Effects of Inclusion of Modified Mixing Devices on Effluent Quality in Aerated Lagoons: Case Study at Wingate, IN WWTP

lochfobbingMechanics

Oct 30, 2013 (3 years and 8 months ago)

252 views

Effects of Inclusion of Modified Mixing Devices on Effluent Quality in
Aerated Lagoons:

Case Study at Wingate, IN WWTP


Ernest R. Blatchley III, Ph.D., P.E, BCEE

Professor, School of Civil Engineering and Division of Environmental & Ecological Engineering

Purdue University

West Lafayette, IN 47907


INTRODUCTION

L
agoons are commonly used for treatment of municipal wastewater in small, rural
communities. The motivations for their use in these settings
include

low operation and
maintenance costs, as well as

availability of inexpensive land. Aerated lagoons are used as an
alternative to
other

lagoon systems (
e.g.
, facultative lagoons). Aerated lagoons typically include
mechanical devices to promote mixing and O
2

transfer, thereby facilitating biochemical
ox
idation of reduced substrates.

Aerated lagoons are relatively simple to operate and accomplish effective removal of
suspended solids (TSS) and
carbonaceous
biochemical oxygen demand (
C
BOD); however,
control of reduced nitrogen, including
ammonia
-
N, can be
problematic in lagoon systems,
especially during periods of prolonged cold weather. This is believed to be attributable to the
relatively slow growth rates that are typical of nitrifying bacteria, as well as their relative
intolerance of cold conditions.

On the other hand, some success in accomplishing nitrification in aerated lagoon systems
has been reported in
cold regions among
systems where attached

growth is promoted. For
example, Richard and Hutchins (1995) reported results of a study in which
a
“ri
nglace”
medium

was

included in an aerated lagoon in
Winter Park, CO
, resulting in significant increases in
nitrification rate (as indicated by an increase in the concentration of nitrate
-
N in the effluent)
,
even under conditions where the water temperature

was just above freezing
. They attributed this
behavior to an increase in
total system biomass, which
was presumed to include

the
community
of
nitrifying

bacteria
. Promotion of attached growth in their system also yielded reductions in
effluent TSS and B
OD.

In an aerated lagoon system,
several possible fates of substrates (including N) can be
identified, including:


1.

Uptake by the microbial community

for incorporation into new cells

2.

Incorporation into settled solids

3.

Liquid


gas transfer

4.

Biochemical oxida
tion (or reduction in the sludge bed)

5.

Effluent discharge.


To varying degrees, all of these fates can be influenced through process design and
control. For example, consider the basic dynamics of liquid


gas transfer, as described by the
“two
-
film” theor
y. Under this model, the rate of transport between the two phases is described
as follows:





(




)


(ㄩ

where:

F

=

net
flux of compound between phases


=

net mass transport rate of compound between phases, per unit interfacial area

K
l

=

overall
mass transfer coefficient, based on liquid
-
phase concentration

C
eq

=

liquid
-
phase concentration that is in equilibrium with (bulk) gas phase

C

=

actual liquid
-
phase concentration.


When C = C
eq
, the system is at equilibrium and no net transport will be observed. When C < C
eq
,
net transport will be from gas


liquid phase. When C > C
eq
,

the opposite will be true (
i.e.
,

net
transport will be from liquid


gas phase
)
.

In general, the difference
between the equilibrium
and actual conditions is used to represent the “driving force” for transport between the two
phases in contact.


Any change to the system that affects
one or more

terms in this equation can be expected
to also affect the net rate of

transport between the gas and liquid phases. For example, the
inclusion of mechanical mixing (normally applied
to

the liquid phase) is known to decrease
resistance to transport on the liquid side of the gas:liquid interface. For volatile compounds, this

can lead to a substantial increase in the overall mass transfer coefficient. In addition, some
mixing devices can increase the gas:liquid interfacial area, thereby
promoting mass transfer
.


Independent of mechanical mixing, it is also possible to influen
ce the rate of mass
transport by changing system chemistry, so as to alter the equilibrium condition. For example,
ammonia
-
N is known to participate in a simple acid
-
base reaction:













(㈩

Li步 all acid
-
扡se reacti潮sⰠ e煵ili扲ium
c潮摩ti潮s f潲 this reacti潮 are esta扬ishe搠 essentially
instantane潵sly
Ⱐ an搠 are g潶erne搠 批 灈

T
he e煵ili扲ium c潮摩ti潮 f潲 this reacti潮 摥termines
the fracti潮 潦 amm潮ia
-
N that is 灲esent as NH
3
,
as well as

the fraction that is present as NH
4
+
.
Th
e equilibrium fo
r

this
acid
-
base
reaction is defined as follows:





[


]
[


]
[



]


(㌩

At T 㴠 ㈰

C, the acid
-
dissociation constant for this reaction has a value of 10
-
9.3

(Stumm and
Morgan, 1996). Therefore, because NH
3

is volatile and NH
4
+

is not, knowledge of equilibrium
for this reaction provides information ab
out the
distribution

of ammonia
-
N, defined as:







[


]

[



]


(㐩

that is 灲esent in the v潬atile f潲m (NH
3
) and the non
-
volatile form (NH
4
+
). Figure 1 illustrates
this
equilibrium distribution.

From this illustration, it is evident that as pH increases

to approach
the pK
a

of equation (3)
, we should expect the efficiency of removal of ammonia
-
N from water to
increase, simply because a larger fraction of the ammonia
-
N wil
l be present in the volatile form,
thereby increasing the “driving force” for
liquid

gas
transfer.



Figure
1
.

Equilibrium

distribution of ammonia
-
N (C
T,N
) as a function of pH at T = 20

C.
For

pH values below 9.3, the majority of ammonia
-
N will be present as NH
4
+
.



Temperature can influence the rate of virtually any physico/chemical or biochemical
process. Specifically, the rate constants and equilibrium constants of reaction
and transport
pr
ocesses
typically demonstrate

temperature dependence. Therefore, seasonal changes in
temperature can be expected to influence many aspects of the behavior of wastewater treatment
systems, which typically depend on a combination of physic
o
/chemical and
biochemical
processes.


In a general sense, biochemical nitrification will proceed when conditions are favorable
for

growth of nitrifying bacteria. Because these organisms are relatively slow
-
growing, they
typically require long (cell) detention times in
the system

(Metcalf and Eddy
,

2003
)
. In addition,
because nitrification can result in expression of substantial oxygen demand, it is necessary to
provide sufficient oxygen to support this process. This usually requires an increase in oxygen
transfer rate
, relative to a system where biochemical nitrification does not take place.



WASTEWATER TREATMENT IN WINGATE, IN

The town of Wingate, IN
constructed
their
wastewater treatment system in 1984 using
funds from a construction grant. The facility, which is
located roughly 1.2 miles northeast of the
town of Wingate, includ
es a three
-
cell aerated lagoon that
discharges treated water to Charles
Ludlow Ditch.
The facility receives septic effluent from residential and commercial activitie
s in
Wingate.

The
Winga
te
wastewater treatment system

was originally configured with two 5
-
HP
“arrow” mixers in the first lagoon, with one 3
-
HP mixer in each of the second and third lagoons

(16 HP total)
.
In this configuration, the system accomplished acceptable treatment with
respect
to BOD and TSS. However, the performance of the system has been
inconsistent or
poor with
respect to removal of ammonia
-
N, particularly during periods of extended cold weather.

Discharge limitations on ammonia
-
N were included in the Wingate NPDES

permit beginning in
the
winter

of 2011.
Therefore, modifications to the system and/or the method of operation will
be required to comply with these pending permit limits.

A conventional approach to this problem involves construction of a
“mechanical”
was
tewater treatment facility to replace the lagoons. Such a system can accomplish reliable
treatment,
such that consistent permit

compliance

can be accomplished
. However, these systems
are more complicated and expensive to operate than lagoons, and the cap
ital costs of such a
system
are likely to

represent a
n unacceptable

financial burden for the community.

Another option is to alter the lagoon system to
improve its performance, particularly as
related to removal of ammonia
-
N
. The specific alteration that
is being examined at Wingate is
the inclusion of alternative mixing devices, and inclusion of media to allow for development of
an attached
-
growth community in the lagoons. This approach is conceptually similar to the
approach reported by Richard and Hutc
hins (1995). As described previously, such a system
should

allow for a substantial increase in the total biomass within the system,

possibly

including
an increase i
n the population of nitrifiers.
Relative to a conventional mechanical (or “package”)
syste
m, this modification to the existing lagoon system has substantially lower capital costs. In
addition, the basic operation of the lagoon system remains largely unchanged.

To examine the effectiveness of this approach, a long
-
term experiment was initiated
at
the Wingate WWTP as a collaborative effor
t involving the Town of Wingate;
Bradley
Environmental (BE)
; Commonwealth Biolabs (CB);

the Indiana Department of
Environmental
Management (IDEM);

and Purdue University (School of Civil Engineering). Participati
on on the
part of Purdue University originally

involved Professor M.K.

Banks. However, Professor Banks
has left Purdue University and is unable to continue her participation in this project.


PROJECT HISTORY

The project was initiated in July 2010 with ins
tallation of a single
BE

1
-
HP pump

(see
Figure 2)

in the third lagoon at the Wingate facility. Data collection was initiated in December
2010, with analyses being performed by CB. In February 2011, six additional
BE

1
-
HP pumps
were installed in the first lagoon. Soon thereafter (February 2011), 1
-
HP enclosed biochemical
reactors (“BOBBER
,


see Figure 3
) were installed in each of lagoons 2 and 3

(one each)
. In
October 2011, the BOBBER in lagoon 3 was moved to lagoon

2, and four additional BOBBERs
were installed.

The 1
-
HP
BE

pumps draw water from the lagoon through

an 8” port and is discharged
back into the lagoon through

an array of radially
-
oriented PVC pipes
(see Figure 2)
. In lagoon 1,
the six 1
-
HP
BE

pumps are d
istributed roughly uniformly across the surface of the lagoon.
Lagoon 2 is now configured with six BOBBERs, which are also distributed roughly uniformly
across the surface of the lagoon. For these systems, water is again drawn toward the device
through a

series of radially
-
oriented PVC pipes. However, in the BOBBER system the water is
discharged into a 6’
-
diameter
black
plastic sphere that contains a medium with a high
specific
surface area which
provides extensive surface area for development of attache
d growth withi
n
the system.





Figure
2
. Schematic illustration of 1
-
HP
BE

mixing devices installed at Wingate WWTP (left);
digital image of 1
-
HP
BE

surface mixing device (
images
provided by
Bradley Environmental
).





Figure
3
.

Digital images of
BE

“BOBBER” devices (photos provided by
Bradley
Environmental
)
.


METHODS


In addition to routine collection and analysis of samples for monthly reporting of system
operation and
performance, sample coll
ection was initiated in December 2010. Effluent samples
from all three lagoons were collected roughly every other week from the Wingate facility and
transported to the CB labs for analysis. Analyses

conducted by CB labs

included the following:




Ammonia
-
N: performed by basification of samples to pH > 11

(to convert all ammonia
-
N
to NH
3
)
, followed by analysis with an ammonia
-
selective electrode. The voltage signal
from analysis of a basified sample was compared with the voltage signals that were
d
eveloped from a series of standards to define the ammonia
-
N concentration in the
sample.



Nitrification rate: 100 mg (as N) NH
4
Cl was added to a 100 mL sample
. The sample was
then aerated for 24 hours, after which the ammonia
-
N concentration was measured,
as
described above.



Media nitrification rate: Twenty randomly
-
selected beads of media were transferred from
a BOBBER to a 100 mL solution of hard synthetic water. The assay described above was
then performed to determine the rate at which ammonia
-
N was re
moved.



Heterotrophic bacteria: These were quantified using a conventional plate method.



Algae: Algal cells were counted under a microscope using a Sedgewick
-
Rafter counting
cell.



NO
2
-
:
Nitrite
was quantified

through formation of an
azo dye produced
at low

pH

by
coupling diazotized sulfanilamide with
N
-
(1
-
naphthyl)
-
ethylenediamine dihydr
ochloride
(NED dihydrochloride)
.

The concentration of the azo dye was measured
spectrophotometrically by comparison with measurements from a set of standards.



NO
3
-

+

NO
2
-
:

Nitrate in a sample was reduced to NO
2
-

using metallic cadmium,
followed
by the complexation and colorimetric analysis described above.

NO
3
-

concentration
was

then estimated by subtraction of the NO
2
-

signal described above.



Other parameters (pH, T, BOD,

TSS, DO) were measured using conventional methods.


RESULTS AND DISCUSSION

Microbial Quality


Figure
4

provides a time
-
course summary of measurements of
microbial quality in the Wingate WWTP. The inclusion of the mixing devices appears to have
resulted in an increase in the heterotrophic bacterial community, especially in lagoons 1 and 2.
This observatio
n is consistent with
the findings

of Richard and Hutchins (1995).

In contrast
, the concentration of algal cells appears to have been reduced by inclusion of
these mixing devices. The changes in
algal content were reflected in measurements of
algal cell
co
unts and chlorophyll a
, and

were most evident in lagoons 2 and 3. Among the factors that
could reduce algal content in a lagoon is mechanical mixing. Efficient mixing of a lagoon will
result in destratification. Under these conditions, algal cells will
be forced
by the mechanical
action of the mixing devices
to move between the upper and lower layers of a lagoon.
Penetration of visible light from the sun
, which is required for photosynthetic activity by algae,

is
likely to be limited
to the upper reache
s of a lagoon. Therefore, algae will experience an
environment in which photosynthesis becomes more difficult than in a stratified lagoon.
In a
stratified lagoon, it is possible for algae to proliferate in the upper portions of the lagoon;
however, algal

growth in the lower layers of a lagoon is likely to be limited by lack of sunlight.

It is possible that other factors may have contributed to the changes in algal cell counts
and chlorophyll a

that were observed in the Wingate lagoons
. However, it appear
s likely that
mechanical destratification may have contributed to these observations
.

A
more detailed
discussion of mixing behavior in the lagoons will be presented later in this report.



Figure
4
.

Time
-
course measurements of microbial quality in effluent samples from the three
lagoons at the Wingate WWTP. For each panel, the vertical dashed lines indicate the last three
modifications to the system. Top panel represents measurements of heter
otrophic bacteria;
center panel represents algal cell counts; bottom panel illustrates measurements of chlorophyll a.

Alkalinity

and pH



These two parameters are intimately linked to each other, and to the
fundamental

biochemistry of the lagoons. In a broad sense, many processes will influence
(carbonate) alkalinity and pH in an aerated lagoon system. However, three important processes
will include oxidation of carbonaceous BOD, oxidation of nitrogenous BOD, and phot
osynthesis.

Biochemically
-
mediated oxidation
of carbonaceous substrates will involve a wide array
of compounds. Using
a simple carbohydrate
as an example

of a carbonaceous substrate
, the role
of inorganic carbon in this process can be illustrated:

























(
5
)

In this reacti潮Ⱐ aer潢oc mi
cr潯oganisms c潭扩ne a car扯by摲ate an搠 潸ygen t漠 yiel搠 CO
2

and
H
2
O as a means of gain
ing access to chemical energy.

The expression of NBOD involves a community of microbes that partic
ipate in a
symbiotic process to oxidize ammonia
-
N to nitrate, with nitrite as an intermediate:










































(
6
)































)






























)

The
Nitroso

bacteria may include species such as
Nitrosomonas

or
Nitrosococcus
, while the
Nitro

bacteria that participate in this process may include
Nitrobacter

or
Nitrospira

(
Metcalf and
Eddy, 2003
).

In addition to oxidation of reduced
substrates, both of these processes also result in
“consumption” of alkalinity, either through production of CO
2

(which functions as an acid) or
through the direct production of H
+
.


In many respects, photosynthesis opposes these oxidation processes
, or wo
rks to
complete the elemental cycles of carbon, oxygen, and nitrogen
.
T
he following expression is
representative of

the stoichiometry of

photosynthesis:




























(
9
)


In this 灲潣essⰠ energy in the f潲m 潦 visi扬e ra摩ati潮 (usually fr潭 the sun) will 摲ive the
灨潴潳ynthetic 灲潣ess t漠 yiel搠 car扯by摲ates an搠 m潬ecular 潸ygen as 灲潤octs⸠
In a摤dti潮Ⱐ
in潲ganic car扯b in the f潲m 潦 CO
2

is “consumed” in this process,
thereby reducing the acidity
of the solution.


Given the complexity of the microbial community

and the soluble substrates

in a system
such as an aerated lagoon, it is likely that other processes will influence alkalinity and pH.
However, the processes lis
ted above
(and their analogs)
are likely to be important contributors to
the overall behavior of alkalinity and pH
.
Therefore, changes in the lagoon environment that
alter the microbial population, particularly as related to
the

organisms that are respons
ible for
BOD expression and photosynthesis, can be expected to influence lagoon alkalinity and pH.

Figure
5

illustrates the time
-
course behavior of alkalinity in the Wingate lagoons.

In the
12
-
month period preceding the completion of the modifications to
the lagoons, a cycle of
alkalinity was evident, whereby alkalinity was generally lowest in mid
-
summer, and highest in
fall and winter. Inclusion of the entire mixing system at Wingate appears to have resulted in a
decrease in the

seasonal

fluctuation of a
lkalinity across the lagoons, relative to the preceding
year.


Figure
5
.

Time
-
course
measurements

of alkalinity in effluent samples from the Wingate
WWTP

lagoons.



Figure
6

provides an illustration of
influent and effluent pH as a function of time (top
panel), as well as illustrations of the difference between influent and effluent pH
(

pH)

across
the system. Effluent pH was higher than influent pH for the entire monitoring

period. If this
interpreted in terms of the processes of biochemical oxidation and photosynthesis, these results
imply that photosynthetic activity has a greater effect on pH than expression of BOD. As
described above, the inclusion of the modified mixi
ng systems has led to a reduction in the
concentration of algal cells, while the concentration of heterotrophic bacteria appears to have
increased. The increase in biomass also has been accompanied by a decrease in effluent BOD
and ammonia
-
N concentration

(to be discussed later). These changes would be expected to yield
a decrease in CO
2

uptake by photosynthesis, along with an increase in CO
2

and H
+

production
resulting from CBOD and NBOD expression. Collectively, these changes would be expected to
yield a decrease in effluent pH

along with a smaller value of

pH. Both of these changes are
evident in the pattern of data illustrated in Figure 4, par
ticularly for the period since October
2011.

However, it is important to recognize that the full configuration of the lagoons with all
mixers operating has only been in place for roughly 6 months, and as such it is not possible to
define the behavior of t
his system in terms of an annual cycle.



Figure
6
.
Time
-
course measurements of influent and effluent pH at the Wingate WWTP (top
panel). Bottom panel illustrates difference between influent a
nd effluent pH
(

pH)
as a function
of time.



The behavior of the bacteria that are responsible for nitrification is known to be related to
pH. Specifically, pH is known to influence nitrifier

activity via changes in the form and
availability of inorganic carbon, activation or deactivation of nitrifying bacteria, and inhibition
by formation of NH
3

or HNO
2
.
Villaverde
et al.

(
1997)

examined nitrifier activity in an
attached
-
growth system and fo
und that the optimum pH for ammonia
-
oxidizing bacteria was
near pH = 8.2 (see Figure 7, left)
.

This observation was consistent with earlier findings of
Alleman (1984). Villaverde
et al.

(1997) also observed that free ammonia (NH
3
) inhibits the
activity o
f nitrite
-
oxidizing bacteria (see Figure 7, right).





Figure
7
.

Observations
of the effect of pH on activity of nitrifying bacteria (from Villaverde
et
al.
, 1997). Left panel illustrates activity of
Nitrosomonas spp.

as a function of pH. Right panel
illustrates accumulation of NH
3
-
N as a function of pH (left vertical axis) as well as accumulation
of NO
2
-
-
N as a function of pH (right vertical axis).


Nitrogen



A primary objective of this project was to examine the ab
ility of the process
modifications to improve removal of ammonia
-
N. Figure
8

illustrates

influent and effluent
ammonia
-
N as a function of time. The inclusion of the complete set of mixing devices, which
was completed in October of 2011, appears to have r
esulted in improved removal of ammonia
-
N
form the lagoons in winter.



Figure
8
.

Influent and effluent ammonia
-
N
(left vertical axis)
at the Wingate WWTP as a
function of time (top panel).
Superimposed on the top panel are records of air and water
temperature at the plant (right vertical axis).
Bottom panel illustrates the difference between
influent and effluent ammonia
-
N (

NH
3
-
N) as a function of time.

The results of these measurements ar
e in qualitative agreement with the report of
Richards and
Hutchins

(1995), in that promotion of attached
-
growth and an overall increase in
biomass within the system appears to have yielded improvement in removal

of ammonia
-
N from
the system.

Also included

in Figure
8

(top panel)

are measurements of air and water temperature at
the Wingate facility. These measurements are included in this graph because the behavior of
nitrifying bacteria is known to be adversely affected by cold temperature. The bottom pa
nel of
Figure
8

illustrate
s

the change in ammonia
-
N concentration
(

NH
3
-
N
) as a function of time.
There is considerable variability in this signal, but a clear seasonal pattern is evident, whereby
removal of ammonia
-
N
diminished

during the winter months. This pattern generally holds
across the entire data set, but the reduction in ammonia
-
N removal was less pronounced in winter
2011
-
2012 than in previous years.

It is important to recognize that the winter of 2011
-
2012 was unusua
lly mild in central
Indiana, in terms of air temperature. On the other hand, water temperature at the Wingate
facility during the winter of 2011
-
2012 was similar to water temperature in the preceding winter
season, yet removal ammonia was improved in wint
er 2011
-
2012 relative to previous years.

One other issue to consider regarding the temperature signals is heat transfer. The
physics of heat transfer are similar to those of mass or momentum transfer. Systems that
increase mass transfer (
e.g.
, by improve
d mixing) are likely to increase heat (and momentum)
transfer. In a general sense, the dynamics of heat transfer between air and (liquid) water can be
described mathematically by a relationship of the following form:







(





)



0
)

where,

F
H

=

flux of heat between air and water


=

rate of heat transfer from air to water per unit air:water interfacial area

K
H

=

overall heat transfer coefficient

T
air

=

air temperature

T
water

=

water temperature.



In general, the rate of heat transfer

between phases will be determined by the product of
the interfacial contact area, the heat transfer coefficient, and the difference between air and water
temperatures. The mixing systems included at the Wingate facility almost certainly increased the
int
erfacial contact area between air and water, as well as the heat transfer coefficient (because of
improved mixing).

Interestingly, water temperature during winter 2011
-
2012 was similar to the
water temperature during winter 2010
-
2011, despite the fact tha
t air temperatures during winter
2010
-
2011 were substantially lower. In other words, the driving force for heat transfer (

T) was
smaller in winter 2011
-
2012. This suggests that heat transfer was improved by the new mixing
devices. If this is true, then

it is possible that water temperature could be substantially reduced
by the system during a period of prolonged cold weather, as is common in central Indiana
winters. It is not clear how this may affect performance of the system with respect to
nitrifica
tion

(or other aspects of treatment), but this is an issue that should be monitored in the
future.


Figure
9

illustrate
s

the time
-
course behavior of ammonia
-
N (top), nitrite (center), and
nitrate in effluent samples from the three lagoons at Wingate. Ammo
nia
-
N
was

removed in all
three lagoons. As described above, inclusion of the full set of mixing devices resulted in
improved ammonia
-
N removal, particularly in the winter months.

Similarly, these changes
appear to have improved removal of nitrite, includ
ing during the winter months.




Figure
9
.

Time
-
course measurements of effluent ammonia
-
N (top), NO
2
-
-
N (center), and NO
3
-
-
N (bottom) at the Wingate WWTP.

The nitrate
-
N signal indicates that NO
3
-

concentrations in all three lagoons are higher
than they were prior to introduction of the mixing devices. This is consistent with promotion of
biochemical nitrification within the lagoons. The pattern of the NO
3
-

signal is such that the
concentration

consistently decreases as water moves through the facility. This may be an
indication of denitrification

activity within the lagoons
. This pattern of behavior appears to be
somewhat more regular after October 2011 than before this date.



Figure
10
.

Time
-
course record of influent and effluent CBOD (top) and change in CBOD
(

CBOD, bottom)
at the Wingate WWTP.

CBOD
-

Figure
10

illustrates the behavior of CBOD at the Wingate WWTP. In general,
effluent CBOD has consistently been below 20 mg/L, and the performance of the Wingate
facility with respect to CBOD removal or control was not substantially affected by inclusion of
the pr
ocess modifications.


Figure
1
1

illustrates the total BOD signal at the Wingate facility. As compared with the
CBOD signal described above, there is an obvious improvement in TBOD as a result of inclusion
of the full set of modifications. This is consist
ent with the improvements in nitrification
described above.

Substantial variations in the TBOD signal are evident in lagoon 1. In absolute
terms, these variations are dampened as water moves through the system.



Figure
11
.

Time
-
course record of TBOD in all three lagoons at the Wingate WWTP.


Particles

-

Figure
1
2

illustrates behavior of total suspended solids (TSS)
in the influent
and effluent of the Wingate facility (top), as well as cha
nges in TSS across the facility (

TSS)
during the monitoring period. The inclusion of the full set of modifications appears to have
yielded an improvement in effluent TSS, in that there appears to be a slight downward trend in
effluent TSS since October 2
011. However, the

TSS signal does not appear to have changed
markedly since October 2011. It is not entirely clear why this is so. The influent TSS signal was
quite variable in samples collected after October 2011, but within this variable signal there

appears to be a slight downward trend in influent TSS. With the relatively long residence time
that characterizes the Wingate lagoons, it is reasonable to expect some dampening of the TSS
signal by simple equalization. It is difficult to conclude from t
his data set that any significant
improvement in TSS removal can be ascribed to the process modifications.



Figure
12
.

Time
-
course record of influent and effluent TSS at the Wingate WWTP (top) and
changes in TSS (

TSS) across the Wingate facility (bottom).


Figure 1
3

provides a more comprehensive summary of the behavior of suspended
particles at the Wingate facility.

Th
e data presented in Figure 1
3
, in which suspended particles
are characterized by measures of TSS (top panel), settleable solids (center panel), and turbidity
(bottom panel), indicate improved particle removal as a result of inclusion of the process
modific
ations. These observations are consistent with those reported by Richard and Hutchins
(1995).



Figure
13
.

Time
-
course record of effluent particle concentrations from the three lagoons at the
Wingate WWTP, as indicated by TSS (top panel), settleable solids (center), and turbidity
(bottom).

Sludge Blanket Depth
-

Collectively, the improvements in NBOD removal and
suspended solids removal imply that sludge production within the Wingate facility should
increase as a result of inclusion of the process modifications.
Figure 1
4

provides a summary of
sludge depth measur
ements that have been performed periodically at the Wingate WWTP
roughly once per month, beginning in May 2011. No obvious trend of increasing sludge depth is
evident from these measurements. Therefore, if changes in sludge accumulation in the Wingate
fa
cility do result from the process changes, it appears that these will be evident on a longer time
-
scale than is illustrated in Figure 1
4
.








Figure
14
.

Time
-
course measurements of sludge de
pth in the lagoons at the Wingate WWTP:
lagoon 1 (top), lagoon 2 (center), lagoon 3 (bottom).


Electrical Power Consumption
-

Figure 1
5

provides a graphical illustration of (daily)
electrical power consumption at the Wingate facility.
Since the inclusion
of the process
modifications, starting in February/March 2011, t
he overall pattern of electrical power usage
has
trended

downward, but within this data set a pattern of seasonal variation in power consumption
is evident. Specifically, daily electrical power usage from April
-
October appears to be
consistently lower than during the period from October
-

April.
To date, the d
ata regarding
electrical power consumption suggest that the new system has lower electrical power
requirements than the original configuration. This is consistent with the fact that the nominal
overall power rating of the new configuration is lower than t
he original configuration.
To be
sure of this trend, it would be beneficial to continue to monitor electrical power usage at the
Wingate facility.



Figure
15
.

Daily electrical power consumpti
on at the Wingate WWTP.




Mixing Behavior
-

Profiles of dissolved oxygen and water temperature were measured
intermittently (roughly once per month), beginning in September 2010. For most of these
sampling dates, measurements were taken at three location
s in each lagoon, which were roughly
equally spaced across a cross
-
section of the lagoon

(see Figure 1
6
)
. At each location,
measurements were collected at the surface, mid
-
depth, and just above the sludge layer.
Therefore, on most sampling dates, nine me
asurements of DO and temperature were collected in
each lagoon.








A

B

C

Lagoon#1

Lagoon # 2

Lagoon # 3

A

B

C

A

B

C


N
ORTH


Figure
16
.

Schematic illustration of sampling locations for lagoon profiling measurements.

A t
e a c h l o c a t i o n, s a mpl e s we r e c o l l e c t e d f r o m t he s u r f a c e, r o u g hl y 5 f e e t be l o w t he s u r f a c e, a n d a t t he
t o p o f t he s l u d g e l a y e r.




Figure
17
.

Results from lagoon profiling measurements.

Top panel illustrates dissolved
oxygen concentration measurements, while bottom panel illustrates water temperature
measurements. Symbols represent the mean of all measurements
(n=9 for most dates),
while
error bars represent
one

standard deviation for this same set of measurements.


Figure 1
7

illustrates the results of the DO and temperature profiling measurements.

For
each entry in these graphs, the point represents the mean of the population of measurements for a
given lagoon
or a given sampling date, while the error bar represents that standard deviation of
that same population of measurements. For most sampling dates, DO and temperature
measurements were recorded at three locations across each lagoon, and at three depths at
each
location. Therefore, for most sampling dates, nine measurements of DO and temperature were
collected in each lagoon.


The DO and temperature data can be used to examine mixing behavior within the
lagoons. In a general sense, a well
-
mixed system is o
ne in which no substantial gradients in
composition are evident within the system.
This will yield a system where system composition,
as defined by constituent concentrations, temperature, etc, show no spatial gradients. In other
words, at any point in t
ime, the composition within each well
-
mixed cell should be the same
everywhere.
These conditions will be met when the processes that are responsible for mixing
within a system are able to move constituents around the system at a rate that is substantially

faster than the rate(s) of processes that affect local concentration.


The DO measurements (Figure 1
7
, top panel) indicate some spatial variability within the
lagoons prior to October 2011, when the current configuration was completed.
This variation is
evident in the magnitude of the standard deviation of measurements on a given date. In theory,
the standard deviation of these measurements should be zero in a well
-
mixed system.
Since that
time, the magnitude of variations in the DO have decreased marke
dly, thereby suggesting th
at
mixing behavior has improved with respec
t to DO.


The temperature measurements suggest that little or no thermal stratification
is evident

within the lagoons. This condition existed before inclusion of the
BE

mixing devices; t
herefore,
there was no real opportunity for change in this behavior.


Well
-
mixed conditions tend to reduce the likelihood of short
-
circuiting, although the
locations of inlet and outlets relatively to each other can also influence this behavior. Well
-
mixe
d systems also are effective for dampening the effects of changes in flow rate or influent
composition; in other words, a well
-
mixed system will equalize flow characteristics, thereby
yielding effluent quality that tends to be relatively consistent. Lastl
y, well
-
mixed conditions can
simplify the analysis of the behavior of a system.


It is important to recognize that the measurements of DO and temperature that are
illustrated in Figure 1
7

are only for DO and temperature within the cross
-
sections of the lag
oon
that are illustrated in Figure 1
6
. It is possible, though unlikely, that quantifiable gradients in DO
or temperature may be evident at other locations in the system. It is also possible that
quantifiable gradients in these (or other) parameters may b
e evident in other parts of the system,
such as within the BOBBERs.

Mass
-
transfer behavior and characteristics within the BOBBER
systems appear to be largely undefined at this time.


The changes in mixing behavior that are evident in the DO profiles appea
r to be related to
the inclusion of the
BE

mixing devices. These changes took place despite the fact that overall
power applied for mixing was reduced. In lagoon 1, overall power was reduced from nominally
10 HP (in the form of two 5
-
HP “arrow” mixers) t
o nominally 6 HP (six 1
-
HP
BE

mixers). In
lagoon 2, power was increased from
3 HP to 6 HP (six BOBBERS). In lagoon 3, power was
decreased from 3 HP to 1 HP (a single
BE

mixer). With respect to lagoons 1 and 2, it is
important to consider not only the no
minal power rating of the mixing devices, but also the
distribution of this power.

In lagoon 1, the original configuration involved two arrow mixers, both located near the
center of the lagoon, pointing in opposing directions. The new configuration involv
es six surface
aerators, which are distributed roughly uniformly across the lagoon

(see Figure 18, left)
.
Similarly, the six BOBBERs in lagoon 2 are roughly uniformly distributed

(see Figure 18, right)
.
This more uniform distribution of mixing energy, as

opposed to the original configuration,
probably results in improved mixing in the lagoons.





Figure
18
.

Digital images of
BE

mixing devices in Lagoon 1 (left) and Lagoon 2 (right) at the
Wingate WWTP.


SUMMARY AND
CONCLUSIONS


The results of sampling and analysis at the Wingate WWTP to date indicate that the
inclusion of the
BE

mixing devices has yielded improvements in effluent quality with respect to
suspended particles and ammonia
-
N. These improvements appear to

have resulted from the
inclusion of an alternative mixing regime within the system, as well as the inclusion of the
BOBBER systems, which promote attached growth. Overall, these systems appear to be using
less electrical power than the previous system.


The new configuration appears to provide for more efficient and more complete mixing
than was accomplished with the previous configuration. This should promote oxygen transfer,
and may also lead to improved stripping of volatile gases, such as NH
3
. On th
e other hand,
conditions that lead to efficient mass transfer also tend to promote efficient transfer of heat. The
data from the winter of 2011
-
2012 indicate the
heat transfer from air to water may have been
improved by the new mixing devices. It is poss
ible that this could lead to substantial reductions
in water temperatures during periods of extended cold weather. The winter of 2011
-
2012 was
the warmest on record in west
-
central Indiana. It is not clear how these systems will perform
during a “normal”

or “cold” winter season, given this apparent improvement in heat transfer.


RECOMMENDATIONS


The inclusion of the
BE

mixing devices appears to have yielded improvements in overall
process performance, both in terms of effluent quality and electrical power

usage. These
improvements indicate the potential for these systems to be used in other, similar applications.
And given the number of lagoon
-
based systems that are in use in the U.S. and elsewhere, there
certainly appears to be a market for this type of

system.


At present, there is no well
-
defined approach to be used in the design of systems based on
this technology. On the other hand, design approaches for other, related systems are in place.
Therefore, it is likely that the principles of design that

have been applied in these related systems
could be adapted to the
BE

mixers. Moreover, the extension of this technology to other
facilities, and possibly other settings, represents a logical opportunity for continued collaboration
among the participants

in this project.


In terms of applied research, several specific topics appear to merit attention. These
include:



Quantification of NH
3

and CBOD uptake rates by the attached
-
growth community in the
BOBBER systems


this behavior of the system is likely t
o be influenced by mass
-
transfer behavior and the composition of the microbial community within the BOBBERs.

By defining this behavior and the process parameters that influence this behavior, it may
be possible to develop a design procedure for the BOBBER

system that can be verified
against field measurements.



Fluid mechanics in aerated lagoons


the BOBBERs and the
BE

mixers appear to have
resulted in improved mixing behavior in the lagoons at Wingate. However, the evidence
to define this behavior is inc
omplete. Numerical simulations (perhaps involving
computational fluid dynamics) and physical tests (
e.g.
, “tracer” tests to allow
measurement of the residence time distribution of lagoon cells) may be beneficial as
methods of validating the effects of the

mixing devices. In addition, these tests may
indicate opportunities for improvement of mixing behavior in lagoons.



Mass and heat transfer


the improved mixing in the Wingate lagoons should yield
increased mass and heat transfer. With respect to mass tr
ansfer, one area of particular
relevance is the ability of these systems to transfer oxygen. However, a closely
-
related
issue is the potential to strip volatile gases, such as NH
3
. Heat transfer is likely to be
most important in the winter months. Detai
led information about heat transfer
characteristics of these systems may provide insights into the behavior of these systems
in cold
-
weather months, as well as opportunities to improve this behavior.


REFERENCES



Alleman, J.E. (1984) “Elevated
N
itrite
O
ccur
rence in
B
iological
W
astewater
T
reatment
S
ystems,”
Water Science & Technology
,
17
, 409
-
419.



Metcalf & Eddy (2003)
Wastewater Engineering: Treatment and Reuse
, Fourth Edition
(G.T. Tchobanoglous, F.L. Burton, and H.D. Stensel), McGraw
-
Hill, New York.



Richard, M. and Hutchins, B. (1995) “
Enhanced Cold Temperature Nitrification in a
Municipal Aerated Lagoon Using Ringlace Fixed Film Media,”
Presented at the Rocky
Mountain American Waterworks Association / Water Environment Association Annual
Conference,
Sheridan Wyoming September 11th, 1995
.



Stumm, W. and Morgan, J.J. (1996)
Aquatic Chemistry: Chemical Equilibria and Rates
in Natural Waters
, Third Edition, John Wiley & Sons, New York.



Villaverde, S.; Garcia
-
Encina, P.A., Fdz
-
Polanco, F. (1997) “
Influence of pH Over
Nitrifying Biofilm Activity in Submerged Biofilters,”
Water Research
,
31
, 1180
-
1186.