Three-dimensional computational fluid dynamics (CFD) method was used to model the boiling two-phase flow in one of the PSBT 5-by-5 rod bundle tests. The rod bundle with all the spacers was modeled explicitly using unstructured computational grids. The six-equation, two-fluid model with the wall boiling model was used to model the boiling two-phase flows in the bundle. The computed void fractions compare well with the measured data at the measuring plane. In addition to the averaged void data, the CFD results give a very detailed picture of the flow and void distributions in the bundle and how they are affected by solid structures in the flow paths such as the spacer grids and mixing vanes.

In the NUPEC PWR Subchannel and Bundle Test (PSBT) International Benchmark exercise [

The commercial CFD software STAR-CCM+ v6.06 [

The computational grid was generated by recreating the rod bundle and the spacers using the 3D CAD package in STAR-CCM+. The geometries of the rod bundle and the 3 different spacers were taken from the problem specification report by Rubin et al. [

CFD grid of spacer with mixing vanes.

CFD grid of spacer without mixing vanes.

CFD grid of simple spacer.

Contacts between rods and spacers.

The completed rod bundle assembly was created by connecting together the rod-spacer sections according to the specification given in [

Rods and spacers (red = MV, yellow = NMV, blue = SS).

The standard six-equation, two-fluid model was used in modeling the boiling two-phase flows considered in this paper. In this model the conservation equations for mass, momentum, and energy are solved for both phases.

The conservation of mass for phase

The conservation of energy for phase

The standard

The drag force between the two phases includes a mean and a fluctuation component. The mean drag force is given by

The fluctuating component of the drag force accounts for the additional drag due to interaction between the dispersed phase and the surrounding turbulent eddies. This force is the turbulent dispersion force or the turbulent drag force:

The drag coefficient in (

The interfacial forces would generally include the lift and wall lubrication forces also. The effects of lift and wall lubrication forces are to move the steam bubbles radially away or towards the rod surfaces. Since in this exercise the computed void distributions are to be averaged across the channel, any information on radial void distribution will be lost in the comparison exercise; hence the inclusion of lift and wall lubrication forces is not important and was left out for simplicity.

At the heated wall, boiling occurs when the wall temperature exceeds the saturation temperature. The steam generation rate is determined by the wall heat partitioning model as follows,

The bubble influence area

The evaporation heat flux can be expressed as

The nucleation site density is obtained from Lemmert and Chawla [

The bubble departure diameter is obtained from Tolubinsky and Kostanchuk [

The bubble departure frequency is obtained from Cole [

A large range of bubble diameters can be expected in the flow. Since bubbles are generated by boiling and removed by condensation, the bubble diameter is expected to be function of the liquid temperature. Kurul and Podowski [

The steady-state bundle test B5 Run 5.1121 was analysed. The flow conditions were given as pressure ^{2}a), mass flux ^{6} kg/m^{2}hr), power

The computed results from the CFD model are shown in Figures

Void fraction.

Void fraction downstream of spacer with mixing vanes.

Effects of vanes on flow.

Void distributions in the 3 measuring planes.

A major advantage of the 3-dimensional CFD method is the level of details it can provide about the flow pattern, void and temperature distributions, and so forth across the whole bundle as shown in Figures

Void distribution above mixing vane spacer around one rod.

Velocity distribution above mixing vane spacer around one rod.

A 3-dimensional CFD model of the PSBT 5-by-5 rod bundle was constructed using the STAR-CCM+ software. Unstructured polyhedral computational cells were used to model the rod bundle and all the spacer grids explicitly. The six-equation two-fluid model together with the wall boiling model was used to model the boiling two-phase flows in the bundle. The steady-state test B5 Run 5.1121 was studied. The computed void fraction averaged over the 4 central subchannels at the upper measuring plane was 0.1576 which is 12% lower than the measured value of 0.1791. This level of agreement between the results is encouraging given the complexity of the geometry and the boiling two-phase flow physics.

A major advantage of 3-dimensional CFD is the level of details it can provide about the flow making it possible to perform detailed design analyses for spacer grid and investigating the effect of mixing vanes. However, before this modelling technology will be accepted for design analysis much more rigorous verification and validation of the models are required. In addition to the comparison of channel-averaged results, comparison against detailed spatial distributions of void, velocity, temperature, and so forth, across the whole bundle is required. Hopefully such detailed measured data will become available soon.