Tracer travel time and model validation

The performance assessment of a nuclear waste repository demands much more compared to the safety evaluation of civil constructions such as dams, or the resource evaluation of a petroleum or geothermal reservoir. It involves the estimation of the low probability (low concentration) of radionuclide transport extrapolated thousands of years into the future. Thus, models used to make these estimates need to be carefully validated. The present paper discusses some issues of validation regarding radionuclide transport. In model validation, it is important to consider the proper performance measures, so that we are not addressing a measure that is too difficult to predict and that is actually not required for the performance assessment. It turns out that point predictions are perhaps not necessary, but averaged line or areal predictions are more feasible and may be all that are needed. This conclusion, which is very tentative, may have significant impact on the design of experiments that will be used for model validation