OECD/NRC Benchmark based on NUPEC BWR Full-size Fine-mesh Bundle Tests (BFBT)


In the past decade, a large amount of effort has gone toward the direct simulation of the boiling transition (BT) for boiling water reactor (BWR) fuel bundles. The most advanced sub-channel codes explicitly take into account droplet along with liquid and vapour. They predict the dry-out process as the disappearance of the liquid film on the fuel rod surface without employing any semi-empirical correlations. Through a series of benchmark comparisons with full length/scale bundle data, it was verified that the codes are reliable in predicting the critical power of the conventional BWR fuel types. However, these sub-channel codes are not yet used in new fuel design. Adequacy of fuel lattice geometries, spacer configurations, etc., is still confirmed mainly by costly experiments using partial- and full-scale mock-ups. The main reason for this situation is a shortage of high resolution and full-scale experimental databases under actual operating conditions.

The detailed void distribution inside the fuel bundle has been regarded as one of the most important factors in the boiling transition in BWRs. Concerning the sub-channel wise void distribution, it is clear that the crossflow across the sub-channel gap dominates void distributions. Most of the well-known sub-channel codes still employ the classical Lahey's Void Drift Model or its modified models. Although there have been substantial efforts to establish a sound theoretical background of detailed void distributions, the numerical models that are verified in a wide range of geometrical and thermal-hydraulic conditions are not yet available. In this sense, this subject still remains the major unsolved problem in the two-phase flow of BWR fuel bundles. The main reason for this lack of resolution is the lack of reliable full bundle databases under operating conditions. Up to now, only partial bundle (3 3 or 4 4) test data under relatively low pressure (1 MPa) conditions have been made available.


It was during the fourth OECD/NRC Boiling Water Reactor Turbine Trip (TT) Benchmark Workshop on 6 October 2002 in Seoul, Korea, that the need to refine models for best-estimate calculations based on good-quality experimental data was discussed. The needs arising in this respect should not have been limited to currently available macroscopic approaches but should have been extended to next-generation approaches that focus on more microscopic processes. It was suggested that this international benchmark be based on data made available from the NUPEC (Nuclear Power Engineering Corporation) Database. From 1987 to 1995, NUPEC performed a series of void measurement tests using full-size mock-up tests for both BWRs and PWRs. Based on state-of-the-art computer tomography (CT) technology, the void distribution was visualised at the mesh size smaller than the sub-channel under actual plant conditions. NUPEC also performed steady-state and transient critical power test series based on the equivalent full-size mock-ups. Considering the reliability of not only the measured data but also of the other relevant parameters (e.g. system pressure, inlet sub-cooling and rod surface temperature), the test series supplied the first substantial database for the development of truly mechanistic and consistent models for void distribution and boiling transition.

This international benchmark, based on the NUPEC Database, encourages advancement in this uninvestigated field of two-phase flow theory with very important relevance to the nuclear reactor's safety margins evaluation. Considering the immaturity of the theoretical approach, the benchmark specification is being designed so that it systematically assesses and compares the participants' numerical models on the prediction of detailed void distributions and critical powers. Furthermore, the following points are kept in mind while the benchmark specification is being established:

  • Concerning the numerical model of void distributions, no sound theoretical approach that can be applied to a wide range of geometrical and operating conditions had been developed.
  • In the past decade, experimental and computational technologies have improved tremendously through the study of the two-phase flow structure. Over the next decade, it can be expected that mechanistic approaches will be more widely applied to the complicated two-phase fluid phenomena inside fuel bundles.
  • The development of truly mechanistic models for critical power prediction is currently underway. These models must include elementary processes such as void distributions, droplet deposit, liquid film entrainment, etc.

The BFBT benchmark consisted of two parts (phases), each part consisting of different exercises.

Benchmark exercises

Phase I - Void Distribution Benchmark

  • Exercise I-1 - Steady-state sub-channel grade benchmark
  • Exercise I-2 - Steady-state microscopic grade benchmark
  • Exercise I-3 - Transient macroscopic grade benchmark
  • Exercise I-4 - Uncertainty analysis of the steady-state sub-channel benchmark

Phase II - Critical Power Benchmark

  • Exercise II-0 - Pressure drop benchmark
  • Exercise II-1 - Steady-state benchmark
  • Exercise II-2 - Transient benchmark
  • Exercise II-3 - Uncertainty analysis of the steady critical power benchmark.

The purpose of this benchmark was not only the comparison of available macroscopic approaches but mostly the encouragement to develop novel next-generation approaches that focused on more microscopic processes. Thus, the benchmark problem included both macroscopic and microscopic measurement data. In this context, the sub-channel grade void fraction data were considered the macroscopic data, while the digitised computer graphic images were considered the microscopic data.

Related data

Data related to BFBT can be requested from the NEA Data Bank:


Mailing list archive

BFBT list


Ian Hill

Publications and reports