Você está na página 1de 17

Aquacultural Engineering 31 (2004) 277293

Analysis of sediment transport modeling using computational uid dynamics (CFD) for aquaculture raceways
Dania L. Huggins, Raul H. Piedrahita , Tom Rumsey
Department of Biological and Agricultural Engineering, University of California, One Shields Avenue, Davis, CA 95616 5294, USA Received 5 December 2003; accepted 23 May 2004

Abstract A simulation model to analyze the water ow and sediment transport in aquaculture raceways was developed using a computational uid dynamics (CFD) software package. The simulation was used to evaluate the efciency of solids settling in the quiescent zone of existing trout raceways. This efciency was based on the percentage of solids removed, which corresponds to the percentage of solids introduced into the raceway that settle in it, with settling taking place primarily in the quiescent zone. The raceway selected for model validation was a rectangular concrete raceway 30.0 m long, 3.0 m wide, 0.9 m deep, with a slope of 0.01. The raceway included a quiescent zone of approximately 5.3 m in length, which was separated from the rearing area by a screen. The water ow rate through the raceway was approximately 0.058 m3 /s. Velocity measurements were recorded at 230 stations along the raceway using an acoustic Doppler velocimeter, for comparison with the results obtained from the simulations. For the purpose of simulating sediment transport, six groups of particles were used to account for the total suspended solids. The sizes of the particles selected were based on an experimentally determined distribution for solids from a similar raceway, and were 692, 532, 350, 204, 61, and 35 m for Groups 16, respectively. The particle density for each size was assumed to be 1150 kg/m3 . Values of the percentage of solids removed for the different particle sizes were 100.0% for the largest particles, and 54.7, 0.9, and 0.1% for the three smallest particles, respectively. This methodology of analyzing the raceway sediment transport in terms of its percentage of solids removed based on CFD simulations can also be used to examine raceway design alternatives for improving the particle removal efciency. 2004 Published by Elsevier B.V.
Keywords: Raceway; Design; Trout; Sediment; Modeling; CFD

Corresponding author. Tel.: +1-530-752-2780; fax: +1-530-752-2640. E-mail address: dhuggins@nccn.net (R.H. Piedrahita).

0144-8609/$ see front matter 2004 Published by Elsevier B.V. doi:10.1016/j.aquaeng.2004.05.007

278

D.L. Huggins et al. / Aquacultural Engineering 31 (2004) 277293

1. Introduction Given the large size of some trout rearing facilities, there is a potential for signicant nutrient discharges (IWMGAO, 1999; NASS, 2001). There are strong economic, social, and regulatory pressures to reduce the release of nutrients from aquaculture operations. However, the large ows and very low constituent concentrations make it very difcult to treat efuents. A potentially effective strategy to reduce the release of solids, which contain nitrogen, phosphorus, and organic matter, is to increase the proportion that settles within a raceways quiescent zone. Understanding how water and sediment particles move in the quiescent zone provides an opportunity to suggest raceway design modications aimed at improving particle settling and improving the quality of the efuents. Therefore, the focus of this study was to develop and validate a computation uid dynamics (CFD) model for a trout production raceway that can be used to evaluate the efciency of solids settling for a particular raceway design. CFD simulations have been used in aquaculture to describe water ow and solids removal in circular tanks (Montas et al., 2000; Veerapen et al., 2002). In their 2002 study, Veerapen and coworkers studied the factors inuencing the removal efciency of swirl separators and double-drain sh tanks. Rasmussen (2002) used CFD modeling to determine the mixing characteristics of a turbot rearing tank and the transport of particulate organic material and the oxygen distribution in rectangular tanks. There are also some studies applying CFD modeling to aquaculture ponds to simulate water ow velocity patterns and sediment conditions (Peterson et al., 2000). The information obtained from pond modeling has also been useful in determining the optimal arrangement of aerators in shrimp growout ponds (Peterson et al., 2001). In the work presented here, a methodology was developed to analyze the efciency of solids removal in a raceway using SSIIM (sediment simulation in water intakes with multiblock option) as the CFD software. CFD simulations were developed to analyze sediment transport for multiple sediment sizes, which can provide information on the distribution and ow of particles and on the proportion of the solids that settle within the quiescent zone, expressed as the percentage of solids removed.

2. Model characteristics The selection of SSIIM for this study was based mainly on its capability to simulate sediment movement. Computing system requirements and licensing fee were some of the other characteristics considered in the selection process. SSIIM is a free CFD program that can be downloaded for academic use from the internet (Olsen, 2002). In addition, SSIIM supports multiple operating systems. It can be executed in computers running under any version of WindowsTM , or OS2. The program is small in terms of its memory requirements (1.3 MB) and is not very resource intensive. Although SSIIM was developed about 10 years ago, it continues to be updated and there are frequent releases of new versions as improvements are made and bugs are identied and corrected. The execution of SSIIM requires two main input les, Control and Koordina. The Control le contains most of the parameters needed for model execution, such as size of arrays or

D.L. Huggins et al. / Aquacultural Engineering 31 (2004) 277293

279

number of grids along the three axes, Mannings roughness coefcient of the raceway walls, complete details of inlet and outlet congurations, ow rate, sediment size, concentration, and average settling velocity. The Koordina le contains the grid geometry. Five output les are written once the program has completed execution (Olsen, 2002) of which the Result and Conres les are used in the sediment transport calculations. The Result le is written by the program when all the iterations specied in the Control le have been completed and when the solutions to the governing equations have converged. The convergence criterion used by SSIIM is when the residuals for all the unknown variables fall below 103 (Olsen, 2002). The Result le contains the results for the velocities Vx (in the direction of the long axis of the raceway), Vy (in the horizontal direction perpendicular to the long axis of the raceway), Vz (in the vertical direction), pressure, turbulent kinetic energy, and diffusivity (, epsilon). The Conres le contains the simulated values of the concentrations for the various sediment groups at each node created in the Koordina le (Olsen, 2002). 2.1. Model equations The water ow velocity calculations are based on the NavierStokes equation for turbulent ow in a general 3D geometry for non-compressible and constant density ow (Olsen, 1991). SSIIM uses the k turbulence model to predict the shear stresses (Versteeg and Malalasekera, 1995). The k model focuses on the mechanisms that affect the turbulent kinetic energy. More details on how the k model is used for calculating the turbulent shear stress are given in Olsens SSIIM manual (Olsen, 2002). 2.2. Model assumptions SSIIM has some limitations in terms of its capabilities with different geometries, hydraulic congurations, and the number of cells (control volumes or elements). Due to these limitations some assumptions had to be made and these are described as follows. 2.2.1. Uniformly distributed inlet As a rst assumption, the model considers that the inuent ow is uniformly distributed along the full depth of the raceway at the rst upstream cross-section. This assumption was made because simulating inuent water splashing on the surface of the water (Fig. 1), as in a real raceway, made convergence in the model slow or unreachable in some cases. This difculty in simulating a surface inlet has been reported by other authors (Bates, 2000) and it appears to be a problem not only with this particular software package, but with CFD algorithms in general (Montas et al., 2000; Bates, 2000). Partly as a result of this assumption, validation of the model was carried out for the downstream end of the raceway, especially in the quiescent zone. 2.2.2. Fish presence As in other studies where CFD modeling has been applied for sh tank design, the presence of sh was not included in the model, due to the complexity that would be introduced and the lack of detailed information that would be needed for model development. For model

280

D.L. Huggins et al. / Aquacultural Engineering 31 (2004) 277293

Outlet Area Quiescent Zone Inlet Area


0.60m

RHS

5.07m 0.22m

0.4m

screen

0.63m

0.9m

1.0m

3.05m

30.0m

LHS
Fig. 1. Schematic diagram of a raceway (slope = 0.01).

D.L. Huggins et al. / Aquacultural Engineering 31 (2004) 277293

281

validation purposes it would have been best to be able to measure velocities in an empty raceway but it was difcult to combine this requirement with the planning and daily activities at the farm where the raceway used to collect calibration data was located. However, this assumption does not weaken the model in the quiescent zone where particle settling analysis is performed, since the sh are excluded from this region (Fig. 1). 2.2.3. Screen conguration A screen separates the rearing area from the quiescent zone. In the raceway the screen is 25.2 m from the inlet wall; however in the simulated system the screen was placed at 24.9 m (Fig. 1) from the inlet wall. The screen in the raceway consisted of two wood framed sections (5 cm 10 cm (2 in. 4 in.), nominal size). Each frame had 50 PVC pipes with an outside diameter (OD) of 2.5 cm (3/4 in. nominal size), which were placed vertically along the whole height of the screen leaving a space of 1.2 cm between the pipes (Fig. 2). The total number of closed spaces was 100 (pipes) with a total of 98 open spaces (the rst and the last pipe in each of the 2 wood frames were very close to the frame). This exact screen conguration was difcult to incorporate in the simulation due to the large number of cells that would have been required and the corresponding number of calculations needed. Therefore, the pipes were replaced with 12 bars (closed space, Fig. 2) to simplify the water

Fig. 2. Divider screen between the rearing area and the quiescent zone.

282

D.L. Huggins et al. / Aquacultural Engineering 31 (2004) 277293

ow calculations. The open spaces (between pipes) in the original system, along the width of the raceway, add up to 19.7% of the total width. However, due to grid size and other software limitations the simulated open area was 22.5% of the total width. The number of openings in the screen (12) was decided based on the maximum allowed number of wall statements that are necessary for dening all the bars in the program script. Other congurations with fewer bars were tested; however, the ow pattern was greatly inuenced by the reduction of the cross-sectional area around the screen. 2.2.4. Location for sediment release Due to the presence of the sh in the system, sediments were released and resuspended throughout the rearing area. Because the sh were not included in the simulation, there was no re-suspension of the particles in the sh rearing area as is observed in the raceways. Therefore, in the simulation, the sediments were released at the water surface extending over an area of 0.75 m 3.0 m (the full width of the raceway) located 0.75 m before the screen. These assumptions probably result in an underestimation of the effect of the quiescent zone since the sediments are released at the surface (before the screen), instead of being resuspended at mid-depth (or closer to the bottom) as they would be in a real raceway. SSIIM allows the user to specify the sediment concentration (volume fraction) for each size group at the water surface. The sediment ux can be calculated at the surface by multiplying the concentration by the particle density and the settling velocity. 2.2.5. Sediment resuspension Due to limitations with the software, sediment transport simulations do not account for resuspension from the bottom of the raceway. This will result in an over-estimation of the solids settled in the raceway. However, due to the cohesive nature of aquaculture particles, the magnitude of the over-estimation is expected to be substantially less than it would be for non-cohesive sediments. 2.3. Grid selection A grid dependency study (Versteeg and Malalasekera, 1995) was performed to eliminate errors due to the coarseness of the grid and also to determine the best compromise between simulation accuracy, numerical stability, convergence, and computational time (Table 1). The number of iterations are those required by the model to converge. The program was executed on a 1.7 GHz Pentium 4 computer with 512 MB of RAM. In this study three grids were tested. These grids were 60 30 12 (21,600 cells), 70 50 16 (56,000 cells), and 80 60 20 (96,000 cells). The rst number (e.g. 60) corresponds
Table 1 Characteristics of model executions for three grids Grid numbering 60 30 12 70 50 16 80 60 20 Time needed to converge (min) 15 110 240 No. of iterations 1609 2423 3637 No. of cells 21600 56000 96000

D.L. Huggins et al. / Aquacultural Engineering 31 (2004) 277293

283

to the number of grids (divisions) along the X-axis of the raceway (cross-sections), the second number (e.g. 30) corresponds to the number of grids along the Y-axis of the raceway (width of the raceway, longitudinal sections), and the third number (e.g. 12) is the number of levels in which the control volume has been divided in the Z-direction. As the number of cells increased so did the number of iterations and the time needed for the model to converge (Table 1). This was due to the increase in the number of cells at which the equations were solved. A qualitative grid analysis for the simulated velocities was made between the different grid sizes (60 30 12, 70 50 16, and 80 60 20). The simulated velocities were very similar for grids 70 50 16 and 80 60 20, but grid 60 30 12 differed from the other two, suggesting that the low resolution grid did not result in accurate simulations. Therefore, grid 70 50 16 was selected as the most appropriate grid for further analysis, considering that with a 80 60 20 grid it would have taken longer for the calculations to converge (Huggins, 2003). The grid spacing was not constant through the raceway. The cell dimensions in the nal grid were 1.0 m in the X-direction by 0.062 m in the Y-direction in areas requiring low resolution because of the uniformity of the ow, such as in the middle section of the raceway. In the areas requiring higher resolution (close to the screen and the outlet) the dimensions of the cells were 0.125 m 0.062 m in the X and Y-directions, respectively.

3. Methods The model was validated against data collected at a commercial trout farm. Details of the raceway in which the date were collected are presented below, followed by details on the model characteristics and simulation conditions. 3.1. Raceway characteristics The concrete raceway used in this study (standard raceway) was 30.0 m long, 3.0 m wide, with a maximum depth of 0.9 m and a slope of 0.01. A coordinate system for the raceway was set as X = distance along the raceway (length), Y = lateral direction, and Z = vertical direction (Fig. 1). The rearing volume, not including the quiescent zone, was approximately 48.4 m3 . Each of the inlet and outlet ows were through two weirs that together add to the total width of the raceway (Fig. 1). The sh stocking density was 11 kg/m3 with an average mass of 49 g per sh. The sides of the raceways are dened as inlet wall, outlet wall, right hand side, and left hand side for description and measurement purposes (Fig. 1). The ow rate for the standard raceway was approximately 0.058 m3 /s, which was calculated using the Francis equations for half-contracted rectangular sharp weirs (Vennard, 1954). 3.2. Data collection using an acoustic Doppler velocimeter (ADV) An acoustic Doppler velocimeter was used to measure velocities (Flow Tracker Handheld ADV, SonTek/YSI, Inc., San Diego, CA). The instrument has a two-dimensional (2D) probe, a velocity range of (0.0015 m/s), accuracy of 1% of the measured velocity, and

284

D.L. Huggins et al. / Aquacultural Engineering 31 (2004) 277293


Z

12

7 30.20

10 11

12

13

14

1516

5.02 m Screen
1=1.6 m

e1

Depths 0.2 h 0.6 h 0.8 h Bottom

5 6 7 8

1 2 3 4

9 10 11 12

Pl an

Detail of sampling points at Plane 1 h = maximum water level at different planes along the raceway

Fig. 3. Schematic diagram for a raceway, including all the planes where velocity measurements were taken.

a resolution of 0.0001 m/s (SonTek, 2002). With the 2D probe it was possible to obtain the horizontal components (Vx and Vy ) of the water velocity vector. The points of sampling were selected based on a preliminary analysis of velocities predicted through simulation. Measurements were concentrated in areas where velocities changed substantially as a function of their location in the raceway. Measurements were taken at 16 planes selected along the length of the raceway (X-axis, Fig. 3) and there were 12 stations for most planes, corresponding to 4 depths (Z-axis, as a fraction of the total depth measured from the surface 0.2 h, 0.6 h, 0.8 h, and at 3 cm from the bottom) and three locations along the width of the tank (Y-axis) (Fig. 3). 3.3. Solids characteristics and their introduction in the system SSIIM has the capability for particle tracking, adjusting the size and density of the input particles as well as the particle injection location. The particle size distribution used in the simulation was obtained from a settling velocity distribution determined in a previous study by Wong and Piedrahita (2000). The diameters of the particle groups were calculated with Stokes law assuming a density of 1150 kg/m3 (Metcalf and Eddy, 1991; Wong, 2001; Timmons et al., 2002) (Table 2). 3.4. Sediment transport The rate of solids exiting the raceway was calculated at the last grid along the X-axis, where the openings for the weirs are located. The rate for each particle group was calculated as:

D.L. Huggins et al. / Aquacultural Engineering 31 (2004) 277293 Table 2 Sediment particle characteristics Particle group (no.) 1 2 3 4 5 6 Settling velocity (m/s) 0.0391 0.0231 0.0100 0.0034 0.0003 0.0001 Particle density (kg/m3 ) 1150 1150 1150 1150 1150 1150 Particle size (m) 692 532 350 204 61 35

285

Inlet concentration (m3 /m3 ) 3.2651 107 5.7799 107 1.3298 107 2.1277 106 2.0745 105 3.1916 106

m e,g =
ExitCells

p,g Vx Cg A

(1)

where m e,g is the mass ow rate of particles from group g exiting raceway (kg/s), p,g the density of the particle group g (kg/m3 ), Vx the water velocity in the X-direction (m/s) at cells where the weirs are located, Cg the sediment concentration of particle group g (m3 /m3 ) obtained from the simulation results, and A the face area of the cells where the weirs are located and through which the sediment ux occurs. The rate of solids settled was calculated for the cells nearest to the bottom of the raceway in a similar manner as the rate of solids exiting the raceway. The rate of solids accumulation for each particle group was calculated as (Olsen, 1999): m a,g =
BottomCells

p,g Vs,g Cg A

(2)

where m a,g is the mass ow rate of particles from group g settled in the bottom of the raceway (kg/s), Vs,g = settling velocity of particle group g (m/s). The sediment calculations were all done using ExcelTM . Since the analysis assumes steady state conditions, the total sediment uxes entering the raceway were calculated as the sum of the particles exiting and settling in the raceway. A check on the calculation for the inuent particle ux was carried out by multiplying the prescribed sediment concentration at the surface times the particle density and settling velocity (Huggins, 2003). The percent of solids removed (PSRg ) was calculated from PSRg = 100 m a,g m a,g + m e,g (3)

4. Results and Discussion Model validation was achieved through a comparison of the simulated and measured velocities. In all cases, comparisons were made for points about 0.5 m from the left wall, 0.5 m from the right wall, and the middle of the raceway. Although velocities were measured at 16 planes along the length of the raceway (Fig. 3), validation was not attempted for the beginning of the raceway and there was a large difference between the simulated and measured velocities in this part of the raceway (Huggins, 2003). The differences were due to

286

D.L. Huggins et al. / Aquacultural Engineering 31 (2004) 277293

the inherent turbulence around the inlet, the presence of sh in the raceway, and the need to simplify the water inlet for the simulations. Other authors have encountered similar results for areas with highly turbulent ows (Bowles, 1999; Shankar et al., 2001). Model validation was carried out for the last 10 m of the raceway, which included the quiescent zone plus the area where the sediments were released in the simulations. Here, the measured Vx and Vy velocities agreed with the simulated velocities (Figs. 46). The measured and simulated velocities in the Y-direction are not shown as they were very small: in the range of 0.1 cm/s for the rst 3.4 m of the raceway, 0.02 cm/s for the middle of the raceway (15.6 m from the inlet), and 0.07 cm/s (25.0 m from the inlet). The data collected with the velocimeter consisted of the average and standard deviation of 3040 measurements made at approximately 1 s intervals. The sensitivity and response speed of the instrument was such that the velocity uctuations caused by turbulence were detected in the measurements giving rise to the standard deviations that are displayed as error bars in Figs. 46. Although, the velocity uctuations in the last section of the raceway diminished substantially when compared to the inlet and middle sections, uctuations were still relatively high around the screen (Figs. 4 and 7). Fluctuations before the screen may be due in part to the activity of the sh around the probe, and in part to the impact of the screen on the ow. Fluctuations after the screen (Fig. 7) are the result the turbulence generated by the reduction of the cross-sectional area due to the screen (Fig. 2). 4.1. Sediment transport and sediment removal analysis The total ow rate of sediments into the raceway was determined by specifying the inlet surface sediment concentrations and settling velocities and was 13,631 g/day (Table 3). The total rate of sediments exiting the raceway was 2485 g/day and the total settling rate was 11,146 g/day. The total rate of solids in corresponded to approximately 26% of the amount of feed applied (42.3 kg/day) (Timmons et al., 2002). The percent removal for particle Groups 13 was approximately 100%. The percent removal for the smaller particle sizes was lower at 54.7, 0.9, and 0.1% for particle size Groups 46, respectively (Table 3). Contour graphs of the sediment uxes were created in MATLAB (Fig. 8) to obtain a more detailed visualization than that available with the graphics option presented in SSIIM. The plots were done for an area near the bottom of the raceway and in the last 7 m of the
Table 3 Sediment calculations for the original raceway system Particle group (no.) 1 2 3 4 5 6 Total Solids entering (g/day) 3247 3396 3484 1840 1583 81 13631 Solids exiting (g/day) 0 0 1 833 1569 81 2485 Solids settled (g/day) 3247 3396 3483 1007 14 0 11146 Solids removed (%) 100.0 100.0 100.0 54.7 0.9 0.1 81.8

D.L. Huggins et al. / Aquacultural Engineering 31 (2004) 277293


100 -12 80 60 40 20 0 -10 -8 -6 -4 -2 0 2 4 6 8 10 13

287

Distance from the bottom of the raceway (cm)

LEFT 20

Velocity (cm/s)
Distance from the bottom of the raceway (cm)
100

MIDDLE
80 60 40 20 0 -8 -6 -4 -2 0 2 4 6 8 10

-10

Velocity (cm/s)

Distance from the bottom of the raceway (cm)

100 80 60

RIGHT 11

-17

22 40 20 0

-10

-8

-6

-4

-2

10

Velocity (cm/s) Simulated M easured

Fig. 4. Measured and simulated Vx velocities. The LEFT corresponds to a location about 0.5 m from the left side of the raceway when facing the inuent. MIDDLE is the center of the raceway, and RIGHT is 0.5 m from the right side of the raceway. The comparison is made for points on Plane 11, 23.9 m from the inlet (Fig. 3). Water depth at that point was 0.85 m. Mean velocity was 2.92 cm/s (calculated from ow rate divided by the cross-sectional area at Plane 11).

raceway. The contour graphs show the rates of sedimentation (g/m2 /day) for the different particle groups. The settling characteristics are similar for Groups 13, as the particles settle before reaching a distance of 28.5 m from the inlet (Fig. 8). Furthermore, a large fraction of the Group 13 sediments settled before the quiescent zone. This could be explained by the

288

D.L. Huggins et al. / Aquacultural Engineering 31 (2004) 277293


100 LEFT 80 60 40 20 0 -10 -8 -6 -4 -2 0 2 4 6 8 10

Distance from thebottomof the raceway (cm)

Velocity (cm/s)
100
Distance from thebottomof the raceway (cm)

MIDDLE 80 60 40 20 0 -10 -86 -4


-2

10

Velocity (cm/s)
Distance from thebottomof the raceway (cm)
100 RIGHT 80 60 40 20 0 -10 -8 -6 -4 -2 0 2 4 6 8 10

Velocity (cm/s) Simulated Measured

Fig. 5. Measured and simulated Vx velocities. The LEFT, MIDDLE, and RIGHT are as described for Fig. 4. The comparison is made for points on Plane 15, 29.7 m from the inlet (Fig. 3). Water depth at that point was 0.91 m. Mean velocity was 2.11 cm/s.

manner in which the particles are introduced into the simulated raceway, and also by the fact that no re-suspension is considered. Also, as was mentioned earlier, the turbulence created around the screen caused some back-ows that would allow some of the larger particles to settle in the rearing area. Although the general settling pattern for Groups 13 is similar, the displacement of particles towards the end of the raceway increases as settling velocity decreases.

D.L. Huggins et al. / Aquacultural Engineering 31 (2004) 277293


100

289

Distance from thebottomof the raceway (cm)

LEFT 80 60 40 20 0 -10 -8 -6 -4 -2 0 2 4 6 8 10

Velocity (cm/s)
100

Distance from thebottomof the raceway cm)

MIDDLE 80 60 40 20 0 -10 -8 -6 -4 -2
100 RIGHT 80 60 40 20 0 -10 -8 -6 -4 -2 0 2 4 6 8 10

10

Velocity (cm/s)

Distance from thebottomof the raceway (cm)

Velocity (cm/s)

Simulated

Measured

Fig. 6. Measured and simulated Vx velocities. The LEFT, MIDDLE, and RIGHT are as described for Fig. 4. The comparison is made for points on Plane 16, 29.9 m from the inlet (Fig. 3). Water depth at that point was 0.91 m. Mean velocity was 2.10 cm/s.

Differences in the sediment deposition pattern were observed for Group 4 relative to Groups 13 (Fig. 8). The lighter Group 4 particles are displaced towards the end of the quiescent zone. The sediment uxes (g/m2 /day) for Group 4 are much lower than for the previous groups, resulting in much lower percent solids removal values. The sediment contours and ux rates for Group 5 are very similar to those for Group 6 as these small

290

D.L. Huggins et al. / Aquacultural Engineering 31 (2004) 277293

Fig. 7. Top view of velocity ow patterns at the outlet area for the raceway. Each drawing has a different scale. The arrows represent velocity vectors proportional to the velocity for each view.

particles have a low settling velocity. In general, it is much harder to capture the Group 5 and 6 particles than the larger particles. Settling for these particles takes place close to the right and left walls of the raceway but at a very low ux rates (Fig. 8).

4.2. Model assumptions When using CFD modeling it is necessary to make some assumptions to simulate the desired system. The assumptions used here, such as those of a simplied inlet structure and uniform ow distribution, and a simplied screen, were essential to obtain convergence of the solutions. The exclusion of re-suspension resulted in an undetermined over-estimation of percent of solids removed. However, the simulations still provide useful information on the settling rates and patterns within a raceway, tools that can be used to analyze the potential impact of raceway modications on solids removal.

D.L. Huggins et al. / Aquacultural Engineering 31 (2004) 277293

291

Fig. 8. Comparative sediment uxes for the different particle groups. The graphs show the last 7.5 m of the raceway, which includes the quiescent zone that is 5.02 m from the outlet wall. The graphs are not to scale. The color represents the sediment concentration as indicated on each of the graphs. The contour graphs were created in MATLAB using the sediment concentration output from Conres le at Level 2.

292

D.L. Huggins et al. / Aquacultural Engineering 31 (2004) 277293

5. Conclusion This study investigated the use of CFD modeling to simulate water ow and sediment transport in aquaculture raceways. Water velocity measurements were used to successfully validate the simulations. The model was used to identify the location in which particles of different densities are likely to settle and the settling rates of those particles. These preliminary results are being used for the evaluation of possible raceway design modications and their potential impact on solids settling in the raceway (Huggins, 2003).

Acknowledgements The authors would like to acknowledge the support of the US Department of Agricultures Western Regional Aquaculture Center (WRAC), participating farms in Idaho, and Dr. Nils Olsen, the author of the CFD software used in this study.

References
Bates, P.D., 2000. Development and testing of a subgrid-scale model for moving boundary hydrodynamic problems in shallow water. Hydrol. Process. 14, 20732088. Bowles, C.B., 1999. An investigation into the ow structure of a generalized open channel intake. Ph.D. Thesis. Nottingham Trent University, 381 pp. Huggins, D.L., 2003. Analysis of sediment transport modeling using computational uid dynamics (CFD) for aquaculture raceways. MS Thesis. University of California, Davis, 265 pp. IWMGAO, 1999. Idaho Waste Management Guidelines for Aquaculture Operations. http://www.deq.state.id.us/ water/gw/Aquaculture Guidelines.Pdf. Metcalf and Eddy, Inc., 1991. Wastewater Engineering. Treatment Disposal and Reuse, 3rd ed. (revised by G. Tchobanoglous, F.L. Burton). Irwin/McGraw-Hill, 1334 pp. Montas, J.H., Prabhakar, G.V.S., Wheaton, F., 2000. CFD analysis of ow in aquaculture tanks. In: Proceedings of the 93rd Annual International Meeting of ASAE. ASAE Paper No. 003111, 24 pp. NASS, 2001. Catsh and Trout Production. National Agricultural Statistics Service. United States Department of Agriculture. http://www.usda.mannlib.cornell.edu/reports/nassr/other/ztp-bb/cfpd0201.pdf. Olsen, N.R.B., 1991. A three-dimensional numerical model for simulation of sediment movements in water intakes. Dr. Dissertation. University of Trondheim, The Norwegian Institute of Technology, 106 pp. Olsen, N.R.B., 1999. Computational uid dynamics in hydraulic and sedimentation engineering. Class notes, 2nd Revision, June 16, 1999. Department of Hydraulics and Environmental Engineering, The Norwegian University of Science and Technology, 65 pp. http://www.bygg.ntnu.no/nilsol/cfd/class2.pdf. Olsen, N.R.B., 2002. A Three-dimensional Numerical Model for Simulation of Sediment Movements in Water Intakes with Multiblock Option. Version 1.1 and 2.0 for OS/2 and Windows. Users Manual. Department of Hydraulics and Environmental Engineering, The Norwegian University of Science and Technology. http://www.bygg.ntnu.no/nilsol/ssiimwin/manual3.pdf. Peterson, E.L., Harris, J.A., Wadhwa, L.C., 2000. CFD modeling pond dynamic processes. Aquacult. Eng. 23, 6193. Peterson, E.L., Wadhwa, L.C., Harris, J.A., 2001. Arrangement of aerators in an intensive shrimp growout pond having a rectangular shape. Aquacult. Eng. 25, 5166. Rasmussen, M.R., 2002. Numerical modeling of hydrodynamics in aquaculture systems: a civil engineer perspective. In: Proceedings of the Fourth International Conference on Recirculating Aquaculture, Roanoke, VA, USA, July 1821, 2002. Virginia Polytechnic Institute and State University, pp. 152162.

D.L. Huggins et al. / Aquacultural Engineering 31 (2004) 277293

293

Shankar, N.J., Chan, E.S., Zhang, Q.Y., 2001. Three-dimensional numerical simulation for an open channel ow with a constriction. J. Hydraulic Res. 39 (2), 187201. Timmons, M.B., Ebeling, J.M., Wheaton, F.W., Summerfelt, S.T., Vinci, B.J., 2002. Recirculating Aquaculture Systems, 2nd ed. Northeastern Regional Aquaculture Center, NRAC Publication No. 01-002. Cayuga Aqua Ventures, Ithaca, NY, 769 pp. Veerapen, J.P., Brooks, M.J., Lowry, B.J., Couturier, M.F., 2002. Solids removal modeling in recirculating aquaculture systems. In: Proceedings of the Fourth International Conference on Recirculating Aquaculture, Roanoke, VA, USA, July 1821, 2002. Virginia Polytechnic Institute and State University, pp. 152162, 491498. Vennard, J.K., 1954. Elementary Fluid Mechanics. Wiley, New York, 893 pp. Versteeg, H.K., Malalasekera, W., 1995. An Introduction to Computational Fluid Dynamics. The Finite Volume Method. Pearson Education Limited, UK, 257 pp. Wong, K.B., Piedrahita, R.H., 2000. Settling velocity characterization of aquacultural solids. Aquacult. Eng. 21, 233246. Wong, K.B., 2001. Enhanced solids removal for aquacultural raceways. Ph.D. Dissertation. University of California, Davis, 370 pp.

Você também pode gostar