Aquaculture 250 (2005) 162 – 174

www.elsevier.com/locate/aqua-online

Do fish enhance tank mixing?

Michael R. Rasmussena,T, Jesper Laursena, Steven R. Craigb, Ewen McLeanb
a
Aalborg University, Department of Civil Engineering, Hydraulic and Coastal Engineering,
Sohngaardsholmsvej 57, DK-9000 Aalborg, Denmark
b
Aquaculture Center, Virginia Polytechnic Institute and State University, Department of Fisheries and Wildlife Sciences,
Blacksburg, VA 24061, USA
Received 1 December 2004; received in revised form 31 January 2005; accepted 1 February 2005

Abstract

The design of fish rearing tanks represents a critical stage in the development of optimal aquaculture systems, especially in
the context of recirculating systems. Poor hydrodynamics can compromise water quality, waste management and the physiology
and behaviour of fish, and thence, production potential and operational profitability. The hydrodynamic performance of tanks,
therefore, represents an important parameter during the tank design process. Because there are significant complexities in
combining the rigid principles of hydrodynamics with the stochastic behaviour of fish, however, most data upon tank
hydrokinetics has been derived using tanks void of fish. Clearly, the presence of randomly moving objects, such as fish, in a
water column will influence not only tank volumes by displacing water, but due to their activity, water dynamics and associated
in-tank processes.
In order to determine the impact of fish presence upon tank hydrodynamics, Rhodamine fluorometry was employed to
examine mixing within a recirculating aquaculture system. Two different methods were compared, traditional, outlet-based
measurements and a technique that employed in-tank data acquisition. Circular tanks were employed during data collection
either in the presence or absence of experimental fish-red drum Sciaenops ocellatus (n =36; 5 kg total wet wt); and at two flow
rates. Irrespective of flow rate, the presence of fish dramatically enhanced the mixing process ( P b 0.001), with mixing times in
tanks with fish being one-third that for tanks without animals. In-tank dispersion coefficients and dispersion numbers also
differed ( P b 0.001) in the presence of fish, irrespective of flow. Presence or absence of fish had no effect upon hydraulic
residence or circulation times. Unlike measurements at the outlet, in-tank observations were more able to isolate the effects of
stochastic, fish-induced mixing, from deterministic, hydrodynamic mixing.
D 2005 Published by Elsevier B.V.

Keywords: Hydrodynamics; Tracer; Hydraulic residence time; Rhodamine; Dispersion number; Red drum

1. Introduction

T Corresponding author. Fax: +45 98142555. Recirculating aquaculture systems (RAS) present
E-mail address: [email protected] (M.R. Rasmussen). the aquaculturist with several advantages. Principal in
0044-8486/$ - see front matter D 2005 Published by Elsevier B.V.
doi:10.1016/j.aquaculture.2005.02.041
 M.R. Rasmussen et al. / Aquaculture 250 (2005) 162–174 163

this regard is the ability to tailor production conditions tively affects feed conversion, influences tank micro-
(water chemistry, photoperiod, etc.) to optimize the biology and elevates stress and consequently the
performance characteristics of the cultured species likelihood of losses due to disease. Poorly regulated
(Skjolstrup et al., 2000). Enhanced control over tank mixing can result in changes in physiology, more
production enables greater harvest flexibility that in pronounced aggression (Griffiths and Armstrong,
turn permits more precise market management which 2000; Odeh et al., 2003) and the formation of social
may increase returns on investment. Additionally, hierarchies, a reduction in sedimentation and solid
because of their reduced consumption of water and waste removal and increased occurrence of dead
heightened management of wastes, RAS provide volumes (Cripps and Bergheim, 2000; Rasmussen et
environmentally acceptable production systems al., 2004).
(Wheaton, 2002; Rasmussen et al., 2004). Moreover, Several studies have examined general mixing
water reuse techniques can reduce energy and labor processes in aquaculture ponds and tanks (Burrows
costs, provide flexibility in the placement of hatch- and Chenoweth, 1955, 1970; Larmoyeux et al., 1973;
eries and offer a high level of biosecurity from Burley and Klapsis, 1988; Gaikowski et al., 2004;
pollution, predator, disease and human viewpoints Rasmussen and McLean, 2004). However, because it
(Skjblstrup et al., 1998). The major drawback of is extremely complex to integrate the more rigid
recirculating aquaculture relates to their overall principles of hydrodynamics with the stochastic
complexity and the large capital costs required for behaviour of fish, most research in this field has
facility start-up. Generally, the latter is offset by employed fishless tanks. In studies that have exam-
increasing stocking densities. Such strategies, how- ined the effect of fish upon tank mixing processes,
ever, are intrinsically dangerous since crowding may results have generally been contradictory or incon-
negatively impact product quality and feed efficiency, clusive (Burley and Klapsis, 1985; Watten and Beck,
while occurrence of and losses due to disease intensify 1987; Watten et al., 2000), most likely due to the
(Wagner et al., 1997; Shoemaker et al., 2000). experimental and/or analytical procedures employed.
Holding tanks are the central feature of RAS. Traditionally, mixing studies estimate dispersion
Tanks may be circular, oval, raceway, D-ended, numbers using tracer techniques. The principal
octagonal, hexagonal, square, conical, or hybrids method employed involves inspection of dye dilution
thereof, in form. A wide variety of materials are used using single-point measurements taken at the tank
in tank construction including fiberglass, plastic, outlet. However, this technique does not provide
concrete, wood, steel and their amalgams. Increas- critical detail upon in-tank hydrodynamic processes
ingly, RAS employ barrier-based earthen ponds, and difficulties are encountered in assessing the
especially in shrimp production. Selection of con- impact of fish upon the mixing process with any
struction material for tanks is generally constrained by degree of certainty. Due to the importance of this field
size, desired shape and site characteristics. For of investigation (Burley and Klapsis, 1985; Watten
example, costs severely limit the use of steel whereas and Beck, 1987; Watten et al., 2000), there remains a
large tank size restricts application of plastics and clear need to intensify research upon the effect of fish
fiberglass. The use of concrete and earthen ponds may presence on tank hydraulics. An increased awareness
be impeded to a certain degree by site topography, of these effects would assist in refining the tank
hydraulic permeability, soil plasticity and prevailing design process.
height of water tables. Irrespective of size or One method of enhancing the current understand-
construction material employed, however, to achieve ing of tank hydrodynamic processes might be to
maximum production potential, a prerequisite to RAS, undertake in-tank measurements of dye dilution. In
tanks must be optimally designed. A complete contrast to outlet-based methods, in-tank techniques
appreciation of tank hydrodynamics represents a vital permit multiple determinations to be made per experi-
part of the engineering process since this characteristic ment. Moreover, because in-tank measurements may
may impair water quality (Burrows and Chenoweth, provide a more accurate evaluation of changes in
1955; Burley and Klapsis, 1988). Poor water quality hydrodynamics over time, the effects of fish presence
reduces stocking potential, decreases growth, nega- upon mixing processes might be more readily
164 M.R. Rasmussen et al. / Aquaculture 250 (2005) 162–174

assessed. The objective of the present study was to USA) for disinfections. The fluidized bed was oxy-
compare outlet-based and in-tank methods for exam- genated using diffusion air lines connected to a 1-hp
ining the mixing process, both in the absence and Sweetwater remote drive regenerative blower
presence of fish, and at high and low flow rates. (Aquatic Ecosystems, Apopka, FL, USA). Water
Circular tanks were used in preference to any other temperature and DO2 were monitored daily using an
form because these units are the most common Y85 Series dissolved oxygen meter (YSI Inc., Yellow
encountered in commercial settings while also permit- Springs, OH). Total ammonia nitrogen (TAN) was
ting acquisition of multiple measurements following monitored daily by spectrophotometric analysis (Hach
single tracer injection. Inc., Loveland, CO, USA). Nitrite and nitrate levels
were quantified once weekly. Lighting was derived
from banks of commercial phosphorescent tubes
2. Materials and methods positioned 6 m above the experimental system.
Salinity was measured with a temperature-compen-
2.1. System configuration sated refractometer (Aquafauna Bio-Marine, Haw-
thorne, CA, USA).
All studies were undertaken using a four-tank Tank 1 (Fig. 1) was used as the experimental tank
seawater recirculating aquaculture system. This sys- whereas tanks 2 and 4 were employed to hold
tem had been in continuous operation for a period of 2 experimental animals. To maintain a constant flow
years for the holding of red drum, cobia and summer into tank 1, water was pumped from tank 3 using a
and southern flounders. Throughout this period, no submersible pump (Little Giant Pump, Oklahoma
mortalities were recorded in any of the tanks. The City, OK, USA). This strategy was used in order to
recirculation configuration (Fig. 1) comprised a KMT- avoid changes in flow due to the accumulation of
based (Kaldnes Miljbteknologi, Tbnsberg, Norway) organic matter within the bead filter. Water flow into
fluidized bed biofilter for conversion of ammonia to tank 1 was carefully monitored using a Dialog MM3
nitrate, a bead filter (Aquaculture Technologies Inc., flowmeter (Master Meter, Mansfield, TX, USA). Flow
Metaire, LA, USA) used to eliminate solids (uneaten rates into tank were controlled using valve adjust-
feed, fecal material, mucus and other fish waste), a ments to the feeder line from tank 3. Maximum flow
protein skimmer for removal of dissolved material and rates of 0.5 l s
1 were attainable using this arrange-
a UV sterilizer (Aquatic Ecosystems, Apopka, FL, ment. Water temperature was maintained at 22 8C and

Fig. 1. Principal design and configuration of the experimental marine recirculating aquaculture system employed during the current
investigations.
 M.R. Rasmussen et al. / Aquaculture 250 (2005) 162–174 165

! injection and measurement at the same point in the

tank.
Inlet
Injection
point Sleeve
Measurements at the outlet permitted analysis of
residence time distribution, whereas the in-tank
Fluorometer configuration facilitated a more thorough investiga-
Rhodamine wt tion of the mixing process. Tank 1 was equipped with
reservoir and
pump Standpipe a fluorometer/injector assembly, which automatically
injected Rhodamine WT into the system and meas-
ured fluorescence. Fig. 2 illustrates the overall layout
of the experimental tank.
Water level was regulated using a central stand-
Fig. 2. Diagram illustrating the overall component arrangement of pipe. The surrounding standpipe sleeve incorporated 2
the experimental tank used during the present studies. holes at the base for egress of particulate materials.
The fluorometer/injector assembly was placed directly
salinity of 8x. Throughout experimental measure- opposite to the inlet. Tank water volume was 440 l.
ments, oxygen concentrations were maintained at z7 The tank had a diameter of 1.21 m and the water
mg l
1. depth was maintained at 0.375 m. Flow was set at
Tank mixing was measured using Rhodamine WT 0.23 l/s (1.9 exchanges/h) and 0.42 l/s (3.4 exchanges/
(Aquatic Ecosystems, Apopka, FL, USA). This tracer h) in order to investigate mixing at high and low flow
was selected due to its low toxicity and sorption rates. The inlet was a single inlet with at diameter of
properties. Fluorescence was measured using a 0.038 m. The inlet was located 0.23 m above the
Cyclops 7 Submersible Flourometer (Turner Designs, bottom of the tank, against the side wall. The inlet
Sunnyvale, CA, USA). nozzle was pointed perpendicular to the radius. The
Experiments were performed using two configu- dimensions of the tank and the fluorometer/injector
rations (Figs. 2 and 3): assembly are noted Fig. 3. The tracer injection point
was situated downstream of the fluorometer to avoid
! injection into the inlet pipe and measurement at the probe interference during injections. Signals from the
tank outlet. fluorometer were measured with a PMD-1208LS

1.22 m Fluorometer/injector
assembly

0.075 m
0.2 m

0.05 m Fluorometer
Injector

Flow direction

0.375 m

0.23 m

0.16 m

Fig. 3. Experimental tank dimensions and sketch of the fluorometer/injector assembly used in determining reactor mixing characteristics.
166 M.R. Rasmussen et al. / Aquaculture 250 (2005) 162–174

analog/digital board (Measurement Computing Corpo- impossible to have residence times that exceed
ration, Middleboro, MA, USA) and stored on a PC. hydraulic residence times.
Signals were sampled with a frequency of 1 Hz. If tank hydrodynamics resemble non-ideal plug
Dedicated software was developed to acquire, store and flow characteristics, additional variables must be
analyze all measurements. Pump-based injection of determined. It is assumed that the one-dimensional
Rhodamine WT tracer into the experimental tank was transport-dispersion model for conservative tracers in
computer controlled. Using 4-h intervals, the software a stationary and uniform flow can be used.
automatically injected 3 ml of dye (100 ppm Rhod-
amine WT) into the tank. Automation of this procedure BC BC B2 C
þU ¼D ð4Þ
minimized fish stress. Although the vast majority of Bt Bx BC 2
Rhodamine WT was removed from the system water by where x is distance, U the mean velocity, D the
the KMT bed and bead filter, measured background dispersion coefficient, and t is time.
concentrations (residual Rhodamine WT) were sub- Assuming that measured tracer concentration as a
tracted from experimental data sets. function of time is proportional to the concentration as
a function of space, a simple relationship between
2.2. Methodological analysis of outlet tracer tracer concentration variance and the dispersion
experiments number can be estimated (Levenspiel, 1999):
 
Measurement of dye concentrations at the outlet rt2 D
¼2 ð5Þ
has traditionally been the preferred method to t¯2 UL
determine the mixing characteristics of non-idealized
where r2t is the variance calculated from the concen-
reactors. Calculation of tracer residence time allowed
tration measured at the outlet (Eq. (6)) and L a
quantification of mixing.
characteristic length. The variance is calculated as
Residence time, t̄, was calculated as
Z l
Z l  2
t
t¯ Cdt
tCdt
rt2 ¼ 0 Z l : ð6Þ
t¯ ¼ Z0 l ð1Þ
Cdt
Cdt 0
0
For higher dispersion numbers (D/UL N 0.01) in a
where C is the concentration and t is time. Residence
closed reactor a correction to Eq. (5) is necessary
time was then compared to hydraulic residence time,
(Levenspiel, 1999):
t h:
   
V rt2 D D 2
ð UDL Þ

th ¼ ð2Þ ¼ 2
2 1
e : ð7Þ
Q t¯2 UL UL
The dispersion number can be determined by
where V is the tank volume and Q is the flow through
solving Eq. (7) iteratively. Variance and residence
the tank.
time is calculated using a numerical trapeze integra-
When the residence time is smaller than hydraulic
tion method. The dispersion number is dimensionless
residence time, this implies that not all of the tank
and indicates the proportion between dispersion and
volume participates in the process. This dead zone
convection. The dispersion number will move asymp-
volume can be calculated as
totically towards infinity as tank mixing approaches
 
t¯ ideal mixed conditions. However, Levenspiel (1999)
Vdead zone ¼ 1
V ð3Þ advocates that this method cannot be employed when
th
the dispersion number exceeds 1. It should be noted
Although difficult to identify physical positions of that it remains impossible to isolate the dispersion
dead zones, their calculated size indicates the level of coefficient from the dispersion number using this
mixing occurring within a reactor or tank. Thus, it is method, as it is unclear which characteristic length
 M.R. Rasmussen et al. / Aquaculture 250 (2005) 162–174 167

12
and velocity should be employed. From theory, it is Measurement
assumed that flow is one-dimensional and the 10 Analytical solution

Concentration (ppb)
characteristic length is the distance between the point
of injection and the point of measurement and the 8

characteristic velocity is the mean axial velocity. 6

When measuring at the outlet in a circular tank, these
values are not readily available. The dye tends to go in 4
a rotational motion and not directly from inlet to
2
outlet.
0
70 75 80 85 90 95 100
2.3. Methodological analysis of in-tank tracer Time (sec )
experiments
Fig. 5. An example curve fit for measured tracer concentrations and
As an option to the traditional methods of analytical solution, R 2 = 0.98.
measurement and analysis of residence time distribu-
tion at the outlet of a circular tank, an alternative is Although it can be argued that tank dye dis-
proposed. Flow in circular tanks tends to be highly persion is three-dimensional, a more detailed model
rotational. If a small amount of dye is injected into the is not justified unless the water velocity is measured
tank, the dye will have a tendency to follow the same more accurately and the transverse and vertical
tangential streamline that it was injected into (Fig. 4). dispersion coefficients are found. The implication
Due to centrifugal forces a secondary current slowly of this simplification is that the dispersion coefficient
rotates the water in a cork-screw fashion. Thus, by contains components of the vertical, transverse and
taking measurements at the same distance from the longitudinal dispersion. However, this is not that
center of the tank, portions of the cloud of dye can be different from real world application, such as pipe
assesed many times at different stages of dispersion. and river flows where Eq. (8) is widely used. The
The mathematical element for this method is the meandering of rivers results in rotational flows at their
one-dimensional equation for transport-dispersion as bends, which resembles the flow present in circular
described by Eq. (4). The analytical solution for Eq. tanks and similar reactors.
(4), with an impulse injection of dye is (Fisher et al., The dispersion coefficient found from Eq. (8) can
1979): be split into separate components depending on the
 ðx
utÞ2  source of mixing:
M
C ð x; t Þ ¼ pffiffiffiffiffiffiffiffiffiffi e
4Dt ð8Þ
4pDt D ¼ Dx þ Dxy þ Dxz þ Di ð9Þ
where M is the mass of dye injected in the cross-
where D x is the longitudinal mean dispersion coef-
sectional area.
ficient, D xy and D xz are the longitudinal dispersion
coefficient as a result of shear in the transverse and
vertical plane, and D i is the longitudinal dispersion
coefficient as a result of internal mixing (e.g., fish).
The present method was employed to determine
the dispersion coefficient, D in a tank with and
without fish. Assuming that hydrodynamic-driven
mixing (D x , D xy, D xz ) are identical in the two
situations, the effect from fish can be determined by
subtracting the dispersion coefficient without fish
from the dispersion coefficient with fish:
Fig. 4. Illustration of the dispersion characteristics of Rhodamine
WT tracer as it circulated around the tank. The positions of the tank
inlet and fluorometer-injector assembly are also noted. Di ¼ Dwith fish
Dwithout fish ð10Þ
168 M.R. Rasmussen et al. / Aquaculture 250 (2005) 162–174

14 circular tank without fish. By calculating the R 2, the

Measurement
best fit between curve and measurement can be found.
Concentration (ppb)

12

10 The fit is completed using each individual peak, until

individual peaks can no longer be observed. Fig. 6.
8
tmix exemplifies that the baseline of harmonic variations
6
increases as a function of time. This baseline is sub-
tcirc
4 tracted from each peak prior to fitting the analytical
2 solution.
0 The circulation time, t circ, can be found by
0 200 400 600 800 1000 calculating the center of gravity of each peak. The
Time (sec )
time period between two peaks represents the
Fig. 6. Diagram illustrating the determination of circulation and circulation time. Mixing time, t mix, represents the
mixing time. Mixing time, t mix, was the time taken for the time taken for harmonic variations to disappear as
disappearance of harmonic variations. Mixing time was selected
illustrated in Fig. 6. For practical purposes, mixing
as the point at which the amplitude of the oscillation was ~10% of
the mean value. time is selected as the point at which the amplitude of
the oscillation is approximately 10% of the mean
value. The dispersion number, which was established
Eq. (4) assumes that the dye is evenly distributed when taking measurements at the outlet can also be
over the cross section of the rotational flow. However, found using this method, provided that a characteristic
in this method the dye is injected at a single point in length and velocity is determined. For flows such a
the tank. Comparison between analytical solutions for pipe and channel flows, the pipe diameter or water
one- and three-dimensional transport-dispersion equa- depth is used.
tions reveals that for small dispersion coefficients The characteristic length in the case of a circular
(D V 0.1 m2/s), no significant differences exist. The tank could be the water depth, Y. The characteristic
one-dimensional equation can therefore adequately velocity can be found as:
describe the mixing. For larger dispersion coefficients, Lcirc
however, a more elaborate analysis is required. The U¼ ð11Þ
tcirc
velocity and the dispersion coefficient are determined
by fitting Eq. (8) to the measurements. Fig. 5 where L circ is the length which the cloud of dye has to
illustrates how the curve fits measurements from the circulate before reaching the fluorometer again.

1 1
0.9 0.9
0.8 0.8
Concentration (ppb)

Concentration (ppb)

0.7 0.7
0.6 0.6
0.5 0.5
0.4 0.4
0.3 0.3
0.2 0.2
0.1 0.1
0 0
0 2000 4000 6000 0 2000 4000 6000
Time (s) Time (s)

Fig. 7. Concentration of Rhodamine WT tracer measured at the tank outlet, either in the absence (left) or presence (right) of fish. The flow into
the tank was 0.42 l s
1.
 M.R. Rasmussen et al. / Aquaculture 250 (2005) 162–174 169

1 1
0.9 Without fish 0.9 With fish
0.8 0.8
Concentration (ppb)

Concentration (ppb)
0.7 0.7
0.6 0.6
0.5 0.5
0.4 0.4
0.3 0.3
0.2 0.2
0.1 0.1
0 0
0 2000 4000 6000 0 2000 4000 6000
Time (s) Time (s)

Fig. 8. Concentration of Rhodamine WT tracer measured at the tank outlet, either in the absence (left) or presence (right) of fish. The flow into
the tank was 0.23 l s
1.

The dispersion number can therefore be calculated presented in Figs. 7 and 8. The figures both
as: demonstrate that mixing within the tanks was excel-
  lent. However, in the presence of red drum, there was
D D
¼  : ð12Þ a noticeable smoothing of the curves, which was
UL Lcirc
Y especially prevalent during the initial phase of the
tcirc
experiment.
All results were analysed using Student’s double- Table 1 summarizes the results of all analyses with
sided t-test. respect to mixing variables. Differences ( P b 0.05)
were detected in hydraulic residence times for both
flow rates examined due to the manual adjustments
3. Results undertaken during the study. However, variances were
low (Table 1). Irrespective of flow rate examined,
Example measured concentrations of Rhodamine increases ( P b 0.001) were observed in mean resi-
WT at the outlet, either in the absence or presence of dence time ratios when red drums were present. Thus,
fish and with flow rates of either 0.42 or 0.23 l s
1 are these data clearly indicate that fish have explicit

Table 1
Mixing variables (FS.D.) determined at the outlet (n = 5) at high and low flow rates and in the absence or presence of red drum in circular tanks
containing a water volume of 440 l
Flow Mixing variable Without fish With fish
High flow Hydraulic residence time, t h (min) 17.40 F 0.005a 17.44 F 0.005a
Mean residence time ratio t c/t h (–) 0.89 F 0.011b 0.97 F 0.03b
Mixing time, t mix (min) 8.06 F 0.58 ND
Circulation time, t circ (min) 0.57 F 0.012 ND
Dispersion number, D/uL (–) 2.77 F 0.71 2.55 F 0.85
Low flow Hydraulic residence, t h (min) time 31.40 F 0.005a 31.37 F 0.005a
Mean residence time ratio t c/t h (–) 0.71 F 0.034a 0.77 F 0.009a
Mixing time, t mix (min) 15.93 F 0.44 ND
Circulation time, t circ (min) 1.09 F 0.019 ND
Dispersion number, D/uL (–) 2.39 F 0.69a 1.37 F 0.27a
ND = not detectable. Presence of superscripts in rows indicates significant differences, with a indicating differences at the 0.05 level or better,
and b indicating differences at a level of 0.001 or better. Mixing and circulation times are calculated.
170 M.R. Rasmussen et al. / Aquaculture 250 (2005) 162–174

4
14
3.5
12
Concentration (ppb)

Concentration (ppb)
3
10
2.5
8
2
6 1.5
4 1

2 0.5

0 0
0 200 400 600 800 1000 0 200 400 600 800 1000
Time (s) Time (s)

Fig. 9. Concentration of Rhodamine WT tracer measured using the in-tank method, either in the absence (left) or presence (right) of fish. The
flow into the tank was 0.42 l s
1.

impact upon tank mixing processes while also tending Example measured concentrations of Rhodamine
to decrease the occurrence of dead zones, as revealed WT determined using the in-tank method, either in the
by increased residency time for the tracer. At higher absence or presence of fish, and with flow rates of
flows, the presence of fish had no impact ( P N 0.05) either 0.42 or 0.23 l s
1 are presented in Figs. 9 and
upon tank dispersion number. However, at low flow 10. Lucid from these results is that a more detailed
rates, addition of fish to the tank resulted in lower analysis of tank hydraulics is possible using in-tank
dispersion number ( P b 0.001). Noteworthy, however, measurements when compared to outlet studies. Thus,
was that the dispersion number was higher than rather than permitting the construction of a single
suggested maxima (Levenspiel, 1999), such that the curvilinear relationship (Figs. 7 and 8), in-tank
significance of these results remains questionable. measurements provided the means to examine the
Calculation of circulation and mixing times was only kinetics of tracer mixing over time. Comparison of in-
possible for tanks without fish. As might be antici- tank measurements with those taken only at the outlet
pated, both circulation and mixing time at flows of illustrate that tank mixing is extremely dynamic
0.23 l s
1 were approximately half those calculated during the first 400–600 s of the process. This
for tank flow rates of 0.42 l s
1. progression is not possible to discern using outlet

4
14
3.5
12
Concentration (ppb)

Concentration (ppb)

3
10
2.5
8
2
6 1.5

4 1

2 0.5

0 0
0 200 400 600 800 1000 0 200 400 600 800 1000
Time (s) Time (s)

Fig. 10. Concentration of Rhodamine WT tracer measured using the in-tank method in the absence (left) or presence (right) of fish. The flow
into the tank was 0.23 l s
1.
 M.R. Rasmussen et al. / Aquaculture 250 (2005) 162–174 171

Table 2
Mixing variables (FS.D.) determined using in-tank determinations (n = 6) at high and low flow rates and in the absence or presence of red drum
in circular tanks containing a water volume of 440 l
Flow Mixing variable Without fish With fish
High flow Hydraulic residence time, t h (min) 17.38 F 0.005b 17.53 F 0.005b
Mixing time, t mix (min) 12.24 F 0.31b 4.41 F 0.38b
Circulation time, t circ (min) 0.67 F 0.0067 0.64 F 0.028
Dispersion coefficient, D (m2/s) 0.0002125 F 0.000064 0.000975 F 000104b
Dispersion number, D/uL (–) 0.00133 F 0.00038b 0.00573 F 0.00059b
Low flow Hydraulic residence time, t h (min) 31.25 F 0.005b 31.97 F 0.005b
Mixing time, t mix (min) 19.29 F 0.69b 6.35 F 0.59b
Circulation time, t circ (min) 1.05 F 0.031 1.11 F 0.024
Dispersion coefficient, D (m2/s) 0.000142 F 000081 0.000706 F 0.000174b
Dispersion number, D/uL (–) 0.00139 F 0.00079b 0.00720 F 0.00186b
ND = not detectable. Presence of superscripts in rows indicates significant differences, with a indicating differences at the 0.05 level or better,
and b indicating differences at a level of 0.001 or better.

only measurements (cf. Figs. 7 and 9 and Figs. 8 and an onus is placed upon the systems engineer to design
10). Nevertheless, like the outlet only determinations, appropriate rearing units that allow optimal produc-
in-tank measurements further highlight the influence tion efficiency for individual species. In this context,
that fish have upon the tank mixing process. attention to tank hydraulics becomes important
Table 2 summarizes the results of all analyses with because fish express a wide range of behaviors that
respect to mixing variables using the in-tank method. may, to a certain extent, be positively influenced by
Although the presence of fish had no impact upon hydrodynamic control. For example, territoriality,
circulation times ( P N 0.05), significant effects were aggressive and boundary responses can be benefi-
observed for all other variables examined. Thus, cially manipulated in salmonids by attention to
regardless of flow employed, the presence of fish stocking density and adjustments to water flow (Ross
was seen to dramatically decrease ( P b 0.001) mixing et al., 1995). Similarly, a detailed understanding of
times by as much as one-third. Increased dispersion tank mixing can be gainfully employed to optimize
coefficients ( P b 0.001) were observed for both flow feeding activities and actions, establish efficient
regimes when red drums were present. sludge removal and ensure correct dissolution of
water treatments (Rasmussen et al., 2004).
Accurate determination of tank mixing, however, is
4. Discussion problematic. Conventionally, researchers have em-
ployed outlet measurements to characterize changes
One of the principal motivations underlying animal in mean residency time and dispersion number.
domestication is to enhance production. An important However, observations from the present studies
component during this process is the selection of provide strong evidence to suggest that the in-tank
animals that are able to perform well in artificial method represents a superior technique for evaluating
environments (Price, 2002). Animal agriculture has the impact of fish upon tank hydrodynamics. This
over a 10,000-year history of selection (Jones and method provided improved data acquisition and also
Brown, 2000) and the performance characteristics of presented the means to more readily discriminate,
terrestrial stocks have been immensely improved. In statistically, differences in mixing caused by the
stark contrast, it has been estimated that only 1–2% of presence of fish. Importantly, the method described
all aquacultured species have experienced genetic herein permitted sampling frequencies high enough to
selection (Gjedrem, 1997). Ostensibly then, the vast detect harmonic variations of tracer within the tank.
majority of aquaculture production is reliant upon The current studies clearly demonstrated that irrespec-
feral, or close-to-wild stocks. The coping styles tive of flow rate, the presence of fish enhanced tank
(Koolhaas et al., 1999) or adaptive capabilities of mixing. Moreover, the time taken to achieve complete
feral populations are often restricted in scope such that tank mixing (t mix) was at least two-thirds less in tanks
172 M.R. Rasmussen et al. / Aquaculture 250 (2005) 162–174

with fish than in tanks without. Employing in-tank may provide partial explanations for the beneficial
measurements determined that the presence of fish effects that have been observed to accrue following
greatly enhanced tank dispersion coefficients for both exercise training of fish (e.g., improved: fitness,
flows examined. In contrast, at high flows and in the growth homogeneity, body composition, food con-
presence of fish, no differences were seen in dispersion version and muscle growth and reduced aggression;
number when measuring at the outlet only. However, Jbrgensen and Jobling, 1993; Hammer and Schwarz,
at low flows, the presence of fish caused reductions in 1994; Azuma et al., 2002). Preservation of position,
dispersion number. Failure to detect differences in especially in higher flows necessarily involves
dispersion number using outlet measurements alone simultaneous use of multiple fins and body flex
has likewise been reported by Watten and Beck (1987), (Breder, 1926) which will result in subtle adjust-
who used circular tanks with channel catfish and ments to swimming forces, resulting in the creation
employed similar residence times to the study of localized jets and vortices. Maintenance of
described herein. Distinct to the investigations here stability within the water column thereby demands
and to those of Watten and Beck (1987) are the control of both external and self-generated alterations
findings of Watten et al. (2000), who stated that the to the aquatic medium. Stability control requires
presence of lake trout decreased mixing in circular precise body and fin movements which will inevi-
tanks. This deduction was made at even lower flow tably create significant, albeit confined perturbations
rates than employed by the present study, using outlet in the water column (Drucker and Lauder, 2003).
measurements only. These movements may provide one explanation for
Comparisons of dispersion number from the the enhanced tank mixing process noted when fish
present experiments without fish illustrate that were present.
irrespective of flow, these remained the same. This Videography undertaken during the present
observation fits well with theory for one-dimensional experiments revealed that changes in flow regimes
flows where dispersion number becomes constant at resulted in red drum repositioning within the water
higher Reynolds numbers (Levenspiel, 1958). Intro- column. At low flow, fish expressed random move-
duction of fish into the system, however, altered this ment and direction, whereas at high flow, animals
dynamic. Fish presence in this instance thus repre- relocated to positions near the tank base and close to
sents an important facet of the hydrodynamic its wall; also, animals positioned in a unidirectional
environment. Although outlet-based (one peak) manner, swimming against the flow. Theoretically,
determination of dispersion number provided accu- changes in fish station, caused by higher flow rates,
rate system evaluation, it was only with in-tank could have reduced fish-tracer interactions which in
determinations that the discrete impact of fish upon turn might have influenced mixing time and dis-
the mixing process could be appreciated. A number persion coefficient. Unequivocal is that the presence
of studies suggest that through hydrodynamic sens- of fish greatly enhanced tank mixing. Although the
ing, teleosts and other aquatic species undertake data from this study illustrates the impact that fish
subtle adjustments in posture, movement and posi- have upon tank hydrodynamics, it is noteworthy that
tioning to facilitate favorable exploitation of varia- together with stocking density, tank design will also
tions in the hydrodynamic environment (Shtaf et al., impact fish hydrodynamics and their overall produc-
1983; Anderson et al., 2001; Webb, 2002). One tion efficiency. The present study illustrated that fish
benefit that arises from hydrodynamic repositioning presence induced measurable and enhanced mixing
is a reduction in the energetic costs of locomotion. in circular tanks. Moreover, in-tank data acquisition
This may be achieved through the fine-tuning of appeared superior to outlet measurements when
standard metabolic rates (SMR; Pettersson and taking account of the impact of fish upon the mixing
Hedenstrom, 2000; Liao et al., 2003), or through process. Clearly, different stocking densities, fish
fish, especially when in schools, taking hydrome- sizes and species will impart diverse influences upon
chanical advantage of vortex streets (Weihs, 1973; tank mixing processes and a more precise under-
Blake, 2004; Tytell and Lauder, 2004). Hydro- standing of such impacts will assist in the design of
mechanical adjustments and or refinement to SMR optimal species-specific rearing units.
 M.R. Rasmussen et al. / Aquaculture 250 (2005) 162–174 173

Acknowledgements Jbrgensen, E.H., Jobling, M., 1993. The effects of exercise on

growth, food utilization and osmoregulatory capacity of
juvenile Atlantic salmon, Salmo salar. Aquaculture 116,
MRR gratefully acknowledges sabbatical travel 233 – 246.
support from the Danish Technical Research Council. Koolhaas, J.M., Korte, S.M., De Boer, S.F., Van Der Vegt, B.J., Van
Reener, G., Hopster, H., De Jong, I.C., Ruis, M.A.W., Blokhuis,
H.J., 1999. Coping styles in animals: current status in behavior
and stress-physiology. Neuroscience and Biobehavioral Reviews
References 23, 925 – 935.
Larmoyeux, J.D., Piper, R.G., Chenoweth, H.H., 1973. Evaluation
Anderson, E.J., McGillis, W.R., Grosenbaugh, M.A., 2001. The of circular tanks for salmonid production. Progressive Fish-
boundary layer of swimming fish. Journal of Experimental Culturist 35, 122 – 131.
Biology 204, 81 – 102. Levenspiel, O., 1958. Longitudinal mixing of fluids flowing
Azuma, T., Noda, S., Yada, T., Ototake, M., Nagoya, H., Moriyama, in circular pipes. Industrial and Engineering Chemistry 50,
S., Yamada, H., Nakanishi, T., Iwata, M., 2002. Profiles in 343 – 346.
growth, smoltification, immune function and swimming per- Levenspiel, O., 1999. Chemical Reaction Engineering, 3rd edition.
formance of 1-year-old masu salmon Oncorhynchus masou John Wiley and Sons, New York, USA. 668 pp.
masou reared in water flow. Fisheries Science 68, 1282 – 1294. Liao, J.C., Beal, D.N., Lauder, G.V., Triantafyllou, M.S., 2003. Fish
Blake, R.W., 2004. Fish functional design and swimming perform- exploiting vortices decrease muscle activity. Science 302,
ance. Journal of Fish Biology 65, 1193 – 1222. 1566 – 1569.
Breder, C.M., 1926. The locomotion of fishes. Zoologica 4, Odeh, M., Schrock, R.M., Gannam, A., 2003. Comparative
159 – 296. hydraulics of two fishery research circular tanks and recom-
Burley, R., Klapsis, A., 1985. Flow distribution studies in fish mendations for control of experimental bias. Journal of Applied
rearing tanks: Part 2. Analysis of the hydraulic performance of 1 Aquaculture 14, 1 – 23.
square m tanks. Aquacultural Engineering 4, 113 – 134. Pettersson, L.B., Hedenstrom, A., 2000. Energetics, cost reduc-
Burley, R., Klapsis, A., 1988. Making the most of your flow. tion and functional consequences of fish morphology. Pro-
IchemE Symposium (111), 211 – 223 (in fish rearing tanks). ceedings of the Royal Society Biological Sciences Series B 267,
Burrows, R., Chenoweth, H., 1955. Evaluation of three types of fish 759 – 764.
rearing ponds. US Fish and Wildlife Service Research Report Price, E.O., 2002. Animal Domestication and Behavior. CABI
#39. 29 pp. Pub, Wallingford, Oxon, UK. 297 pp.
Burrows, R., Chenoweth, H., 1970. The rectangular circulating Rasmussen, Michael R., McLean, Ewen, 2004. Comparison of two
rearing pond. Progressive Fish-Culturist 32, 67 – 80. different methods for evaluating the hydrodynamic perform-
Cripps, S.J., Bergheim, A., 2000. Solids management and removal ance of an industrial-scale fish-rearing unit. Aquaculture 242,
for intensive land-based aquaculture production systems. Aqua- 397 – 416.
cultural Engineering 22, 33 – 56. Rasmussen, M.R., Laursen, J., McLean, E., 2004. Development of
Drucker, E.G., Lauder, G.V., 2003. Function of pectoral fins in efficient sludge cones for the concentration of raceway-derived
rainbow trout: behavioural repertoire and hydrodynamic forces. solids in recirculating aquaculture systems. In: Rakestraw, T.T.,
Journal of Experimental Biology 206, 813 – 826. Douglas, L.S., Correa, A., Flick, G.J. (Eds.), Proceedings of the
Fisher, H.B., List, E.J., Koh, R.C.Y., Imberger, J., Brooks, N.H., 5th International Conference on Recirculating Aquaculture,
1979. Mixing in Inland and Coastal Waters. Academic Press, Roanoke, VA, USA, July 22–25, 2004. Virginia SeaGrant
New York, USA. 483 pp. Publication 04-08, pp. 400 – 410.
Gaikowski, M.P., Larson, W.J., Steuer, J.J., Gingerich, W.H., 2004. Ross, R.M., Watten, B.J., Krise, W.F., DiLauro, M.N., 1995.
Validation of two dilution models to predict chloramine-T Influence of tank design and hydraulic loading on the behavior,
concentrations in aquaculture facility effluent. Aquacultural growth, and metabolism of rainbow trout (Oncorhynchus
Engineering 30, 127 – 140. mykiss). Aquacultural Engineering 14, 29 – 47.
Gjedrem, T., 1997. Selective breeding to improve aquaculture Shoemaker, C.A., Evans, J.J., Klesius, P.H., 2000. Density and
production. World Aquaculture 28, 22 – 45. dose: factors affecting mortality of Streptococcus iniae
Griffiths, S.W., Armstrong, J.D., 2000. Differential responses of infected tilapia (Oreochromis niloticus). Aquaculture 188,
kin and nonkin salmon to patterns of water flow: does 229 – 235.
recirculation influence aggression? Animal Behaviour 59, Shtaf, L.G., Pavlov, D.S., Skorobogatov, M.A., Barekyan, A.Sh.,
1019 – 1023. 1983. The influence of flow turbulence on fish behavior. Journal
Hammer, C., Schwarz, G., 1994. The effect of endurance swimming of Ichthyology 23, 129 – 140.
on growth, body composition and calorific content of 0-group Skjblstrup, J., Nielsen, P.H., Frier, J.O., McLean, E., 1998. Novel
whiting (Merlangius merlangus, Gadidae). ICES Council Meet- design to evaluate performance characteristics of fluidised bed
ing Papers, Copenhagen, Denmark. 16 pp. biofilters for production of rainbow trout in aquaculture
Jones, M., Brown, T., 2000. Agricultural origins: the evidence of recirculation systems: nitrification rates, oxygen consumption
modern and ancient DNA. Holocene 10, 769 – 776. and sludge collection. Aquacultural Engineering 18, 265 – 276.
174 M.R. Rasmussen et al. / Aquaculture 250 (2005) 162–174

Skjolstrup, J., McLean, E., Nielsen, P.H., Frier, J.O., 2000. The Watten, B.J., Honeyfield, D.C., Schwarts, M.F., 2000. Hydraulic
influence of dietary oxilinic acid on fluidised bed biofilter characteristics of a rectangular mixed-cell rearing unit. Aqua-
performance in a recirculation system for rainbow trout. cultural Engineering 24, 59 – 73.
Aquaculture 183, 255 – 268. Webb, P.W., 2002. Control of posture, depth, and swimming
Tytell, E.D.T., Lauder, G.V., 2004. The hydrodynamics of eel trajectories of fishes. Integrative and Comparative Biology 42,
swimming: I. Wake structure. Journal of Experimental Biology 94 – 101.
207, 1825 – 1841. Weihs, D., 1973. Hydromechanics of fish schooling. Nature 241,
Wagner, E.J., Jeppsen, T., Arndt, R., Routledge, M.D., Bradwisch, 290 – 291.
Q., 1997. Effects of rearing density upon cutthroat trout Wheaton, F., 2002. Recirculating aquaculture systems: an
hematology, hatchery performance, fin erosion, and general overview of waste management. In: Rakestraw, T.T., Doug-
health and condition. Progressive Fish-Culturist 59, 173 – 187. las, L.S., Flick, G.J. (Eds.), Proceedings of the 4th International
Watten, B.J., Beck, L.T., 1987. Comparative hydraulics of a Conference on Recirculating Aquaculture, Roanoke, VA,
rectangular cross-flow rearing unit. Aquacultural Engineering USA, July 18–21, 2002. Virginia SeaGrant Publication 02-11,
6, 127 – 140. pp. 57 – 68.