COMPUTATIONAL METHODS FOR PASSIVE SOLAR SIMULATION

The use of network models and a range of computational methods for the simulation of passive solar buildings are described, compared, and assessed. The following methods are considered: (i) Steadystate methods; (ii) Finite difference methods, explicit and implicit; (iii) Modal or analytic spectral methods; (iv) Fourier series methods. Methods (ii) to (iv) are compared by applying them to a series of numerical examples using the same data. In particular, the effect of network nodes with small thermal mass and the effect of increasing network size are considered. With explicit methods it is shown that it is important to approximate building components with small thermal mass by zero-mass nodes. Generally, the modal and implicit finite difference methods are the most efficient. © 1990.