A Field Study of Frazil Ice Accumulation and Adhesion on Trash Racks

Div. of Water Resources Engineering

Luleå University of Technology

S-951 87 Luleå, Sweden

Lars-Olof Andersson

Div. of Water Resources Engineering

Luleå University of Technology

S-951 87 Luleå, Sweden

and

SKEGA AB

S-934 81 Ersmark, Sweden

November 1992

ABSTRACT

A field experiment to study frazil ice blockage of water intake trash racks in hydropower plants is described. The experiment, conducted during the 1990/1991 winter using a submerged video camera, was designed to investigate ice accumulation on trash racks and compare this with laboratory tests. Differences between ice accumulation and adhesion on steel bars with and without a rubber coating were examined. The submerged video camera functioned well and was found to be an excellent tool to document and evaluate these ice phenomena. Frazil ice problems occured in air temperatures as high as -4°C. Frazil ice accumulation started on the upper section of the trash racks and progressed downwards. Ice was mainly deposited on the upstream face of the racks. The rubber coating material, with its poor adhesion to ice, appeared to mitigate frazil ice problems in the intakes in two ways. Ice was much more easily removed from trash rack elements coated with rubber and frazil did not stick to coated trash racks under small depress of supercoolings.

1. INTRODUCTION

Frazil ice is the name given to small ice particles formed in turbulent, supercooled water. The particles are primarily discoid in shape and supercooling of only a few hundredths of a degree is enough to favour frazil ice formation. The particles grow as long as the water is supercooled. During this `active' stage they easily adhere to objects with which they collide (Carstens, 1970). Observations in the field have shown individual frazil ice particle sizes between 1 and 15 mm (Wigle, 1970, Ashton, 1979, Osterkamp et al., 1982).

Water velocities higher than 1 m/s cause enough turbulence to keep the frazil ice particles well mixed (Carstens, 1970, Matousek, 1984). Layered frazil ice and slush ice are formed when the flow velocity is between 0.6 and 1 m/s. Skim ice and an ice cover are formed at velocities lower than 0.6 m/s.

A fundamental study of frazil ice particle formation and growth was carried out by Daly (1984). In order to simulate the motion of frazil ice crystals, a numerical model is under development (Svensson et al. 1988, Andersson et al. 1991).

Laboratory tests to achieve a better understanding of the prevention of frazil ice problems have been conducted (Daly 1987a, Daly 1987b). Small scale experiments using trash racks at field sites have been carried out (Mussalli 1987, Haynes et al., 1991).

The accumulation of frazil ice on water intakes is a serious problem in many hydro-power plants operated in cold climates (Ashton, 1986, Daly, 1991). A complete blockage resulting in total shutdown, preceded by a large reduction in power production is not unusual. Production losses due to ice problems may be costly for the hydro-power industry. At just one Swedish plant during a 10 year period, the average cost of ice problems and heating the trash rack was 1 million SEK/year, equivalent to $0.2 million/year. The costs of having staff on a 24-hour alert were not included.

2. OBJECTIVES

To improve the understanding of frazil ice formation and growth, a research project was initialized in 1987 at the University of Luleå, Sweden. The long term goal of the project is to increase the knowledge of water intake blockage due to frazil ice and to find materials and methods that will reduce ice adhesion to trash racks.

The aims of this study were:

a) To observe the weather conditions and temperature changes before frazil ice problems occurred.

b) To record the frazil ice accumulation on single bars and on the entire trash rack, with respect to water depth, flow rate and time.

c) To estimate the adhesion forces between ice and trash rack elements with different surface materials at field sites.

We know of no earlier investigations in full scale power plants that provide answers with regard to the influence of water depth, the single bar geometry, side walls etc. Experiments made in the laboratory with model trash racks might give different answers than trials with trash racks in field. The field tests were also used to indicate the practical value and limitations of laboratory tests. Controlled experiments in a flume located in a cold room can be conducted and repeated. In the field, frazil ice formation is highly variable. Conclusions drawn from results on a small scale need to be compared with real situations in the field, in order to evaluate the practical merits of using laboratory experiments to find the solution to the prevention of ice problems in a hydro-power plant.

It was hoped to verify laboratory results by comparing estimates of adhesion between ice and rubber with the adhesive forces between ice and steel surfaces.

3. EXPERIMENTAL

Experiments were conducted during the 1990/91 winter at three hydro-power plants in Sweden: Finnfors on the Skellefte River; Untraverket on the River Dalälven, and Olidan on the Göta River, listed from north to south (see locations in Figure 3.1). These three power plants were chosen because they often experience frazil ice problems. The design of Untraverket and Olidan permits the easy installation and maintenance of test equipment (see figure 3.2). Each intake can be drained in a few minutes.

General data for the power-plants are shown in Table 3.1.

Table 3.1: General data for the hydro-power plants tested.

---------------------------------------------------------
Power Plant   Head [m]      Flow [m3/s]   Power         
                                          Production    
                                          [MW]          
---------------------------------------------------------
Finnfors          20.0      250           50            
Untraverket       13.5      350           42            
Olidan            32.0      550           135           

3.1. Test equipment

The experiment was documented with a video camera, Sony CCD 500, installed in a marine pack MPK-F340. This equipment makes recordings possible under water down to a depth of 40 metres. An underwater lighting system, IKELITE UVHC-3, was used to improve the light conditions. The power of the light was 20 W which was enough to give a clear and bright picture without too much light scatter from particles in the water. Both the marine pack and the light were set in a steel box to avoid damage to the equipment due to impact from timber or ice blocks. Descriptions of the test equipment for each power plant are given below.

Finnfors, Skellefte River

At Finnfors one trash rack section (1.2 x 7.0 m) was replaced by an identical rack coated with a 3 mm thick rubber material. The location is shown in Figure 3.3. A qualitative evaluation, by operating the trash rack cleaner, of different ice accumulations was achieved by comparing the performances of the surfaces of the original and coated racks.

Untraverket, River of Dalälven

Upstream of the plant, a 1 km long canal runs between a dam and the front of the intake. The first 700 m runs in a westsouthwest to northnortheast direction and the last 300 m runs east to west. The canal has a width of 80 m and a depth of 6 m. No test equipment was set up on the existing trash rack. The video camera and its steel box were mounted on the trash rack cleaner (Figure 3.4). The ice accumulation across the whole depth down to 8 and 9 m, could then be recorded.

Olidan, Göta River

A trash rack rake moves up and down in front of the trash rack. The rake is used to rinse the intakes from small objects like grass, leaves and other trash that causes head losses. The rack has a 6 m long shaft which permits rinsing down to 4 m below the water surface. In our experiments the rinsing part of the rake was replaced by the steel box, housing the underwater video camera. Consequently recordings could be done down to 4 metres.

Three steel test bars with a rectangular cross section 50 mm in depth and 30 mm wide were mounted in front of the existing, heated, trash rack (figure 3.5). The middle bar was coated with the same rubber material as that used at Finnfors. A rubber coated wire with a triangular cross section was mounted next to the bars, intended to evaluate the influence on ice accretion of self-induced vibrations combined with a rubber coating. The 60 mm spaced test specimens were electrically insulated from the trash rack and mounted approximately 0.1 m in front of it. This design assured that water past the test specimen first. Heat from the control trash rack would then not affect the ice build-up on the test bars but assured that the ice accretion started on the test sections. In this test design, mechanical rinsing of the test section was not possible.

Underwater video recordings showed the ice accumulation at different depths and also the difference in ice growth on the steel bars and coated bars. Particle behaviour in front of the rack was also recorded. Video recordings were used to evaluate the adhesion of ice to the different specimens.

3.2. Experimental variables

3.2.1. Temperature measurements and weather conditions

The temperature was regarded as the most critical parameter in the test and it was decided that it should be determined by as many different independent systems as possible. The strategy included the easy replacement of damaged of temperature probes, preferably without draining the intake. The temperature was measured at three different depths in order to find eventual temperature gradients during the freezing process. The different systems used were thermistors, copper/constantan thermocouples and Pt 100 platinum resistance thermometers. The Cu/constantan probes were used to measure water temperature at the test site, air temperature outside the intake approximately 2 m above the water level and air temperature at the test set up, inside the trash rack house.

Thermocouples of copper/constantan were individually shielded in polyethylene tubes, sealed with a PVC-resin. The shielded gauges were inserted in steel tubes, fastened on the trash racks for protection. Care was taken to place the tubes at the required depth. Each thermocouple was individually checked at 0°C in a mixture of ice/snow and water, before and after the field tests were conducted. The necessary corrections were made during data analysis. It was then possible to measure the temperature with an accuracy of a few hundredths of a degree.

Thermistors or NTC (Negative Temperature Coefficient) resistors are non-linear and show an exponential relationship between resistance and temperature in

R=exp(a+bT)

where T is the temperature in degrees Kelvin, R is the resistance in ohms and a and b are constants. The exponential relationship means that the sensor is sensitive to small changes in temperature and, since thermistors are cheap, they are an interesting alternative to other measuring systems for temperature, especially if the temperature has to be monitored at many points.

Thermistors used in this test have been NTC-Resistors 60-284-50, with a nominal resistance of 50 000 W at +25 °C supplied by ELFA, Sweden. Since the overall resistance in the wire and connections is much smaller, only a few ohms, the system will be fairly insensitive to changes in resistance in the wire due to temperature changes. Each thermistor was soldered to filmshielded wires in a KAHL 55-752-04 cable from the same supplier. The two conductors were individually insulated with rubber tape and placed inside a degreased polyethylene tube with an inside diameter of 4 mm. The thermistor glass bead was centred and the PE-tube subsequently filled with a two component epoxy resin. After the epoxy had been allowed to cure, the PE-tube was slowly dipped into and withdrawn from a PVC-resin twice to form an approximately 1 mm thick watertight layer. A junction box in which the resistance was converted to a voltage proportional to the temperature was constructed. The voltage thus obtained was then used as an input to the data logger.

Pt 100 temperature probes, with an accuracy of 1/10 DIN, shielded by stainless steel tubes were used during calibration at the field site. Using a Thermolyzer S2541 between the Pt-probe and the data logger, the required resolution was achieved in the vicinity of 0 °C.

Finnfors. Thermocouples were used to record water and air temperature. The water temperature was measured 2 m below the water surface and 0.5 m behind the trash rack.

Untraverket. Air and water temperature were measured with thermocouples. The water temperature was measured continuously at a water depth of 6 m and on some occasions at water depths of 4 and 8 m. Once a day, at 7:30 a.m., the air temperature and water temperature were recorded manually with an accurate mercury thermometer and a Pt-100 Platinum resistance thermometer respectively.

Synoptic and automatic weather stations are located around Sweden. Data on air temperatures, amount of precipitation, wind speeds and wind directions were obtained from stations near to each power plant.

3.2.2. Water velocity

Finnfors. Water velocities were measured at 10 m intervals on straight lines perpendicular to the intake at three locations 50 m upstream of the power plant. At the time of measurement, the river was partly covered with ice.

Untranverket. The water velocity profile across the entire trash rack was measured when no frazil ice was present and the flow rate through the plant was normal for the season.

4. RESULTS

4.1. Temperature measurements and weather conditions

Finnfors. Frazil ice formation appeared suddenly at Finnfors on 13 November 1990. Temperatures recorded before and after the frazil problems occurred are tabulated in Table 4.1. Head losses larger than 2 m exposed the temperature gauge to the air which caused the peak shown in Figure 4.1.

Table 4.1: Temperatures observed at Finnfors power plant.

---------------------------------------------------------
    Date      Time       T (water) °C   T (air) °C   
---------------------------------------------------------
   901106     15:30         1.04          - 9           
   901107     07:30         0.77                        
   901112     07:30         0.18          - 6           
              15:30         0.27          - 2           
   901113     07:30         0.08          - 8           
              15:30         0.03          - 8           
   901114     07:30         0.01          - 14          
              15:30         0.04          - 14          
   901115     07:30         0.01          - 8           
              15:30         0.02                        
   901116     07:30         0.04          - 1           
              15:30         0.03                        

Untraverket. Figure 4.2 shows air temperatures. At Gävle and Films Kyrkby the average daily temperatures were measured in the automatic and synoptic station. The third curve represents the temperature recorded at the power plant at 7:30 each morning. In general all three temperature curves follow the same pattern.

Air and water temperatures recorded each morning at Untraverket power plant from November 1 to the end of December, are shown in Figure 4.3. The water temperature declined while air temperatures were below 0 °C. Air temperatures fluctuated more rapidly than water temperatures. Water temperature did not vary with depth.

Frazil ice problems occurred during December,1990, coincident with snowy weather (Figure 4.3). The wind direction was northnortheast to northnorthwest during the first part of December, i.e. parallel in the downstream direction to the intake canal, and during the rest of the month the wind direction was southwest. The average wind speed was 3.5 m/s.

It should be noted that air temperatures of only -4 °C were sufficiently low to cause frazil ice problems (see Dec 5 1990 in figure 4.3).

Olidan. Olidan experienced frazil ice problems during February 1991. Average daily air temperatures were measured at a nearby weather station (Såtenäs, figure 4.4). Frazil ice problems occurred on the February 5 at an air temperature of -7.0 °C. Little precipitation occurred during that day and the average wind speed was 5 m/s from the northeast.

4.2. Water velocity

Finnfors. The water velocities, measured upstream of the intake, ranged between approximately 0.3 m/s at the centre of the river, to 0.8 m/s about 20 m from the shore, figure 4.5.

Untraverket. Immediately upstream of the intake, the velocity profile was constant with depth except close to the surface and bottom, see figure 4.6. The velocity at the surface was low (0.2 m/s) because of the skimmer wall which extended 1.5 m below water level, i.e. 2.5 m below the top of the trash rack (Figure 3.2). A wake was caused that affected the flow in front of the rack. Below the skimmer wall, the water velocity reached a constant value of 1.2 m/s. Upstream of the intake canal, the water velocity varied between 0.2 and 0.6 m/s depending on the flow through the plant (100 and 300 m3/s respectively).

Olidan. The water velocity was not measured but was calculated to 0.8 m/s when the flow rate was 40 m3/s.

4.3. Frazil ice accretion

Untraverket. Untraverket experienced frazil ice problems several times at the beginning of the 1990/91 winter season. The freezing process under water was recorded on two different occasions on a third occasion, the gate upstream of the intake was lowered to the bottom and the entire trash rack was drained and exposed in the trash rack house. The frazil ice blockage and the thickness of the accumulated ice was estimated. Figures 4.7 and 4.8 show the results of ice accretion measurements on two of the trash rack bars on two occasions. These were considered representative of other samples. Figure 4.7 indicates that ice accumulation started at a depth of approximately 2 m. 1 hour and 20 minutes later, ice had reached to a depth of almost 7 m at intake 2 and approximately 5 m at intakes 3 and 5.

Figure 4.8 indicates that on this occasion the ice build-up had been in progress for a longer time than that shown in the previous figure. The ice growth reached a depth of 7 m in the middle of the day and to the bottom, 8m, at approximately 17.00, resulting in complete blockage.

The bars were visible through the ice accumulation. The ice remained transparent, but became more opaque when complete blockage occurred. This can be explained by a high porosity of the accumulated ice. During accumulation the deposited ice became more and more dense, and hence increasingly opaque. The accumulated ice thickness was 10 to 15 cm immediately before the intake was drained on 21 December.

During this winter period, water was discharged through the spillways resulting in energy production losses. Figure 4.9 shows the spillage during the period when frazil ice problems occurred. The spillage peak on 9 December occurred because of drifting ice problems not because of frazil ice. The plant turbine had to shut down to enable the intake to be cleaned from surface ice, transported from upstream. The remaining peaks (marked with arrows) were caused by frazil ice problems.

Olidan. A video, recorded while the hydro-power plant was in operation, showed the ice accumulation at different depths and also the differences in ice growth between the steel bars and coated bars. Particle behaviour in front of the rack was also recorded. Olidan experienced frazil ice problems on the night of 4/5 February. Ice accumulation on only the uppermost 1 m of the rectangular steel bars of the test section could be recorded because a large amount of block-ice from upstream collected above most of the rack (see figure 4.10). The ice blocks, together with frazil ice deposition, caused a head loss of approximately 2.5 m, rather than complete blockage. The ice blocks originated from the break up of the ice sheet on the river, which is kept open for navigation.

Frazil ice accumulated on the steel bars to a depth of 1 m with an ice thickness of 5 cm. Ice samples were taken from both the test bar and the block ice. Thin sections were prepared and evaluated in cross-polarized light. The block-ice consisted of frazil ice. The size of the individual frazil ice particles in the blocks ranged between 5 and 8 mm. The trash rack was drained and exposed to warm air and the ice blockage could be recorded. Frazil ice problems continued on several occasions over a period of approximately 2 weeks.

4.4. Frazil ice adhesion

Finnfors. At 00.30 on November 14, the engineer on duty was informed by the Power company's Central Control Board that power production was very low at Finnfors. When he arrived at the power station, the head loss was 14 m.

No difference in accumulated ice thickness on the original trash rack and the coated test rack was observed before the cleaning. The flow patterns in front of both trash racks were also similar. It should be noted that the trash rack cleaner measured 3 m in width, but the test section was only 1.2 m wide. The rack cleaner therefore covered a further 0.9 m, on each side of the non-coated trash racks, during cleaning.

A significant difference in the de-icing performance of the rinsing equipment was observed when the rubber coated trash rack was cleaned, compared to the original trash rack elements. Ice was removed from the coated rack in large pieces (approximately 1 by 1.2 m). This effect was observed several times. After cleaning, a substantially greater flow was indicated through the rubber coated trash rack than through the control rack, despite a reduced spacing between the trash rack bars of approximately 8% due to the coating. Ice removed from the un-coated rack was fragmented. The problem continued for about 24 hours until all ice was mechanically removed and full production could be achieved.

Untraverket. Recordings taken during the drainage of the intake on December 21, showed the difficulties involved in the removal of ice from the conventional steel trash rack using the trash rack cleaner. Although the trash rack was exposed to temperatures of approximately 10 °C, the ice was difficult to remove. Half of the trash rack area was equipped with electrical heating which caused the ice to melt and fall off after some time.

Olidan. No ice accumulated on the bar coated with rubber material. The rubber coated test wire did not accumulate ice either. Whether this was due to self induced vibrations or to the rubber material itself is not clear. Ice on the original trash rack was difficult to remove shortly after it had been drained and exposed to warmer air.

5. DISCUSSION

Frazil ice accumulation as a function of depth was the main parameter studied at Untraverket. The video recordings indicated that the ice accumulation started at the upper part of the trash rack and progressed downwards. Frazil ice started to accumulate approximately 2 m below the water surface. This can be explained by the location of the skimmer wall which reached 1.5 m below the water line and caused wake currents and low velocities at the upper part of the trash rack. Below 1.5 m depth the velocity was constant and the frazil ice particles could be transported directly towards the trash rack, to form the first ice growth.

Several reasons may account for the origin of ice accumulation in the upper region of the trash racks. Frazil ice is more concentrated close to the water surface than at a depth below the surface. When the upper section of the trash rack becomes blocked, water carry-ing the frazil ice particles is forced downwards. The effect on pressure of the freezing water may contribute to a small extent to the phenomenon. A change in pressure equivalent to a 10 m increase in water depth causes a change of -0.0074 °C (Hobbs, 1974) in the freezing point. Depths of this magnitude are common in front of an intake and the supercooling is not more than a few hundredths of a degree.

Ice on the trash racks is composed mainly of frazil ice particles, carried by the water and deposited on the bars. The trash racks in this study were all located indoors where the temperature was well above freezing point (around 10°C), so that ice growth due to the transfer of latent heat from the rack to the air will not take place. Calculations of the magnitude of this form of ice growth have shown that, even when the air temperature is below freezing point, the mass growth rate is negligible compared with the mass rate of deposition of the frazil ice suspended in the flow (Daly, 1987a). Growth resulting from the transfer of the latent heat to the supercooled water is 100 times greater than growth due to latent heat transferred from the rack to the air. This is still negligible compared with deposited frazil ice.

Video recordings showed that the accumulated ice was highly porous, i.e. had a high water content which caused transparency. Adhesive strength between the ice and the steel bar was nevertheless high. This was corroborated by the inability of the trash rack cleaner to remove the ice.

Frazil ice may not be the only ice problem in a hydro-power plant such as Olidan. Navigation on the river produced a large amount of block ice that originated from broken ice cover. Ice blocks together with frazil ice caused blockages of the entire intake.

The video recordings showed frazil ice particles very clearly. The discoid particles which moved irregularly due to turbulence, could be seen when they reflected light. Particle sizes up to 8 mm indicated that they have been transported long way in supercooled water.

In 1990, severe frazil ice problems occurred at Finnfors, in the early morning of November 14, coincident with supercooling down to -0.05 °C, recorded with the acquisition system. A high level of supercooling may favour adhesion, since the bars will cool to a lower temperature relative to the freezing point (Daly, 1987a).

Table 5.1: Thermal conductivities for different materials (1*Hands et.al., 1977, 2*Broberg et.al., 1976, 3*Weast et.al., 1984, 4*Hobbs, 1974)

------------------------------------------------------------------------
Material                             Material                     
                         (W/mK)                                 (W/mK)        
------------------------------------------------------------------------
Iron (Fe)                80 1*         Carbon black filled      0.33 3*       
                                       (60 phr) rubber                        
Stainless steel,         15 2*         Water                    0.56 4*       
SS-2333                                                                       
Copper                   401 1*        Ice (solid)              2.2 4*        
Rubber (pure gum)        0.18 3*                                              

A difference in performance between materials of different thermal conductivities has been observed at Olidan. Submerged copper and a steel plates are used as frazil ice indicators.

In Olidan, no frazil ice accumulated on the rubber coated specimen. This was not expected since no difference in ice accretion between the tested materials has been observed in laboratory experiments (Jensen et al., 1989). At Finnfors no difference in accumulated ice thickness was observed after the icing had taken place. The probable reason for this discrepancy may be a combination of three factors.

Firstly, the supercooling at Finnfors was approximately -0.05 °C. The channel has a large width-to-depth ratio which results in a large cooling rate it being regarded as one of the coldest reaches in Sweden. In Olidan, frazil ice problems occurred at only one intake and they were not severe. A maximum head loss of about 2 m was recorded.

Secondly, the depth of the clearance between the test specimen, extension in the flow direction, was greater in Finnfors (120 mm) than in Olidan (50 mm). Hence, the probability for an ice particle to attach should be greater at Finnfors. A possible explanation of the time lag in Figure 4.1 could be that the rubber coated part stayed open for a longer time than the uncoated part. If this was the case, an increased water velocity, due to a decreased open area during the ice accumulation, would further prevent the ice particles from adhering to the coated bars. Unfortunately, the icing process could not be documented below the water surface.

Finally the lower accretion of ice to rubber may be explained by the lower surface energy of rubber, approximately 30 mJm-2 compared with metals, that usually show values of hundreds of mJm-2 in practice, due to natural contamination (clean machined metal surfaces have surface energies of 1000 to 3000 mJm-2). If the adhesive forces are small enough, the drag force may be high enough to remove the deposited ice before bridging occurs.

A significant difference in ice adhesion between rubber-coated trash racks and non-coated trash racks was expected. The choice of rubber material, designated P1-13534, in the field test was based on materials tested in the laboratory which had earlier shown low ice adhesion properties (Andersson, 1989). It was however modified to comply with the requirements of tear strength, wear resistance, good adhesion to metals etc. Video inspection at the coated rack element one year after the installation showed minimal or no wear of the rubber surface, despite the coated trash rack. Due to the coating extending 3 mm in front of the adjacent trash rack, elements have been subjected to extra high strains.

It should also be emphasised that materials that show low ice adhesion values do not automatically show a resistance to ice accretion. Ice attached to the rubber coated trash rack was however removed much more easily than ice from the non-coated trash racks, even though the test section measured only 1/3 of the width of the mechanical rake device. If the test section had been de-iced separately, the difference might have been even more pronounced.

Earlier investigations (Sayward, 1977, Andersson, 1989) have shown that adhesion forces always exist between ice and substrate since all surfaces have a certain surface energy. Although these forces are small for polymers, ice may well be difficult to remove from such surfaces. In the case of trash racks, an optimum solution to blockage due to frazil formation would be that, as the frazil ice builds up, the reduction in section area would build up a pressure that at a specific level would shear the accumulated ice away. This is what we think happened in Olidan. This tendency has also been found for this material in laboratory tests (not published). Beside pure shear stresses, the ice build-up may disturb the normal flow pattern across the rack and create vibrations. Personnel at the power plants explain that the trash racks sometimes start to "sing" during the frazil ice build-up. This extra energy input may be enough in combination with a low energy coating to allow for "self rinsing". If this is not enough, then mechanical rinsing is required.

By carefully designing the trash rack elements and selecting the proper surface treatment many of the problems that exist today could be to a great extent reduced.

There is, as far as we can see, no simple solution to icing on submerged structures.

6. CONCLUSIONS

Conclusions that can be drawn based on the field experiments are:

* The frazil ice accretion on trash rack starts at the upper section, progresses downward and ends in a complete blockage.

* The ice adhesion is lower on rubber coated materials than on steel.

* In one test, frazil ice accumulated on the test steel bar but not on the rubber coated test bar.

* A submerged video camera is a very powerful tool to register frazil ice formation at field sites.

7. FURTHER WORK

Further work should include the following:

* Repeat the recordings of ice accumulations on the trash rack at Untraverket. Try to capture the start and the course of the ice blockage as well as the drainage on the same occasion. Analyse statistics on the course of frazil ice problems on intake trash rack.

* Improve and extend the test equipment in Olidan. Build a system to record frazil ice accretion on a single steel bar and compare this with laboratory experiments.

* Record temperature data and weather conditions that correspond to frazil ice problems.

* Replace a larger section of the trash rack with a coated section to be able to operate the trash rack cleaner within the coated section at Finnfors. Evaluate the effect of the trash rack cleaner on the coated section.

* Extend the field test to instal coated trash rack sections in other power plants.

8. ACKNOWLEDGEMENTS

Special thanks are due to Dr. Lennart Billfalk for his advice and for reviewing this report. Sincere thanks are extended to the personnel at the electric plants of Finnfors, Untraverket and Olidan. The work has been financed by SKEGA AB, Swedish State Power Board, Stockholm Energy, Skellefteå Kraft, The Swedish National Industrial Board, Coldtech and Luleå University of Technology.

9. REFERENCES

Andersson, A. and Svensson, U. (1991) A Numerical Simulation of an Ice Particle Trajectory. In Proceedings at the 4th International Phoenics Users Conference, Miami Beach, USA, pp. 241-288.

Andersson, L.O., (1989). Ice Adhesion to Polymer Materials. Proceedings at the 10th International POAC Conference, Luleå, Sweden, pp.786-795.

Ashton, G.D. (1979) Frazil Ice. American Scientist.

Ashton, G.D. Ed (1986) River and Lake Ice Engineering. Littleton, Colorado: Water Resources Publications. ISBN 0-918334-59-4.

Ashton, G.D. (1988) Intake Design for Ice Conditions. Developments in Hydraulic Engineering-5 (P. Novak, Ed.). New York: Elsevier Applied Science.

Broberg, B. et al (1976) Formelsamling i hållfasthetslära. Publication no. 104. Department of solid Mechanics, Royal University of Technology, Stockholm. Printed in Stockholm, pp 186-193.

Carstens, T. (1970) Heat Exchanges and Frazil Formation. Proceedings of the Symposium on Ice and its Action on Hydraulic Structures, Reykjavik, Island, IAHR Paper no. 2.11.

Daly, S.F. (1984) Frazil Ice Dynamics. CRREL Monograph 84-1, US Army CRREL, Hanover, NH.

Daly, S.F. (1987 a) Modelling trash rack freeze up by frazil ice. Proceedings of the first International Symposium on Cold Regions Heat Transfer, June 4-7, Edmonton, Alberta.

Daly, S.F. (1987 b) The evolution of frazil ice in rivers and streams. Research and Control. Proceedings of the first International Symposium on Cold Regions Heat Transfer, June 4-7, Edmonton, Alberta.

Daly, S.F. (1991) Frazil Ice Blockage of Intake Trash Racks. Cold Regions Technical Digest, No. 91-1, March, US Army CRREL, Hanover, NH.

Hands, D. and Horsfall, F. (1977) Thermal Diffusivity and Conductivity of Natural Rubber Compounds. RAPRA Members Report No. 9. The Rubber and Plastics Research Association.

Haynes, F.D.,Daly, S.F., Garfield, D.E. and Clark, C.H. (1991) Field Test of Trash Rack Heating to Prevent Frazil Ice Blockage. Preliminary Report, US Army CRREL, Hanover, NH.

Hobbs, P.V. (1974) Ice Physics. Printed in Great Britain by J.W. Arrowsmith Ltd, Bristol Bs3 2NT, England. ISBN 0-19-851936-2, p357.

Jensen, H. (1989) Testing of ice growth on models of bars for trash racks. Report No. STF60 F89047. Norwegian Hydrotechnical Laboratory, Trondheim, NORWAY.

Matousek, V. (1984) Types of Ice Run and Conditions for their Formation. In Proceedings at IAHR Ice Symp., Hamburg, Vol I, pp. 315-328.

Mussalli, Y.G., Gordon L.S. and Daly, S.F. (1987) Frazil ice control using electromechanical vibrators and ice resistance coating. Presented at ASCE WATERPOWER 87, Portland, Oregon, August 19-21.

Osterkamp, T.E. and Gosink, J.P. (1982) Selected Aspects of Frazil Ice Formation and Ice Cover Development in Turbulent Streams. Proceedings of the Workshop (2nd) on Hydraulics of Ice-Covered Rivers, NRCC Associate Committee on Hydrology and Subcommittee on Hydraulics of Ice-Covered Rivers, Civil Engineering Department, University of Alberta, Edmonton, Alberta T6G 2G7, pp. 131-146. (Also see discussions and replies, p. 147)

Svensson, U. and Andersson, A. (1988) A Numerical Model of Ice Accretion on Structure, Proceedings at IAHR Ice Symposium, Sapporo, Japan, Vol. 2, pp. 436-445.

Weast, R.C., Astle, M.J. and Beyer, W.H. (1984) CRC Handbook of Chemistry and Physics. 64th edition. Library of Congress Card No. 13-11056. Printed in U.S.A. ISBN 0-8494-0464-4, pp E9-E10.

Wigle, T.E. (1970) Investigation into Frazil, Bottom Ice and Surface Ice Formation in the Niagra River. Proceedings of the Symposium on Ice and its Action on Hydraulic Structures, Reykjavik, Island, IAHR Paper no. 2.8.