Author(s): Rui Aleixo; Sandra Soares-Frazao; Yves Zech
Keywords: No Keywords
Abstract: The author would like to express his thanks to Prof. Chanson for starting the discussion and for providing results from other well-known transient cases, that allow for a broader view of the problem. It is never too much to stress the importance of assessing the repeatability of a phenomenon by performing the same experiment several times in the exact same conditions. This is often a problem, since sometimes an experiment cannot be physically repeated (e.g. field measurements or destructive test), or because repeating the same experiment several times is expensive. It must also be mentioned that repeating an experiment is important to determine the type A uncertainties (GUM, 2008; Muste et al., 2017; VIM, 2012). When repeating the same experiment is possible, the number of repetitions must be assessed, namely by using some pre-established criterion, for example, the convergence of a given parameter such as the mean values. Several expressions for this can be found in literature, namely, Yanta and Smith (1973), Hitching and Lewis (1999) and Durst et al. (1996). These expressions, obtained for stationary flows, lead usually to a number of samples of hundreds to thousands, which is often not practical for transient cases even in a laboratorial environment. The suggestion made by the Discusser of 25 repetitions seems a good compromise, provided there is some degree of convergence of the variable of interest (mean velocity, Reynolds stresses, etc.), a fact also recognized by the Discusser.