Discussion of code predictions for COVA experiments

The computer codes ARES, ASTARTE and SEURBNUK have been developed within the European community for the analysis of the mechanical consequences of a hypothetical core disruptive accident in a fast reactor. ARES was developed in Germany by Interatom, ASTARTE was developed in the UK and SEURBNUK was jointly developed by the UKAEA and JRC Ispra. ARES and ASTARTE are Lagrangian codes whereas SEURBNUK is an Eulerian code. All three codes use finite difference formulations to solve the equations of motion accounting for dynamic coupling between a compressible fluid and thin flexible structures under axisymmetric conditions. These codes and other comparable codes contain non-verified simplifications and assumptions. Therefore, an extensive model assessment has been made in recent years by comparing code predictions with small scale explosion test results provided by the COVA experiments. The COVA experiments were jointly performed by the UK and JRC Ispra. These experiments progress from simple bare assemblies to tests incorporating the main axisymmetric features of loop and pool reactor designs with one new feature introduced at a time. The vessels were partly filled with water leaving an air gap above the water surface. An energy release was provided by a low density explosive (LDE) positioned on the axis. This paper briefly describes the main results and trends found in the comparison of the calculated and measured results in terms of loads on structures and thin vessel deformations. (orig./GL)