Biogeosciences, 4, 269–282, 2007
www.biogeosciences.net/4/269/2007/
© Author(s) 2007. This work is licensed
under a Creative Commons License.
Biogeosciences
Structure of mass and momentum fields over a model aggregation of
benthic filter feeders
J. P. Crimaldi1 , J. R. Koseff2 , and S. G. Monismith2
1 Department
2 Department
of Civil and Environmental Engineering, University of Colorado, Boulder, CO, 80309-0428, USA
of Civil and Environmental Engineering, Stanford University, Stanford, CA, 94305-4020, USA
Received: 18 January 2007 – Published in Biogeosciences Discuss.: 9 February 2007
Revised: 19 April 2007 – Accepted: 19 April 2007 – Published: 16 May 2007
Abstract. The structure of momentum and concentration boundary layers developing over a bed of Potamocorbula amurensis clam mimics was studied. Laser Doppler
velocimetry (LDV) and laser-induced fluorescence (LIF)
probes were used to quantify velocity and concentration profiles in a laboratory flume containing 3969 model clams.
Model clams incorporated passive roughness, active siphon
pumping, and the ability to filter a phytoplankton surrogate
from the flow. Measurements were made for two crossflow
velocities, four clam pumping rates, and two siphon heights.
Simultaneous use of LDV and LIF probes permited direct
calculation of scalar flux of phytoplankton to the bed. Results show that clam pumping rates have a pronounced effect on a wide range of turbulent quantities in the boundary
layer. In particular, the vertical turbulent flux of scalar mass
to the bed was approximately proportional to the rate of clam
pumping.
1
Introduction
Shallow estuarine communities are commonly dominated by
suspension feeders that filter phytoplankton and other particles from the overlaying flow. The extent to which these
feeders can effectively filter the bulk phytoplankton biomass
depends on the vertical distribution and flux of phytoplankton in the water column. Phytoplankton generally reproduce near the surface where there are high levels of incident light. Density stratification or low levels of vertical
mixing can isolate phytoplankton in the upper layers of the
water column (Koseff et al., 1993). On the other hand, turbulent mixing can distribute phytoplankton throughout the water column, where they become accessible to benthic grazers. But grazing acts as a near-bed sink of phytoplankton,
Correspondence to: J. P. Crimaldi
([email protected])
and, in the absence of sufficient phytoplankton replenishment
from above through mixing, can produce a phytoplanktondepleted, near-bed region called a concentration boundary
layer. The severity of this concentration boundary layer depends on a balance between the rate at which grazers remove
phytoplankton and the rate at which phytoplankton is delivered to the bed by turbulent mixing processes. The nature of
the turbulence, and hence of the vertical mixing, depends on
tidal energy, waves, bed geometry, and the presence of the
benthic feeders themselves, whose geometric roughness and
siphonal currents can alter the flow.
In the present study, we investigate how aggregations of
benthic suspension feeders can alter the structure of both the
overlaying momentum and concentration fields. The physical roughness associated with benthic communities alters the
turbulent velocity field above them (Butman et al., 1994; van
Duren et al., 2006), significantly enhancing both turbulence
intensities and Reynolds stresses. Active siphonal currents
associated with filter feeding also impact the overlaying flow
structure (Ertman and Jumars, 1988; Larsen and Riisgard,
1997), and have been shown to enhance turbulence intensities (Monismith et al., 1990; van Duren et al., 2006). The
presence of suspension feeders also changes the structure of
the concentration field above them (O’Riordan et al., 1993;
Widdows and Navarro , 2007). Near-bed concentrations are
reduced by the filtering action of the community, but this effect can be mitigated by additional mixing due to both physical roughness and siphonal pumping.
We used model aggregations of the Asian clam Potamocorbula amurensis for this study. San Francisco Bay in
California, USA is abundantly populated by this invasive
species (Carlton et al., 1990), and it has become clear that the
grazing pressure exerted by these clams alters the dynamics
of phytoplankton blooms in the San Francisco Estuary. The
goal of the study is to quantify changes to the momentum and
concentration fields produced by the passive siphon roughness and active siphonal pumping of the clam aggregations.
Published by Copernicus Publications on behalf of the European Geosciences Union.
270
J. P. Crimaldi et al.: Mass and momentum fields over filter feeders
Crimaldi et. al: Mass and momentum fields over filter feeders
3
LDV/LIF
optics
LIF PMT
To LIF amplifier
Glass flume sidewall
False Plexiglas sidewall
Laser Beams
60 cm
Measuring Volume 20 cm
Model Clam Plates
Flow
False Plexiglas sidewall
x
Glass flume sidewall
LDV PMT
To LDV tracker
Fig.
1.1.Top
sidewalls, and
and the
theLDV/LIF
LDV/LIFoptical
opticalmeasurement
measurementsystem.
system.
Fig.
Topview
viewofofthe
theflume
flumetest
testsection
sectionshowing
showingthe
themodel
model clam
clam plates,
plates, false
false sidewalls,
Normalized mean velocity
2
1.2
8.75
Methods
1.0
2.10.8 Flume
0.6
U/U∞
The
experiments were performed in an V/U
open-channel,
recir∞
0.4
W/U∞Mechanics Labculating flume in the Environmental Fluid
0.2
oratory
at Stanford University. The flume is constructed of
0.0
stainless
steel and Plexiglas, with glass sidewalls in the test
section
to reduce laser refraction at the glass/water interface.
-0.2
-10
-5 in normal0 operation 5is approximately
10
The flume
capacity
distance from
flume
centerline (cm)by a digital
8000 liters. Lateral
A centrifugal
pump
commanded
frequency controller draws water from a downstream reservoir
charges
a constant-head
theofflume.
Fig.
2. and
Lateral
profiles
over a smoothtank
bed upstream
at z = 15ofcm
mean
An overspill
excess
water
the constantstreamwise
(U ), pipe
lateralreturns
(V ), and
vertical
(W )from
velocities,
normalized
by tank
the mean
stream
velocity Ureservoir.
sidewalls
are
head
backfree
to the
downstream
Water
from the
∞ . The false
located
at 10 and -10
fromthe
the flume
centerline.
constant-head
tankcm
enters
through a full-width diffuser and then passes through three stilling screens with decreasing coarseness to remove any large-scale structure in the
geous
smaller amounts
of dye are
flow because
and homogenize
the turbulence.
Theused.
flow then passes
Excurrent
flows consisted
of ambient
flume fluid
witha
through
a 6.25:1,
two-dimensional
contraction
andwith
enters
a rectangular
dose of dyechannel.
added toAit3-mm
(Fig.rod
5). spanning
This dosing
process
was
the flume
floor
at
the beginning
channelas
section
trips the boundary
layer
done
continuallyofinthe
real-time
the experiment
is conducted.
2m
upstream ofcontrolled
the test section.
The pump
flow passes
through
An
electronically
centrifugal
drew ambient
the from
test section,
then through
an exit section,
over
fluid
the flume’s
constant-head
tank intoand
thefinally
excurrent
an adjustable
weirfluid
backininto
downstream reservoir.
supply
line. The
thethe
constant-head
tank was Freewell
streamand
velocities
the test
section of 10
cmconcentration
s−1 to 40 cm s−1
mixed
had theinsame
background
dye
as
arefluid
usedthat
forwas
this about
study. to enter the flume. An electronically
the
The test
section
3 m pumped
long andconcentrated
0.6 m wide, with
a rectcontrolled
gear
pumpisthen
dye from
a
angular
cross
section
and
a
nominal
flow
depth
of
25 cm
20-liter reservoir into to the excurrent stream. The concentrated dye entered the excurrent stream via the dye injector,
and
was then mixed
into the excurrent
supply first by passBiogeosciences,
4, 269–282,
2007
www.biogeosciences.net/bg/0000/0001/
(Fig. 1). In the middle of the test section floor is a 1.8-m
long by 20-cm wide, removable section that can accommodate either a set
model clam
smooth plate
Topof
View
6.40 plates or a single
3.20
for baseline flow measurements. A pair of thin Plexiglas
sidewalls border the lateral edges of the model clam plates.
1.60
These false walls extend vertically from the bed through the
free surface and act as symmetry planes3.20
to effectively model
a wide bed of clams. The boundary layers developing on
the false vertical walls grow to only about 1 cm
3.20thick at the
Side View
downstream edge, and thus have little effect on flow over the
plates (Fig. 2). The flow is quite uniform across the inner
test section, with no significant variation or secondary flow
Fig. 3. Top and side views of a single model clam siphon pair in
structure.
the raised position, with the rubber siphon material shown in gray.
for this siphons
study were
on theofflume
cenTheMeasurements
exhalant and inhalant
(withmade
diameters
1.6 mm
and
terline
the model
plates
(or over
smooth
3.2
mm,over
respectively)
areclam
shown
as white
circlesa in
the topplate
view,
for baseline
The streamwise
location
thethe
and
as dashedmeasurements).
lines in the side view.
The side view
shows of
that
measurements
is denonted
which
the
raised
siphon models
protrudeby
3.2x,mm
into is
themeasured
flow. The from
overlaying
upstream layer
edge flow
of the
platesleft
(Fig.
1). The
plates
extend
from
boundary
is from
to right
in the
figure.
Dimensions
are
mm.
x=0in at
the upstream edge to x=180 cm at the downstream
edge.
ing
centrifugal excurrent supply pump, and then
2.2 through
Model the
Clams
through an in-line static mixer. A flow meter measured total
excurrent
flow rate, and
a manifold
the supply
Models of supply
clam aggregations
were
placed in split
the removable
floor nine
section
of the flume.
The models
mimicked
threethat
clam
into
streams.
Individual
flow meters
ensured
the
feautures:
(1)
the
physical
roughness
associated
with
siphons
nine streams had equal flow rates. Each of the nine streams
that areinto
raised
flow,
(2) clam
the incurrent
and excurpassed
one into
of thethe
nine
model
plates, where
a series
rent
siphonal
flows
associated
with
filter
feeding,
andof(3)
of internal channels routed excurrent fluid to each one
the
the ability
of the
to plate.
filter mass
fromportion
the flow.
Two
441
excurrent
jetsclams
in each
A small
of the
excurrent stream was diverted to the calibration jet for use as a
reference for calibrating
the LIF probe.
www.biogeosciences.net/4/269/2007/
Biogeosciences, 0000, 0001–14, 2007
To LDV tracker
est section
showing
the
model
clam
false
sidewalls,
and
the LDV/LIF
optical
measurement
system
1. Top view
of the
flume
test
section plates,
showing
model
plates, false
sidewalls,
and the LDV/LIF
optical measurement
system.
J.Fig.
P. Crimaldi
et al.:
Mass
and
momentum
fieldsthe
over
filterclam
feeders
271
Normalized mean velocity
1.2
8.75
8.75
1.0
0.8
Top View
0.6
U/U∞
V/U∞
W/U∞
0.4
Top View
6.40
0.2
U/U∞
V/U∞
-0.2
W/U
∞
-10
6.40
3.20
1.60
3.20
3.20
0.0
-5
0
5
10
Lateral distance from flume centerline (cm)
Side View
3.20
1.60
3.20
Fig. 2. Lateral profiles over a smooth bed at z = 15 cm of mean
Fig. 2. Lateral
over
a smooth
z = 15 cm
of mean
streamwise
(U ), profiles
lateral (V
), and
verticalbed
(Wat
) velocities,
normalstreamwise
(U ),free
lateral
(V ),velocity
and vertical
(W )false
velocities,
normalized
by the mean
stream
U∞ . The
sidewalls
are
ized
by
the
mean
free
stream
velocity
U
.
The
false
sidewalls
are
∞
located at 10 and −10 cm from the centerline.
located at 10 and
-10
cm
from
the
centerline.
5
10
Fig. 3. Top and side views of a single model clam siphon pair in
the raised position, with the rubber siphon material shown in gray.
The exhalant and inhalant siphons (with diameters of 1.6 mm and
3.20
3.2 mm,
respectively) are shown as white circles in the top view,
Side
View
0
and as dashed lines in the side view. The side view shows that the
raised siphon models protrude 3.2 mm into the flow. The overlaying
types
of clam models
from flume
centerline
(cm) were built, one with raised siphons
boundary layer flow is from left to right in the figure. Dimensions
(that were therefore rough) and one with flush siphons (that
geous because smaller amounts of dye are used.
mm.
Fig.are
3. inTop
and side views of a single model clam siphon pair in
presented a smooth surface except for the presence of the
Excurrent
flows
consisted
of
ambient
flume
fluid
with
with
Fig.
3.
Top
and
side
views
ofrubber
a single
modelshown
clam
siphon pai
the
raised
position,
with the
siphon material
in gray.
siphon orifices). Tests were run with only one type of clam
smooth model
bed
z
=
15
cm
of
mean
a doseat
of
dye
added
to
it
(Fig.
5).
This
dosing
process
was
The
exhalant
and
inhalant
siphons
(with
diameters
of
1.6
mm
and
(siphons raised or siphons flush) present the
in theraised
flume position, with the rubber siphon material shown in g
continually
in real-time
as the experiment
is conducted.
the centrifugal
excurrent
supply
pump,
and then
3.2 ing
mm,through
respectively)
are shown as
white circles
in the
top view,
and vertical
) Each
velocities,
atdone
one(W
time.
model
typenormalconsisted
of nine identical
20The
exhalant
and
inhalant
siphons
(with
diameters
ofthe1.6
and
as
dashed
lines
in
the
side
view.
The
side
view
shows
that
An electronically
controlled
centrifugal
drew ambient
through an in-line static mixer. A flow meter measured
total mm
cm×20-cm
square
plates
that fill
theare
1.8 mpump
long cutout
in the
velocity U
.
The
false
sidewalls
∞
raised
siphon
models
protrude
3.2
mm
into
the
flow.
The
overlaying
fluid from
flume’s
tankarray
into3.2
the
excurrent
excurrent supply
flow rate,as
andwhite
a manifold
split the
mm, respectively)
are shown
circles
in supply
the top vi
flume
floor. the
Each
plate constant-head
contains a 21×21
of individlayerstreams.
flow is from
left to right
in the
figure.ensured
Dimensions
supply
fluid
in the plus
constant-head
was in
well boundary
into nine
Individual
flow
meters
that the
the centerline.
ual
clamline.
siphonThe
pairs
(inhalant
exhalant) , tank
resulting
in
mm. in the side view. The side view shows that
and as dashed
lines
mixed and had the same background dye concentration
as are nine
streams had equal flow rates. Each of the nine streams
441 clams per plate, and a total of 3969 clams in4the strip of
the fluid
thatThe
wasclam
aboutmodels,
to enterwhile
the flume.
Anraised
electronically
passed into
one of the 3.2
nine model
clam the
plates,
whereThe
a series
siphon models
protrude
mm into
flow.
overlay
nine
plates.
idealized,
were fullcontrolled
gear pumpofthen
concentrated
dye with
from a
of internal channels routed excurrent fluid to each one of the
scale
representations
theirpumped
biological
counterparts,
boundary
layer441flow
is from left to right in the figure. Dimensi
20-literboth
reservoir
into todimension
the excurrent
stream. The
excurrent jets in each plate. A small portion of the exregard
to siphonal
and pumping
rates.concenDiunts of dye
are
used.
are
mm. current stream was diverted to the calibration jet for use as a
trated dyeand
entered
the excurrent
stream
via the
dyein
mensions
flow rates
were based
on observations
ofinjector,
real
and
was
then
mixed
into
the
excurrent
supply
first
by
passreference for calibrating the LIF probe.
Potamocorbula
amurensis
clams
made
by
Cole
et
al.
(1992)
d of ambient flume fluid with with
and Thompson (personal comm.).
9.6
9.53
Fig. 5). This
dosing
wassiphons had the same
Clam models
withprocess
raised and flush
www.biogeosciences.net/bg/0000/0001/
Biogeosciences, 0000, 0001–14, 2007
and exhalant
geometry (Fig. 3), but
latter
e as the inhalant
experiment
isorifice
conducted.
ingthethrough
the centrifugal excurrent supply pump, and th
did not protrude into the flow. Thus, the flush siphon model
d centrifugal
pump clam
drew
ambient
for an individual
consisted
simply of a pairthrough
of holes in an in-line static mixer. A flow meter measured to
a
smooth
plate
through
which
siphonal
currents
flowed.
tant-headModels
tankofinto
the excurrent
excurrent supply flow rate,
and a manifold split the sup
9.53
9.6
individual clams were arrayed on plates placed
the constant-head
well
intowasnine
in the flume testtank
section was
(Fig. 4).
This geometry
re- streams. Individual flow meters ensured that
peated across the entire array of 3969 clam models for both
Fig.had
4. Top equal
2×2
arrayrates.
of model Each
clam siphon
showackground
dye concentration as
nine
streams
flow
ofpairs
the
nine strea
the raised and flush siphon model types.
Fig. 4.
Top view
of view
aspacing
2 of×a 2used
array
model
clam
siphon
ing the clam
for theof
study.
The resulting
clam
array pairs sh
Model clam
plates
were cast in a mold using apassed
firm rubber;intoconsisted
enter the flume.
An
electronically
ofof 3969
the
nine
model
clam
plates,
where
a ser
siphon
pairs
in a 21×189
pattern.
Theresulting
overlaying
ing
the clamone
spacing
used
for
the
study.
The
clam
a
details of the construction process are given by O’Riordan
boundary-layer flow is from left to right in the figure. Dimensions
pumped(1993).
concentrated
dye
from
a
of
internal
channels
routed
excurrent
fluid
to
each
one
of
consisted
3969
siphon pairs in a 21×189 pattern. The overla
The interior of the plates included a series
of chan- of are
in mm.
nels
that
linked
all
of
the
incurrent
siphons
together
and
all
e excurrent stream. The concen441
excurrent jets
each left
plate.
A small
of the
boundary-layer
flow in
is from
to right
in theportion
figure. Dimens
of the excurrent siphons together. The back of each plate
rrent stream
via
the dye
injector,
current
stream
was
diverted
to the calibration
jet for use a
areand
in one
mm.
had a pair
of outlets:
one for
all incurrent siphons,
Filter
feeders
inhale phytoplankton-laden
fluid through
for
all
excurrent
siphons.
Irregularities
in
the
casting
process
their
incurrent
siphons
and
then
exhale
fluid
with
some
frache excurrent
supply first by passreference for
calibrating the LIF probe.
led to variations in the flow rates of individual clams. The
tion of the phytoplankton removed through their excurrent
siphons. Thus, while the volume of water entering and leavmeasured average deviation of the vertical excurrent velociIncurrent
generated
a similar
ties across a row of individual clams was approximately
25%
ingflows
the clamwere
is constant,
the amountby
of suspended
scalarsystem,
of
the
mean.
Although
this
was
not
an
intentional
feature
of
mass
(e.g.
phytoplankton)
is
not.
We
model
this
concentrag/0000/0001/
Biogeosciences,
0000,
0001–14,
20
without dyetioninjection.
An electronically
controlled
centr
the design, it is likely more representative of the real-world
change by labeling the excurrent flows with a fluorescent
situation than uniform flow.
dye. Thefluid
model in
clams
inhale ambient
from the flume,
gal pump drew
through
thewater
incurrent
siphon orifi
www.biogeosciences.net/4/269/2007/
while nine flowmeters ensured that each of the nine mo
Biogeosciences, 4, 269–282, 2007
clam plates had the same incurrent flow rate. A mani
then combined the nine incurrent streams, and the resul
272
J. P. Crimaldi et al.: Mass and momentum fields over filter feeders 5
Crimaldi
et. al: Mass and momentum fields over filter feeders
Incurrent Pump
Return to Flume Reservoir
Flume Constant-Head Tank
Flowmeter
Dye Pump
Controller
Excurrent
Pump
Controller
123.45
123.45
From Incurrents
9 Clam Plates
Dye
Reservoir
Dye Pump
To Excurrents
Incurrent
Pump
Controller
Dye
To Flume
From
Flume
Pump
123.45
Excurrent Supply
Dye Injector Excurrent Pump Passive Mixer
Flowmeter
Manifold
Fig.
the incurrent
incurrent and
and excurrent
excurrentsiphon
siphonflows,
flows,and
andfor
fordosing
dosingthe
theexcurrent
excurrentflows
flows
Fig.5.5.Schematic
Schematicofofthe
theplumbing
plumbingsystem
systemresponsible
responsible for
for driving
driving the
with
system, and
and the
the top
top half
halfisisthe
theincurrent
incurrentsystem
systemAAtotal
totalofof3969
3969model
model
withfluorescent
fluorescentdye.
dye.The
Thebottom
bottomhalf
half of
of the
the figure
figure is
is the excurrent system,
clam
driven
byby
this
system.
clamsiphon
siphonpairs
pairs(grouped
(groupedininnine
nineplates)
plates) are
were
driven
this
system.
but
exhale water that has a known concentration of dye.
1000they
8
Thus,6 in our system, clear fluid respresents phytonplanktonu 2 /u 2τ
4
laden water in the real system, and dyed fluid respresents
wa2
2
w 2/u
ter in the real system that has had phytoplankton
filtered
from
τ
8
it.100
This
inverse system, which was first used by Monismith
6
z +et al.4 (1990) and then later by O’Riordan (1993) is advantageous2 because smaller amounts of dye are used.
108
Excurrent
flows consisted of ambient flume fluid with a
dose64of dye added to it (Fig. 5). This dosing process was done
continually in real-time as the experiment was conducted.
0
2
4
6 drew ambient
8
An electronically
controlled
centrifugal
pump
Normalized
turbulence
intensities
fluid from the flume’s constant-head tank into the excurrent
supply line. The fluid in the constant-head tank was well
Fig.
6. Normalized
streamwise
(u) and vertical
turbulence inmixed
and had the
same background
dye (w)
concentration
as
tensities.
LDV
Reflume.
and lines are
θ = 1320,
the fluidSymbols
that was are
about
to data
enteratthe
An electronically
corresponding
DNSpump
results
at Re
from Spalartdye
(1988).
θ = 1410
controlled gear
then
pumped
concentrated
from a
20-liter reservoir into to the excurrent stream. The concentrated dye entered the excurrent stream via the dye injector,
and was then mixed into the excurrent supply first by passBecause the dye fluoresces in an omni-directional pating through the centrifugal excurrent supply pump, and then
tern, we were able to place the LIF receiving optics in the
through an in-line static mixer. A flow meter measured total
backscatter configuration without any loss of signal (as opexcurrent supply flow rate, and a manifold split the supply
posed
to thestreams.
LDV receiving
optics,
which
were
placedthat
preferinto nine
Individual
flow
meters
ensured
the
entially
in
the
strong
forward-scatter
lobes).
The
LIF
receivnine streams had equal flow rates. Each of the nine streams
ing
opticsinto
andone
theofLIF
directly
within
the
passed
thePMT
nine were
modelmounted
clam plates,
where
a series
LDV
front
optics
(using
a
backscatter
module
intended
for
of internal channels routed excurrent fluid to each one of the
making
backscatter
LDV
measurements).
Thus,
the
receiv441 excurrent jets in each plate. A small portion of the exing
opticsstream
for thewas
LIFdiverted
automatically
moved with
current
to the calibration
jet the
for measuruse as a
ing
volumefor
as the
LDV/LIF
was traversed throughout
reference
calibrating
thesystem
LIF probe.
the test section of the flume, maintaining consistent alignment. The PMT for the LIF had a pinhole section that
Biogeosciences, 4, 269–282, 2007
www.biogeosciences.net/bg/0000/0001/
Incurrent flows were generated by a similar system, but
1000
8
6 dye injection. An electronically controlled centrifuwithout
4
gal pump drew fluid in through the incurrent siphon orifices
2
while nine flowmeters ensured
uw / u 2τthat each of the nine model
1008
clam plates
had
the
same
incurrent flow rate. A manifold
6
∂U 2
z + combined
4
/u streams, and the resulting
ν
then
the nine incurrent
∂z τ
stream2 returned into the flume reservoir downstream of the
flume
108test section. There was a great deal of turbulence in
6
the flume
reservoir (generated by the plunging action of the
4
flume flow spilling over the weir), and this turbulence con0.2 fluid in0.4
0.6
1.0 therestantly0.0
mixed the
the reservoir.
The0.8reservoir
fore served as a well Normalized
mixed source
of
fluid
for
the
flume
stress
pump and the excurrent supply pump.
The
of background
dye stresses.
in the flume
greware
Fig.
7. concentration
Normalized Reynolds
and viscous
Symbols
LDV time
data atdue
Reto
1320,
and lines
are corresponding
DNSflows.
results
with
the
constant
dosing
of the excurrent
θ =
at Resystem
Spalartsuch
(1988).
θ = 1410
The
wasfrom
designed
that this concentration growth
was extremely linear in time (Crimaldi, 1998). Excurrent
flows contained a fixed amount of dye added on top of the
masked stray
light from
anywhere
other than
section.
existing
background
flume
concentration,
so the
the test
difference
More details
on construction
and operation
ofthe
theexcurrent
LIF probe
between
the background
concentration
and the
are given by Crimaldi
concentration
remained(1998).
constant with time.
The smallest scalar fluctuations in a flow occur at the scale
at which
viscous diffusion acts to smooth any remaining con2.3
Instrumentation
centration gradients. Batchelor (1959) defines this scale as
2.3.1 Velocity measurements
ηB = ηK Pr−1/2
(2)
A Dantec, two-component, laser-Doppler velocimeter (LDV)
where
Pr to
is the
Prandtl
number. According
to Barrett
was
used
measure
velocities.
This instrument
was (1989),
operthe Prandtl
number
Rhodaminefluorescence
6G, the dye(LIF)
usedprobe
in the
ated
in tandem
with afor
laser-induced
study, is 1250. Therefore, the smallest concentration scales
ηB = 3 µm were about 35 times smaller than the smallest
www.biogeosciences.net/4/269/2007/
Biogeosciences, 0000, 0001–14, 2007
Fig. 5.
5. Schematic
Schematic of
of the
the plumbing
plumbingsystem
systemresponsible
responsiblefor
fordriving
drivingthe
theincurrent
incurrentand
andexcurrent
excurrentsiphon
siphonflows,
flows,and
andfor
fordosing
dosingthe
theexcurrent
excurrentflows
flows
Fig.
with
fluorescent
dye.
The
bottom
half
of
the
figure
is
the
excurrent
system,
and
the
top
half
is
the
incurrent
system
A
total
of
3969
model
with fluorescent dye. The bottom half of the figure is the excurrent system, and the top half is the incurrent system A total of 3969 model
clam
siphon pairs
pairs
(grouped
in nine
nine
plates)are
arefields
drivenover
bythis
this
system.
J.clam
P. Crimaldi
et al.:
Mass in
and
momentum
filter
feeders
273
siphon
(grouped
plates)
driven
by
system.
10008
1000
8
10008
1000
8
6
6
4
4
22
+
zz+
6
6
4
4
2 /u22
uu2 /u
ττ
22
2/u22
ww2/u
ττ
1008
100
8
66
44
+
uw/ u/ 2u 2τ
uw
τ
∂U 2
∂U
2
ν
τ
ν ∂z/u/u
∂z τ
1008
100
8
z
z+
66
44
22
22
108
10
8
1010
88
66
44
66
44
00
22
44
66
88
0.0
0.0
Normalizedturbulence
turbulenceintensities
intensities
Normalized
0.2
0.2
0.4
0.4
0.6
0.6
0.80.8
1.01.0
Normalizedstress
stress
Normalized
Fig.6.6.
6.Normalized
Normalizedstreamwise
streamwise(u)
(u)and
andvertical
vertical
(w)
turbulence
inFig.
(w)
turbulence
inNormalized
streamwise
(u)
and
vertical
(w)
turbulence
intensities.Symbols
Symbols
are
LDV
data
Reθθ =
= and
1320,
and
lines
are
tensities.
areare
LDV
datadata
at Re
=1320,
lines
arelines
corretensities.
Symbols
LDV
atatθ Re
1320,
and
are
sponding
DNS results
at Reθ at
=1410
from
Spalart
corresponding
DNS results
results
at Re
Reθθ =
= 1410
1410
from(1988).
Spalart(1988).
(1988).
corresponding
DNS
from
Spalart
Fig.
Normalized
Reynolds
and
viscous
stresses.
Symbols
are
Fig.
Normalized
Reynolds
and
viscous
stresses.
Symbols
Fig.
7.7.7.Normalized
Reynolds
and
viscous
stresses.
Symbols
areare
LDV
data
atRe
1320,
andlines
lines
arecorresponding
corresponding
DNSresults
results
LDV
data
are
LDV
data
atatRe
=1320,
and and
lines
are corresponding
DNSDNS
results
at
θ θ==1320,
θRe
Reθ θ==from
1410Spalart
fromSpalart
Spalart
(1988).
atat
1410
from
(1988).
Re
=1410
(1988).
θ Re
to measure mass fluxes. The LDV was driven with an argonBecause
the dye
dye
fluoresces
in an
an
omni-directional
patthe
in
omni-directional
patIonBecause
laser operated
in fluoresces
the 514.5 nm
single-line
mode, with
we were
were
able
to place
place
themeasuring
LIF receiving
receiving
optics
the
we
able
the
LIF
optics
the
atern,
nominal
output
of to
1.0
W. The
volume
wasininelbackscatter
configuration
without
any
loss of
ofin
signal
(asopopbackscatter
configuration
any
loss
signal
(as
liptical
in shape,
with the without
long axis
oriented
the crossposed to
the
optics,
which
were
prefertodirection.
the LDV
LDV receiving
receiving
optics,of
which
wereplaced
placed
preferchannel
The
dimensions
the measuring
volume
entially
in
the
forward-scatter
lobes).
LIF
receiventially
in intensity
the strong
strong
forward-scatter
lobes). The
The0.1
LIFmm
receiv(to
the e−2
contour)
were approximately
in
ingvertical
optics
LIF
were
within
the
optics and
and the
the
LIF PMT
PMT
weremounted
mounted
directly
within
the
the
streamwise
directions,
and 1directly
mm
in the
crossLDV front
optics
(using
aa backscatter
intended
for
front
opticsThe
(using
backscatter
modulein
intended
for
channel
direction.
smallest
scales of module
motion
the flows
making
LDV
the
making backscatter
backscatter
LDV
measurements).
Thus,
thereceivreceivmeasured
in this study
can measurements).
be determined byThus,
estimating
the
ing optics
the
automatically
moved
optics for
for
the LIF
LIF(Tennekes
automatically
movedwith
withthe
themeasurmeasurKolmogorov
scale
as
and Lumley,
1972)
ing volume
volume as
as the
the LDV/LIF
LDV/LIFsystem
systemwas
wastraversed
traversedthroughout
throughout
!1
the test
section
testκzν
section
of the
the flume,
flume, maintaining
maintaining consistent
consistent alignalign3 4 of
ηment.
(1)
The
PMT
for
the
LIF
had
a
pinhole
section
that
K ≈
The
PMT
for
the
LIF
had
a
pinhole
section
that
u3τ
masked
stray
lightfrom
fromanywhere
anywhere
other
thanthe
the
test
section.
masked
light
test
section.
station,
atstray
a sample
rate
of
80 Hz. other
The than
mean
free
stream
−1and
Moredetails
details
on
construction
and
operation
of
the
LIF
probe
More
construction
operation
of
the
LIF
probe
velocity
was Uon
=11.6
cm
s
,
which
resulted
in
a
calcu∞
aregiven
givenby
byCrimaldi
Crimaldi(1998).
(1998).
are
lated
momentum-thickness
Reynolds number of Reθ =1320.
fluctuations
inina aflow
atatthe
Thesmallest
smallest
scalar
fluctuations
flowoccur
occur
thescale
scale
We The
compared
thescalar
data to
Spalart’s DNS
results
simulated
at
atat
which
viscous
diffusion
acts
to
smooth
any
remaining
conwhich
viscous
diffusion
acts
to
smooth
any
remaining
conRe
=1410.
Although
the
Reynolds
numbers
differ
by
6%,
θ
centration
Batchelor
(1959)
defines
scale
centration
gradients.
Batchelor
(1959)
definesthis
this
scale
the
variationgradients.
of normalized
turbulence
parameters
with
Reasθas
is quite weak, enabling a valid comparison.
−1/2
−1/2
ηηBB(Fig.
==ηηK6)
Pr
Pr
(2)
K
Turbulence intensities
were normalized by the(2)
square of the shear velocity uτ that was obtained by fitwhere
Pr
Prandtl
number.
According
totoBarrett
(1989),
where
Prisisthe
the
Prandtl
number.
According
Barrett
ting
the mean
velocity
profile
to the
law of the
wall. (1989),
The
the
Prandtl
number
for
Rhodamine
6G,
the
dye
used
ininthe
the
Prandtl
number
for
Rhodamine
6G,
the
dye
used
the
agreement between the LDV results and DNS calculations
study,
isis1250.
Therefore,
the
smallest
concentration
scales
study,
1250.
Therefore,
the
smallest
concentration
scales
by Spalart (1988) was excellent for both the turbulence inηηBB == 33µm
were about
35 times
than
smallest
about
timessmaller
smaller
thanthe
the
smallest
tensities
andµm
thewere
viscous
and 35
turbulent
stress profiles
(Fig.
7).
where
κ is the Kolmogorov constant, z is distance from the
www.biogeosciences.net/bg/0000/0001/
www.biogeosciences.net/bg/0000/0001/
bed, ν is the viscosity, and uτ is the shear velocity. The smallest value of ηK for the flows in this study (corresponding
to uτ =1.7 cm s−1 and z=0.1 cm) was approximately 0.1 mm.
Although this value of ηK is comparable to the dimension
of the measuring volume in the vertical and streamwise directions, it is smaller than the dimension in the cross-flow
direction. Nonetheless, the LDV easily captures the larger
scales responsible for the transport of mass and momentum.
The LDV laser and optics were mounted on a motorized,
computer-controlled three-axis traverse which permitted the
LDV measuring volume to be positioned anywhere within
the test section. The traverse system was accurate to within
approximately 200 µm.
For validation purposes, LDV measurements of boundary
layer turbulence were taken in the flume over a smooth plate
that was installed in the same location where the model clam
plates were later installed. The LDV results were then compared with direct numerical simulations (DNS) of boundary
layer turbulence over a smooth bed by Spalart (1988). Data
were recorded at 23 logarithmically-spaced vertical stations
between z=0.7 mm and z=180 mm, measured from the bed.
Approximately 20 min of velocity data were recorded at each
2.3.2
www.biogeosciences.net/4/269/2007/
Concentration measurements
Biogeosciences,
Biogeosciences,0000,
0000,0001–14,
0001–14,2007
2007
We developed a laser-induced fluorescence (LIF) probe to
make non-intrusive measurements of dye concentrations in
the flow above model clam beds. The LIF probe used the
same measuring volume as the LDV, ensuring that the velocity and concentration measurements were made in the same
location. This is particlarly important for the scalar flux measurements, which resulted from correlations of velocity and
concentration measurements. The laser light in the combined LDV/LIF measuring volume was absorbed by fluorescent dye in the flow and re-emitted at a different wavelength.
The fluoresced light was optically filtered and converted to a
electrical current with a photomultiplier tube (PMT). Finally,
the current is converted to a voltage using an ideal currentto-voltage converter.
Because the dye fluoresces in an omni-directional pattern, we were able to place the LIF receiving optics in the
backscatter configuration without any loss of signal (as opposed to the LDV receiving optics, which were placed preferentially in the strong forward-scatter lobes). The LIF receiving optics and the LIF PMT were mounted directly within the
LDV front optics (using a backscatter module intended for
Biogeosciences, 4, 269–282, 2007
274
J. P. Crimaldi et al.: Mass and momentum fields over filter feeders
6
Crimaldi et. al: Mass and momentum fields over filter feeders
probe. The time response of the LIF probe is extremely fast;
without
dyeassociated
(other than
background
dye)with
represented
the time
constants
withthe
the dye
fluorescence,
phytoplankton-laden
fluid.
Using
the
LIF
probe,
0.8
the PMT, and with the LIF signal amplifier are extremelywe calcua nondimensional
in thisin“inverse”
syssmall lated
relative
to the time scalesconcentration
of turbulent motion
this
0.6
flow. tem as
− CB
∗ modelC
During experiments over the
Cinv
= clams, the, LIF calibra(3)
0.4
C
CB the measureE −
tion jet remained in the flume, positioned
above
0.2
ment where
region in
freeoutput
streamofofthe
theLIF
flow.probe
A small
portion
C the
is the
at the
measurement
of thelocation,
dyed flowCmixed
for
the
excurrent
supply
was
diverted
B is the output due to the background dye (mea0.0
through
the in
calibration
jet. The and
LIF C
probe
was periodically
sured
the
free stream),
E is the output of pure excur0.0
0.2
0.4
0.6
0.8
1.0
positioned
in
the
free
stream
and
behind
the
calibration
jet this “inrent fluid (measured using the calibration
jet). To put
Normalized calibration jet concentration
during experiments to maintain the probe calibration as backverse” measurement in a more intuitive framework, we then
ground and excurrent concentrations rose during the experia complementary nondimensional concentration
Fig.8.8.Normalized
Normalized
calibration
curve
for LIF
the LIF
showing
Fig.
calibration
curve
for the
probeprobe
showing
the thementsdefined
due to dye accumulation in the flume.
linear
system
response.
linear system response.
Normalized LIF output
1.0
2.3.3
∗
C ∗ = 1 − Cinv
Concentration normalization
(4)
∗
such that C = 1 now corresponded (in the real system) to
making
backscatter
measurements). Thus, the receivscales of
motion ηLDV
was conK . Thus, the LIF probe could not measureAs discussed earlier, a known concentration of dye
fluid with full phytoplankton load, and C ∗ = 0 corresponded
ing
optics
for
the
LIF
automatically
moved
with
the
measurthe smallest scales of motion present in the studied flows.tinuously added to the excurrent jets. The dyed fluid repto fluid that had its entire phytoplankton concentration reing volume as the LDV/LIF system was traversed throughout
However, the LIF probe could easily measure the larger con-resented filtered fluid devoid of phytoplankton, and fluid
the test section of the flume, maintaining consistent alignmoved by filtration. These nondimensional concentrations
centration scales that are responsible for the vast majority ofwithout dye (other than the background dye) represented
ment. The PMT for the LIF had a pinhole section that
could be converted
dimensional
concentrations
phytoplankton-laden
fluid. toUsing
the LIF probe,
we calcu-for a real
the scalar flux.
masked stray light from anywhere other than the measureby considering
the ambient
phytoplankton
concentralated system
a nondimensional
concentration
in this
“inverse” sysTovolume.
validateMore
the LIF
probe,
we measured
dye
ment
details
on construction
and known
operation
of con-tem astion and the filtering efficiency of the bivalve.
centrations
potential
core(1998).
of a jet flowing from a
the
LIF probe in
arethe
given
by Crimaldi
C − CB
7.5
mm
diameter
the at
flume
test sec- ∗
The
smallest
scalarcalibration
fluctuationstube
in a within
flow occur
the scale
,
(3)
Cinv =
− CB
measurements
oftothe
jet fluid
were made
1 mm
attion.
whichLIF
viscous
diffusion acts
smooth
any remaining
con3 CEResults
downstream
of the Batchelor
jet orifice(1959)
(x/Ddefines
= 0.13),
centration
gradients.
thiswhich
scale asensuredwhere C is the output of the LIF probe at the measurement
Vertical
ofdue
velocity
and concentration
that the measuring
volume was well within the potential corelocation,
CB is profiles
the output
to the background
dye (mea-data were
ηof
Pr−1/2
(2)from
B =
taken
for
two
crossflow
velocities
and
four
clam pumping
theηKjet.
Eight different dye concentrations ranging
sured in the free stream), and CE is the output of pure excur-
rates,
for eachusing
of the
clam model
((Table
0 to 100
were pumped
theto jet,
and(1989),
the result-rent fluid
(measured
thetwo
calibration
jet). Totypes
put this
“in- 1). For
where
Pr isppb
the Prandtl
number. through
According
Barrett
vertical profile,
simultaneous
LDV and
ingPrandtl
LIF data
are linear
(Fig 8). Calibration
jetused
concentrations
verse”each
measurement
in a more
intuitive framework,
weLIF
thendata were
the
number
for Rhodamine
6G, the dye
in the
acquired
at approximately
20 vertical
stations. The stations
were normalized
by the maximum
value
used in the
test (100defined
a complementary
nondimensional
concentration
study,
is 1250. Therefore,
the smallest
concentration
scales
were
logarithmically
spaced,
usually
starting
at z = 0.5, and
LIF
output
was
normalized
so
that
the
output
from
the
ηppb).
=3
µm
were
about
35
times
smaller
than
the
smallest
B
∗
C ∗ =ending
1 − Cinv
(4) of data
at
z
=
120
mm.
Typically,
100,000
samples
40 ppb
was η0.4.
A
least-squares
estimate
of
the
slope
of
scales
of jet
motion
.
Thus,
the
LIF
probe
could
not
resolve
K
the
in thedye
studied
flows.
wereCacquired
approximately
from
each LDV
chanthesmallest
line wasscales
1.006of±motion
0.003.present
The actual
concentrations
∗ =1 nowatcorresponded
such that
(in80
theHz
real
system)
to
However,
the
LIF
probe
could
easily
measure
the
larger
con- ex-fluid with
∗
nel and
from
the
LIF
probe
for
each
vertical
station,
although
used in the measurements over the model clams rarely
full phytoplankton load, and C =0 corresponded
centration
that are
responsible
for the vast range
majority
shorter
records
were phytoplankton
used near the concentration
upper edge ofrethe boundceeded 5 scales
ppb, well
within
the demonstrated
of of
linear-to fluid
that had
its entire
the
layer
where
the
variance
of
the
signals
was
small.
ityscalar
of theflux.
LIF probe. The time response of the LIF probe ismovedary
by filtration. These nondimensional concentrations
To validate
thethe
LIFtime
probe,
we measured
known
dye
presented
in thisconcentrations
paper focus for
ona perturbations
extremely
fast;
constants
associated
with
thecondye flu-could beResults
converted
to dimensional
real
centrations
in
the
potential
core
of
a
jet
flowing
from
a
by considering
the ambient
concentrato the momentum
andphytoplankton
scalar concentration
fields by the
orescence, with the PMT, and with the LIF signal amplifiersystemmade
7.5 mm diameter calibration tube within the flume test secthe filtering
the bivalve.
presence
of theefficiency
clams. of
These
perturbations come from two
are extremely small relative to the time scales of turbulenttion and
tion. LIF measurements of the jet fluid were made 1 mm
sources: the presence of roughness (in the case of the model
motion in this flow.
downstream of the jet orifice (x/D=0.13), which ensured
clams with raised siphons), and the presence of siphonal curDuring
experiments
over
the
model
clams,
the
LIF
calibrathat the measuring volume was well within the potential core
3 Results
rents and filtering. The flush siphonal orifices without clam
in the flume,
positioned above
the from
measureoftion
thejet
jet.remained
Eight different
dye concentrations
ranging
pumping
alter the
(Figs. 9 and
region
in the
free stream
A small
portionVertical
0ment
to 100
ppb were
pumped
throughofthethe
jet,flow.
and the
resulting
profilesdid
of not
velocity
and flow
concentration
data10).
were The data
these twovelocities
figures compares
boundary-layer
flow
of the
mixed
LIF
datadyed
wereflow
linear
(Fig. for
8). the excurrent supply was divertedtaken shown
for twoincrossflow
and four the
clam
pumping
a smooth
withmodel
the flow
over
the flush
clam modthrough
the calibration
jet. Thewere
LIF probe
was periodically
Calibration
jet concentrations
normalized
by the
rates, over
for each
of the plate
two clam
types
(Table
1). For
els, withprofile,
no siphonal
currents.
experiments
maximum
the test
ppb).theLIF
output jeteach vertical
simultaneous
LDVBoth
and LIF
data werewere perpositionedvalue
in theused
freeinstream
and(100
behind
calibration
was
normalized
so
that
the
output
from
the
40
ppb
jet
was
acquired
at
approximately
20
vertical
stations.
The
formed
with
a
free
stream
velocity
of
U
10 cm s−1 , corduring experiments to maintain the probe calibration as back∞ =stations
0.4.
A least-squares
estimate
of the slope
the line
logarithmically
spaced,
usually
startingshow
at z=0.5,
andflow over
to Re
Results
that the
ground
and excurrent
concentrations
roseofduring
thewas
experi-were responding
θ = 560.
1.006±0.003.
The
actual
dye
concentrations
used
in
the
ending
at
z=120
mm.
Typically,
100
000
samples
of
data
the flush siphon orifices was indistinguishable from flow over
ments due to dye accumulation in the flume.
measurements over the model clams rarely exceeded 5 ppb,
were acquired
at approximately
80 Hz
from each LDV
the smooth
plate. Thus, the
perturbations
to thechanflow demonwell
within
the
demonstrated
range
of
linearity
of
the
LIF
nel
and
from
the
LIF
probe
for
each
vertical
station,
strated later are due only to the presence of although
siphon roughness
2.3.3 Concentration normalization
As discussed earlier,
a known
Biogeosciences,
4, 269–282,
2007concentration of dye was continuously added to the excurrent jets. The dyed fluid represented filtered fluid devoid of phytoplankton, and fluid
and/or siphonal pumping.
The effects ofwww.biogeosciences.net/4/269/2007/
siphon roughness and pumping on the shear
velocity are shown in Fig. 11. Shear velocity uτ is a measure
of the bed shear stress τw , where uτ = (τw /ρ)1/2 . As is
Crimaldi et. al: Mass and momentum fields over filter feeders
Crimaldi
al: Mass
momentum
fields fields
over
filter
feeders
J.Crimaldi
P.et.Crimaldi
et and
al.:
Mass
momentum
fields
over
filter
feeders
et. al:
Mass
andand
momentum
over
filter
feeders
7
275 7 7
Table 1. Parameters varied in the experiments
1.1.Parameters
Table 1.Table
Parameters
variedvaried
in
theinexperiments
Table
Parameters
varied
inthe
theexperiments.
experiments
Parameter
Symbol
Values
Comments
Parameter
Symbol
Values
Comments
Parameter
Symbol
Comments
Parameter
Symbol
Comments
Free stream velocity
U∞ValuesValues
10, 40 cm s−1
−1
−1
Clam
pumping
rate
Q
0,
0.030,
0.045,
0.060
ml
s
rate per clam
Free
velocity
10,
4040
cm
s s−1
Free stream
velocity
cm
s−1
Freestream
stream
velocityU∞ U∞
U∞10, 40
10,
cm
−1
Clam
siphon
height
h
0,
3.2
mm
“flush”
and “raised” in text
s
0,0,
0.030,
0.045,
0.060
s s−1rate
raterate
perclam
clam
Clampumping
pumping
rate Q QQ 0, 0.030,
0.030,
0.045,
0.060
ml
per
clam
Clam Clam
pumping
rate rate
0.045,
0.060
mlmls−1
per
Clam
siphon
height
h
0,
3.2
mm
“flush”
and
“raised”
in text
sh
Clam siphon height
0, 3.2 mm
“flush” and
“raised”
in text
Clam siphon height
hs
s 0, 3.2
mm
“flush” and “raised” in text
1000
22/u22
ττ
u 2 /u 2τ uw /u
2
2
w 2/u 2τ w /u τ
100
2
z +100
100
6
4
26
4
62
4
2
10
10
U∞ (cm/s)
z + 10
26
4
62
4
6
4
40
1
30
20
2
0
1
2
0
0
2
2
4
6
20
20
4
6
10
8
10
4
6
8
Normalized
turbulence
intensities
1.2
1.1
1.0
0.9
0.8
8
Normalized turbulence intensities
1.5
1.3
30
40
1.5
1.3
1.4
1.2
40 40
(a) siphons
(a) siphons
flushflush
1.4
30 1.4
10
2
1
40
(b) siphons raised
1.5
0.7
00.6
U∞ (cm/s)
U∞ (cm/s)
U∞ (cm/s)
62
4
40
U∞ (cm/s)
U∞ (cm/s)
6
4
z+
u 2 /u 2τ
1000
1000
(a) siphons flush
6
4
1.3
1.1
1.2
1.0
30
1.1
0.9
1.0
0.8
0.9
0.7
0.8
0.6
20
0.7
0.5
0.6
0.5
0.5
0
.030
.045
Q (ml/s)
.030
.045
.060
.060
10
1.7
siphons
raised
(b) (b)
siphons
raised
1.6
1.7
1.6
1.5
1.4
30
1.3
20
1.2
1.1
20
1.0
0.9
10
0.8
0 0.7
0.6
10
0.5
30
0
1.5
1.7
1.4
1.6
1.3
1.5
1.2
1.4
1.1
1.3
1.0
1.2
0.9
1.1
0.8
1.0
0.7
0.9
0.6
0.8
0.5
0.7
0.6
0.5
.030
.045
Q (ml/s)
.030
.045
.060
.060
0
.030
.045 .060
0
.030
.045 .060
Q (ml/s)
Q (ml/s)
Fig. 9. Comparison
of streamwise
and vertical turbulence intensiNormalized
turbulence intensities
Fig. 11. Contours
of shear velocity uτ as a function
of free stream
Q
(ml/s)
Q
(ml/s)
ties for flow over non-pumping flush siphon orifices (closed symvelocity U∞ and clam pumping rate Q for flow over (a) flush
Fig.
9. Comparison
of streamwise
and vertical
turbulence
intensiFig.
Comparison
streamwise
vertical
turbulence
intensibols)9.with
flow over of
a smooth
plate and
(open
symbols).
Fig.
11. and
Contours
of shear
velocity
uτforasua function
of
siphons
(b) of
raised
siphons.
Units
arefree
cmstream
s−1 .
Fig. 9. ties
Comparison
of
streamwise
and
vertical
turbulence
intensifor
flow
over
non-pumping
flush
siphon
orifices
(closed
symτ contours
ties for flow over non-pumping flush siphon orifices (closed sym- Fig. 11. Contours
shear
velocity
uττrate
as
function
ofoffree
Fig. 11.
Contours
of shear
velocity
u
as aaQ
function
freestream
stream
velocity
U
clam
pumping
for
flow
over
(a)
flush
∞ and
over
plate
(open
symbols).
ties for bols)
flow
over
non-pumping
flush
siphon
orifices
(closed symbols)with
withflow
flow
overa asmooth
smooth
plate
(open
symbols).
−1
velocity
U∞
and
clam pumping
pumping
rate Q
over
(a)
flush
velocity
U∞
and
Qforfor
for
flow
overare
(a)
flush
siphons
and
(b)clam
raised
siphons. rate
Units
uτ flow
contours
cm
s
.
−1−1
bols) with flow over a smooth plate (open symbols).
siphons
and
(b)
raised
siphons.
Units
for
u
contours
are
cm
s
.
τ
siphons and (b) raised siphons. Units for u contours are cm s .
3.1
1000
6
4
uw/u 2τ
62
4
uw/u 2τ
Profiles
τ
3.1 Profiles
present
vertical profilesflow
of velocity
and contwoFigures
figures12-18
compares
the boundary-layer
over a smooth
3.1
Profiles
1000
centration
data
in
a
common
figure
format.
Because
the in100
plate
with 12-18
the flow
over vertical
the flush
clam of
models,
with
Figures
present
profiles
velocity
andnocon6
2
26
uw/u
fluence
of
the
clams
is
greatest
in
the
near-bed
region,
dis4
τ
4
siphonal
currents.
experiments
performed
with
centration
data
inBoth
a vertical
common
figure were
format.
Because
the
Figures
12-18
present
profiles
ofinvelocity
and
con-inz + 100
2
tance
from
the
bed
(the
vertical
axis
the
plots)
is
shown
−1
2
6
a free
stream
velocity
ofisUgreatest
=10 cm
sthe ,near-bed
corresponding
∞figure
fluence
of the
informat.
region,
discentration
data
in aclams
common
Because
theto
in100 z + 104
on
a logarithmic
scale.
Each
figure
contains
four
plots
Re
=560.
Results
show
that
the
flow
over
the
flush
siphon
θ
tance
from
the
bed
(the
vertical
axis
in
the
plots)
is
shown
6
26
fluence
of the clams
is greatest
in the near-bed
region, velocdisrepresenting
different
combinations
free the
stream
4
4
orifices
indistinguishable
fromfigure
flowof over
smooth
z+
onfrom
a was
logarithmic
scale.
Each
contains
four
plots
10
tance
the
bed
(the
vertical
axis
in
the
plots)
is
shown
2
ity U
positionto (flush
or demonstrated
raised). For later
profiles
62
∞ and
plate.
Thus,
thesiphon
perturbations
the flow
representing
different
combinations
of
free
stream
veloc4
1
onare
aoflogarithmic
scale.
Each
figure
contains
plots
10
concentration-related
quantities,
there
is no four
data
for
the
tosiphon
the
presence
siphon
roughness
and/or
2
6
itydue
U∞only
and
position of(flush
or free
raised).
For
profiles
0.0
0.2
0.4
0.6
0.8
1.0
representing
different
combinations
of
stream
velocQ
=
0
case
since
clam
pumping
is
required
for
concentra4
siphonal
pumping.
1
of concentration-related
quantities, there is no data for the
Normalized Reynolds stress
2
ity UShear
siphonu position
(flushof or
raised).
For
profiles
tion
measurements.
∞ and
velocity
a measure
the
bed shearfor
stress
τw ,
0.0
0.2
0.4
0.6
0.8
1.0
τ isclam
Q
=
0
case
since
pumping
is required
concentra1
1/2
of
concentration-related
quantities,
there
is
no
data
for
The
influence
of
clam
pumping
on
U
was
small
andthe
limwhere
u
=(τ
/ρ)
.
As
is
typical
in
turbulent
boundary
Normalized Reynolds stress
τ
w
tion measurements.
Fig.10.
10.Comparison
Comparison
ofReynolds
Reynolds
stress
correlations
flow
over
Fig.
of
stress
correlations
forfor
flow
over
0.0
0.2
0.4
0.6
0.8
1.0
to thesince
near-bed
region
(Fig.
12).
The
effect
was
largest
Q layer
=ited
0 flows,
case
clam
pumping
is
required
for
concentrauτ increases
with
the freeon
stream
velocity,
U∞ lim.
The influence
of clam
pumping
U was
small and
non-pumpingflush
flush
siphon
orifices
(closed
symbols)
with
flow
over
non-pumping
siphon
orifices
(closed
with
flow
over
Normalized
Reynolds
stresssymbols)
forpresence
the slowof(Uthe
= 10 siphons
cm s−1 produced
) flow over
raised
siphons,
tion
measurements.
∞ raised
Fig.
10.
Comparison
of
Reynolds
stress
correlations
for
flow
over
The
an
increase
in
smoothplate
plate
(open
symbols).
The
lines
DNS
results
Spalart
ited to the near-bed region (Fig. 12). The effect was largest
a asmooth
(open
symbols).
The
lines
areare
DNS
results
byby
Spalart
where
near-bed
values
of U were
as Qsiphons
increased.
non-pumping
flush siphon orifices (closed symbols) with flow over u
The
of clam
U to
was
and
limat influence
any
of Upumping
thesmall
flush
−1on retarded
τ for
∞ as compared
(1988)atatReRe
(1988)
=670.
θ = 670.
the given
slow value
(U
) flow
over
raised
siphons,
θ
∞ = 10 cm s
Fig. 10. aComparison
of
Reynolds
stress
correlations
for
flow
over
This
was
consistent
with
the
u
contours
in
Fig.
11,
where
smooth plate (open symbols). The lines are DNS results by Spalart ited
τ only
(Fig.
11).
Siphonal
pumping
caused
modest
increase
in
to
the
near-bed
region
(Fig.
12).
The
effect
was
largest
where near-bed values of U were retarded as Q increased.
non-pumping
orifices (closed symbols) with flow over
−1
(1988)flush
at Resiphon
the
greatest
sensitivity
toswas
changes
inpronounced
wall raised
stress(i.e.,
weretheseen
θ = 670.
,This
although
the
the
increase
more
foruτthe
slow
(U
=
10
cm
)
flow
over
siphons,
∞
was consistent
with the uτ contours in Fig. 11, where
a smooth plate (open symbols). The lines are DNS results by Spalart
fornear-bed
slow
overwas
raised
siphons.
For
flush
siphons
or faster
slope
of
theflows
contours
for
the
raised
siphon
clam
shorter records were used near the upper edge of the bound- where
values
of greater)
U
retarded
asstress
Q increased.
the
greatest
sensitivity
to were
changes
in wall
were
seen
(1988) at Reθ = 670.
flows,
the
effect
of
Q on
U uwascontours
negligible
milmodels
at
low
free
stream
velocities.
ary layer where the variance of the signals was small.
This
was
consistent
with
the
in beyond
Fig.
11,a few
where
τ
for
slow
flows
over
raised
siphons.
For flush
siphons
or faster
limeters from the bed.
typical
turbulent boundary
layerfocus
flows,onuτ perturbations
increases with
Resultsinpresented
in this paper
the greatest
in wall stress
were
seen
flows, thesensitivity
effect of Qtoonchanges
U was negligible
beyond
a few
milthe free
stream
velocity,
U∞ . concentration
The presencefields
of the
For
the slow flow with flush siphons, clam pumping attenProfiles
made
to the
momentum
and scalar
byraised
the
slow
flows
overthe
raised
limeters
from
bed. siphons. For flush siphons or faster
typical
in turbulent
boundary
layer flows, uτ increases
with for3.1
siphons produced
an increase
in uτ at any given
of
U∞
uated the streamwise turbulence intensities (expressed as the
presence
the clams.
These
comeofvalue
from
two
flows, the
of Q
onwith
U was
negligible
a few milFor effect
the slow
flow
flush
siphons, beyond
clam pumping
attenthe free of
stream
velocity,
U∞perturbations
. The presence
the raised
uu)present
in a narrow
region
centered
aroundand
z =con4 mm
as compared
to the flush
siphons
(Fig.
11).
Siphonal
pump- Figures
variance
sources:
the
presence
of
roughness
(in
the
case
of
the
model
12-18
vertical
profiles
of
velocity
the
bed. turbulence
produced
an increase in
uτ atuany
given value
of U∞ limeters
uatedfrom
the streamwise
intensities (expressed as the
typical siphons
in
boundary
flows,
with
ingturbulent
caused
only
modestlayer
increase
in uττ, increases
although
the
the
(Fig. 13a).
This
to theformat.
height atBecause
which the
verclams
with raised
siphons),
and the presence
of siphonal
cur-in- centration
data
in aacorresponds
common
figure
the
uu)
in
narrow
region
centered
z =atten4being
mm
compared
to
the
flush
siphons
(Fig.
11).
Siphonal
pumpvariance
the freeas
stream
velocity,
U
.
The
presence
of
the
raised
For
the
slow
flow
with
flush
siphons,
clamaround
pumping
∞
tical excurrent
achieve
a horizontal
crease
more pronounced (i.e., the slope of the contours influence
rents
andwas
filtering.
of the jets
clams
is greatest
in the trajectory
near-bed after
region,
caused an
only
modest
in given
uτ , although
thefree
in- uated(Fig.
13a).by
This corresponds
to the
height
at
which as
theet
versiphonsing
produced
increase
inincrease
uτsiphon
at any
value ofthe
Ulow
the over
streamwise
turbulence[see
intensities
the
∞
bent
Fig.in
4 (expressed
inplots)
O’Riordan
al.
was
for
theorifices
raised
distance
from thethe
bedcrossflow
(the vertical axis
the
is shown
Thegreater)
flush siphonal
withoutclam
clam models
pumpingatdid
not
crease
was
more
pronounced
(i.e.,
the
slope
of
the
contours
tical
excurrent
jets
achieve
a
horizontal
trajectory
after
being
as compared
to
the
flush
siphons
(Fig.
11).
Siphonal
pumpuu)
in
a
narrow
region
centered
around
z
=
4
mm
variance
stream
Since the
excurrent
were
laminar,
streamwise
alter
the velocities.
flow (Figs. 9 and 10). The data shown in these
on(1995)].
a logarithmic
scale.
Each jets
figure
contains
four
plots
was only
greater)
for the
raised in
siphon
clam models
at low
by corresponds
the crossflowto[see
in which
O’Riordan
et al.
13a).over
This
the Fig.
height4 at
the vering caused
modest
increase
uτ , although
the the
in- free (Fig.bent
(1995)].
Since
the
excurrent
jets
were
laminar,
streamwise
stream
velocities.
tical excurrent jets achieve a horizontal trajectory after being
crease was more pronounced (i.e., the slope of the contours
www.biogeosciences.net/bg/0000/0001/
Biogeosciences,
0001–14,
www.biogeosciences.net/4/269/2007/
Biogeosciences,
4,O’Riordan
269–282,
2007
bent over by the crossflow
[see Fig. 4 in0000,
et 2007
al.
was greater)
for the raised siphon clam models at low free
stream velocities.
(1995)]. Since the excurrent
jets were laminar,
streamwise
www.biogeosciences.net/bg/0000/0001/
Biogeosciences,
0000, 0001–14,
2007
1000
276
J. P. Crimaldi et al.: Mass and momentum fields over filter feeders
8
Crimaldi et. al: Mass and momentum fields over filter feeders
(a) Siphons flush, U∞ = 10 cm/s
(b) Siphons flush, U∞ = 40 cm/s
z (mm)
100
100
Q = 0.000
Q = 0.030
Q = 0.045
Q = 0.060
10
10
1
1
0
2
4
6
8
10
0
z (mm)
(c) Siphons raised, U∞ = 10 cm/s
100
10
10
1
1
2
4
6
8
20
30
40
(d) Siphons raised, U∞ = 40 cm/s
100
0
10
10
U (cm/s)
0
10
20
30
40
U (cm/s)
Fig. 12.
Effect
of clam
Q on Q
vertical
profiles
of mean
streamwise
velocity
U for
combinations
of of
free
stream
Fig.
12. Effect
ofpumping
clam pumping
on vertical
profiles
of mean
streamwise
velocity
U different
for different
combinations
free
streamvelocity
velocityU∞
−1 and
−1=
U∞ and
siphon roughness.
Theright
left and
right correspond
columns correspond
to Ucm
cm
s−1=40
and
40 cm s−1 , respectively.
The
and siphon
roughness.
The left and
columns
to U∞ =10
U∞
cmUs∞
, respectively.
The top and bottom
∞ s= 10
top and bottom
rows correspond
to siphons
flush (h
(hss=3.2
= 0)
and respectively.
raised (hs = The
3.2 mm),
respectively.
The bottom
row plotsdotted
containline
a at
rows correspond
to siphons
flush (hs =0)
and raised
mm),
bottom
row plots contain
a horizontal
horizontal
linethe
at location
z = hs of
= the
3.2raised
mm tosiphon
denotetops
the (below
locationwhich
of the data
raised
siphon
(below which
couldocclusion
not be acquired
z=hs =3.2
mm todotted
denote
could
nottops
be acquired
due todata
optical
of the indue toEach
optical
the instruments).
plot of
contains
color-coded
each of(Q=0,
the different
clam
pumping
struments).
plotocclusion
containsof
color-coded
profilesEach
for each
the different
valuesprofiles
of clamfor
pumping
0.030,values
0.045,ofand
0.060
ml s−1 ).
−1
(Q = 0, 0.030, 0.045, and 0.060 ml s
).
representing
different combinations of free stream velocity
turbulence intensities were locally reduced. The effect is
U∞ and
siphon
position
(flush
raised).
Fortheprofiles
of
similar for the
fast flow
case or
(Fig.
13b), but
attenuation
concentration-related
there
is no
data
formore
the Q=0
was closer to the quantities,
bed as the jets
were
bent
over
rapidly
case since
pumping
is required
for slow
concentration
by theclam
stronger
crossflow.
For the
flow overmearaised
surements.
siphons (c), there was a similar near-bed attenuation by the
pumping,
but it pumping
was now accompanied
by a strong
enTheclam
influence
of clam
on mean velocity
U was
the bed.
for the
flow
small hancement
and limitedfurther
to the from
near-bed
regionFinally,
(Fig. 12).
Thefast
effect
over the
siphons
(d),
thecm
effect
clam over
pumping
was
was largest
forraised
the slow
(U∞
=10
s−1of) flow
raised
minimal
as
the
turbulence
intensities
were
dominated
by
the
siphons, where near-bed values of U were retarded as Q inflow
and
roughness.
creased. This was consistent with the uτ contours in Fig. 11,
where the
to changes
in wall
stress intensiwere
Thegreatest
effect ofsensitivity
clam pumping
on vertical
turbulence
tiesslow
is opposite
fromraised
that seen
for theFor
horizontal
streamwise
seen for
flows over
siphons.
flush siphons
or
14).ofThe
vertical
energy
imparted
by the ainfaster intensities
flows, the(Fig.
effect
Q on
U was
negligible
beyond
current andfrom
excurrent
flows enhanced the vertical turbulence
few millimeters
the bed.
intensities. The effect was strongest for the slow flows (a, c)
For the slow flow with flush siphons, clam pumping atwhere the relative strength of the clam pumping was stronger.
tenuated the streamwise turbulence intensities (expressed as
In these cases, the influence of the pumping extended deep
the variance
in a narrow
around
into theuu)
boundary
layer. region
For thecentered
faster flows
(b, d)z=4
the mm
effect
(Fig. 13a).
This
corresponds
to
the
height
at
which
the
verwas minimal except close to the bed.
tical excurrent jets achieve a horizontal trajectory after being
The
of roughness
due to
theO’Riordan
raised siphons
probent over
bypresence
the crossflow
(see Fig.
4 in
et al.,
duced
increased
Reynolds
stress
relative
to
the
flush
siphon
1995). Since the excurrent jets were laminar, streamwise turcase, especially for the fast flow cases (Fig. 15). Clam pumpbulence intensities were locally reduced. The effect is similar for the fast flow case (Fig. 13b), but the attenuation was
0000,
2007
closerBiogeosciences,
to the bed as the
jets0001–14,
were bent
over more rapidly by
the stronger crossflow. For the slow flow over raised siphons
Biogeosciences, 4, 269–282, 2007
(c),
there was a similar near-bed attenuation by the clam
ing also produced an increase in Reynolds stress, with the
pumping,
butdramatic
it was now
accompanied
by aover
strong
effect more
at slower
flows and
the enhanceraised
ment
further from the bed. Finally, for the fast flow over the
siphons.
raised siphons (d), the effect of clam pumping was minimal
In all
cases, theintensities
mean nondimensional
concentration,
C ∗ and
,
as the
turbulence
were dominated
by the flow
was reduced near the bed (relative to the free stream value
roughness.
of C ∗ = 1) due to the filtering action of the clams (Fig. 16).
The effect of clam pumping on vertical turbulence intensiThe reduction was significantly more pronounced for slower
ties
is opposite
that more
seen so
forfor
thethe
horizontal
streamwise
flows
(a, c), andfrom
slightly
raised siphons
(c,
intensities
(Fig.
14).
The vertical
energy imparted
by the ind). Increased
clam
pumping
also enhanced
the concentration
current
andbut
excurrent
flows
enhanced
vertical turbulence
reduction,
the effect
on the
near-bedthe
concentrations
was
intensities.
The effect
was pumping
strongestproduced
for the slow
flows
(a, c)
relatively weak.
Stronger
a larger
conwhere
the relative
strength
of thethe
clam
pumping
stronger.
centration
reduction
throughout
depth
of thewas
boundary
Inlayer.
these cases, the influence of the pumping extended deep
into the boundary layer. For the faster flows (b, d) the∗ effect
The nondimensional concentration fluctuations, c c∗ ,
was
minimal
exceptlarger
closefor
to the
the slow
bed. flow cases relative to
were
significantly
The
presence
of
roughness
due
to thetoraised
siphonstheprothe fast flows since there was less mixing
homogenize
duced
increased
Reynolds
stress
relative
to
the
flush(a,c),
siphon
concentration field (Fig. 17). For the slow flow cases
case,
especially
for
the
fast
flow
cases
(Fig.
15).
Clam
where the excurrent jets penetrated farther into the flow,pumpthe
ing
also
produced
an increase
in was
Reynolds
stress,
peak
in the
concentration
variance
above the
wall.with
Thethe
effect
more dramatic
at the
slower
and overin the
raised
peak moved
farther from
wall flows
and decreased
magnitude as pumping increased. For the fast flow cases (b,d), the
siphons.
peak concentration variance was at the tops of the siphons,
www.biogeosciences.net/bg/0000/0001/
www.biogeosciences.net/4/269/2007/
J. P. Crimaldi et al.: Mass and momentum fields over filter feeders
277
(a) Siphons flush, U∞ = 10 cm/s
z (mm)
100
(b) Siphons flush, U∞ = 40 cm/s
100
Q = 0.000
Q = 0.030
Q = 0.045
Q = 0.060
10
10
1
1
0
1
2
3
4
5
0
z (mm)
(c) Siphons raised, U∞ = 10 cm/s
100
10
10
1
1
1
2
3
uu (cm/s)
4
10
15
20
(d) Siphons raised, U∞ = 40 cm/s
100
0
5
5
0
5
2
10
15
uu (cm/s)
20
2
Fig. 13. Effect of clam pumping Q on vertical profiles of streamwise turbulence intensity uu for different combinations of free stream
velocity U∞ and siphon roughness.
(a) Siphons flush, U∞ = 10 cm/s
z (mm)
100
(b) Siphons flush, U∞ = 40 cm/s
100
Q = 0.000
Q = 0.030
Q = 0.045
Q = 0.060
10
10
1
1
0.0
0.2
0.4
0.6
0.8
1.0
0
z (mm)
(c) Siphons raised, U∞ = 10 cm/s
100
10
10
1
1
0.2
0.4
0.6
ww (cm/s)
0.8
2
2
3
4
5
(d) Siphons raised, U∞ = 40 cm/s
100
0.0
1
1.0
0
1
2
3
ww (cm/s)
4
5
2
Fig. 14. Effect of clam pumping Q on vertical profiles of vertical turbulence intensity ww for different combinations of free stream velocity
U∞ and siphon roughness.
www.biogeosciences.net/4/269/2007/
Biogeosciences, 4, 269–282, 2007
278
J. P. Crimaldi et al.: Mass and momentum fields over filter feeders
Crimaldi et. al: Mass and momentum
fieldsUover
filter
(a) Siphons flush,
cm/sfeeders
∞ = 10
100
z (mm)
z (mm)
100
Q = 0.000
Siphons flush, U∞ = 10 cm/s
Q(a)
= 0.030
Q = 0.045
= 0.000
QQ
= 0.060
Q = 0.030
Q = 0.045
Q = 0.060
100
10
(b) Siphons flush, U∞ = 40 cm/s
100
10
10
10
1
1
1-0.5
-0.4
-0.3
-0.2
-0.1
(c) Siphons
∞ = 10 cm/s
-0.5
-0.4 raised,
-0.3 U-0.2
-0.1
100
0.0
-4
1
100
(c) Siphons raised, U∞ = 10 cm/s
-3
-2
-1
0
(d)
-4 Siphons
-3 raised,-2U∞ = 40-1cm/s 0
0.0
100
z (mm)
z (mm)
11
(b) Siphons flush, U∞ = 40 cm/s
(d) Siphons raised, U∞ = 40 cm/s
100
10
10
10
10
1
1
1
-0.5
-0.4
-0.5
-0.4
-0.3
-0.2
-0.1
2
uw
-0.3(cm/s)
-0.2
-0.1
0.0
1
-4
0.0
-3
-2
-3uw (cm/s)
-2
-4
2
2
-1
0
-1
0
2
uw (cm/s)
uw (cm/s)
Fig. 15. Effect of clam pumping Q on vertical profiles of Reynolds stress uw for different combinations of free stream velocity U∞ and
siphon roughness.
Fig. 15. Effect of clam pumping Q on vertical profiles of Reynolds stress uw for different combinations of free stream velocity U∞ and
siphon roughness.
(a) Siphons flush, U∞ = 10 cm/s
z (mm)
100
(b) Siphons flush, U∞ = 40 cm/s
100
Q = 0.030
Q = 0.045
Q = 0.060
10
10
1
1
0.5
0.6
0.7
0.8
0.9
1.0
0.5
z (mm)
(c) Siphons raised, U∞ = 10 cm/s
100
10
10
1
1
0.6
0.7
0.8
*
C
0.9
0.7
0.8
0.9
1.0
(d) Siphons raised, U∞ = 40 cm/s
100
0.5
0.6
1.0
0.5
0.6
0.7
0.8
0.9
1.0
*
C
16. Effect
of clam
pumping
on vertical
profiles
meannondimensional
nondimensional concentration
combinations
of free
stream
Fig. 16.Fig.
Effect
of clam
pumping
Q onQvertical
profiles
ofof
mean
concentrationCC∗ ∗forfordifferent
different
combinations
of free
stream
velocity
U
and
siphon
roughness.
velocity U∞ and∞siphon roughness.
www.biogeosciences.net/bg/0000/0001/
Biogeosciences, 4, 269–282, 2007
Biogeosciences, 0000, 0001–14, 2007
www.biogeosciences.net/4/269/2007/
J. P. Crimaldi et al.: Mass and momentum fields over filter feeders
279
(a) Siphons flush, U∞ = 10 cm/s
z (mm)
100
(b) Siphons flush, U∞ = 40 cm/s
100
Q = 0.030
Q = 0.045
Q = 0.060
10
10
1
1
0.000
0.005
0.010
0.015
0.020
0.000
z (mm)
(c) Siphons raised, U∞ = 10 cm/s
100
10
10
1
1
0.005
0.010
0.015
0.010
0.015
0.020
(d) Siphons raised, U∞ = 40 cm/s
100
0.000
0.005
0.020
0.000
0.005
c* c *
0.010
0.015
0.020
c* c *
Fig. 17. Effect of clam pumping Q on vertical profiles of nondimensional concentration variance c∗ c∗ for different combinations of free
stream velocity U∞ and siphon roughness.
(a) Siphons flush, U∞ = 10 cm/s
z (mm)
100
(b) Siphons flush, U∞ = 40 cm/s
Q = 0.030
Q = 0.045
Q = 0.060
100
10
10
1
1
-0.030
-0.020
-0.010
0.000
-0.030
z (mm)
(c) Siphons raised, U∞ = 10 cm/s
100
10
10
1
1
-0.020
-0.010
wc* (cm/s)
-0.010
0.000
(d) Siphons raised, U∞ = 40 cm/s
100
-0.030
-0.020
0.000
-0.030
-0.020
-0.010
0.000
wc* (cm/s)
Fig. 18. Effect of clam pumping Q on vertical profiles of vertical scalar flux wc∗ for different combinations of free stream velocity U∞ and
siphon roughness.
www.biogeosciences.net/4/269/2007/
Biogeosciences, 4, 269–282, 2007
Fig. 18. Effect of clam pumping Q on vertical profiles of vertical scalar flux wc∗ for different combinations of free stream velo
siphon roughness.
280
J. P. Crimaldi et al.: Mass and momentum fields over filter feeders
rates fluxes
by infaunal
Biology, 113, 219
Turbulent scalar
can be bivalves,
expressed Marine
as a nondimensional correlation
coefficient,
defined asstructure of velocity and scal
Crimaldi,
J.: Turbulence
100
6
a bed of model bivalves, Ph.d. thesis, Stanford Univ
4
wc∗
ρw,c = √ Ertman,
√
S. C. and Jumars, P. A.: Effects of(5)bivalve siph
wwonc∗the
c∗ settlement of inert particles and larvae, Journ
z (mm)
2
10
6
4
2
1
6
-0.5
-0.4
-0.3
rw,c
-0.2
-0.1
0.0
Research,
797–813,
1988. formulation
where −1≤ρw,c
≤1. The 46,
correlation
coefficient
removes the
effect ofP.the
variances, reJonsson,
R.,individual
Petersen,w J.and
K.,c Karlsson,
O., Loo, L. O
sulting in a true
measure
the correlation
between
two
son,
S. F.: of
Particle
depletion
abovethe
experimental
biv
(Fig. 19). Thesitu
profiles
collapse
into
a
relatively
tight
band,
measurements and numerical modeling of biva
with a common peak correlation coefficient of approximately
in
the
boundary layer, Limnology and Oceanograph
-0.38. Note that the vertical location of the peak correlation
1998, 2005.higher in the flow than the correcoefficient is significantly
Koseff,
sponding peak
of wcJ.,∗ . Holen, J., Monismith, S., and Cloern, J.: Co
of vertical mixing and benthic grazing on phytopla
lations in shallow, turbid estuaries, Journal of Mari
4 Discussion51, 843–868, 1993.
Fig. 19. Vertical profiles of the correlation coefficient ρw,c for all
Fig. 19.
Vertical profiles of the correlation coefficient ρw,c for all
experimental conditions listed in Table 1. Flush siphon cases are
P. S. and Riisgard, H. U.: Biomixing generate
experimental
conditions
in Table
1. Flush siphon cases The
are resultsLarsen,
shown in red,
and raisedlisted
siphon cases
in black.
of this study add to a growing body of literafilter
feeders:
A diffusion
modelalter
forcharnear-bottom ph
shown in red, and raised siphon cases in black.
ture that demonstrates
how benthic
filter feeders
of concentration
Sea Research,
37, in81–90, 1997
acteristics of depletion,
momentum Journal
and scalar
fields
In all cases, the mean nondimensional concentration, C ∗ ,
Monismith,
S., Koseff,
Thompson,
O’Riordan, C
the water column.
The results
share J.,
some
qualitative J.,
simiwas reduced near the bed (relative to the free stream value
larities with previous
studies,
despite
the
fact
that
different
H.: A study of model bivalve siphonal currents, Lim
of C ∗ =1) due to the filtering action of the clams (Fig. 16).
species
were
involved.
Our results
that streamwise
Oceanography,
35,show
680–696,
1990. turButman,
A., Frechette,
M., Geyer,
W. R., andfor
Starczak,
R.:
TheC.
reduction
was significantly
more pronounced
slower V.bulence
intensities are relatively insensitive to clam pumping
O’Riordan, C.: The effects of near-bed hydrodynamic
Flume
on food-supply
bluesiphons
mussel
flowsExperiments
(a, c), and slightly
more so for to
thethe
raised
(c,Mytilusand siphon roughness, whereas vertical turbulence intensibivalve
filtration
rates,
Ph.d. thesis,
Stanford
Univer
Edulis-L
as
a
function
of
boundary-layer
flow,
Limnology
and
d). Increased clam pumping also enhanced the concentration
ties increase with
pumping
rate and
roughness.
This result
is
reduction, but 39,
the 1755–1768,
effect on the near-bed
O’Riordan,
C.,inMonismith,
S., and
Koseff,
Oceanography,
1994. concentrations was
consistent with
the increase
turbulent kinetic
energy
(TKE, J.: A stud
relatively
weak. Stronger
pumping L.,
produced
a larger conboundary-layer
over a bed of mo
turbulence
intensities in formation
all three directions)
Carlton,
J., Thompson,
J., Schemel,
and Nichols,
F.: Thethe
re-sum of thetration
centration
reduction
throughout
the Bay,
depthCalifornia
of the boundary
over beds ofand
shutOceanography,
and open mussels
van
Limnology
38,by
1712–1729,
199
markable
invasion
of San
Francisco
(USA) bydemonstrated
the
layer.
Duren
et
al.
(2006).
Our
measured
increases
in
Reynolds
Sobral,
P.,
Widdows,
J.:
Effects
of
increasing
current
AsianThe
clamnondimensional
Potamocorbula
amurensis,
Marine
Ecology
Progress
concentration fluctuations, c∗ c∗ ,
stress due to siphon roughness are qualitatively similar to
bidity and particle size selection on the feeding activ
Series,
66,
81–94,
1990.
were significantly larger for the slow flow cases relative to
measurements over mussels by Butman et al. (1994) and van
for growth
of Ruditapes
from
Cole, the
B., fast
Thompson,
J.,
and
Cloern,
J.:
Measurement
of
filtration
flows since there was less mixing to homogenize the
Duren et al. (2006).
However,
van Durendecussatus
et al. (2006)
did Ria Formo
concentration field (Fig. 17). For the slow flow cases (a, c),
where the excurrent jets penetrated farther into the flow, the
peak in the concentration variance was above the wall. The
www.biogeosciences.net/bg/0000/0001/
peak moved farther from the wall and decreased in magnitude as pumping increased. For the fast flow cases (b, d), the
peak concentration variance was at the tops of the siphons,
as the excurrent jets were bent over almost immediately by
the crossflow.
Turbulent fluxes of scalar concentration in the vertical direction, wc∗ , were always negative, meaning that mass (i.e.,
phytoplankton) had a net flux towards the bed as a result of
near-bed turbulent processes (Fig. 18). The fluxes tended towards zero at the bed and in the free stream, with a peak
near z=10 mm. The magnitude of the peak flux increased
approximately linearly with clam pumping, Q, and was also
significantly larger when the raised siphons were present. A
surprising result is that the fluxes were largely insensitive to
the mean free stream velocity U∞ .
Biogeosciences, 4, 269–282, 2007
not see any significant change in Reynolds stress for inactive versus actively feeding mussels; our results over model
Biogeosciences,
clams show an increase in Reynolds stress
as pumping ac-0000,
tivity increases, especially at slow crossflow velocities. This
disparity is likely due to functional differences in the feeding
mechanisms between the two species.
In a study over an array of artificial siphon mimics in a
natural channel, Jonsson et al. (2005) found that near-bed
Chl a concentration depletion increased with shear velocity. This finding went counter to the expectation (shared by
the authors of the study) that increased mixing at higher values of uτ would reduce Chl a depletion through enhanced
vertical mixing. The results of our study indicate that concentration depletion decreases dramatically with flow speed
(and thus uτ – see Fig. 11). However, increases in uτ due
to bed roughness had little effect on concentration depletion,
and we did indeed find situations where the concentration
depletion was larger for the raised-siphon case as compared
to the flush-siphon case, even though uτ was larger with the
siphons raised. It appears that uτ by itself may not be a reliable metric for concentration depletion.
www.biogeosciences.net/4/269/2007/
000
J. P. Crimaldi et al.: Mass and momentum fields over filter feeders
When mixing rates are low and phytoplankton replenishment to the bed is poor, the maximum concentration
depletion can be located some distance above the bed,
due to vertical momentum of the exhalant jets that moves
phytoplankton-depleted fluid away from the bed (Sobral and
Widdows , 2000; Widdows and Navarro , 2007). It has been
suggested that this scalar stratification is beneficial to the filter feeders since it would reduce the degree of re-filtrartation
of phytoplankton-depleted water. In the present study, the
maximum concentration depletion was always at the lowest
measured location. Presumably, this is due to the fact that our
flow conditions were always relatively energetic compared to
the previous studies.
Our study is the first of its kind to directly measure turbulent vertical mixing of mass above a bed of bivalves. The
profiles of wc∗ in Fig. 18 show that the peak turbulent flux
of mass to the bed is approximately 50% larger when the
clam siphons are raised. This increase in flux might be expected to decrease the near-bed concentration depletion, but
it does not (Fig. 16). One explanation is that the bivalves are
able to access higher concentrations by raising their siphons
(thus depleting more mass), and this effect overwhelms the
roughness-induced increase in turbulent mass flux to the bed.
This idea is supported by O’Riordan (1993), who found that
incurrent flows had a lower percentage of previously filtered
fluid when siphons were raised.
The turbulent mass flux measurements present a detailed
picture of how and where mass is transfered to the bed. For
our study, the turbulent mass flux goes to zero at approximately 100 mm. This height indicates the extent of the water
column that is directly impacted by the presence of the filter
feeders. Supply of mass to the bed from regions above this
distance would need to rely on large-scale turbulent structures that are not present in our flume. Below 100 mm,
mass is actively transported to the bed by organized momentum structures in the presence of the concentration boundary
layer. This flux is largest near z=10 mm. The magnitude of
this flux increases with siphonal pumping rate and roughness
due to raised siphons. Closer to the bed, the turbulent mass
flux wc∗ goes back to zero due to the hydrodynamic requirement that ww go to zero at the bed. However, this does not
mean that the overall mass flux is being reduced. Instead,
the mass flux in the near-bed region is accomplished through
mean processes associated with the steady (although spatially inhomogenous) siphonal currents. These fluxes are not
captured in wc∗ , but have been demonstrated by O’Riordan
(1993) and others.
On a final note, there are experimental simplifications in
this study that are not representative of real bivalve communities. These include (1) the uniformity of the model bivalve
array spacing and geometry, (2) spatially consistent siphon
heights, and (3) steady siphonal pumping rates. The effect
of these approximations is not known, and future research
would be valuable in these areas.
www.biogeosciences.net/4/269/2007/
5
281
Conclusions
In this paper we presented a set of measurements in a laboratory flume over a bed of model bivalves. The model bivalves
incorporated the effect of roughness in the form of raised
clam siphons, incurrent and excurrent flows, and siphonal
filtering of ambient scalar mass in the overlaying flow. We
measured profiles of velocity and mass concentration for different free stream velocities, clam pumping rates, and siphon
positions (flush or raised). The results show that clam pumping rates had a pronounced effect on a wide range of turbulent
quantities in the boundary layer. In particular, the vertical
turbulent flux of scalar mass to the bed was approximately
proportional to the rate of clam pumping. However, the formation of a concentration boundary layer above the clams
was only weakly sensitive to the pumping rate. When the bivalves pump more vigorously, the increased turbulent scalar
flux of phytoplankton towards the bed mitigates the decrease
in concentration of available food. The results demonstrate
an important mechanism whereby bivalves are able to effectively filter a broad depth range of the water column rather
than just re-filtering the layer of water adjacent to the bed.
Acknowledgements. This work was supported by the National
Science Foundation under Grant No. OCE950408. The authors
wish to acknowledge work done by C. A. O’Riordan on design and
construction of the model clam plates. The manuscript benefited
greatly from detailed editing by P. Jumars, and from a review by an
anonymous referee.
Edited by: E. Boss
References
Barrett, T.: Nonintrusive optical measurements of turbulence and
mixing in a stably-stratified fluid, Ph.D. thesis, University of California, San Diego, 1989.
Batchelor, G.: Small-scale variation of convected quantities like
temperature in turbulent fluid, J. Fluid Mech., 5, 113–133, 1959.
Butman, C. A., Frechette, M., Geyer, W. R., and Starczak,
V. R.: Flume Experiments on food-supply to the blue mussel
Mytilus-Edulis-L as a function of boundary-layer flow, Limnol.
Oceanogr., 39, 1755–1768, 1994.
Carlton, J., Thompson, J., Schemel, L., and Nichols, F.: The remarkable invasion of San Francisco Bay, California (USA) by
the Asian clam Potamocorbula amurensis, Mar. Ecol. Prog. Ser.,
66, 81–94, 1990.
Cole, B., Thompson, J., and Cloern, J.: Measurement of filtration
rates by infaunal bivalves, Mar. Biol., 113, 219–225, 1992.
Crimaldi, J.: Turbulence structure of velocity and scalar fields over
a bed of model bivalves, Ph.d. thesis, Stanford University, 1998.
Ertman, S. C. and Jumars, P. A.: Effects of bivalve siphonal currents
on the settlement of inert particles and larvae, J. Mar. Res., 46,
797–813, 1988.
Jonsson, P. R., Petersen, J. K., Karlsson, O., Loo, L. O., and Nilsson, S. F.: Particle depletion above experimental bivalve beds: In
Biogeosciences, 4, 269–282, 2007
282
J. P. Crimaldi et al.: Mass and momentum fields over filter feeders
situ measurements and numerical modeling of bivalve filtration
in the boundary layer, Limnol. Oceanogr., 50, 1989–1998, 2005.
Koseff, J., Holen, J., Monismith, S., and Cloern, J.: Coupled effects
of vertical mixing and benthic grazing on phytoplankton populations in shallow, turbid estuaries, J. Mar. Res., 51, 843–868,
1993.
Larsen, P. S. and Riisgard, H. U.: Biomixing generated by benthic
filter feeders: A diffusion model for near-bottom phytoplankton
depletion, J. Sea Res., 37, 81–90, 1997.
Monismith, S., Koseff, J., Thompson, J., O’Riordan, C., and
Nepf, H.: A study of model bivalve siphonal currents, Limnol.
Oceanogr., 35, 680–696, 1990.
O’Riordan, C.: The effects of near-bed hydrodynamics on benthic
bivalve filtration rates, Ph.d. thesis, Stanford University, 1993.
O’Riordan, C., Monismith, S., and Koseff, J.: A study of concentration boundary-layer formation over a bed of model bivalves,
Limnol. Oceanogr., 38, 1712–1729, 1993.
Biogeosciences, 4, 269–282, 2007
Sobral, P., Widdows, J.: Effects of increasing current velocity, turbidity and particle size selection on the feeding activity and scope
for growth of Ruditapes decussatus from Ria Formosa, southern
Portugal, J. Exp. Mar. Biol. Ecol., 245, 111–125, 2000.
O’Riordan, C., Monismith, S., and Koseff, J.: The effect of bivalve
excurrent jet dynamics on mass transfer in a benthic boundary
layer, Limnol. Oceanogr., 40, 330–344, 1995.
Spalart, P.: Direct simulation of a turbulent boundary layer up to
Reδ2 = 1410, J. Fluid Mech., 187, 61–98, 1988.
Tennekes, H. and Lumley, J.: A First Course in Turbulence, The
MIT Press, Cambridge, 1972.
van Duren, L. A., Herman, P. M. J., Sandee, A. J. J., and Heip,
C. H. R.: Effects of mussel filtering activity on boundary layer
structure, J. Sea Res., 55, 3–14, 2006.
Widdows, J. and Navarro, J. M.: Influence of current speed on clearance rate, algal cell depletion in the water column and resuspension of biodeposits of cockles (Cerastoderma edule), J. Exp. Mar.
Biol. Ecol., 343, 44–51, 2007.
www.biogeosciences.net/4/269/2007/