PERSEO Benchmark - UNIPI results

In the framework of the OECD/NEA/CSNI/WGAMA group, an activity on the "Status report on thermal-hydraulic passive systems design and safety assessment" has been conducted. Within this activity, a benchmark exercise, based on the experimental data developed in the full scale PERSEO (in-Pool Energy Removal System for Emergency Operation) component separate effect test facility has been proposed and carried out. An "OPEN" benchmark exercise, hosted by ENEA, has been conducted. PERSEO is a full scale facility aimed at studying a new passive decay heat removal system operating in natural circulation. It was built at SIET laboratory in Piacenza (Italy) modifying the existing PANTHERS IC-PCC facility. It should be underlined that there is no a direct scaling relation between PERSEO and the PRHR of the AP-600, IC of SBWR or other designs. However, the heat transfer data developed in PERSEO facility could be useful for the generic assessment of system codes for the full-scale heat transfer phenomena in HX in pool (in-tube side and pool side) of passive systems. Among the four actual tests performed in the PERSEO experimental campaign, Test 7, which is divided into two parts, has been chosen for its completeness: the system stability and the long term cooling capability. The main targets identified for the benchmark are: • Investigation of the exchanged power from the HX to the HXP versus the HXP level; • Prediction of the mass transfer between the two pools when the triggering valve is open, steam flow accelerated into the OP through the injector and characterization of possible instabilities due to the steam/cold water interface in the Injector; • Test the 1D nodalization, fictitious 2D nodalization and 3D component, available in some code, against natural circulation and thermal mixing in large pools; • Characterization of “user effect” for the simulation of heat transfer in HX submerged in large pool and thermal hydraulic behavior of large pool. Eighteen organizations worldwide expressed their interest to participate in the PERSEO Test 7 benchmark. In the present report, nine results received by eight organizations are presented. The benchmark of PERSEO Test 7 Part 1 and 2 has been very successful and valuable to highlight some of the main capabilities and criticalities of the code used by the benchmark participants in the simulation of a passive system operating in natural circulation. In general, it has been found that almost all codes are qualitatively able to predict the phenomena occurred in PERSEO Test 7. However, some qualitative and quantitative discrepancies with respect to the experimental data have been found. The main limitation, observed in almost all the adopted codes, is the significant underestimation of the heat transfer between the HX and the HXP. This has been attributed by most of the participants to the underestimation of the condensation heat transfer coefficient on the tube-side, and in some cases, also to the underestimation of the nucleate boiling heat transfer coefficient on the pool-side. With the exception of IBRAE, which adopted SOCRAT code, all the other participants decided to modify the code or the nodalization to correctly predict the heat transfer: - GRS (ATHLET) used a developer version of ATHLET introducing Kutateladze correlation (for laminar wavy condensation) and Chen correlation (for turbulent condensation); - KAERI_a (MARS), KAERI_b (SPACE), POLITO (RELAP5-3D), UJV (RELAP5) and UNIPI (RELAP5) multiplied the heat transfer coefficient by a factor ranging from 2.2 to 3.0; - NRG (SPECTRA) set the characteristic length for the film condensation to 5 cm, corresponding to a multiplier of 2.44 for the heat transfer coefficient. - UofU (MELCOR) modified several sensitivity coefficients in the Heat Structure Package and set a multiplier of 180 instead of 120 for the number of tubes. The other phenomenon which presented a significant criticality is the prediction of the thermal stratification in the pools, which has been underlined in particular by POLITO, UJV and UNIPI. However, it should be mentioned that in PERSEO Test 7 Part 1 and 2 the thermal stratification is present mainly at the beginning and at the end of the transients when the passive system is not in operation. It should be noted that the missing prediction of the thermal stratification did not prevent to correctly predict the behavior of the passive system during the operation phase. The coupling between the two pools and the mass transfer when the triggering valve is open has been qualitatively simulated by all the participants with minor quantitative discrepancies. The instabilities at the injector, which can be seen in the injector differential pressure, have been partially predicted only by few participants. However, the missing prediction of the instabilities at the injector did not prevent to correctly simulate the overall transient behavior. Considering the nodalization strategy for the pools, the majority of the participants adopted a fictitious 2D nodalization for the pools: 2 vertical hydraulic regions laterally connected through cross-flow junctions. This approach resulted in a more accurate prediction of the main phenomena occurring in the pools in PERSEO Test 7. KAERI_b modeled the HXP in test 7 Part 2 with the only 1-D approach, showing relatively poor predictions. KAERI_a adopted a 3D approach for the OP, but no evident improvement with respect to the 2D approach is shown in the calculation results, taking into account the two OP temperatures considered in the analysis. UofU adopted one control volume for the OP and three vertical control volumes to simulate the HXP to avoid the definition of artificial, non-physical natural circulation loops in MELCOR. The user effect has been investigated considering the submitted results. User effect can be relevant in particular regarding the nodalization strategy of the pools (1D, fictitious 2D and 3D approaches), as underlined by some participants, and in setting the initial and boundary conditions. Considering the three participants which adopted RELAP5 (UJV and UNIPI) and RELAP5-3D (POLITO), it is possible to note that similar choices have been made in the development of the nodalization (e.g. primary pressure imposed as boundary condition, HXP and OP modeled with two vertical pipes connected by cross flow junctions). The main difference is the adoption of a single pipe for the modeling of the HX by POLITO and UJV and fifteen pipes by UNIPI. The results of the three calculations are similar both considering the qualitative and the quantitative analysis and the same limitations emerged (in particular the underestimation of the heat transfer between the HX and the HXP and the missing prediction of the thermal stratification). Therefore, the user effect related to the nodalization strategy has a minimal influence in this case. The use of a higher level of detail for the simulation of the HX tubes seems to not give any particular advantage. The analysis of a single calculation for a specific code does not allow an immediate characterization of the user effect for that specific code. However, as underlined in the previous section a 2D approach for the pools nodalization provided in general a more accurate results in the calculation. Some discrepancies emerged in the comparison of the initial conditions obtained by the various participants. However, comparing the calculated results during the transient, few discrepancies related to the initial conditions emerged and they are limited mainly to PhW1 of Test 7 Part 1 and 2 when the system is not in operation. Therefore, a discrepancy on the initial conditions that may be considered unacceptable in other cases, may be considered acceptable in this case.