The solar system went into operation in April 2000. Table 1 shows the heat balances for the years 2001 to 2003.

In the first winter (2000/2001) there were some start-up problems with the heat pump. For this reason it was hardly possible to discharge heat from the ATES. This led to a higher fraction of heat that had to be delivered by the gas condensing boiler. The solar fraction still reached 32 % in 2001 due to a high direct usage of solar heat and the winter season 2001/2002 where the heat pump worked more reliably.

The total heat demand for space heating and domestic hot water in the first regular year of operation 2002 was 597 MWh/a. This was 20 % more than calculated during design (497 MWh). The solar collectors delivered a usable heat input of 278 MWh/a; 119 MWh/a were used directly, 158 MWh/a were provided by way of the ATES which worked with an energy return-ratio of 64 %. The electricity demand of the heat pump was 44 MWh/a, the gas boiler delivered 322 MWh/a. Referred to end energy use the solar fraction resulted to

43 %.

Figure 4 shows the monthly heat balances for the years 2002 and 2003. Due to optimized hydraulic adjustments and some improvements in the control system it was possible to increase the solar fraction to 49 % in 2003. Especially in the summer month the direct usage of solar heat could be increased (Figure 4). The summer heat load can still not be covered completely by solar energy. This is mainly caused by the strictly seasonal operation of the aTeS (see also Figure 6). During summer the ATES is in charging mode; to avoid blockings of the well screens the flow direction should not be changed frequently to discharging mode and back. Therefore in the summer months only the storage volume of the buffer heat store can be used for a solar heat delivery during night or during days without sun.

110

100

90

80

70

60

50

40

30

20

10

The yearly heat balance is indicated in an energy flow (sankey) diagram based on the results of the year 2003 in Figure 5.

balances of the ATES for the years 2002 and 2003 are illustrated. The already mentioned seasonal operation in a summer (charging) and a winter (discharging) mode can clearly be seen. Also an ATES-typical temperature decrease in the discharging mode can be observed. Only in the beginning of the discharging period a direct usage of the heat is possible. Afterwards the heat is discharged via the heat pump which has very good operating conditions in the beginning with COPs between 6 and 7 decreasing to approx. 3.5 at the end of the discharging period. The yearly mean values for the COPs can be found in Table 1.

For monitoring reasons, seven additional boreholes have been drilled to be able to place more than 50 temperature sensors into the storage volume. Figure 7 shows ground temperatures of one complete storage cycle starting from the end of the discharging period in spring 2003 (01.03.2003) to the end of the charging period (01.10.2003) and the following discharging period until the end of 2003. A small part at the top of the aquifer layer can be observed, where temperature changes occur much faster then in the other parts. Obviously in this narrow part the hydraulic conductivity is much higher than in the rest of the aquifer layer and by this the groundwater exchange takes place preferably in this part. Other temperature values show that this effect is not symmetrical around the well but with a stronger tendency in the direction shown in Figure 7. The consequence of this is an irregular shape of the storage volume and a slightly lower efficiency due to higher losses to the surroundings because of a bigger surface. This has been investigated more detailed by GTN by calibrating a 3-dimensional finite-element model for coupled flow and heat transfer with the monitoring data /2/.

In February 2001 a breakthrough of groundwater to the ground surface occurred at the cold well while discharging the store with a high flow rate. An investigation of the well showed large blocked parts in the screen which caused a higher pressure in the well. Additional failures in the connection of the well piping caused the breakthrough to the surface. An operation with a by 20 % reduced flow rate was still possible until the problem was fixed in August 2002 by cleaning the screen and installing a new well piping inside the old one.