Complex Flow Measurements in the INL MIR Facility using PIV

These experiments were performed at the Idaho National Laboratory (INL) Matched Index of Refraction (MIR) flow system in support of the KNERI project "Advanced computational thermal fluid physics (CTFP) and its assessment for light water reactors and supercritical reactors."  The goal of the experimental portion of the study is to answer the scientific needs, guide code development and assess code capabilities for treating the generic forced convection problems in ALWRs and SCRs.  The  data provides  benchmark velocity and turbulence measurements for the portion of the study dwelling on forced convection in complex reactor geometries.  For the representative geometry, the experimental model provides a generic simulation of flow along fuel rods separated by periodic grid spacers as in an SCWR concept.

Download the report (MS Word)

Barton L. Smith, PI

The model represents two fuel rods from this core

The model is a two-rod configuration which was selected to include some flow features of thermal SCWR concepts suggested by Forschungszentrum Karlsruhe, INL and Prof. Oka of U. Tokyo.  The geometry is scaled to be six to seven times larger than typical fuel pins.  The rod diameter is 2.50 inches (63.5 mm) and the axial pitch of the ring-cell spacers is 17.5 inches (444.5 mm).  The outer diameter of these spacers is 3.025 inches (76.8 mm) so the nominal dimensions of the rectangular flow channel containing the simulated fuel rods are 3.025 by 6.050 in^2 (76.8 X 153.7 mm^2).  Consequently, the pitch-to-diameter ratio (p/D) is about 1.21 for the simulated fuel rods.  These spacers have lengths of 1.75 inches (44.5 mm) and inside diameters of 2.85 inches (72.4 mm).  In the measuring region the rods and side and end walls are fabricated of quartz to match the refractive index of the light mineral oil employed as the fluid.

Cross Sectional View
Model Cross Section

MIR Model

Measurements of velocity are made using a state-of-the-art Particle Image Velocimetry (PIV) system
by LaVision Inc.  PIV generates a planar 2-component velocity vector field at an instant in time.  The flow is seeded with 10 micron diameter silver-coated glass spheres. Each measurement requires a pair of digital images of these particles.  The seeds are chosen to be small and neutrally buoyant, and are illuminated by a laser sheet. Cross-correlations are performed on small pieces (interrogation windows) of the image pairs to determine the most likely velocity vector in the plane for that sub region. The  interrogation window initially consists of 64 pixels in each direction and these may overlap one another by 50%. The camera is placed perpendicular to the laser sheet. The resolution and accuracy of the result can be improved by shifting the 2nd window in the estimated direction of the velocity vector by a known amount. A first pass with no shift provides the estimate of how much to shift the window on the second pass.  Multiple passes  make it possible to reduce the interrogation windows to 16 pixels, quadrupling the spatial resolution.  Data were acquired at two streamwise (x) stations and 33 spanwise (z) planes in between two spacers.  Each plane consists of 400 samples.


The entire set of statistics is a 1.6 million data point, 155 MB ASCII (TechPlot format) file that you can
download here if you have the network muscle. 

Some feedback that I've been getting is that no one is interested in plowing through a file that size.  The three files below are y-z planes at full resolution for four locations and are in the same format as the large file.

The file has columns of x, y, z, U, V, u'u', v'v', and u'v'.   The streamwise (x) origin is at the downstream end of the upstream spacer.  The domain extends slightly upstream of that point.  The second spacer is at x = 400 mm.  The spanwise (z) origin is centered on the lower rod.  The cross stream (y) axis origin is centered in the channel.  Note that the rods may not be precisely vertically  centered in the channel due to deformation of the rod spacers. 
Other known issues with this data:
1) While acquiring the data, we were able to precisely find the center of the lower rod.  Determining the locations of the spanwise edges of the channel was not as easy.  You will notice that data at the extreme z values is sketchy, possibly due to the laser sheet being blocked or refracted. 
2) Data taken looking through the cylinders (i.e. large positive values of z) are distorted.
3) "Masks" were applied to the raw data at the locations of the rods, spacers, and the channel walls.  These regions will have the value 0.

Some jpg images and avi animations of the mean streamwise (U) and cross-stream (V) velocity as well as the TKE are shown below.  Blue regopms are masked, and are presumably solid surfaces.  The fence furthest right (upstream) is partially blocked by the upper spacer, which was not made from index-matched material. 

Streamwise Velocity
Streamwise Velocity (U)

Cross-stream velocity