into the absorber channels: for a higher cell density (600 cpsi), the cell size is smaller and the rays penetrate less than for an absorber with lower cell density (500 cpsi). This behaviour is related to the volumetric effect, since the radiation is absorbed in the first millimetres from the entrance and not only on the aperture surface. Absorbers with wider channels have more “volumetricity” since the volume irradiated is bigger, but the specific surface to absorb and transfer the heat are both smaller, resulting in lower temperature values.

The behaviour for different solar radiation fluxes under the same mass flow rate is as expected: higher radiative fluxes lead to higher absorber temperatures, but the temperature distribution is similar in all cases. In fig. 4 only the results for the cup with 600 cpsi are presented, the profiles of the other absorber are analogue to those shown in the previous chart.

Air temperature distribution at the absorber outlet:

To obtain the temperature distribution for the air exiting the absorber, the data from the thermocouples located in several positions at a cross section 2 cm behind the absorber have been interpolated in Matlab with the function called Griddata, which fits a surface to the data in nonuniformly-spaced vectors. With the ‘v4’ method, griddata produces smooth surfaces. Once the data are interpolated for the defined grid, the level contours of air outlet temperature are plotted.

The temperatures extrapolated near to the edges of the square absorbers must not to be considered because there are not experimental points near the edges to function as boundary conditions. For this reason the calculated temperatures are too low with respect to the real temperatures in the square corners.

The main difference between the distribution of the two absorbers is caused by the temperature at the absorber centre: in the 600 cpsi absorber the central temperature is in the same temperature range as the thermocouples next to it, while the central temperature of the 500 cpsi absorber is always lower than the temperatures at each side of the center. The colder central region in the 500 cpsi absorber suggests a non-uniform air mass flow distribution through this cup. The air flow distribution through the 600 cpsi absorber seems to be more uniform. From this, it is possible to conclude that the air flow distribution for both cups is different.

On the other hand, the warmer region that is observed at the left upper part of the 500 cpsi absorber may be due to the fact that the solar radiation was focused at the centre of the Solair, which is located in that position regarding to the 500 cpsi absorber cup (position 22).

Other important aspect is the great difference between the absorber wall temperature and the air outlet temperature. In volumetric solar receivers the air is supposed to reach the same temperature of the absorber matrix before exiting the channel. However, in these tests between the wall temperature at 20 mm from the absorber exit and the air temperature measured 20 mm after the exit, there is always a difference about 150°C for the 600 cpsi absorber and about 70°C for the 500 cpsi absorber. This means that the air does not reach the same temperature as the absorber in the first centimeters of the absorber channel, as some simulations suggest. Air temperature is lower than the absorber temperature all along the channel. Thus, air is always cooling the absorber. Nevertheless, in the first part of the
absorber channel, radiative heating up dominates over convective air cooling. From the point where radiation has no more effect over the channel walls, the absorber temperature begins to drop, more rapidly when that point is nearer to the entrance.

The extrapolation of the temperatures inside the absorber, considering only the last falling slope, gives quite a good estimation for the air temperatures.

A way of comparing the relative "heat transfer efficiency” from the absorber monolith to the air for each absorber module is to calculate the relative difference from the last absorber wall temperature (measured in the central axis at 20 mm from the exit) to the air outlet temperature in the same axis. The results show a better relative efficiency for the 500 cpsi absorber, but the absolute values are lower than for the other absorber. The variation in mass flow rates does not affect to the relative temperature decrease, while the variation in radiative fluxes has a slight effect: for higher radiation, worse heat transfer.

Conclusions

— The temperature distribution along the channel for a determined channel geometry (or absorber cell density) is the same for different mass flow rates or solar radiation fluxes. It achieves a maximum value inside the channel and then the temperature decreases due to the heat transfer to the cooling air.

— For a smaller cell or channel diameter, the maximum wall temperature is reached earlier, and then the temperatures decrease with a stronger slope. On the other hand, for absorber with a larger cell size, the maximum temperature is reached deeper in the channel.

— — The great difference between the absorber wall temperature and the air outlet temperature indicates that the studied absorbers are not sufficiently efficient to transfer all their energy to the air. The relative heat transfer efficiency is better for lower cell densities.

To sum up, the studied absorbers do not behave as it is expected for an ideal volumetric solar receiver, where the maximum matrix temperature would be at the end of the channel and the air would reach the same temperature of the absorber matrix before exiting the channel. These experimental results can be used to validate theoretical (numerical or analytical) models that reproduce the heat transfer in duct volumetric solar receivers.

Acknowledgements

The authors acknowledge the collaboration with the German enterprise Emitec, and the efforts of Mr. Arndt-Udo Rolle in the manufacturing of the samples are specially appreciated. Results discussions with Ma Jesus Marcos are also very appreciated.