See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/286053996

Hydrodynamics in the 'Cornell-type' dual-drain tank

Article · January 2000

CITATIONS READS
19 2,619

3 authors:

Steven T. Summerfelt John Davidson

Superior Fresh LLC The Conservation Fund
177 PUBLICATIONS   4,815 CITATIONS    52 PUBLICATIONS   1,570 CITATIONS   

SEE PROFILE SEE PROFILE

Michael B. Timmons
Cornell University
108 PUBLICATIONS   2,755 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Indepdendent text View project

Fish Welfare and Performance in Recirculating Aquaculture Systems (RASALMO) View project

All content following this page was uploaded by Steven T. Summerfelt on 08 December 2015.

The user has requested enhancement of the downloaded file.

Pages 160-166 In: Proceedings of the Third International Conference of Recirculating Aquaculture, July
20-22, 2000, Roanoke, Virginia.

Hydrodynamics in the ‘Cornell-Type’ Dual-Drain Tank

Steven T. Summerfelt, Research Engineer, and

John Davidson, Research Associate
The Conservation Fund’s Freshwater Institute
Shepherdstown, West Virginia

Michael B. Timmons, Professor

Agricultural and Biological Engineering Department
Cornell University, Ithaca, NY 14853

Introduction

Circular fish culture tanks can be managed as “swirl settlers,” settling basins with two
effluents, because of their capability to concentrate solids at their bottom and center.
Solids that concentrate at the bottom center can be removed in a small flow stream by
using a bottom-drawing center drain, while the majority of flow is withdrawn at an
elevated drain. The two drains have been typically located at the tank’s center, which
then takes advantage of both the ‘tea-cup effect’ and the strength of the overall flow
when it drains through the tank center, as reviewed by Timmons et al. (1998). However,
the ‘Cornell-type’ dual-drain culture tank differs significantly from the other dual-drain
designs, because it only places the bottom drain at the tank’s center and removes the
majority of flow through an elevated drain located on the tank’s side-wall. This paper
reports an investigation of water mixing and settleable solids fractionation within a
'Cornell-type' double-drain circular culture tank.

Materials and Methods

Settleable solids fractionation and tank mixing were studied within a 3.66 m (12 ft) i.d. x
1.22 m (4 ft) tall circular ‘Cornell-type’ dual-drain tank at the Freshwater Institute.
Replicated tests were conducted under two hydraulic exchange rates (1 and 2 rearing tank

1
volume exchanges per hour), at three diameter to depth ratios (12:1, 6:1, 3:1), three
bottom drain flow percentages (5, 10, and 20% of total flow), and in the presence and
absence of rainbow trout. Some of the trials could not be conducted, however, because
the flow rates through the tank’s bottom drain were too low to flush pellets through the
3.8 cm (1.5 in) diameter drainpipe. Trout were maintained at densities exceeding 60
kg/m3 and fish were fed 1% body weight per day. The experimental protocol and the
methods described were in compliance with the Animal Welfare Act (9CFR)
requirements and were approved by the Freshwater Institute’s Institutional Animal Care
and Use Committee.

Bottom drain flows were set and calibrated using a calibrated bucket and a stop watch.
Total flow was measured using a pipe-mounted flow meter; these flows were checked
using a stop watch and a larger basin of known volume. Total suspended solids (TSS)
concentrations were measured during all trials where fish were present. TSS samples
were collected on the tank’s inlet flow, both tank outlet flows, and the tank’s bottom-
center drain outlet flow after it has passed through a microscreen filter.

Tank Mixing Tests

Rhodamine WT dye was added to the culture tank in a single pulse to determine the
effectiveness of tank mixing under the different conditions tested. Dye-pulse flushing
was monitored by collecting water samples over a period of time (Figure 1).

120

100
3

80
Absorbance x 10

25.0 minute = ‘mixing-cup’ calculated mean HRT

60

40

20

0
0 20 40 60 80 100 120 140

Time after dye-pulse input, min

Figure 1. An example plot showing the absorbance data for each sample taken after
the dye-pulse was added to the culture tank.

The mean hydraulic retention time (HRT) of the water flowing through the culture tank
was estimated using the “mixing-cup” method described on page 63.3 of Levenspiel
(1989), as follows:

2
 n
∑ t i ⋅ Ci ⋅ Δt i
measured mean residence time (min) = i =1 (Eq. 1)
n
∑ Ci ⋅ Δt i
i =1
where: Ci is the absorbance of sample i, ti is the length of time the sample i was collected
after the dye-tracer was input into the tank, and Δti is the interval between adjacent
samples when sample i was collected. The ‘mixing-cup’ method accounts for unequal
sample intervals within the data.

The ideal mean residence time was also calculated by dividing the water volume in the
tank (V) by the water flowrate through the tank (Q). Tank turnover efficiencies were
estimated from the ratio of the two residence times found above:
measured mean residence time (min)
turnover efficiency = (Eq. 2)
ideal mean residence time (min)
If mixing within the tank is perfect, the tank turnover efficiencies equal 1.0. Turnover
efficiencies of less than 1.0 indicate less than perfect mixing and that some water short
circuits through the culture tank. Turnover efficiencies greater than 1.0 are an indication
of problems with the precision of water flow rate or tank volume measurements, or
inconsistencies within the samples taken during the dye-tracer test.

Pellet Flushing Tests

Pulses of sinking PVC cylindrical pellets (each about 3 mm in length by 3 mm o.d.) were
added to quantify solids flushing kinetics and the relative strength of the radial flow. The
plastic pellets had a specific gravity of 1.05, similar to the specific gravity reported for
trout fecal matter. The plastic pellets also exhibited a settling velocity of 3.8 cm/s, which
was similar to the settling velocity reported for fecal matter 1.7-4.3 cm/s (Warrer-Hansen,
1982). Feed pellets tested in this study settled more rapidly (e.g., 14 cm/s) than the
plastic pellets, but the feed pellet settling velocities measured were similar to those
reported by others.

During a pellet flushing test, 1000-grams of dry pellets were weighed out, wetted in a
bucket to remove air bubbles from the pellets, and placed into the tank water in a pulse.
The plastic pellets settled fairly rapidly and were transported by radial flow towards the
bottom-center drain. The plastic pellets were then collected from the discharge from the
bottom-center drain at specific time intervals. Baskets containing 1-mm mesh stainless
steel screen were used to capture the pellets from the flow during each sampling interval.
Another screened basket was placed to capture all pellets flushed through the side-wall
drain. Following collection, the pellets were oven dried and then weighed. Minimums of
three pulsed-pellet tests were run for each set of conditions.

An unsteady-state mass balance was developed to quantify pellet flushing, assuming zero
pellet inflow, reaction, or accumulation, i.e., Loss = Inflow – Outflow – Reaction –
Accumulation. The outflow term in the mass balance was broken into a component
representing simple mass action of flow to the bottom drain, and a component

3
representing pellet enrichment at the bottom drain due to a combination of sedimentation
and radial flow (assuming that the pellet outflow at the tank’s sidewall is negligible); this
was done to distinguish between the two different mechanisms transporting pellets to the
bottom center drain, i.e., Loss =– Outflow (mass action) – Outflow (enrichment). Or
more specifically,
dC
V⋅ = −Qout , b ⋅ Cout , b − k ⋅ Cout , b ⋅ V
dt (Eq. 3)
= V ⋅ {− (Q / V + k ) ⋅ Cout , b }
Where: k is a 1st order rate constant characterizing bead enrichment at the bottom-center
drain and Qout,b and Cout,b are the flowrate and concentration of pellets flushed through the
bottom center drain at a given time. Integration provides an equation that can be used to
model pellet flushing through the bottom drain in real time:
Cout ,b (@ t ) = Cout,b (@ t = 0) ⋅ e{−(Q / V + k )⋅t} (Eq. 4)
Note that the flushing of a homogeneously distributed dye pulse is only due to mass
action and would result in a similar equation, but without k (1st order enrichment
constant). Therefore, the k-value increases the rate of solids flushing relative to the
culture tank exchange rate. After manipulation of Eq. 4, the k-value for each pellet
flushing test was calculated from the slope of the line produced by plotting the pellet
flushing data (Figure 2) after manipulation to the following form:
y-axis = -LN(fraction of solids remaining) (Eq. 5)
x-axis = t (Eq. 6)
slope of the linear regression line = (Q/V + k) (Eq. 7)

4/28/99 2' Depth

T. Flow = 55.2 gpm B. Flow = 11.7 gpm
k=0.690/min 3.57% not included t0 = 1.43 min
5
Data Excluded from
4 Linear Regression
-LN (fraction solids

3
remaining)

Data Used for Linear

2 Regression
1 Linear (Data Used for
Linear Regression)
0
0 20 40 60 80 100 120 y = 0.7224x - 1.035
Time (min) R2 = 0.9554

Figure 2. An example plot showing pellet-flushing data plotted as the natural logarithm
of the ‘fraction of pellets remaining’ for each outlet sample taken after a pulse of pellets
was added to the culture tank. The data not included in the linear regression only
represented 3.57% of the total mass of pellets flushed from the tank.

4
A sample pellet-flushing test is shown in Figure 2. Note that not all the data fits the
assumption that pellet enrichment at the bottom center drain due to settling and radial
flow could be approximated by 1st order kinetics. In most tests, effective pellet flushing
occurred and only a small portion (e.g. less than 15%) of the pellets did not flush
according to the 1st order kinetic equation.

Results and Discussion

Solids Partitioning Between Side-wall and Bottom-Center Drains

Less than 5% of the sinking pellets were flushed through the side-wall drain during all
trials. However, larger fractions of pellets were flushed through the side-wall drain when
fish were present (e.g., < 4.3%) than during trials conducted in the absence of fish (e.g., <
1.4%). Increasing the culture tank exchange rate from 1 ex/hr to 2 ex/hr also increased
the fraction of pellets flushed through the sidewall drain.

The k-values estimated from the pellet-tracer tests are an indication of the rate that pellets
can be flushed from the bottom-center drain and also of the relative strength of the radial
flow. When fish were present, the rate that pellets were flushed through the bottom-
center drain was greater at 2 ex/hr than at 1 ex/hr and at the smaller diameter:depth ratios.
All the trials at 2 ex/hr and diameter:depth ratios of 3.1:1 and 6:1 produced strong radial
flows, rapid solids flushing, and exhibited no problems with solids flushing.

A very important observation was that the settleable solids frequently deposited about the
tank’s bottom-center drain during most trials at 1 ex/hr, but only at the diameter:depth
ratio of 12:1 during trials at 2 ex/hr. During these trials, the radial flow transported the
settleable solids to the center portion of the tank, but water velocities were so low in the
middle of the tank that a good portion of these solids settled within a torus-shaped region
about the center drain. Fortunately, this accumulation of settled solids were usually
sufficiently near to the center drain that pulling the external stand-pipe regulating the
bottom-center drain flow, even for an interval of < 1 min, was sufficient to flush the
accumulated solids. The presence of fish improved the rate that pellets were flushed
through the bottom-center drain (i.e., the K-values in Table 1) for a given diameter:depth,
underflow percentage, and overall tank exchange rate.

The relative importance of the two pellet flushing mechanisms, i.e., mass action (i.e.,
Q/V) versus enrichment at bottom-center drain (k), was illustrated by calculating the
percentage of pellet flushing due to enrichment alone:
k
% flushing due to enrichment = ⋅ 100 (Eq. 8)
k + Q/V
The radial flow mechanism played a much larger role than the mass transport mechanism
in the transport of pellets to the bottom-center drain during the trials at 2 ex/hr and
diameter:depth ratios of 3.1:1 and 6:1 and during the 1 ex/hr trials at diameter:depth
ratios of 3.1:1 (Table 1). Pellets were not flushed effectively during the remainder of the
trials, so the k-values and the enrichment percentage shown in Table 1 could not be
calculated under these conditions.

5
Table 1. The relative amount of flushing produced by enrichment (i.e., k/[k+Q/V]*100)
at the bottom-center drain was calculated for each trial that a k-value (listed before
enrichment-value) could be estimated.
k-value ± standard error (min) / enrichment ± standard error (%)
Dia:depth = 3.1:1 Dia:depth = 6:1 Dia:depth = 12:1
2 ex/hr (fish present)
5% bottom flow 0.39±0.02 / 93±0 0.13±0.05 / 71±14 NA
10% bottom flow 0.48±0.01 / 94±0 0.34±0.05 / 91±2 0.02±0.02 / 58±14
20% bottom flow 0.87±0.24 / 96±1 0.58±0.06 / 95±1 0.07±0.03 / 0
1 ex/hr (fish present)
5% bottom flow 0.06±0.04 / 84±2 NA NA
10% bottom flow 0.06±0.03 / 62±24 0.08±0.04 / 65±17 NA
20% bottom flow 0.10±0.04 / 80±7 0.20±0.02 / 91±1 NA
2 ex/hr (no fish)
5% bottom flow 0.08±0.04 / 61±17 NA NA
10% bottom flow 0.53±0.14 / 93±2 NA NA
20% bottom flow 1.37±0.02 / 98±0 0.00±0.01 / 12±12 NA
1 ex/hr (no fish)
5% bottom flow -0.01±0.00 / 83±0 NA NA
10% bottom flow 0.01±0.04 / 29±9 NA NA
20% bottom flow 0.08±0.00 / 0 NA NA
Note, an enrichment-value of zero indicates that the term (k/[k+Q/V]*100) was negative, e.g., the beads
took longer to flush than the hydraulic exchange rate through the tank. NA indicates that either bead
flushing was so poor that k-values could not be calculated, or that bottom-drain pipe velocity was too low
to flush beads and the tests were not run.

Additionally, not all of the data was included in the linear regression of the bead-pulse
data as plotted according to the –LN(fraction of solids remaining) versus time, as in the
example shown in Figure 2. This data was excluded because it did not fit the 1st order
kinetics assumption for solids enrichment at the bottom-center drain. We think that when
large portions of the pellets did not exhibit enrichment at the center drain according to 1st
order kinetics, it was due to insufficient velocity in the rotational flow, i.e. the actual
water velocity starting from the outside wall. Under these low velocity conditions, the
pellets did not flush rapidly because they settled on the tank floor just before reaching the
tank’s center drain. Velocities at the outside walls appear to require velocities of at least
15 cm/s (0.5 ft/s) to promote solids departure from the center drain.

The pellet-pulse data when plotted according to the –LN(fraction of solids remaining)
versus time (Figure 2) also provides a graphical estimate of the time required for the
beads to travel from the tank’s surface to the bottom center drain, i.e., the dead time (t0)
for solids flushing from the tank. The x-intercept of this data provides the estimate of the
tank’s dead time. When fish were present, average dead times ranged from 1.0 to 2.6 min
for all conditions tested.

6
 Tank Mixing
When water rotates about a circular tank, a torus-shaped region about the center drain can
become an irrotational zone with lower velocities and poor mixing (Timmons et al.,
1998). However, according to the dye-tracer tests, the ‘Cornell-type’ dual-drain tank
exhibited good mixing characteristics during all trials with fish present, as illustrated by
tank turnover efficiencies near 1.0. However, mixing was often better during the 2 ex/hr
trials than at the 1 ex/hr trials.

Settleable Solids Fractionation and Waste Management

Due to variability in the TSS data, the TSS data collected during all trials was combined
according to its sampling location. This data indicates that the flow discharged from the
‘Cornell-type’ tank’s bottom-center drain contained TSS concentrations (i.e., 19.6 ± 3.6
mg/L) that were more than 10-times greater than the TSS concentration discharged from
the side wall drain (i.e., 1.5 ± 0.2 mg/L). For these trials, the concentration of TSS
discharged from the side-wall drain was sufficiently low that it would not require further
solids treatment in order to pass even many of the more stringent discharge limits
imposed by environmental regulatory agencies. On the other hand, the concentrated and
relatively small flow discharged from the bottom-center drain would definitely require
treatment before discharge. However, this flow was treated effectively using a
microscreen filter where, on average, capture of the TSS discharged from the circular
culture tank’s bottom-center drain was greater than 80%.

Acknowledgments
The work reported in this publication was supported in part by the Northeastern Regional
Aquaculture Center at the University of Massachusetts Dartmouth through grant number
96-38500-3032 from the Cooperative State Research, Education, and Extension Service
(CREES) of the United States Department of Agriculture (USDA). Any opinions,
findings, conclusions, or recommendations expressed in this publication are those of the
authors and do not necessarily reflect the view of the USDA or any of their sub-agencies.

Literature Cited

Levenspiel, O. 1989. Chemical Reactor Omni Book. Oregon State Univ., Corvallis, OR.

Timmons, M.B., S.T. Summerfelt, and B.J. Vinci. 1998. Review of circular tank
technology and management. Aquacultural Engineering 18(1): 51-69.

Warren-Hansen, I. 1982. Methods of treatment of waste water from trout farming. Pages
113-121 In J. Alabaster (ed.), EIFAC Technical Paper No. 41. Report of the EIFAC
Workshop on Fish-Farm Effluents, Silkeborg, Denmark, 26-28 May, 1981, FAO Rome.

View publication stats