Ice in the Environment: Proceedings of the 16th IAHR International Symposium on Ice Dunedin, New Zealand, 2nd–6th December 2002 International Association of Hydraulic Engineering and Research FRAZIL-ICE INGESTION BY A SUBMERGED WATER INTAKE:

NUMERICAL-MODEL AND ICE-TANK FINDINGS

Robert Ettema1, Zhiming Chen1 and Yong Lai1

ABSTRACT This paper addresses the problem of frazil-ice blockage by an offshore, submerged water intake in frigid waters. It is well known that such intakes are potentially prone to ice blockage by frazil ice, and possibly by slush or brash ice drawn with flow to the intake. However, little is known about how intake geometry and size influence intake performance with respect to frazil ice. Presented are findings from a numerical model, and an ice-tank model, of flow-field behavior and frazil-ice ingestion by an intake located offshore. The intake form considered comprises a conical inlet, fitted optionally with an elevated cap, and connected to a shoreward pipeline. The ice-tank experiments are unique, insofar that experiments involving frazil ice have not before been done at the scale of the present experiments. The numerical and the ice-tank findings show that an elevated cap placed above the intake may substantially reduce the amount of frazil ice ingested by a submerged, conical intake. INTRODUCTION Submerged, offshore water intakes are used quite commonly for withdrawing water from coastal waters. For example, they are used to meet the diverse water needs of thermal power stations, oil refineries, and communities located near coastal waters. Presently, there are few guidelines (and little actually known) about the hydraulic performance of such intakes. Especially lacking is information about intake performance in frigid-water conditions. Under such conditions, submerged intakes are prone to ice blockage by frazil ice. Herein are the results of a numerical model and a series of ice-tank experiments on frazil-ice ingestion by an offshore, submerged intake. The intake is located in moderately deep water (Fig. 1). The layout of the intake is fairly representative of offshore intakes used in the U.S. Great Lakes. The intake comprises a simple cone fitted with an elevated cap above the cone. The findings show how a cap, elevated to variable heights, influences frazil-ice ingestion by the intake. Of practical design importance are the merits of a cap, and an indication of the optimum height of the cap above the rim of the intake (Fig. 1.). The paper extends flow-field insights presented earlier by Ettema et al. (2001), who investigated the flow field at a conical intake.

1 Iowa Institute of Hydraulic Research, University of Iowa, Iowa City, Iowa City, IA

Figure 1: Typical prototype layout and dimensions of generic intake condition investigated. Intake rim diameter, D, and cap height, H, variable.

The ice-tank experiments are unique insofar that there appear to be no prior laboratory experiments entailing frazil-ice formation at the size scale of an ice tank. Prior laboratory experiments involved much smaller volumes of chilled water, usually of the order of 1m3 or considerably less. BACKGROUND Reliable techniques for mitigating ice blockage of submerged intakes have yet to be developed. The various frazil-mitigation techniques suggested for partially submerged intakes (e.g., Daly, 1991) have not been tried for submerged intakes, and some methods may be inappropriate. A difficulty for offshore, submerged intakes is the distance from shoreline to intake. Typically that distance is of the orders of a 1000 m, thereby complicating the use of mechanical (e.g., scraping; circulation of heated water) or even electrical techniques for ice control. Consequently, it is likely that considerable reliance will have to be placed on designing the intake flow field so as to minimize intake entrainment of frazil ice. Remoteness and inaccessibility have precluded opportunities to directly observe (not to mention measure) frazil-ice accumulation on offshore, submerged intakes. It is thought that frazil ice accumulates on intake trash-rack bars and appurtenant fixtures of intakes. If they develop as they do on trash racks for riverside or lakeside intakes, accumulations usually grow into the flow by continual deposition, extending upstream and increasing in width until the space between adjacent bars is bridged and then completely blocked (e.g., Daly, 1991). Most water intakes are operated so as to produce a constant, pumped flow rate, irrespective of their depth of submergence; essentially, as submergence depth decreases, and ambient pressure head decreases, a pump has to work harder to create the pressure head needed to deliver the same flow rate. Frazil-ice (and possibly slush-ice) accumulation increases headloss associated with flow from the intake to the on-shore

pump station. Operators of offshore intakes commonly notice the onset of frazil-ice accumulation at an intake by the drop in water level at the pump sump; the drop often occurs rapidly, within a few hours. NUMERICAL MODEL Advances in computational fluid dynamics now make it relatively convenient to determine flow fields generated by submerged intakes. The present study used a three-dimensional numerical model, U2RANS (Lai et al., 2000). U2RANS is a comprehen-sive general-purpose code for modeling 3-D fluid flow. It solves the Reynolds averaged Navier-Stokes (RANS) equations. In the present study, the standard ε−k turbulence model with a wall function was chosen because of its success in similar applications. Ettema et al. (2001) give the model’s governing equations. The simulated discharge was at a rate of 3 32.24 10 m /s,-¥ with an average velocity of flow through the intake pipe is equal to 0.478 m/s. The flow field around and into the intake is unknown and was determined by numerical simulation. The water-air interface is treated as a plane of symmetry, for which the hydraulic-roughness height assumed for the ice-cover underside is 5 % of the ice-cover thickness considered. The calculations were carried out for different diameters of the intake rim (D = 3.05 m, 2.10 m, and 0.90 m) with and without a cap, and for different elevations between the cap and the intake. The computations were completed on a SGI Power Challenge workstation. The first calculations were made on two grids, with 300,000 and 500,000 grid points. The results differ by no more than 4 % with respect to the maximum velocity at a distance of 0.5D and 0.75D respectively from the intake’s centerline axis. Although this is not a rigorous measure of the grid dependency of the results, further grid refinement was not possible on the available computers, and further coarsening of the grid resulted in loss of the resolution of the flow field. The final calculations were made on the finer grid. For that case, a converged solution was obtained in about 1,500 iterations; errors reduced by about 4~5 orders of magnitude. This accuracy is considered adequate for the present study. ICE-TANK EXPERIMENTS The ice-tank experiments involved IIHR’s ice tank. Fig 2 shows the location of the intake (geometry as in Fig. 1, but size reduced by a scale of 10) within the tank. Fig. 3 is a view of the tank. The ice tank is 21 m long, 5 m wide, and 1.5 m deep. Flow through the intake was withdrawn by means of a pump, and was set at 3 32.43 10 m /s-¥ , about 15 % higher than the discharge scaled following Froude-number similitude, to compensate for the fact that the density of the frazil ice produced in the basin cannot be scaled. The flow was re-circulated to the basin by way of a return manifold that released the flow beneath the floor at the center of the basin. A cap was placed at selected heights (0.3D and 0.7D) above the intake. Before starting an experiment, the ice was skimmed from the tanks and retained at the tanks two ends. Two rows of fans placed along the tank’s walls were used to simulate wind action on water. The fans produced small waves (about 25 mm in height) that served to agitate the water surface, prevent an ice cover from readily forming, and thereby to force the water to super-cool. The air temperature above the ice tanks had to be below –10 °C for the tank to produce frazil ice in a measurable quantity.

Figure 2: Layout of ice tank and location of intake.

Figure 3: Ice-tank water surface agitated by airflow generated using fans. An underwater video camera, placed at the bottom of the basin, about 2 m away from the intake, was used to observe frazil ice formation and accumulation on a mesh placed over the intake’s mouth. To aid viewing of frazil ice, an underwater light was used to illuminate the area around the intake. Fig. 4 is a camera view showing frazil accumulating on the mesh across the intake mouth. Each test began by starting the pump used to withdraw flow through the intake. The volumetric concentration of frazil ice in the water drawn subsequently into the intake,

Figure 4: Frazil-ice accumulation on mesh across intake mouth. The rods were used to support an elevated cap placed above the intake rim. and the total number of frazil-ice particles drawn to the intake were estimated using the mass of frazil ice collected on a mesh placed across the intake’s mouth. The average volumetric concentration of frazil ice in the water column was estimated by dividing the weight of the collected frazil ice by the volume that the frazil ice occupied; i.e.,

ice ice icev

water ice

Vol W /γCVol QT+

= = (1)

where vC is volumetric concentration; Volice is the volume of the collected frazil ice; Volice+water is the total volume of the frazil ice and water entered the intake; Wice is the weight of the frazil ice collected on the mesh in the period T; iceγ is the density of ice, 917 kg/m3; and, Q is water discharge into the intake, 3 32.43 10 m /s.-¥ The number of frazil particles collected was estimated as

ice

ice

3 2 3

8v

VolNVolume of each frazil ice particle

Vol1( 2 10 ) π (0.1 10 )4

6.37 10 C

-

=

=¥ ¥ ¥ ¥ ¥

= ¥ ¥

(2)

Eq. (2) assumes a representative frazil-ice particle discoid in shape, with a diameter of 2 mm, and a thickness of 0.1 mm. Those dimensions coincide with the average size of frazil particle measured in the present experiments; the frazil was notably uniform in size.

The experiments were conducted to measure frazil-ice accumulation on the intake over four periods, four periods of time, 5, 10, 15 and 30 minutes. Fig. 5 shows the temporal increase in the mass of frazil ice collected by the mesh placed over the intake mouth. The mass of frazil ice collected on the intake mesh increased slowly with time, then increased rapidly as the frazil concentration in the tank increased. The mass of collected frazil ice asymptotically approached a maximum, which coincides with significant blockage of the intake by frazil ice; the blockage reduces the intake discharge, and in turn entrains less frazil ice into the intake. Frazil ice accumulation nearer the center of the mesh was higher than at the brim. This indicates that more frazil ice particles are drawn into the center of the intake. This observation matches the results obtained from the numerical model.

00.5

11.5

22.5

33.5

44.5

0 5 10 15 20 25 30 35

Time (minutes)

Col

lect

ed fr

azil

ice

( kg)

Figure 5: Temporal accumulation of frazil ice on the collection mesh across the intake mouth (no cap); intake flow started at time 0. Cap Influence on Frazil Ingestion The presence of a cap above the intake reduced the amount of frazil ice drawn to the intake. The experiment findings, for two cap heights (H/D of 0.3, 0.7, D = 3.05 m) and for no cap (or H/D effectively = 1.43), were conducted to obtain a trend of frazil ice entrainment with respect to cap height above the intake. The findings are given in Fig. 6, which relates the mass of frazil accumulated on the intake fitted with a cap to the mass accumulated when the intake had no cap. The ice-tank data are for the first 5 minutes of intake operation, during which time the mesh was not blocked, and thereby the influence of the cap on frazil ice entrainment best demonstrated. The ice-tank findings are supported by those from the numerical simulation, which also are plotted in Fig. 6. The numerical-model data are mass of collected ice normalized by mass collected by the intake without a cap. The ice-tank data show the cap to be somewhat more effective than does the numerical model. This finding may be attributable to the finer size of the frazil-ice particles in the numerical model relative to the intake size; being finer, the numerical frazil is less affected by buoyancy.

Figure 6: Influence of intake-cap height (H/D) on quantity of frazil ice ingested by intake.

Figure 7: Ice-tank experiment in which frazil-ice accumulation plugged the intake. CONCLUDING COMMENTS The findings presented in this paper show that an elevated cap placed above the intake helps isolate the intake from the water surface, and thereby may minimize frazil-ice ingestion by a submerged water intake of conical shape. The ice-tank experiments also show that comparatively large-scale experiments may be conducted on frazil ice;

Frazil Ingestion/Frazil Ingestion (No Cap)

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

H/D

0.0

0.5

1.0

1.5

Numerical modelIce tank

Note: H/D = Y/D = 1.43(no cap)

heretofore, frazil-ice experiments involved a much smaller volume of water than used in the present study. It is of interest to comment that experiments also showed how frazil ice accumulation can lead to the collapse and ingestion of a trash rack placed over the intake. In this regard, Fig. 7 shows a severe accumulation of frazil ice that completely plugged the test intake. REFERENCES Daly, S.F. Frazil Ice Blockage of Intake Trash Racks. USA Cold Regions Research and

Engineering Laboratory, US Army Corps Engineers, Cold Regions Technical Digest, No. 91-1 (1991).

Ettema, R., Chen, Z. and Lai, Y. Hydraulic performance of offshore water intakes: some CFD findings. In Proc. (CD) Ports and Ocean Engineering under Arctic Conditions Conference, August, Ottawa, Canada (2001).

Lai, Y.G., Weber, L. and Patel, V.C. U2RANS: A comprehensive hydraulic flow simulation code – its development and applications. In Proc. 4th Int. Conf. on HydroInformatics, Cedar Rapids, IA (2000).