Advanced nuclear reactor concepts heavily rely on the availability and efficiency of passive cooling systems. This especially holds for advanced containment designs with passive decay heat removal systems that function by natural phenomena. Also, the development of catalyst modules for hydrogen mitigation measures is based on natural basic principles leading to hydrogen reduction and additional atmospheric mixing.

To prove the functionability and availability of passive systems and their respective components, demonstration experiments at different scales are mandatory. In addition, it is the general perception that many more improved computational tools are needed for this purpose, where present lumped-parameter analysis methods are insufficient to provide the necessary information about local details and spatial distributions. Therefore, the next step in the development of analytical/numerical models is the transition/extension from lumped-parameter to multidimensional models and containment analysis codes.

Also, recent posttest lumped-parameter analyses of the Heiss Dampf Reaktor H2 distribution experiment E11.2 with preexisting atmospheric stratification show a number of deficiencies compared with the data, indicating a need for more detailed modeling.

The GOTHIC thermal-hydraulic containment code provides this required extension of the lumped-parameter model by incorporating multidimensional submodels for selected nodes (subcompartments). Applications of both model types to simulate hydrogen dispersion experiments in the Battelle Model Containment (BMC) demonstrate the limitations of the traditional approach and the improvement achieved by the multidimensional simulation. The importance of thermal and hydrogen concentration stratifications, the interactions with structural heat conductors, and the requirements to set up a consistent model when coupling lumped-parameter and multidimensional representations are discussed.

Several hydrogen-mixing experiments performed in the BMC more than a decade ago were simulated with multidimensional GOTHIC models.

Three types of modeling concepts have been tested:

1. lumped-parameter model

2. each compartment modeled two-dimensionally with the intercompartment connections simulated as flow path junctions

3. full three-dimensional nodalization of the BMC, intercompartment connections simulated as gaps.

The results of these GOTHIC calculations are compared with the experimental data and demonstrate the improvements that can be achieved by performing multidimensional containment simulations.