A thermal network based on a control volume finite difference discretisation of the house was used. The model used was implemented in MATLAB [4]. Two zones were used: a basement zone and the rest of the building. MATLAB subroutines were written to model the performance of the BIPV/T system, the heat exchanger/heat pumps group, the ground source/HPs interaction, and the TES reservoir.

The BIPV/T model [1, 3] was based on dividing the channel lengthwise in a given number of sections, assuming uniform temperatures at the top and bottom surfaces. An exponential profile is assumed within each section; temperatures and heat transfer rates are found iteratively using a 1-D thermal network between the air and the surfaces. The exit temperature of each section is used as the entrance temperature of the next section.

A simple model based on the variation of effectiveness of the heat exchanger with varying flow rates and manufacturer’s data was used to model the performance of the HX and one or both HPs

[2] . This model was modified to use an adjusted temperature of the ground as an input.

Finally, a 4-node simple model, as described in [5], was used to model the TES reservoir. In this model, the water in the TES is divided into four nodes (a convenient model since there are also four internal divisions in the tank). The water entering the tank (either from the radiant floor heating system or the HPs) is assumed to go to the node having the highest temperature below that of the incoming water. Simple water exchanges are then assumed to take place between the nodes to guarantee mass flow rate balance. According to [5], the use of 3 or 4 nodes is a good compromise between an accurate model and a conservative design (which would assume only one node).