LABORATORY OBSERVATIONS OF FRAZIL ICE

Transcription

1 Ice in the Environment: Proceedings of the 16th IAHR International Symposium on Ice Dunedin, New Zealand, 2nd 6th December 22 International Association of Hydraulic Engineering and Research LABORATORY OBSERVATIONS OF FRAZIL ICE S. Clark 1 and J.C. Doering 1 ABSTRACT The supercooling of river water leads to active frazil ice production. This phenomenon can be detrimental to hydroelectric generating stations as active frazil can block trash racks, restricting or totally blocking the flow to the turbine. When frazil ice leads to the formation of anchor ice, the geometric and hydraulic properties of the flow can be altered. Increased staging caused by anchor ice formation in the tailrace can result in a reduction of the net operating head of the generating station. These effects can result in a significant reduction in potential revenue. Numerous field and laboratory studies of frazil ice production have been conducted in the past, although the number of published photographs of frazil ice particles is relatively small. The purpose of this paper is to illustrate several stages and shapes of frazil ice particles during their development. Observations made during experiments in the Hydraulics Research & Testing Facility at the University of Manitoba are presented. Typical results from the Digital Image Processing System (DIPS) used to analyze frazil ice production are also presented. EXPERIMENTAL APPARATUS Historically, frazil ice research has been conducted from a Eulerian frame of reference, meaning observations and measurements are made from a stationary position while the water and frazil particles flow by. In a laboratory, this setup necessitates the use of pumps to recirculate the water, which shears the frazil particles making it difficult to observe their growth. In the Hydraulics Research & Testing Facility at the University of Manitoba a counter-rotating flume is used to circumvent this problem. The circular flume bed is 2 cm wide, 1.2 m in diameter, and is supported by a turntable. The bed is lined with 6 plates, which are covered with a uniform layer of gravel to simulate a river bed. Sets of plates with different sized particles can be used to test various bed roughnesses. Ducts beneath the bed plates, as well as in the inner and outer flume walls are kept slightly above freezing to simulate a geothermal gradient. The flume walls are supported by a separate structure, allowing the walls and bed to rotate in opposite directions. After balancing the shear forces caused by the walls and the bed, the vertical water velocity distribution becomes similar to that of a natural river when viewed from 1 Hydraulics Research & Testing Facility, Civil Engineering, University of Manitoba, Winnipeg, Manitoba. Canada R3T 5V6 Jay_Doering@UManitoba.ca

2 the frame of reference of the bed, while the water remains virtually stationary from an absolute frame of reference. This setup results in a Lagrangian frame of reference, thus allowing undisturbed observation of the frazil ice as it forms. The entire flume is located in a coldroom that controls the temperature to within. 1 C. Water temperature is measured using a Guildline 954 Platinum RTD with a sampling rate of 1 Hz. The precision of the thermometer is.1 C, with and accuracy of.5 C. It is regularly calibrated using a distilled ice bath to ensure its accuracy. Images are collected using a Sentech STC-11a monochrome progressive scan CCD camera with a 25 mm lens and a shutter speed of 1/2 th of a second. They are then converted to digital format with a DIPIX FPG-44 Power Grabber board. The light source is a 5 W halogen floodlight located within the inside wall of the flume. The light goes through a polarizing sheet before entering the water, and then goes through a second polarizer fixed to the camera s lens. When the lens polarizer is oriented perpendicular to the wall polarizer, light will only reach the camera if it is refracted by an object in the field of view. In our case by frazil particles. The field of view of the camera is approximately 55 mm by 41.3 mm and is converted to a digital image having 64 by 48 square pixels. This results in a single pixel being approximately 86 µm square. Thousands of images are collected during an experiment and are analyzed using a Digital Image Processing System (DIPS) algorithm developed using Matlab. (Doering and Morris, 22) This algorithm subtracts an image containing ice particles from a reference image without particles. The intensity of frazil particles in the difference image is then greater than the background, allowing the algorithm to recognize the particles and convert the image to binary form. The particles in the binary image are then dilated once, eroded twice, and dilated once more. This procedure smoothes the edges of the particles, fills in any gaps, and also removes any noise that may appear in the images. Several functions in Matlab are then used to count the number of particles in the images, determine their size (in pixels), and determine where they appear in the vertical profile. These results can be used to create histograms of the relationship between time, frazil size, frazil location, and number of particles. OBSERVATIONS During a typical experiment the smallest aperature size is used in order to increase the depth of field as much as possible. This creates relatively dark images. Although the frazil particles in the images can be seen with the eye as well as easily detected with the DIPS, the images do not show the texture of the frazil particles. An experiment was conducted with a larger aperature size, thus creating brighter, clearer images. From these images several observations and comparisons to previous research can be made. Particle Shape Disks Although the length of one pixel is about 86 µm, the smallest particles that can be seen with certainty are around 6 pixels in diameter, or about.5 mm. The largest uniform frazil disks are approximately 5 mm in diameter. Figure 1a shows a typical disk shaped frazil particle. Frazil particles with diameters greater than 5 mm tend to have both nonuniform thicknesses and shape. It is common to see a small circular area of different light intensity in the middle of the larger frazil particles. In extreme cases, the edges

3 seem to form a ring around an otherwise circular particle. Small, saw-tooth shaped protrusions extend from the edges, and these particles are also susceptible to breakage. These observations are consistent with findings made by Arakawa (1954), and Osterkamp (1982). The diameter to thickness ratio is difficult to measure since the frazil particles rarely travel exactly perpendicular to the flow with respect to either of their axes. It is more common to see the particles as oval shaped because of their skewed orientation in the water. a. b. Figure 1: a) Typical frazil disk, 4.15 mm in diameter; b) Sintering and flocculation of particles. As the amount of frazil increases, the particles flocculate and sinter together. It is more common to see larger, irregular shaped particles sintered together than small perfectly rounded particles. Both the size and the density of these flocs increase as an experiment progresses. It is common at the end of an experiment to have a few very large flocs with a low concentration of individual frazil particles. Particle sintering and flocculation is shown in figure 1b. Stars and Hexagons Six pointed star shaped particles can be sometimes seen in the flume, but have not been observed exceeding about 1.5 mm from point to point. Hexagonal particles have been observed much larger however. Figure 2a shows a large hexagonal particle with sawtooth edges and non-uniform thickness. These particles appear later in the experiments. Needles Spicule or needle particles have been observed, most commonly at the beginning of an experiment. These particles most often occur with 3 or more needles sintered together in random orientations. A combination of spicule and discoid particles sintered together has also been observed, as shown in figure 2b, which was taken using a small aperture.

4 When looking at the surface of the water at the beginning of a typical experiment several needle particles can usually be seen. It is hypothesized that single needles develop at the surface and are brought into suspension through turbulence. a. b. Figure 2: a) Hexagonal particle 1.5 mm in diameter; b) Sintering of needle and discoid particle. Left hand needle is 11.4 mm long, discoid is 3.4 mm in diameter. Nucleation The water used in the experiments is Winnipeg tap water, which does not pass through a water treatment plant. It is therefore common to have small particles suspended in the water. Reports of ice formation while the bulk water temperature was above C are rare, but have been reported in the literature. In our experiments spicules of surface ice have been seen when the water temperature is.7 C, and small frazil disks have been seen at.5 C. These observations were made during experiments with air temperatures of 15 C. It is important to note that the thermometer is located 6 cm from the bottom of the flume, while typical water depths range from 1 to 2 cm. Since the heat loss is from the surface, the water temperature must be slightly cooler near the surface even though the water is typically well mixed by the counter-rotating motion. There is generally also intermittent border ice along the flume walls that develops during an experiment. It is not clear whether the border ice acts as a seeding material for the frazil formation, but this seems likely. There is no deliberate seeding of the water. It is not clear whether mass exchange processes originating in the water, such as splashing, bubble bursting etc. causes the frazil ice nucleation, although since particles can first be seen at the surface this is also possible. It is generally accepted that secondary nucleation is the cause of the rapid proliferation of frazil ice particles. Figure 3 shows a broken piece of frazil. It is likely that this piece will continue to grow and perhaps become unstable, shear, and become the nucleus of further particles. DIGITAL IMAGE PROCESSING SYSTEM As previously mentioned, when acquiring images for the purpose of analysis, darker images with higher depths of field are preferred. Figure 4a shows a typical image that has had its corresponding reference image subtracted. This removes any scratches or imperfections in the flume walls. Figure 4b shows the resulting binary image after

5 analysis. The DIPS program then uses this binary image to count the number of particles, the size of the particles, as well as their vertical distribution in the flume. Figure 3: Jagged, triangular particle broken from larger particle. a. b. Figure 4: a) Digital image with reference image subtracted. b) Binary image produced by DIPS. Note: largest particle 4.3 mm in diameter. Since the time the image was acquired at is known, graphs illustrating the temporal variation of particle size and number of particles can be generated. Figure 5 shows one such graph for an experiment conducted with an air temperature of 15 C, water velocity of 5 cm/s, water depth of 15 cm, and a bottom with pebbles having a D 5 of about 1 cm. This 3 dimensional plot shows that at the beginning of supercooling both the number and size of frazil particles is small. The number of particles increases dramatically when the water reaches its maximum supercooling value. At this point there is many more small particles than large particles, but the small particles soon floc together. To further illustrate the effect that the water temperature has on the number of

6 frazil particles, figure 6 shows a graph of water temperature versus time superimposed on a graph of the number of frazil particles versus time s le P artic o f e r b m N u P ar ticl e 2 S ize (Pixels) Time (Minutes) 4 5 Figure 5: Three dimensional frazil size distribution Temperature ( C) Number of Particles Time (Minutes) Figure 6: Measured water temperature and number of frazil particles versus time. FUTURE ENDEAVORS While high resolution images of various frazil ice formations have been acquired, there are still particles that cannot be detected by the current system. Daly (1986) was able to view particles as small as 3 µm. The ability to view particles as they just begin to form

7 would help to accurately determine the temperature at which nucleation occurs. This small field of view would nevertheless limit the ability to have an overall view of what is happening in the flume. One possible solution would be to have two or more cameras acquiring images simultaneously. This would allow one camera to be acquiring a very small field of view while the second would acquire images covering the entire height of the flume. Additional cameras could be used as needed. The ability to use two identical cameras with two identical frame grabber boards acquiring simultaneous images has sparked the question as to whether stereoscopic images of the frazil particles can be obtained. This would help to resolve the third dimension (width into the flume) and perhaps help solve the problem of overlapping particles appearing as a single particle in two-dimensional images. Either of the aforementioned additions to the experimental setup would improve the current DIPS and allow an even more detailed characterization of the formation of frazil ice. SUMMARY Several different types of frazil ice formations have been discussed and illustrated. Needle shaped particles have been typically seen first, followed by disk particles, stars, and hexagons. The maximum size of uniform frazil disks has been observed to be about 5 mm in diameter, after which they become unstable, non-uniform, and jagged. A sample of the results from the DIPS developed at the Hydraulics Research and Testing Facility has been presented, and several methods of improving the system have been briefly discussed. ACKNOWLEDGEMENT The authors wish to thank Manitoba Hydro s R&D committee and NSERC for funding this research. Further thanks are extended to R. Hartle, M. Morris, S. Ye, J. Malenchuk, and R. Murrillo. REFERENCES Arakawa, K. Studies on the freezing of water: II; formation of disk crystals. In Journal of the faculty of science, Hokkaido University, series II (physics) 4(3): (1954). Daly, S.F. Frazil Ice Dynamics. Monograph 84-1, US Army Corps of Engineers, CRREL (1984). Daly, S.F. and Colbeck, S.C. Frazil ice measurements in CRREL s flume facility. In Proceedings of the IAHR Ice Symposium International Association for Hydraulic Engineering and Research, Iowa City, Iowa, USA (1986) Doering, J.C. and Morris, M.P. A digital image processing system to characterize frazil ice. Canadian Journal of Civil Engineering; Special Issue (in press 22). Osterkamp, T.E. Frazil ice formation: a review. Journal of the Hydraulics Division (1978). Osterkamp, T.E. and Gosink, J.P. Frazil Ice Formation and ice cover development in interior Alaska streams. Cold Regions Science and Technology 8: (1983).