A full-scale, laboratory building section at Silsoe Research Institute was adapted to simulate cross-ventilating air#ow patterns representative of those that develop in naturally ventilated, intensive livestock buildings. In the fully controlled, steady-state environment of the building section, the distribution of aerial pollutants and/or tracers was shown to depend on source location and ventilation rate. No single location within the building showed a consistently representative concentration of tracer that would be suitable for calculating ventilation rate by the constant release tracer method. Estimates of ventilation rate using the constant release tracer method, based on the average concentration of groups of six sampling locations within the building section, ranged from 93 to 119% of the ventilation rate measured using calibrated fan-wheel anemometers placed in the ventilation ducts. The best estimates of ventilation rate were obtained when the tracer concentration was measured at the perimeter (in the outlets and inlets) of the building section*these ranged from 99 to 109% of the ventilation rate measured using the fan-wheel anemometers.
Air#ow and pollutant distribution in the building section were predicted using computational #uid dynamics (CFD). However, these predictions were of limited help as a tool to select sampling locations within the building section, because the use of CFD requires the selection of parameters which entails veri"cation of predictions against experimental data. Visualization of the air#ow by smoke was simpler and more suitable for the present purpose.