Describes a computer program called LIGHTS that may qualify as a 'numerical test cell' for lighting/HVAC interaction studies. States the program is general in that it allows a room to be defined in terms of an arbitrary number of surfaces, enclosed air masses and interconnecting airflow paths. Diffuse, short and long-wave radiation is rigorously modelled, as are conduction, surface-to-air convection and bulk-flow convection. Heat balances at each surface and air node form a differential algebraic equation set that is solved using standard numerical techniques. Modelling control parameters allow either steady-state or transient solutions and different boundary conditions without changing the model itself. Demonstrates exact solutions to simple problems, followed by comparisons with preliminary National Institute of Standards and Technology test cell results. Argues the results are within the range of experimental discrepancies. Shows in addition that the program can be used to generate weighting factors for simplified models often used in design-oriented tools. Presents arguments for the use of numerical rather than physical test cells for schematic investigation of complex systems, and that the proper role of experimentation is determination of model coefficients.

KEYWORDS: Calculating, computer programs, testing, lighting, heating, ventilation air conditioning.