This paper outlines the results of a physical simulation (at a 1:700 -1:1000 geometric scale) of a thunderstorm downburst. Three different methods are examined in order to generate the time dependent nature of a downburst: directly controlling the fans and via two different types of opening apertures. Similarities are shown to exist between each method, although the results obtained from one approach are favoured since they appear to be independent of the downdraft velocity. Significant run-to-run variations between each experiment are discovered and in general it is found beneficial to interpret the results in terms of 10 run ensemble averages. An attempt to simulate a translating downburst is also undertaken and the results are shown to compare favourably with full-scale data.