A Review of the Meetings of the Working Group on the Comparison of Calculational Models

The purpose of the Calculational Model Group was to discuss the models being used to determine the consequences of sodium fuel interaction in reactor incidents and to compare the results when the codes were applied to specific test cases. This review is intended to compare and comment on the various treatments and to present the results of the comparison of predictions when the codes were applied to specific test cases. The results of the first comparison calculation by the Working Group assumed instantaneous fuel fragmentation and mixing and was reported to the 2. Specialist Meeting on SFI at Ispra. Since the Ispra meeting additional results for the original comparisons have been presented to the Working Group as new codes were introduced or developed. However one of the two test cases considered at the last meeting in April 1975 was substantially the same as the case originally considered in that instantaneous fragmentation and mixing was assumed although some of the other parameters such as the dimensions of the interacting zone and length of sodium column were different. It is therefore appropriate to consider the recent cases only. Codes used for parametric surveys indicate how the pressures, expulsion velocities and work energy depend on various assumptions. It is implicit in all the codes that at the time of initiation of the interaction the coolant is in the liquid state and heat transfer from the fragmented particles is by conduction. For UO2/Na systems it is the conductivity of the UO2 which limits the heat transfer initially. The assumption of uniform sodium temperature is made in many of the codes. This stage of the interaction has been defined as Stage A and it is considered separately from Stage B when the interacting zone has expanded sufficiently to permit vapour to be present. The mass of fuel fragmenting with time was assumed to increase with time and calculations were carried out for two cases Tf = 0 and Tf = 10 ms where Tf is the fuel fragmentation time. Results for these two test cases were submitted by Caldarola, Antonakas (TEXAS), Fishlock (EXPEL), Scararo (FUS-PEC), Costa (CORFOU) and Mizuta. The results were not presented in identical form and detailed comparison of the time variation of pressure in the interacting zone, fuel and sodium temperatures, the length of the interacting zone and work done is not possible although in many cases pressure time histories were given and, where not available, peak pressures in Stage A and Stage B were given. The time of expulsion of sodium from the channel and the work done up to this point were also available in most cases. In conclusion: The wide range of different solutions to the two problems reflect the different treatments in the various models although it was not always possible to identify precisely which features of any individual models contributed to a particular result. The exercise did not uncover any unusual results and in this respect the comparison has been useful in demonstrating that there do not appear to be any important errors in the computational techniques in the codes. The exercise did not suggest that any one model had any special merit over any other. It is possible that some of the assumptions common to all models are not a good representation of what actually happens in any thermal interaction and the division into Stages A and B although convenient has not been demonstrated convincingly in any experiment. Shock tube experiments provide the best evidence for high pressures in the early stages of the interaction suggesting a compressed liquid state followed later by pressures associated with a two phase state but it requires to be demonstrated that such interactions arise with molten fuel and sodium. When the processes in a thermal interaction are more precisely defined then most models will be able to describe them and to determine the consequences with sufficient accuracy. If expulsion times arc the main interest the exercise has demonstrated that the various treatments do not lead to very different predictions and have demonstrated that expulsion times of about 10 ms can arise with only a small fraction of the energy available in the fuel being converted to work. Although it would have been possible to identify a number of other factors which are known to affect the results there does not appear to be any need to make a comparison of their effects between the various codes