Nominal run and uncertainty analysis

Within the toolbox two types of simulation runs are distinguished: the nominal run and the uncertainty analysis loop.

The nominal run represents a single simulation which is aimed to compute the most likely, unbiased estimates for the specified model. E.g., when a dietary exposure assessment is requested, in the nominal run a single exposure distribution is estimated using nominal values for all data and parameters.

In the uncertainty analysis loop, each simulation run is repeated a large number of times. Each run starts with a different scenario using data obtained with bootstrapping, parametric resampling and/or re-calculation of uncertain values. As a result, a large number of uncertain dietary exposure distributions is estimated which are used to estimate uncertainty limits (p5, p95).

Running a nominal run first has the advantage that the user may evaluate these modelling results before doing the final analysis. The model specification of rather complex simulation models and the corresponding output results are evaluated to detect any errors or misspecifications. Possible errors in the data and/or model settings are identified and corrected. In the final exposure assessment, the uncertainty analysis is included and the uncertainty of estimates is assessed.

Figure 2 Uncertainty analysis settings.