Statistical distributions of validation performance indexes for experimental data uncertainty

The behavior of a set of performance indexes used for model validation is considered and analyzed with Monte Carlo simulations and determination of distributions of output quantities and their statistical estimates. A relevant result regards an offset appearing at index output, when considering input data uncertainty, for cases of good correspondence between model and experimental results and consequently very low index values near zero. Such offset represents a systematic error of indexes considered as measuring tools. Additionally, index output is also characterized in terms of dispersion.