Experiments were performed in a scaled annular cavity design, to investigate melt dispersal from the reactor pit when the reactor pressure vessel (RPV) lower head fails at low system pressure of less than 2 MPa. The fluid dynamics of the dispersion process was studied using model fluids, water, or bismuth alloy instead of corium, and nitrogen or helium instead of steam. The effects of different breach sizes and locations and different failure pressures on the dispersion were studied, specifically by testing central holes, lateral holes, horizontal rips, and complete unzipping of the bottom head.With holes at the base of the bottom head, the most important parameters governing the dispersion of melt are the hole size and the burst pressure. The fraction dispersed into the reactor compartments increases with larger holes and higher pressures. Values up to 76% have been found for both melt simulant liquids, water, and metal. With lateral breaches the liquid height in the lower head relative to the upper and lower edge of the breach is an additional parameter for the dispersion process, and usually not all the liquid is discharged out of the RPV. The liquid fraction entrained out of the RPV can be higher with a small breach than with a large one because of the longer blowdown time. With lateral failures, maximum dispersed fractions of 50% were found with water as melt simulant and less than 1% with liquid metal. It follows from similarity considerations that the results from the liquid metal tests represent the lower bound for the dispersed melt fractions; however, they are probably closer to the expected values than the results from the water tests, which represent the upper bound. So, significantly less dispersion of melt can be expected for lateral breaches at pressures below 2 MPa, probably less than 10%. If higher dispersion occurs, due to higher pressure at failure or with failures near the bottom center, simple devices to reduce the dispersion out of the cavity may be feasible.