Indepth diagnosis of a secondary clarifier by the

Water Science & Technology Vol 54 No 8 pp 83–92 Q IWA Publishing 2006

application of radiotracer technique and numerical
modeling
H.S. Kim*, M.S. Shin*, D.S. Jang* and S.H. Jung**
*Department of Environmental Engineering, Chungnam National University, 220 Gungdong Yusunggu,
Daejeon 305-764, Korea (E-mail: [email protected]; [email protected])
**Korea Atomic Energy Research Institute, P.O.Box 105, Yuseong, Daejeon 305-353, Korea
(E-mail: [email protected])

Abstract To make an indepth diagnosis of a full-scale rectangular secondary clarifier, an experimental and
numerical study has been performed in a wastewater treatment facility. Calculation results by the numerical
model with the adoption of the SIMPLE algorithm of Patankar are validated with radiotracer experiments.
Emphasis is given to the prediction of residence time distribution (RTD) curves. The predicted RTD profiles
are in good agreement with the experimental RTD curves at the upstream and center sections except for the
withdrawal zone of the complex effluent weir structure. The simulation results predict successfully the well-
known flow characteristics of each stage such as the waterfall phenomenon at the front of the clarifier, the
bottom density current and the surface return flow in the settling zone, and the upward flow in the exit zone.
The detailed effects of density current are thoroughly investigated in terms of high SS loading and
temperature difference between influent and ambient fluid. The program developed in this study shows the
high potential to assist in the design and determination of optimal operating conditions to improve effluent
quality in a full-scale secondary clarifier.
Keywords Density current; density waterfall; residence time distribution; SS; temperature

Introduction
Nowadays, the wastewater treatment plant (WWTP) is the one of the significant facilities
inevitably required for the conservation of water quality due to the rapid industrialization
in Korea. Since the end of 2003, 242 facilities have been operated nationwide (Korea
Ministry of the Environment, 2004). The installation plan is scheduled to handle about
80% clarification of sewage generated from the municipal regions. However, most of the
WWTPs are over-designed excessively large and further not operated effectively accord-
ing to the recent statistics of the Korea Ministry of Environment. It is time to evaluate
seriously the performance of the WWTP process because of the cost penalty increase for
sewage treatment.
The activated sludge process is the widely used form of biological wastewater
treatment. Activated sludge has a two-step process related with the biological oxidation
and the solids separation. Following the oxidation step, mixed liquor suspended solids
(MLSS) from mixed liquor is separated by gravity in a secondary clarifier, which is a
common and extensively used application process in WWTP. A portion of the settled
biological solids in the clarifier is returned to the aeration tank to maintain an appropriate
MLSS concentration. The role of clarification is very important because the effluent
quality is directly affected by the separation of the biological solids. Thus, improvement
of solid separation in a secondary clarifier will have a great impact on effluent quality,
considering the more and more stringent effluent quality standard.
doi: 10.2166/wst.2006.857 83
 Design of clarifiers is mainly based on the parameter of the surface overflow rate
(Q/Ac, where Q is the flow rate and Ac is the clarifier cross-sectional area) of the tank.
This design variable assumes the uniform plug flow through the tank. Most commonly
used design criteria incorporate this relatively simple concept. However, the feature of
many full-scale sedimentation tanks does not follow the ideal flow behavior because sus-
pended solids removal in a clarifier is not only a function of overflow rate but also other
variables such as density flow, turbulence and multi-dimensional features of flow, etc.
H.S. Kim et al.

In more detail, the degree of non-uniformity is known as a complex function of the

characteristics of the SS (particle size, density, settling velocity, and concentration), the
flow field (inlet flow velocity, flow rate, and turbulence intensity), the geometrical
configuration of a clarifier and meteorological conditions (McCorquodale and Zhou, 1993;
Stamou, 1995). In order to assist in advanced design for sedimentation tanks the effect of
the variables mentioned above on the hydraulics of tank flow must be carefully evaluated.
The causes of density currents formed in a clarifier have been still debated in a lively
fashion due to the complicated flow pattern. Therefore, residence time distribution (RTD)
curves were measured with the application of a radiotracer in order to elucidate the
internal flow phenomena in a full-scale secondary clarifier in WWTP. This real time
RTD curve can provide various important hydrodynamic parameters but it is not enough
to visualize and resolve the detailed flow pattern inside the system. For that reason
numerical modeling is performed to figure out the flow pattern and the distribution of
SS concentration. The special emphasis of this study is given mainly to analyze the
fundamental mechanisms of flow caused by density current in a secondary clarifier with
high SS loading.

Experiments
Experiments were carried out in the full-scale rectangular clarifier with a surface area of
120 m2 and a depth of 3.38 m at Suyoung wastewater treatment plant in Busan, Korea.
RTD was measured by making use of I-131 (8.2 mCi as liquid phase) as a tracer.
I-131 (8.04 days) is routinely produced from HANARO, the research reactor in Korea
Atomic Energy Research Institute (KAERI), in order to supply domestic hospitals
for medical purposes. It may have too long a half-life for this application but it is
advantageous in that I-131 is in anionic form in NaI solution and it does not need to be
processed any further to make it stable in the water and easy to handle due to relatively
low gamma energy.
A radiation detector was developed for practical application in the field because the
most popular scintillation detector (NaI) required a heavy lead shield to eliminate back-
ground radiation from the surrounding environment. The radiation absorbed into detectors
is converted into the scintillation light and the light is amplified by a photo-multiplier
tube. Each detector is coupled with a rate meter, which counts the number of pulses
generated by a detector and transmits the result to a computer for data recording. The
result is converted into a RTD by background subtraction, possible half-life correction
and normalization.
Detection was performed at 20 points in the full-scale clarifier, including the inlet and
outlet opening. The tracer detection was measured far away from the centreline because
the scrapper circulated the whole clarifier to remove scum on top and settling solids on
the bottom. Figure 1 shows the schematic of a full scale secondary clarifier and radiotra-
cer detection points for measurements.
Table 1 shows the flow rate introduced into clarifiers during the experiment. The flow
rate was fluctuated as time and the average theoretical velocity was calculated as
84 0.0023 m/sec, the value of which was used to evaluate the strength of density currents.
 H.S. Kim et al.
Figure 1 The location of the detector for RTD study in the secondary clarifier

In the field the average MLSS and effluent SS concentrations are respectively about
1,300 mg/L and 5 mg/L.
The temperature between the tank inlet and outlet was measured together with the
on-site atmosphere temperature because surface heat transfer would become significant
by the effects of solar radiation. During the experiment, solar radiation on the water
surface induced temperature increases of 0.5–1.0 8C between the tank inlet and outlet.
Solar radiation in summer can eventually produce the temperature increase inside the
clarifier, which causes the density current and likewise SS loadings.

Numerical method
Governing equation and solution
The flow patterns for density-stratified fluids in a secondary clarifier are usually different
from those simple fluid flows of plug type with uniform density. Even if the relative
importance of inertial and gravity forces in the clarifier can be simply characterized in
terms of the inlet momentum and buoyancy flux defined by the Froude number (Fr), a
computer model for the resolution of complex stratified flow requires a set of conserva-
tion equations for continuity and momentum, turbulent kinetic energy k, dissipation of
turbulence energy 1 and solids concentration (Patankar, 1980). These can be expressed as
 
›ðrfÞ › › ›f
þ ðruj fÞ ¼ Gf þ Sf ð1Þ
›t ›xj › xj ›xj

where f denotes general dependent variables expressed as a physical quantity per unit
mass. Further, u, v, r, Gf and Sw stand for x, y velocity components, density, turbulent
diffusion coefficient and source term corresponding to f, respectively.

Table 1 Influent flow rate during experiment

Date Hour Flow rate to 36 clarifiers Avg. theoretical velocity

(m3/3 hrs) (m/s)

1st day 15:00 –18:00 23,000 0.002329

18:00 –21:00 26,600 0.002694
21:00 –24:00 28,100 0.002846
2nd day 00:00 –03:00 21,400 0.002167
03:00 –06:00 17,500 0.001772
06:00 –09:00 16,100 0.001631
09:00 –11:00 23,300 0.002360
Avg. theoretical velocity* (m/s) 0.002257
*Average theoretical flow through velocity, U ¼ Q / WD 85
 The concentration equation for suspended solid is
   
›C ›C i ›C i › ›Ci › ›C i
þu þv ¼ nsx þ nsy þ V s Ci ð2Þ
›t ›x ›y ›x ›x ›y ›y

in which nsx and nsy are eddy diffusivity of suspended solids in the x- and y-direction;
and Vs is particle settling velocity. The eddy viscosity is calculated from the standard k-e
H.S. Kim et al.

turbulence model which relates to the turbulence kinetic energy of k and the turbulence
dissipation rate of 1 (Orszag, 1993). It has been widely used for many industrially
relevant flows.
The SIMPLEC (semi-implicit method for pressure-link consistent equations) algor-
ithm, with a power-law difference scheme presented by Patankar, was applied for the
pressure-linked momentum equation (Patankar, 1980). A system of discretized linear
equations as shown in Eqn (3) was solved iteratively due to the nonlinear feature of the
implicit equation.

ap fp ¼ aE fE þ aW fW þ aN fN þ aS fS þ b ð3Þ

where aE, aW, aN, aS, and aP are coefficients of east, west, north, south and main grid
nodes.

Local density
The local fluid density is related empirically to the local values of temperature and
sediment concentration given by

r ¼ rT þ DrS ð4Þ

where rT ¼ density as function of temperature, kg/m3

DrS ¼ density increase by suspended solids, kg/m3
An empirical equation of density proposed by Gill (1982) was used to incorporate the
temperature effect on density. For density-driven flows, the increase of local fluid density
is related to the local values of sediment concentration by


Drs ¼ Cs Ss12 £ 1023 ð5Þ

where C ¼ solid concentration and Ss ¼ the specific gravity of the solid particles, 1.3
(Larsen, 1977; Tchobanoglous and Burton, 1991).

Settling model
A number of empirical formulas have been proposed to describe the relationship between
solid concentration and solid settling velocity (Mazzolani et al., 1998). For low SS
concentration, settling velocity may be calculated with a discrete settling model (DSM).
The settling velocity of particles is dependent on particle size, showing that the larger
particle shows up the higher settling velocity. From the photographic measurements, Li
and Gancsarczyk (1987) presented the following settling velocity.

V s ¼ 0:35 þ 1:77Di ð6Þ

where Di ¼ cross sectional diameter of SS (in the range of 0.05– 1.4 mm).
In this study, DSM was applied to the four groups of particle size for the calculation
of settling velocity because DSM could estimate outflow SS concentration more correctly
than any other models proposed for a secondary clarifier (Takacs et al., 1991; Kim, 2005;
86 Kim et al., 2005).
Heat flux model
The net surface heat flux was computed by the following Eqn (7) from Ryan et al.
(1974).

fn ¼ 0:94wsc ð1 2 0:65C 2 Þ þ 5:15 £ 10213 ðT a þ 273Þ6 ð1 þ 0:17C2 Þ

2 5:51 £ 1028 ðT s þ 273Þ4 2 ½2:7ðT v;s 2 T v;a Þ1=3
 
Ts 2 Ta

H.S. Kim et al.

þ 3:2W 2
ðes 2 ea Þ 1 þ 0:61 ð7Þ
es 2 ea

where wsc ¼ clear sky solar radiation, W/m2

Ta ¼ air temperature, 8C
Ts ¼ surface temperature, 8C
es ¼ saturated vapor pressure at water surface temperature, mbar
ea ¼ vapor pressure of the air, mbar
C ¼ cloud cover fraction from 0 to 1
W2 ¼ wind speed, m/s
Tv,s, Tv,a ¼ virtual surface and air temperatures, respectively, K
The equilibrium temperature, that was, the theoretical temperature that the water body
would reach at steady state, was computed by solving for Ts, in the above equation when
wn was set equal to zero.

Boundary condition
A uniform, parallel inlet flow was imposed with horizontal velocity of u0 and vertical
velocity, v ¼ 0. The turbulence energy level was assumed with specified values of
m ðk =lm Þ. The inlet SS loading was assumed to be in a well mixed
k ¼ 0:03u20 and 1 ¼ C3=4 3=2

state with dimensionless concentration unit. The water surface was modelled as a sym-
metric plane where the vertical velocity v and the normal gradients of velocity component
u and k were set to zero. For e and concentration C, empirical boundary conditions were
used (Imam et al., 1983). The effluent outlet is in the top cells at the end of the tank.
Outlet boundary values were computed from outlet grid points by satisfying the overall
mass continuity. The clarifier bottom was treated as a perfect-absorbing boundary, where
particles might not be resuspended by the fluid flow. At the rigid walls, the wall function
approach was applied for the flow parameters, while for the concentration a zero gradient
condition was used (Launder and Spalding, 1972). Near the wall k and e could be calcu-
lated from the assumption of local equilibrium conditions. Some simplification had to be
accepted here with the application of the 2-D model because the longitudinal effluent
weirs had 3-D structures.

Results and discussion

Model validation
Figure 2 shows the measured and calculated RTD profiles, where solid and dotted lines
represent the experimental and predicted RTD curves, respectively. Channel number
(CH) specifies the detection location of the radiotracer along the horizontal flow direction
with three different depths as shown in Figure 1.
In the inlet zone, RTDs at CH1 and CH4 show a trend of slow decrease in concen-
tration due to the surface return flow, while the RTDs at CH3 and CH6 rapidly decrease
in a quasi-exponential fashion by the existence of strong bottom density currents. In the
settling zone, RTD profiles show rather smooth curves like left-handed Gaussian profiles,
where rapid bottom density current gradually decreases and thereby SS is favourable to
settle down. In the withdrawal zone, the initial arrival time of tracer is delayed compared 87
H.S. Kim et al.

Figure 2 Model validation from RTD comparison with experiment

to other RTDs. Especially in CH16 and CH17, a minor difference between prediction and
measurement is noted in RTDs. However, the greater the depth, the more this difference
is reduced by the decrease of the effect of top effluent weirs. CH20 is located in
the outlet of the clarifier in which the RTD curve represents the retention time of flow.
Maximum concentration time of the predicted CH20 appears about 80 minutes longer
than that of the experimental RTD. This is because the overflow from four-side weir
boards is combined and flows to the exit location. In conclusion, the predicted RTD
curves were in good agreement with the experimental RTD curves at the upstream and
88 the middle stream sections except for the location of the discharged weirs.
Density effect by suspended solids
Figure 3 shows the comparison of the flow field due to density effect caused by the SS
inlet loading. As shown in Figure 3(a) of neutral density case, the influent flow is
deflected downward after impinging on the baffle. The following downward current
impinges on the tank bottom below the baffle and then goes along the clarifier bottom
and forms a visible recirculation zone near the end wall. In neutral density conditions,
plug-type, uniform flow occurs except in the neighborhood of baffle.

H.S. Kim et al.

Considering the density effect on the flow field as shown in Figure 3(b), the horizontal
inlet jet does not reach and strike the inlet baffle, but rather plunges downward to the
tank bottom as a density waterfall due to the low densimetric Fr. Under the submerged
lip of the baffle, a counter-flow re-entered the inlet zone to provide the entrainment of
fluid by the density waterfall. And then the flow forms a forward bottom current over the
settling zone and goes upward in the withdrawal zone. The strength of the bottom density
current, the upward flow in the withdrawal zone, and the recirculation are increased with
a decreasing inlet velocity and higher SS inlet loading (say, low Fr). Therefore, the calcu-
lation result shows clearly that SS inlet loading affects significantly the flow pattern in
the clarifier.

Density effect by temperature

Both influent and ambient temperatures affecting the flow pattern in the tank were
selectively evaluated except for SS loading. Density currents induced by temperature
differentials were often found in a primary clarifier with relatively low SS loading.
Figure 4(a) shows the flow condition under the effect of surface cooling in winter.
In Figure 4(a), an influent temperature of 2 8C above the ambient temperature caused a
strong buoyant flow and significant short-circuiting due to the low hydraulic loading and
the large size of the clarifier. A strong forward surface current and the recirculating
bottom flow behind the baffle and near the end wall were found in Figure 4(a). On the
other hand Figure 4(b) considered the positive heat flux by daytime solar radiation on a

Figure 3 Flow characteristics of a secondary clarifier 89

H.S. Kim et al.

Figure 4 Density currents induced by temperature differences

bright summer day. Cold influent that was 2 8C lower than the ambient temperature
created a cascading flow, bottom density current, and surface return flow, which is
comparable with the flow pattern by SS loading presented in Figure 4(b).

Behavior of radiotracer
After the flow with SS loading was reached at steady state, radioisotope in aqua-phase
was injected at the end of aeration tank as a type of impulse. Thus two component
mixtures of tracer and water were modelled to simulate the experiment.
Figure 5 shows the overall spatial distribution of radiotracer concentration with the
elapsed time. Tracer forms a high concentration in the bottom of the clarifier due to the
strong density current and flows toward the exit along the bottom current as time passes

90 Figure 5 Radiotracer distributions with elapsed time

as shown in Figure 5(a) and (b). In Figure 5(c) and (d), radiotracer shows relatively
high concentration behind the inlet baffle by the surface return flow and also the high
concentration slowly disappears in the bottom of clarifier. When the radiotracer reaches
the withdrawal zone as presented in Figure 5(e) and (f), it diffuses upward and discharges
into the effluent weirs. It is expected that lots of tracer leaves is the clarifier approxi-
mately after 3 hours, which is in good agreement with the value of the designed residence
time. Based on the simulated behaviour of radiotracer, it is determined that the tracer will

H.S. Kim et al.

be detected earlier in the bottom region with higher concentration than any other loca-
tioner due to the effect of strong bottom density current.

Conclusions
Based on this study, several useful conclusions can be drawn: numerical simulation pre-
dicted successfully the measured RTD curves of radioactive tracer. In detail, the simu-
lated RTD profiles were in good agreement with the experimental RTD curves especially
at the upstream and the middle-stream sections but a minor difference was observed at
the location of discharge weirs with complex geometry.
This study has shown that clarifier performance is strongly related to hydraulic perform-
ance such as density current. For example, injected radiotracer as a pulse type appears to
move along the bottom line of the clarifier because of the strong density bottom current. The
arrival time of tracer is earlier in the bottom but is delayed on top by the surface return flow.
Most tracer leaves out of the clarifier after 3 hours, which is quite a similar value to the
designed residence time. The model predicted the well-known flow characteristics in a clarifier
such as the waterfall phenomenon at the front end of the clarifier, the strong bottom density
current and surface return flow in the settling zone, and the upward flow in the withdrawal
zone. In general, it is believed that the numerical model is considered as a viable tool for the
proper design and determination of optimal operating conditions of a full-scale clarifier.

References
Gill, A.E. (1982). Atmosphere-Ocean Dynamics. Academic Press, New York.
Imam, E., McCorquodale, J.A. and Bewtra, J.K. (1983). Numerical modelling of sedimentation tanks. Journal
of Hydraulic Engineering, ASCE, 109(12), 1740 –1754.
Kim, H.S. (2005). Numerical modeling of SS separation, thermal drying and incineration pollutant abatement
in a sludge and wastes related combined system. Ph. D. thesis, Department of Environmental
Engineering, Chungnam National University, Republic of Korea.
Kim, H.S., Shin, M.S., Jang, D.S., Jung, S.H. and Jin, J.H. (2005). Study of flow characteristics in a
secondary clarifier by numerical simulation and radioisotope tracer technique. Applied Radiation and
Isotope, 63, 519 –526.
Korea Ministry of the Environment (2004). Operation trend of sewage treatment plant. http://www.me.go.kr
(accessed 10 December 2004).
Larsen, P. (1977). On the Hydraulics of Rectangular Settling Basins: Experimental and Theoretical Studies,
Department of Water Resources Engineering, Lund Institute of Technology, University of Lund, Lund,
SwedenReport No.1001.
Launder, B.E. and Spalding, D.B. (1972). Mathematical Models of Turbulence. Academic Press, New York,
USA.
Li, D.H. and Gancsarczyk, J.J. (1987). Stroboscopic determination of settling velocity, size and porosity of
activated sludge flocs. Water Research, 21(3), 257 – 262.
Mazzolani, G., Pirozzi, F. and d’Antonoi, G. (1998). A generalized settling approach in the numerical
modeling of sedimentation tanks. Water Science and Technology, 38(3), 95 – 102.
McCorquodale, J.A. and Zhou, S. (1993). Effect of hydraulic and solids loading on clarifier performance.
Journal of Hydraulic Research, 31(4), 461 – 478.
Orszag, S. (1993). Introduction to the renormalization group and turbulence modeling. Guest lecture for
users’ meeting in Fluent Inc., SRN 713, pp. 1– 11. 91
 Patankar, S.V. (1980). Numerical Heat Transfer and Fluid Flow. McGraw-Hill Company, Hemisphere
Publishing Corporation, USA.
Ryan, P.J., Harleman, D.R.F. and Stolzenbach, K.D. (1974). Surface heat loss from cooling ponds. Water
Resource Research, 10, 930.
Stamou, A.I. (1995). Modelling of Settling Tanks – a critical review. 3rd International Conference, Water
Pollution 95, Porto Carras, Greece.
Takacs, I., Patry, G.G. and Nolasco, D. (1991). A dynamic model of the clarification – thickening process.
H.S. Kim et al.

Water Research, 25(10), 1263 –1271.

Tchobanoglous, G. and Burton, F.L. (1991). Wastewater Engineering. 3rd edn, McGraw-Hill Publishing Co.,
New York, USA.

92
 Reproduced with permission of copyright owner.
Further reproduction prohibited without permission.