Simulation of the solar circuit for the Federal Environmental Agency's solar cooling system Dessau, Germany

The solution for the problem mentioned in the previous paragraph seems to be minding the night behaviour of an evacuated tube solar collector, working with the heat pipe principle, in the case of a mass flow is allowed.

Heat pipes, which are responsible for the heat transfer during the day, could work as insulators during the night. During periods of no solar radiation, fluid inside heat pipes is in the liquid phase in the bottom part and does not evaporate. The top of the heat pipe is supposed to be a vacuum or with a little amount of vapour; this uncertainty does not influence the final conclusion. If the hot fluid from the circuit flows into the double-pipe heat exchanger placed in the top of the solar panel, a reverse heat transfer seems to appear.

This is not realistic: heat pipes cannot work in reverse condition. If the top of the heat pipe (the only part in contact with the circuit pipes) is a vacuum ambient, heat transfer will be avoided; if vapour is accumulated into the top, it will receive heat but the natural convection to the bottom will not appear, due to the gravity.
The only way to lose heat from the circuit to heat pipes during the night is conduction into the structure of the heat pipe and radiation.
These considerations can be summarized writing that the heat loss coefficients (a1, a2) coming from solar collector tests are not corrected for the night. They considered heat losses during sunny conditions, from all the surface area of the panel. During the night, these coefficients as well as the use of overall area introduce an error, as above justified. Unfortunately, tests are only for incident radiation conditions; no data are available when solar panels work in the reverse situation.

What is possible to do for a complete simulation of solar collectors is to use two different models. The simulation run using the standard model agrees with the reality during the hours of incident solar radiation; data from test reports can be used. For the remaining hours of the day, a different model should be employed. The model for the lumped heat losses (Type 908) could fit the problem if the right value of the parameter “heat losses” is set.
Because there is no availability of heat loss coefficients for the night, a rough value for the parameter can come from the Equation 3.4.4, setting GT equal to zero and using a decreased value of Ac, which excludes the heat pipes surface and considers only the surface of the heat exchanger on the top of the collectors. The overall model should include a controller based on the solar radiation which automatically switches the use of the two models during the day.