An evaluation of a computer code for predicting indoor airflow and heat transfer.

The evaluation of a code can be done by investigating two items: solving the correct equations and solving equations correctly and eficiently. An indoor airflow code VentAirI has been developed and is evaluated here. An evaluating procedure is suggested. The code is characterized by the standard high-Reynolds-number k-E model with wall function, the two-band radiation model and the SIMPLE algorithm. Test examples are: 1. A three-dimensional forced convection problem (Re=5000), 2. A natural convection problem (Ra=5 *10^10), 3. A natural convection-radiation interaction problem (Ra= 1.45 *10^9). All calculations are compared with experimental results and published numerical solutions. Grid refinements are used to improve the accuracy of the predictions. The applicability of the Boussinesq approximation is confirmed. The prediction of heatflux through the boundaries are, however, less accurate. The code exhibits a relatively low convergence rate; the finer the grid, the slower the convergence. A fast multi-grid solver combined with local grid refinements is suggested. Consequently, another indoor airflow code Ven&irII is developed.