from a hopper with a revolving conveyor belt at
Test procedure
Water flow through the model was set in
its base. The rate of bead discharge into the model
approximate accordance with the conditions
was regulated by means of a sliding gate at the
depicted in Figure 9 (19 January 1994), such that
edge of the belt.
the Mississippi and Missouri rivers conveyed
48,400 and 60,600 ft3/s (1370 and 1715 m3/s),
PIV system for measuring water and
respectively. Water levels were adjusted by means
ice velocities
The layout of the PIV system used for the model
of a tailgate at the downstream end of the model.
is given in Figure 11. A color CCD camera (Sony
The ice hopper was set in motion, releasing beads
SSC-C374) with a wide-angle lens, having an 8-
onto the Missouri River in a single-thickness layer.
mm-focal length, is used to capture the images.
Most tests were run with the beads discharged at
The camera is mounted 5.4 m above the model, at
maximum areal concentration. Some runs, though,
an oblique angle such that a 5.7- by 4.1-m area forms
were made with lesser concentrations to qualita-
the camera's field of view. The area is delineated in
tively determine how ice concentration affected
Figure 8. Sixteen 500-W halogen lamps uniformly
ice movement through the confluence. Dye visu-
alization was also used to establish the principal
through careful positioning of the lights.
water velocity distributions.
Video images were recorded using a Sony EVO-
Once the PIV system was set up, model ice
9650 video recorder. They were subsequently digi-
pieces (plastic beads) at a low areal concentration
tized by means of a frame-grabber (Matrox
were released into the modeled flow to determine
Meteor RGB). The digitized images (i.e., video
and map the water velocity field in the confluence.
frames) were composed of 640 by 480 pixels hav-
Subsequently, model ice was discharged into the
ing 8-bit gray-level resolution. A 133-MHz
modeled Missouri River channel at a set rate.
Pentium personal computer was used to process
Model ice velocities were measured using white
and store digitized video images. Custom software
plates placed among the black beads.
was written in the C language to control the VCR
and the frame-grabber during playback and frame
Image recording and processing
grabbing. The subsequent image processing soft-
Prior to videotaping of water flow and ice
movement, two aspects of the PIV system had to
ware, essentially a cross-correlation algorithm
be calibrated: the dimensions of the field of view,
including software to correct the areal distortion
and the PIV-measured velocities. A start-up seg-
of the video image, was written in Fortran 77. A
ment of video was made to correct the image dis-
standard color monitor was used for on-line dis-
tortion introduced by the video camera's lens and
play of the video imaging.
angle of view. The start-up segment entailed vid-
Particles with the same specific weight, but dif-
eotaping distance markers and a grid of known
ferent color and shape, were used as flow tracers
spacing over the area of interest. The markers were
for the open-water and ice-covered flows. Low
concentrations of polypropylene beads (spaced
several beads per 100 mm2) released onto the
relative to the position of the video camera. The
water surface were used as tracers for mapping
start-up video images allowed corrections to be
the open-water flows. For the ice-covered flows,
made for areal distortions resulting from the video
in which a more or less continuous layer of black
camera's oblique view of the model.
beads moved through the confluence, ice vectors
were determined using a low concentration mix-
the accuracies of the water and ice velocities
ture of square plates (10, 10, and 1 mm) of white
obtained using the PIV system. This was com-
polypropylene and the black beads. The plates
pleted by comparing PIV velocities with those
were needed to ensure that adequate patterns were
measured by timing hand-released tracers as they
formed in the layer of moving model ice. The size
traveled known distances in the model. As the
of both tracer particles, the beads and the squares,
dimensions of the area of interest and the period
were sufficiently small that they followed the
between images are set earlier, velocities follow
movement of water and ice. Importantly, they also
directly.
were large enough to meet the size constraint (i.e.,
Before model ice was discharged into the model,
to occupy at least one pixel of the video image)
video images were taken of the open-water flow
needed for adequate image resolution.
conditions to determine the open-water flow field.
17