WEFTEC.

06

Modeling of Sedimentation in Stormwater Detention Basins

Masatsugu Takamatsu1), Michael Barrett2), and Randall J. Charbeneau3)
1. Doctoral Candidate, Dept. of Civ. Engr., The University of Texas, Austin, TX.
CRWR Pickle Research Campus Bldg 119, MC R8000 University of Texas, Austin, Texas 78712
2. Research Associate Professor, Dept. of Civ. Engr., The University of Texas, Austin, TX
3. Professor, Dept. of Civ. Engr., The University of Texas, Austin, TX
ABSTRACT
To minimize the non-point source pollution from roadway surface, the Texas Department of
Transportation (TxDOT) initiated a research project for design of a small footprint
non-proprietary detention basin, which consists of pipes and box culvert sections with
specialized inlet and outlet system. A conceptual model was developed to estimate particle
removal efficiency of the rectangular detention basin for the treatment of stormwater runoff by
extending ideal horizontal tank theory under the condition in which water level is varied. A
physical model was built in 1/5 scale of the prototype to measure its particle removal
performance and verify the conceptual model. Several experimental runs were conducted using
the physical model with steady inflow conditions, but various inflow rates, durations, and
suspended solid concentrations (SSC). Measured time series outflow SSC and particle removal
efficiency conformed closely to the calculated results from the conceptual model although the
measured particle removal efficiency in the physical model was always few percent better than
the predicted result using the conceptual model.
KEYWORDS
Conceptual modeling, sedimentation, stormwater, BMP, detention basin, removal efficiency.
INTRODUCTION
Treatment of stormwater runoff is required in many areas to reduce the discharge of pollutants.
One of the most common facilities installed to accomplish this treatment is an extended detention
basin. In this research project funded by the Texas Department of Transportation (TxDOT), a
simple and cost effective stormwater detention basin has been developed to treat polluted runoff
from highways. The detention basin is a dry basin composed of off the shelf materials including
pipe and box culvert sections, the schematic of which is shown in Figure 1.
Figure 1 - Schematic of rectangular stormwater detention basin

Water level is changeable

Orifice
Outlet Zone

Inlet Zone

Copyright 2006 Water Environment Foundation. All Rights Reserved

5559

WEFTEC.06

Water level is increased while highway runoff is flowing into the detention basin and decreased
by draining through outlet orifice at the other end of the basin. Water overflows an outlet weir
structure if the runoff volume is larger than basin capacity. Sedimentation occurs in the
sedimentation basin by gravitational force.
It is important to be able to estimate particle (or Suspended Solids (SS)) removal efficiency to
determine appropriate design parameters such as dimensions and drainage time. Many models
were proposed for stormwater facilities such as extended detention basins and wet ponds. A
methodology was suggested by EPA in 1986 for the analysis of detention basins for the control
of urban runoff pollution (Urbonas 1993). The methodology combines probability concept with
theoretical removal efficiency modeling concepts. For wet ponds, Su (2006) presented a removal
efficiency model using plug flow reactor (PFR) and continuous-stirred tank reactor (CSTR)
model to estimate the characteristics of first flush effect reduction. Luyckx (2005) proposed a
normalized particle efficiency and showed its strong relationship with Hazen number for several
CSO treatment facilities such as storage sedimentation tank, high side weir overflow, and vortex
overflow. However, none of these models deal with the change of water level in the basin
throughout a storm.
Therefore, the research objective was set as follows:
1) Develop a conceptual model for estimating particle removal efficiency for a rectangular basin
with vertical side slopes that incorporates unsteady inflow and outflow rate and
corresponding water level fluctuation.
2) Build a physical model to evaluate whether the conceptual model predicts the removal
efficiency.
3) Use the conceptual model to address typical design questions.
METHODOLOGY
CONCEPTUAL MODEL
The removal efficiency of particles with a known settling velocity in a primary settling basin can
be simply calculated by ideal horizontal flow reactor theory. A conceptual model to calculate
particle removal efficiency was developed in the same manner of the ideal horizontal flow
reactor theory by employing similar assumptions used in the theory. This section covers 1) how
the conceptual model was developed to calculate removal efficiency of particles and 2) how the
model can be applied in simple problems such as a runoff with constant inflow rate and SSC.
Detailed derivation and application can be available at CRWR online report 0605 (Takamatsu
2006).
1) MODEL DESCRIPTION
This section describes how a rectangular detention basin was modeled and how the modeling
process was similar and/or different from the ideal horizontal flow reactor theory. As shown in
Figure 1, stormwater runoff from a roadway surface flows into the basin through inlet pipes and
drains out of the basin through the outlet orifice. Water level increases when the inflow rate is
larger than the outflow rate and starts to decrease when inflow stops or outflow rate is larger than
inflow rate. Flow process here is very different from a primary settling tank in the following two
points: first, inflow and outflow rates are not always constant and they are not equal; second,
water level varies depending on the relative inflow and outflow rates.

Copyright 2006 Water Environment Foundation. All Rights Reserved

5560

WEFTEC.06

Next, assumptions used for the conceptual model are explained here. Water level is assumed to
be zero when runoff starts. Inflow SSC is assumed to be uniformly distributed over the depth of
the basin. Also, flow in the basin is assumed to be the plug flow and resuspension is negligible as
in the ideal horizontal flow reactor theory. Outflow rate is controlled by an orifice set at the
bottom of the end wall. Particles which settle before reaching the outlet zone are removed. Water
overflows the basin when the runoff volume is larger than the basin capacity. However, the
overflow condition is not investigated in this research.
First, start from a simple mass balance in the rectangular sedimentation basin. Water level change
can be determined as follows if both the inflow rate, Qin(t), and outflow rate, Qout(t), are known.
h(t) is the water level as a function of time, B is width, L is the length of the basin, and t
represents time.
dh ( t )
dt

1
BL

(Qin (t ) Qout (t ))

(1)

Mass balance was considered in a control volume at the right hand side of the detention basin
including the outlet orifice as shown in Figure 2. The local flow rate at x and t, Q(x,t), is equal to
the sum of the outflow rate, Qout(t), and upflow rate (or downflow rate when the tank is
emptying) within the control volume as shown in Equation (2).
Q( x , t ) = ( L x ) B

dh ( t )
dt

+ Qout ( t )

(2)

Figure 2 - Local convectional velocity at storm runoff period

(L x )
dh

dh ( t )
dt

h(t)
Qout(t)

Q(x,t)
u(x,t)

The corresponding local horizontal flow velocity at x and t, u(x,t), is Q(x,t) divided by the
submerged cross sectional area at t as shown in Equation (3). Vertical flow velocity at the water
surface is the velocity of water level change, dh(t)/dt , and 0 at bottom. Vertical velocity in
between, v(z,t), is linearly distributed over depth;

Copyright 2006 Water Environment Foundation. All Rights Reserved

5561

WEFTEC.06

u (x, t ) =
v ( z, t ) =

Q( x , t )
B h(t)
dh ( t )
dt

(3)

h(t)

Particle is known to have a constant settling velocity, vs, (e.g., Stokes velocity). Velocity vector
of particles can be specified as follows where up(x,t) is longitudinal particle velocity and vp(z,t)
is downward particle velocity.
u p (x, t ) = u (x, t )
v p ( z, t ) = v ( z , t ) v s

(4)

Critical settling velocity is the velocity where all the particles with higher settling velocities will
settle out in a reactor. In the ideal horizontal flow reactor, critical settling velocity is simply the
overflow velocity which is a constant flow rate divided by the surface area. However, in this
rectangular detention basin, calculating critical settling velocity was not that simple because of
unsteady inflow and outflow rate and variable water level. Figure 3 shows the schematic of
trajectories of water molecules and a particle which has a critical settling velocity in an ideal
horizontal flow reactor and the rectangular detention basin. Trajectories of a water molecule and
a particle are curved while the trajectory of a particle in an ideal horizontal flow reactor is
straight.
Figure 3 - Trajectories of water molecules and a particle with critical settling velocity in an ideal
horizontal flow reactor (above) and rectangular stormwater detention basin during the filling
period (below)
Water molecule trajectory
SS particle trajectory with critical settling velocity

Trajectories, or pathlines, of particles were calculated from the velocity field. A pathline is a line
which is traced out in time by a given fluid particle as it flows, while a streamline is an
instantaneous line whose tangents are everywhere parallel to the velocity vector (Currie 1974). A
pathline is calculated using the following equation.
Copyright 2006 Water Environment Foundation. All Rights Reserved

5562

WEFTEC.06

dx

= u (x, t )
dt
dz
= v( z, t )
dt

(5)

The benefit of using the pathline concept in this study is the position of a particle can be traced
with time. In this section, both water molecules and particle pathlines are described. Longitudinal
and vertical position of a water molecule can be derived in (6) by integrating Equation (5) after
the substitution of Equation (3) and (4).
x w ( t ; t in ) =
z w ( t ; t in ) =

t
Qin ( t )dt
B h ( t ) tin
h(t )

(6)

z ( t in ; t in )
H ( t in ) w

where xw(t; tin) and zw(t; tin) is the longitudinal and vertical position of the traced water molecules
at time t, flowing into the tank at time tin respectively. In the same manner, pathline equations for
a particle can be derived as follows using particle own settling velocity, vs.
x p ( t ; t in ) = x w ( t; t in )

t 1 dt + z p ( t in ; t in )
z p ( t ; t in ) = H ( t ) vs tin

h(t)
h ( t in )

(7)

Finally, detention time of the particle is calculated here. Since the position of the suspended solid
particle, (xp(t;tin), zp(t;tin)), is known, time to reach the end of the tank, tout, for the particle can be
calculated by substituting xp(t;tin)=L in Equation (7) and solving for tout. Detention time of a
particle can be simply calculated from tout- tin. As shown in Figure 3, a particle with critical
settling velocity, which enters the basin at the very top of the water column, will settle to the
bottom on the right hand side when reaching the outlet. Therefore, the focus was on a particle
released from the water surface at the inlet. Initial vertical position of such a particle is the same
as the water level at tin, or simply h(tin). Therefore, vertical position of the particle is shown in
equation (8) by substituting zp(tin;tin)=h(tin) in (7), the pathline equation of a particle.

t 1 dt + 1
z p , top ( t ; t in ) = h ( t ) vs tin

h(t )

(8)

Then, the minimum particle which meets z p , top ( t out ; t in ) > 0 establishes the critical settling
velocity, vs,c ( t in ) as shown in equation (9).
vs,c ( t in ) =

1
tout 1 dt
tin
h(t)

Copyright 2006 Water Environment Foundation. All Rights Reserved

5563

(9)

WEFTEC.06

Some particles settle out, but some dont if the settling velocity of a particle is smaller than
critical velocity. Figure 4 shows the trajectories of a particle with lower settling velocities. The
figure shows a particle released from the water surface that didnt settle out, but the other particle
released from the middle of the water level just settled out at the outlet. This implies that
particles with higher initial position than this particle will all escape, and those with lower initial
position will all settle out. The ratio of settling can be calculated as described below.
Figure 4 - Trajectory of a particle with lower settling velocity than critical settling velocity and
the particle ratio of captured and escaped
Water molecule trajectory from surface
SS particle trajectory from surface
SS particle trajectory which runs into right bottom

Ratio escaped
Ratio captured
First, the pathline equation can be calculated for a particle which exactly runs into the bottom
orifice by putting (t,zp)=(tout,0) into the particle pathline equation (7). Then the settling ratio can
be calculated by solving

z p ( t in ; t in )
h ( t in )

since z p ( t in ; t in ) is the maximum initial height for

particles to settle out. The ratio captured can be simplified as the ratio of settling velocity to the
critical settling velocity.
z p ( t in ; t in )
h ( t in )

vs
tout 1 dt =
= v s tin
h(t )
v s, c ( t in )

(10)

The ratio captured, also called reactor settling potential function (Lawler 2006), is shown in
equation (11), and this equation suggests that all particles with settling velocity larger than
vs
critical settling velocity will settle out. The fraction,
, will settle out for particles with
v s ,c ( t in )
settling velocity smaller than critical settling velocity.
1 for vs > vs,c ( t in )

F( vs , t in ) = vs
v ( t ) for vs < vs,c ( t in )
s,c in

Copyright 2006 Water Environment Foundation. All Rights Reserved

5564

(11)

WEFTEC.06

Assume the settling velocity of inflow particles has its own probability density function, e(vs).
Then, removal ratio for each inflow time, Rout(tin), can be calculated as a function of inflow time.

R out ( t in ) = F( vs ) e( vs )dvs

(12)

Overall removal ratio of suspended solids can be calculated by the total mass of particles flowing
out divided by total mass of particles flowing in the basin as follows:

R=

Ts

Q( t in ) C( t in ) R out ( t in )dt in

Ts

Q( t in ) C( t in )dt in

(13)

where Ts represents duration of storm runoff.

Time series outflow SSC can be estimated by equation (14).
C( t out ) = C( t in ) (1 R out ( t in ) )

(14)

2) CONCEPTUAL MODEL APPLICATION

An example problem and its solution using the developed conceptual model are shown in this
section to show how the conceptual model could be applied. The problem is a simple example
which has a constant inflow rate and SSC. The values shown below are taken from the run A,
one of the actual runs using the physical model described shortly.

Assume that there is constant inflow of Qin=0.53 L/s for the duration of Ts=40 min and the
inflow SSC is Cin=202 mg/L. Outflow is governed by a small orifice installed at the bottom of
the end wall. The effective area of the orifice is Ae=0.43 cm2, the multiplication of orifice
coefficient and the area of orifice. Assume the particle size distribution of inflow suspended solid
can properly described by lognormal distribution and the mean of ln(dp), p, is 2.286 (dp (m))
and the standard deviation of ln(dp), p, is 0.908 (dp (m)). Buchan (1989) suggests that soil
particle size distribution can be well described by lognormal distribution. Table 1 summarizes the
input parameter values used in the example.

Copyright 2006 Water Environment Foundation. All Rights Reserved

5565

WEFTEC.06

Table 1 - Input parameters, their units, and values

Parameters
Ae
L
B
Qin
Ts
Cin
p
p
p
w

Unit
2
cm
m
m
L/s
min
mg/L

g/cm
3
g/cm
g/cm/s

Value
0.43
6.96
0.62
0.53
40
202
0.908
2.286
2.65
1
0.01

Stokes law was applied to calculate particle settling velocity, vs m/hr, from the diameter of a
particle, dp m.
vs =

)(

2 3600
g p w
d p *10 4 *
18
100

(15)

= 3.24 *103 d p 2

For the parameter values in this equation, gravitational acceleration, g, is 981cm/s2, density of
particles, p, is 2.65 g/cm3, density of water, w, is 1.00 g/cm3, and viscosity of water is 10-2
g/cm/s. The random variable transformation was used to describe the distribution of settling
velocity. Settling velocity will have a lognormal distribution vs=LN(v,v) when particle size has
a lognormal distribution dp=LN(p,p). The mean of lognormal settling velocity, v, is -1.186 and
the standard deviation of lognormal settling velocity, v, is 1.816 using the formula for random
variable transformation. Figure 5 shows Particle Size Distribution (PSD) and settling velocity
distribution of inflow runoff in the forms of a Cumulative Density Function (CDF).

Copyright 2006 Water Environment Foundation. All Rights Reserved

5566

WEFTEC.06

Figure 5 - CDF of particle size distribution (left) and settling velocity (right)

The graph shows that about 80% of particles are smaller than 20m while the particle size ranges
from 0 to 50m, and settling velocity of more than 70% of particles is less than 1m/hr.
The example problem was solved by writing a code in MATLAB. How the MATLAB code was
built is described here. Mass balance equation (1) was rewritten by substituting an orifice
equation for outflow rate.
dh ( t )
dt

1
Q A e 2gh ( t )
B L in

(16)

This was discretized and explicitly solved. Calculated water level change, given inflow rate, and
calculated outflow rate are shown in the Figure 6. As the graph shows, it takes about 6 hours to
drain all the water after runoff stops and the drainage time is about 9 times longer than the 40
minute runoff period.

0.3

0.6

0.25

0.5
Flow rate (L/s)

Water level (m)

Figure 6 - Calculated water level change in the sedimentation basin (Left) and Inflow and
outflow rate (Right)

0.2
0.15
0.1
0.05

Outflow
Inflow

0.4
0.3
0.2
0.1

0
0

Time (hr)

Time (hr)

Copyright 2006 Water Environment Foundation. All Rights Reserved

5567

WEFTEC.06

An imaginary water column is released from the inlet every time step, dt, from the beginning to
the end of the runoff. The water column travels through the sedimentation basin and eventually
reaches the outlet. Time to reach outlet, tout, corresponding to each inflow time, tinn, was
calculated by substituting xp(tout;tin)=L in Equation (7).
L=
=

1
B h ( t out )

tout
tin Qin dt

(t out t in ) Qin

(17)

B h ( t out )

End of the travel time, tout, of a water column which started traveling at tin, can be calculated
iteratively from the equation (17). Figure 7 shows the calculated detention time, tout-tin of water
columns released at tin. The graph shows that traveling time or detention time of the first quarter
(0 to 10 minutes) is less than an hour while detention time of the last quarter (30-40 minutes) is
around 4 hours.
Figure 7 - Theoretical detention time of water columns with respect to their inflow time
Theoretical detention time (min

350
300
250
200
150
100
50
0
0

10

20

30

40

Inflow time (min)

Critical settling velocity, vp,c(tin), was calculated from equation (9). The integral, in the
denominator of equation (9), was calculated using the MATLAB function trapz, which
conducts a trapezoidal numerical integration. Figure 8 shows the critical particle size change
with respect to the inflow time, tin. This implies that particles larger than 8 m would all settle
throughout the runoff-drainage process.

Copyright 2006 Water Environment Foundation. All Rights Reserved

5568

WEFTEC.06

Critical particle size (m)

Figure 8 - Critical particle settling velocity change with respect to inflow time

9
8
7
6
5
4
3
2
1
0
0

10

20

30

40

Inflow time (min)

As shown in equation (12), PDF of settling velocity, e(v), and potential settling function, F(v),
should be known to calculate the overall particle removal ratio. PDF of settling velocity, e(v)
can be described as follows using the calculated mean and standard deviation value of natural
logarithm of particle diameter, ln(dp):
e( v s ) =

1
e
v s v 2

1 ln( v s ) v

2
v

(18)

Settling potential function, F(vs), can be calculated from the calculated critical particle velocity.
For this application case, Stokes velocity, shown in equation (15), can be put into (11) and can be
further transformed as follows:
1 for vs > vs,c ( t in )

2
F( vs ) = v
dp
s
for vs < vs,c ( t in )
=

v
(
t
)
d
(
t
)
s,c in
p,c in

(19)

This settling potential function varies depending on inflow time. This function can be visually
understood by Figure 9. This graph shows the particle size distribution of inflow and the fraction
removed in the form of Probability Density Function (PDF). The group of particles shown in the
graph flowed into the detention basin at tin=5 min and flowed out at tout=40 min. Calculated
critical particle settling velocity, vs,c, is 0.195 m/hr and the corresponding critical particle size,
dp,c, is 7.8 m. The area surrounded by the solid and dotted lines corresponds to the fraction
flowing out, and the area under the dotted line is the fraction settled out.

Copyright 2006 Water Environment Foundation. All Rights Reserved

5569

WEFTEC.06

Figure 9 - Particle size distribution of inflow suspended solid and the fraction removed.
0.08
0.07
0.06
PDF

0.05

Original PDR
Treated fraction

0.04
0.03
0.02
0.01
0.00
0

10

20
30
40
Diameter (micron)

50

60

Equation (19) and (18) was put into (12) to calculate outflow SSC and particle removal
efficiency. Figure 10 shows the calculated time series outflow SSC. This graph shows the
outflow SSC gradually increased until approximately 50 min, which is 10 minutes after runoff
stopped. Then, the concentration dropped steadily with increasing time until water was
completely drained. The calculated total particle removal efficiency was 87.5%.
Figure 10 - Time series inflow and outflow SSC change
250

SSC (mg/L)

200
150
100
50
0
0

20

40

60

80

100 120 140 160 180 200 220 240 260 280 300 320 340
Time (min)

Timeseries outflow SSC from the outlet may not be necessary to be estimated for stormwater
management purpose since the total particle removal ratio is more important in the general case.
However this model can determine the particle size distribution change in the outflow. This
implies that this study could be extended for estimating removal ratio of other important
pollutants in stormwater such as nutrients, heavy metals, and organic materials in particulate
form since hydrophobic pollutants are likely to attach to the surface of suspended solids.

Copyright 2006 Water Environment Foundation. All Rights Reserved

5570

WEFTEC.06

Adsorption is usually highly dependent on the surface area of sediment, which can be estimated
from the particle size distribution.
PHYSICAL MODEL
To verify the conceptual model, a 1:5 scale (model:prototype) physical model was built with the
dimensions approximately 24 feet long, 2 feet wide, and 1.2 feet deep. The full basin will drain
in 4.8 hours, which corresponds to 24 hour drainage time for the prototype after receiving 1
inche of runoff from a 1 acre roadway surface. The scaling was followed by Hazens law which
states that sedimentation is controlled by the overflow rate (Thompson 1969). Artificial silica
particles, which have diameters ranging from 0 to 50 m, were used as suspended solids and
injected from a mixing tank. During experiments, water level was recorded by bubble flow meter
every minute and water samples were taken from inflow and outflow periodically for measuring
SSC. Figure 11 shows a picture of the physical model.

Figure 11 - 1/5 scale physical model of a dry detention basin

Particles were delivered to the system using a 5 gallon mixing tank and peristaltic pump with
flow rate controller.
Here is the experimental procedure that took place at each run.
1. Target inflow rate, runoff duration, and inflow SSC were determined prior to each run.
2. Determine SSC in the mixing bucket and the speed of peristaltic pump.
3. Inflow runoff with a given SSC was introduced to the sedimentation basin and maintained for
a predetermined duration.
4. Samples were taken periodically from the inlet and outlet using 250 mL polyethylene bottles.
Inflow samples were taken periodically so that at least 4 or 5 samples can be taken. Outflow
samples were taken periodically (5 to 10 minutes interval) from the time runoff started until
10 to 20 minutes after the runoff stopped. After that, intervals of taking samples became
longer, up to 30 minutes, since the SSC change was small.

Copyright 2006 Water Environment Foundation. All Rights Reserved

5571

WEFTEC.06

5. Water level in the basin was recorded every minute for the entire runoff and drain process
using a bubble flow meter (ISCO 3230). Acquired data were used to calculate inflow and
outflow rates.
6. SSC of each sample was measured by the filtration method after the entire process was
finished.
RESULTS AND DISCUSSION
Five runs were conducted using the physical model with different outlet orifice sizes. Table 2
shows the experimental conditions of all these runs. Run A is the example used in the section of
conceptual model application.

Table 2 - Setup conditions and inflow hydraulic conditions

Effective area of
orifice, Ae (cm2)
0.43
0.43
0.48
0.48
0.48

A
B
C
D
E

Basin length, Inflow rate,

L (m)
Qin (L/s)
6.96
0.53
6.96
0.32
6.96
0.9
6.96
1.17
6.96
0.73

duration, Ts
(min)
40
80
20
15
30

Theoretical overflow Mean inflow

rate, Qin/BL (m/hr) SSC, Cin (mg/L)
0.44
202
0.27
267
0.75
246
0.98
258
0.61
212

Figure 12 shows the measured and calculated water level for run A and B among these five runs.
These graphs show that water level was appropriately calculated with equation (16). The
calculated results for the rest three runs are as good as them.
Figure 12 - Measured and calculated water level for four runs.
(A ) Q in = 0 .5 3 (L / s), T s= 4 0 (m in ), lo n g basin

(B ) Q in = 0 .3 2 (L / s), T s= 8 0 (m in ), lo n g basin

M e asu re d

0 .2 5

C alc u late d

0 .2

M e asu re d

0 .3

0 .1 5
0 .1
0 .0 5

W ate r le ve l (m )

W ate r le ve l (m )

0 .3

0 .2 5

C alc u late d

0 .2
0 .1 5
0 .1
0 .0 5

0
0

50

100

150

200

250

T im e (m in )

50

100

150

200

250

T im e (m in )

Figure 13 shows the measured inflow SSC, and measured and calculated outflow SSC for these
five runs. Most of them show the calculated time series outflow SSCs are close to the measured
results while the deviations in run B is not very small. A possible reason to explain the deviation
is the inflow SSC. Although inflow SSCs were tried to be constant throughout the runoff
duration, some of them fluctuated to some extent as shown in the graph especially run B. This
might affects the conceptual model results since average inflow SSC values were used for the
conceptual model inputs. Some of the intense runoffs such as Run C and D have high outflow
SSCs at the beginning of the runoffs. This happened because of the resuspention effect and it was
not considered in the conceptual model.

Copyright 2006 Water Environment Foundation. All Rights Reserved

5572

WEFTEC.06

Figure 13 - Measured inflow and outflow SSC, and calculated outflow SSC
B: Qin=0.32 (L/s), Ts=80 (min)

A: Qin=0.53 (L/s), Ts=40 (min)

200
150

Inflow
Outflow(measured)

100

Outflow(calculated)

SSC (mg/L)

SSC (mg/L)

250

50
0
0

50

100
150
Time (min)

200

Outflow (measured)
Outflow (calculated)

100

150

200

250

D: Qin=1.17 (L/s), Ts=15 (min)

300

Inflow

50

Time (min)

SSC (mg/L)

SSC (mg/L)

Inflow
Outflow (measured)
Outflow (calculated)

250

C: Qin=0.9 (L/s), Ts=20 (min)

300
250
200
150
100
50
0

350
300
250
200
150
100
50
0

Inflow
Outflow (measured)

250
200

Outflow (calculated)

150
100
50

50

100

150

200

250

Time (min)

50

100

150

200

250

Time (min)

SSC (mg/L)

E: Qin=0.73 (L/s), Ts=30 (min)

300
250
200
150
100
50
0

Inflow
Outflow (measured)
Outflow (calculated)

50

100

150

200

250

Time (min)

Figure 14 shows the Measured and calculated particle removal efficiencies for all these five runs
with the order of theoretical overflow rate. The graph shows that calculated results
underestimated the removal efficiency for a few percent, but the trend is consistent.

Copyright 2006 Water Environment Foundation. All Rights Reserved

5573

WEFTEC.06

Figure 14 - Measured and calculated particle removal efficiencies

Particle removal efficiency

1.00
0.90
0.80
0.70

Measured
Conceptual

0.60
0.50
A

Runoffs

DESIGN QUESTIONS
The developed conceptual model can also answer important design questions.
1. How does the density of suspended solids affect removal efficiency?
To answer this question, particle removal efficiency was calculated for particles with different
densities, and simulations with different flow rates were compared while keeping the runoff
volume the same. Of primary interest are particles with densities less than 2.65 g/cm3. Figure 15
shows the removal efficiency of the particles with the same particle size distribution used in this
research but different densities ranging from 1.2 to 2.65 g/cm3, using the assumption that density
is uniform for all the particle sizes. Removal efficiency is lower if density is lower, and the
decrease is more significant for lighter particles. The density of smaller particles in stormwater
runoff can range from 1.1 g/cm3 (Christina 2001) for typical highway organic soil to around 3.0
g/cm3 for larger particles (Zanders 2005). In Austin, TX, the density of particles smaller than 75
m ranged from 1.4 to 2.4 g/cm3 with the mean of 1.8 g/cm3 for four different storm samples
from a highway, and 1.4 to 2.8 g/cm3 with the mean of 1.8 g/cm3 for three different storm
samples from the inlet of an extended detention basin (Karamalegos 2006). When combined with
the influence of density on removal efficiency, these results suggest that characteristics of
particle density should be considered in analysis and design of stormwater treatment systems.

Copyright 2006 Water Environment Foundation. All Rights Reserved

5574

WEFTEC.06

Figure 15 - Calculated removal efficiency changes with different particle densities and different
1
Removal ratio

0.8
0.6

Q=0.5 (L/s)
Q=1 (L/s)
Q=2 (L/s)

0.4
0.2
0
1.0

1.5

2.0
2.5
Density (g/cm3)

3.0

2. How does drainage time affects removal efficiency?

The process to determine the drainage time is equivalent to the process used to determine the
outflow orifice size. The theoretical drainage time, td (hr), is defined as the time to drain all the
water from full (h=H(m)) to empty.
Ae =

B L 2h
td
g

( )

0.62 * 6.96 *104

2 * 0.34
1
3.16
A e cm2 =
*

=
3600
9.81 t d (hr) t d (hr)

(20)

Then, removal efficiencies of particles were calculated for four different orifice sizes (0.41, 0.82,
1.23, and 1.64 cm2), which correspond to theoretical drainage times of 7.7, 3.9, 2.6, and 1.9
hours for three different flow rates (Qin=0.5, 1.0, and 2.0 L/s). However, the runoff volume is the
same for all cases. The calculation results are shown in Figure 16. Removal efficiency decreased
significantly with increasing orifice size when compared to increasing flow rate.

Copyright 2006 Water Environment Foundation. All Rights Reserved

5575

WEFTEC.06

Figure 16 - Removal ratio change with different effective orifice sizes

Removal efficiency

1
0.8
Q=0.5 (L/s)
Q=1 (L/s)
Q=2 (L/s)

0.6
0.4
0.2

Drainage time (hr)

6.3
3.3
2.1

0
0

1.6

0.5
1
1.5
Effective orifice area (cm2)

These analyses are very important for designing a detention basin because one can determine
appropriate orifice size from design storm and target particle removal efficiency. Also, these
calculation results can be scaled up to the prototype scale using Hazen number scaling, which is
not discussed much in this paper.
CONCLUSIONS
A conceptual model used to calculate outflow SSC and particle removal efficiency was
theoretically developed based on the assumptions and concepts that are used in the ideal
horizontal flow reactor theory, and a program to apply the theory to simple cases was developed
in MATLAB.

To see if the conceptual model can estimate outflow SSC and particle removal efficiency, a
physical model was built and five runs were conducted with different orifice sizes and runoff
conditions. Outflow SSCs were accurately calculated by the conceptual model while particle
removal efficiencies were underestimated by a few percent for all cases.
Several simulations were conducted with the conceptual model to answer two typical design
questions. When the drainage time is decreased, by increasing the area of outlet orifice, the
removal efficiency will decrease. Also, particle removal efficiency was calculated when particle
density is lower than 2.65 g/cm3. The decrease of particle density will affect removal efficiency
especially when the density is smaller than 2 g/cm3, which is likely for smaller particles with a
high organic carbon fraction.
ACKNOWLEDGEMENT
The financial support for this project was received from Texas Department of Transportation
(TxDOT). The writers are thankful to TxDOT for their financial support and advisory.

Copyright 2006 Water Environment Foundation. All Rights Reserved

5576

WEFTEC.06

REFERENCES:
Buchan, G.D. (1989). Applicability of the simple lognormal model to particle-size distribution
in soils. Soil Sci. 147:155161.
Camp, T. R. (1946). Sedimentation and the design of settling tanks. Trans. ASCE, 111(2285),
895-936.
Cristina, Chad., Tramonte Jarrod., Sansalone John J. (2002). A granulometry-based selection
methodology for separation of traffic-generated particles in urban highway snowmeld runoff.,
J. Water, Air, and Soil Pollution, 136: 33-53..
Currie, I.G. (1974). Fundamental Mechanics of Fluids., McGraw-Hill, Inc, New York.,
pp38-39.
Lawler, D. F., and Benjamin, M.M. (2006). "Gravity Separations," Chapter 13 (draft) in "Water
Quality Engineering: Physical-chemical Treatment Processes." to be published by
McGraw-Hill, New York.
Luyckx, G., Vase, G., and Berlamont, J. (2005). Solids separation efficiency of combined sewer
overflows., Water Science and Technology Vol51 No2 pp71-78. IWA Publishing 2005.
Su, Y. and Mitchell, G.F. (2006). Characteristics of First Flush Effects in Storm Water Runoff
via a Retention-Detention System., TRB 2006 Annual Meeting CD-ROM.
Takamatsu, M. (2006). Physical and Conceptual Modeling of Sedimentation Characteristics in
Stormwater Detention Basins. CRWR online report 0605.
http://www.crwr.utexas.edu/online.shtml
Thompson, D. M. (1969). Scaling laws for continuous flow sedimentation in rectangular tanks.,
Proc. Instn civ. Engrs, 43, May, pp. 453-461.
Urbonas, Ben. and Stahre, Peter. (1993). Stormwater-Best management practices and detention
for water quality, drainage, and CSO management., PTR Prentice Hall, Englewood Cliffs,
New Jersey.
Zanders, J.M. (2005). Road sediment: characterization and implications for the performance or
vegetated strips for treating road run-off. J. Science of Total Environment., ELSEVIER, 339,
41-47.

Copyright 2006 Water Environment Foundation. All Rights Reserved

5577