Mean concentrations of ammonia gas released as a tracer from an isolated lo
w-rise building have been measured and predicted. Predictions were calculat
ed using computational fluid dynamics (CFD) and two dispersion models: a di
ffusion model and a Lagrangian particle tracking technique. Explicit accoun
t was taken of the natural variation of wind direction by a technique based
on the weighted summation of individual steady state wind direction result
s according to the probability density function of the wind direction. The
results indicated that at distances >3 building heights downstream the weig
hted predictions from either model are satisfactory but that in the near wa
ke the diffusion model is less successful. Weighted solutions give signific
antly improved predictions over unweighted results. Lack of plume spread is
identified as the main cause of inaccuracies in predictions and this is li
nked to inadequate resolution of flow features and mixing in the CFD model.
Further work on non-steady state simulation of wake flows for dispersion s
tudies is recommended.