A GFR benchmark: Comparison of transient analysis codes based on the ETDR concept

摘要

In preparation for the transient analysis of the Generation IV gas fast reactor (GFR) and experimental technology demonstration reactor (ETDR) designs, a transient benchmark exercise (in the frame of the Generation IV GFR project) was performed to compare the capabilities and limitations of the different code systems to analyse these new reactor concepts. The benchmark was based on the ETDR concept and was performed in three phases, i.e. a loss-of-flow (LOF) transient with reactor scram for phases 1 and 2 and a small-break loss-of-coolant accident for phase 3. The organizations which participated in the benchmark were AREVA, France; CEA, France; NRG and TUD, The Netherlands; AMEC, United Kingdom; INL, USA (phase 1); CIRTEN, Italy; JRC-IE EURATOM and PSI, Switzerland. Phase 1 of the benchmark was performed in a "blind" manner in that all participants were provided with the same information, but were free to make their own judgement on the use of this information. Following a review of the results from phase 1, the conclusion was made that the agreement between all the partners was acceptable, but better agreement was to be preferred. The reason for the discrepancies was identified, namely: different heat transfer correlations for the main and DHR heat exchangers and reactor core, modelling of the vessel wall thermal capacity, core and vessel pressure drop calculations and flow resistance in the DHR helium and water loops. The phase 1 transient was repeated as (phase 2) but with more 'strict boundary conditions' to identify which of the above gave the largest contribution to the differences in the phase 1 submissions. The second phase results were much improved. However, to demonstrate the capabilities of the codes at low pressures a SB-LOCA transient with reactor scram (phase 3) was defined to model the ETDR DHR system during the depressurization from 70 bar to the 3 bar containment pressure. The phases 1 and 2 results showed that the use of different heat transfer correlations for the main heat exchanger does not significantly affect the core coolant temperatures, while the modelling of ETDR heat structures has a major impact on the core inlet temperatures. Even though three participants used the RELAP5 code and two participants used CATHARE, there was limited consistency between the results for each specific code. This was particularly true for phase 1, where the wide spread in the results came from different input and boundary assumptions by the codes users, i.e. the so-called "user-effect". The results of the third phase (SB LOCA with reactor scram), showed sensitivity to the extrapolation of the core pressure drop calculation to low pressures and low Reynolds numbers, with the result that there was an extreme sensitivity of the core flow rate to small changes in core geometry, grid spacer losses, etc. resulting in large differences in coolant, clad and fuel temperatures. The most important message to be taken from the benchmark exercise is that to obtain consistent results, it is especially important to have consistency in the use and interpretation of the plant data as well as the availability of clear and unambiguous information and it is only when these features have been resolved is it possible examine the importance of differences in physical modelling. On this basis all the codes used demonstrated an ability to model gas-cooled systems under natural circulation and low pressure low Reynolds numbers' conditions.