Measurement of steam explosion loads: the ECO experiments

Full text of publication follows: Steam explosions might occur if, during an LWR core melt accident, molten corium comes in contact with water. They are still a safety concern during accident analysis. In principle, steam explosions can load reactor structures (in-vessel or ex-vessel) with high pressures. In addition, they may convert part of the thermal energy of the melt into mechanical energy, i.e. destructive work. The ECO experiments are being conducted to measure the explosion pressures and actually measure the Energy COnversion with relevant melt masses (of the order of 10 kg), which has never been done before. The occurrence of a steam explosion is assured by applying an external trigger. A real measurement of the mechanical energy is achieved by confining the explosion in a closed volume (a cylinder) equipped with a piston (that contains the water pool). In response to the explosion pressure, the piston moves downward, compressing a stack of crushing material with a known force. So the piston movement gives a direct measurement of the mechanical energy. The expected explosion pressures are high and conservative estimates of the possible energy conversion give quite high values. This requires a very strong facility. So, e.g. the piston alone weighs 2.7 t. In order to further characterize the events, the interaction vessel is equipped with a large number of pressure transducers, thermocouples and void sensors (to observe the development of the multiphase region during pre-mixing). The alumina from a thermite reaction (2600 K) is used to simulate the corium (the iron is held back). It should be noted that, with respect to volume and thermal energy, 10 kg of alumina correspond to about 30 kg of a corium that largely consists of urania. Therefore the scale of ECO is comparable to that of FARO rather than that of KROTOS. Also the test vessel has a diameter of 0.59 m. In the seven tests performed up to now, the water mass was always around 280 kg, the largest melt mass released was 16.4 kg. In this test the water mass interacting directly with the melt was reduced by a restriction tube with only 0.3 m diameter. This test resulted in explosion pressures of the order of 100 MPa - the pressures were beyond 45 MPa for more than a millisecond - and an energy conversion of 2.4 %. In a repetition test, quite similar conditions were achieved but the duration of the high pressures was shorter and the energy conversion was only 0.9 %. Detailed results will be reported. It might be that the relatively small mechanical energies are due to an over-strong confinement of the explosion (by the heavy piston plus the crushing material) and fast quenching of supercritical fluids by the large amount of cold water present. These topics are still under investigation. Another area of theoretical investigation are the rather peculiar pressure histories observed during the pre-mixing phase when the melt is released into the interaction vessel. The typical initial pressure is about 0.25 MPa. Directly after melt-coolant contact a very steep pressure rise (several ten mega-pascals per second) is observed that is followed by a more slow pressure rise, an almost constant pressure, or even a slight decrease at a level of about 1.0 MPa. In the two tests discussed above, another (not as) steep pressure rise occurs towards the end of the pre-mixing phase, so that the pressure at triggering is around 2.0 MPa. A deeper understanding of these pressure histories comes from detailed simulations of the experiments with the MATTINA code. The results will be reported. (authors)