Simulation of IB-LOCA in TRACE : A semi-blind study of numerical simulations compared to the PKL test facility

This thesis studied the performance of the thermal hydraulic software TRACE applied on an intermediate sized break (IB) happening on the cold leg in a pressurized water reactor (PWR), causing a loss-of-coolant accident (LOCA). The same accident has previously been simulated in the PKL Test Facility, which is a scaled version of a PWR and is used to simulate transients stemming from different accidents. The thesis was performed as a semi-blind study: firstly, the accident was simulated without any knowledge of the PKL results. When a final blind model was chosen, the PKL results were revealed, and the TRACE model was improved. Before the simulations of the IB-LOCA took place, the new internal parts in the upper parts of the reactor pressure vessel in PKL had to be modelled, and the steady state had to be tuned to attain the correct initial conditions. The simulations were performed by using the software SNAP together with TRACE, providing a graphical interface. TRACE achieved steady state with satisfying results regarding water levels, pressure losses and mass flows. The temperatures in TRACE deviated from the PKL temperatures but an explanation is uncertainties in PKL. To verify TRACE’s core output power, the calculation of the power was done by using mass flow rate and specific entropy and comparing to the heaters’ specified power. This resulted in lower output power meaning that the coolant was not heated enough. This indicated non-physical energy losses in the TRACE model and should be further investigated.The blind transient simulation, modelled with default choked flow and no offtake model, resulted in TRACE overestimating the break mass flow and the peak cladding temperatures, compared to the PKL reference solution. This resulted in the pressure decreasing too quickly and too early activation of the safety system. The modified simulations showed that it is important that the offtake model, which accounts for different flow regimes, is activated. Default choked flow multipliers were the multipliers that performed the best. However, none of the transient simulations could be completed due to fatal errors and memory problems, but some conclusions could be drawn from the observed trends. This concluded in the offtake model being most important due to stratified flow occurring