The solar system considered in this work is a large hot water production system suitable for a small hotel, blocks of flats, offices or similar applications. Although the FDS system developed can be applied to small systems as well it is thought that the expenditure required would not balance the extra benefits incurred in such cases and in domestic applications the users are usually more sensitive to the maintenance of their own system in comparison with the maintenance staff of a hotel for example or the tenants of a multi building installation where everybody but really nobody is responsible. The system schematic is shown in Fig. 1. The system consists of 40 m2 of collectors, a differential thermostat (not shown in Fig. 1) and a 2000 L storage tank. The system is also equipped with a data acquisition system which measures the temperatures at four locations of the SWH system; the collector array outlet (T1), the storage tank inlet (T2), the storage tank outlet (T3) and the collector array inlet (T4). The global solar radiation, the ambient temperature, and the pump state (on/off) are also recorded.

In fact, a TRNSYS model is used to simulate the system. Two draw off profiles have been used. The first one is repeated day after day, the second one is computed using the free generator developed at the Univeristat Marburg [7] and used in [8]. Being totally different, these profiles will show that the results do no depend on them. In a real application, the temperature readings would be affected by noise. So, it has been decided to add random noise to each computed temperature. It has to be noted that the time step is one hour.

Four drifts or defaults are taken into account in this study. For the collector, two parameters are considered: F’ (linked to the fin efficiency), and UL (linked to the thermal insulation); for details see TRNSYS manual. For F’, a progressive decrease is computed so that the value decreases from 0.7 to 0.6. For UL, a sudden increase is considered (from 3 to 4 W/m2.K) followed by a progressive increase (from 4 to 5 W/m2.K). For the connecting pipes, the variations of the U value are similar to the variations of the UL value. These drifts will be shown in the last section when presenting the results. It has to be noted that it has been necessary to write our own components for TRNSYS to be able to read the values from files, which allow continuous variations of the parameters (one value corresponding to one hour).

The drifts have been both considered separately and combined. In the latter case, which leads to the lowest performance of the system, the yearly increase of the auxiliary electrical power needed to deliver the hot water is less than 7.5%. It can be concluded that if the FDS is able to detect the faults before the end of the drifts, it is sufficiently efficient.