The PACTEL GDE-24 experiment was calculated with APROS 4.02 code [18], [19], [20],

[21] . The models of the CMT, PBL and IL were included to the APROS model of PACTEL

created for SBLOCA calculations. The CMT was modelled with a standard pipe module of

APROS divided in 30 equal length nodes.

• APROS calculated successfully the recirculation phase.

• Due to numeric diffusion the temperature profile was not as steep as in the experiment.

• Problems occurred during the draining mode. Injection flow started to oscillate continuously never reaching the anticipated full magnitude. The vapour entering to the top node condensed directly to the subcooled water and caused a flow stagnation due to rapid pressure drop. The injection was possible only after the water had reached saturation temperature in the boundary node. Hence, the explanation for these flow oscillations was lack of continuous existence of saturated liquid layer, which would prevent direct contact of vapour and cold water.

• Oscillating characteristics caused delay to the timing of main events.

• Important parameters were maximum time step, amount of nodes and hydraulic diameter in the CMT. To reduce the condensation it was possible to manipulate the condensation heat transfer coefficient by giving higher values to the hydraulic diameter in the CMT nodes. To get more accurate calculation results, it was necessary to use dense CMT nodalization (30 nodes) and small maximum time step (0.025 s) during the first 1000 s transient period.

For the calculation of the GDE-34 experiment a smaller CMT model with 30 nodes was created and the CMT water was initiated with warm water. Some minor modifications for the PBL pipework were also made.

• The recirculation mode existed in the calculation though it was not observed in the experiment. Hence, the density difference between PBL and CMT was enough to initiate the flow in the calculation.

• Injection flow oscillated, but with smaller amplitude than in GDE-24, because of less condensation. The average injection flow rate was quite near to the measured one. Also, flow injection stagnated once due to temporary water level increase in the vertical section of cold leg 2, which is connected to the PBL.

• The increase of hydraulic diameter reduced condensation also in this case.

• The calculation was not very sensitive to any other parameters.

• Timing of the main events agreed well with the experiment.

For the calculation of the GDE-43 experiment a smaller CMT model with 30 nodes was also used.