TREATMENT WETLANDS - CHAPTER 17 docx
Bạn đang xem bản rút gọn của tài liệu. Xem và tải ngay bản đầy đủ của tài liệu tại đây (607.69 KB, 27 trang )
627
17
Sizing of FWS Wetlands
A performance-based procedure is unavoidable for FWS wet-
lands because they address too many applications with vary-
ing degrees of source strength and pretreatment. In general,
the targets are also variable because of the differences in reg-
ulatory requirements among countries, states, and discharge
recipients. Loading specications, whether as pollutant kilo-
grams per wetland hectare per year or as wetland hectares
per person equivalent, have not, and, in general, cannot be
developed to deal with the spectrum of situations created by
FWS diversity of applications. It is not an exaggeration to say
that each FWS wetland is unique.
Design of FWS wetlands may be roughly divided into
two categories: sizing calculations and physical specica-
tions. Sizing requires characterization of the incoming water
and regional meteorology as well as the goals of wetland
treatment, as discussed in Chapter 16. Here, a comprehen-
sive sizing strategy is presented, based on the information
assembled. Chapter 18 will deal with physical consider-
ations, including the number of cells, layout, depth, bathym-
etry, soils and plants, structures, and lining. It is recognized
that wetlands are almost never stand-alone treatment devices,
but rather form part of a treatment train. Other components
may be mechanical, such as clariers or lters, or more natu-
ral, such as settling basins or lagoons. More than one type
of wetland may be involved at the same site—FWS, VF, and
HSSF. This chapter focuses on only the FWS component of
treatment systems.
Based on the performance-based sizing algorithm intro-
duced in Chapter 15, the key features described in this chap-
ter are the following:
Parameter selection and calculation
6
.
Select rate constants and seasonality.
(Select plant community type.)
7. Select DTD (hydraulic) efciency P-value.
(Select compartmentalization.)
8. Adjust area till goals are all met.
Constraint checks and iteration
9. Set estimated growth cycle.
10. Check biogeochemical cycles for consistency.
11. Check chemical constraints.
12. Check loading graph for risk assessment.
13. Adjust for seasonality.
Repeat this procedure as needed to meet design goals.
The complexity of wetland behavior and, hence, of the
sizing calculation is such that a single equation for wetland
area cannot be written. At this point in the evolution of the
treatment of wetland design, the sizing procedure must move
out of the realm of the pocket calculator and into the world
of spreadsheet computations. The single formula, black-box
approach will not sufce.
17.1 POLLUTANT REDUCTIONS AND
PERFORMANCE COMPUTATIONS
The information collected is utilized as a basis to forecast
the area needed to achieve the goals of the project. The pro-
cedure outlined here is for a steady-ow situation. Different
ows may need to be considered to deal with daily or sea-
sonal peaking. Different seasons may need to be considered,
so that the “bottleneck” period of the year is identied and
analyzed.
Many literature sources provide single equations into
which anyone may insert numbers to compute a wetland
area. Although such an option would be very convenient,
it is not realistic except for a few highly specialized cases.
Accordingly, the procedure given here explores the effect of
changing wetland area (or equivalently, detention time) on
the forecasted efuent concentrations of contaminants. It is
presumed that the designer will conduct the design calcula-
tions via a computer spreadsheet so that design changes may
be easily explored.
Wetland hydrology is rst determined as a necessary precur-
sor to area calculations. Annualized calculations are considered
rst; the effects of season will be considered later in the chapter.
WATER BUDGET
The annual water budget forms the basis for understand-
ing pollutant reductions and the area required. Assuming
a constant water level, the ows out of the wetland may be
computed from the inow and meteorological data. Some
inltration is also assumed here if leakage is present in the
wetland. The relevant equations are
QQAPETI
oi
()
(17.1)
qq PETI
oi
()
(17.2)
where
wetland area, m
evapotranspiratio
2
A
ET
nn rate, m/d
infiltration rate, m/d
preci
I
P
ppitation rate, m/d
hydraulic loading rateq ,, m/d
flow rate, m /d
3
Q
© 2009 by Taylor & Francis Group, LLC
628 Treatment Wetlands
The inlet hydraulic loading will be increased by rainfall
(≈0.5–1.5 m/yr) and decreased by evapotranspiration (ET) (≈
0.5–1.5 m/yr). However, there may be seasonal imbalances.
These amounts are important if the wetland is to have a very low
hydraulic loading or, correspondingly, a long detention time.
The ratio a (P ET)/q
i
is the atmospheric augmentation.
Evapotranspiration (rain) has two effects: lengthening
(shortening) of detention time and concentration (dilution) of
dissolved constituents. The use of an average ow rate com-
pensates for altered detention time but not for dilution or
concentration. The fractional error in a rst-order model
prediction of concentration due to ow averaging is approxi-
mately equal to a, for a −0.5. Thus, if 25% of the inow
evaporates, use of a rst-order model with average ow pre-
dicts concentrations 25% lower than required by the mass
balance. If rain adds 25% to the ow, use of a rst-order
model predicts concentrations 25% higher.
A possibly important feature of ET is that the transpira-
tion carries water into the root zone (see Part I, Chapter 4).
Therefore, if the pollutant mass balances are done on surface
water, then the transpiration component is the same as inl-
tration: it carries materials into the soil. For a fully vegetated
wetland, somewhat more than half of ET is attributable to
transpiration. Transpiration is typically about one half to two
thirds of ET (Kadlec, 2006c):
TETE ET()A
(17.3)
where
evaporation, m/d
transpiration, m/d
E
T
AAtranspiration fraction of , dimensionlET eess
The pollutant mass balances will be conducted on a cells-
in-series basis (see Figure 17.1). Results of the overall water
mass balance are apportioned to the cells according to the
chosen number of TIS. For the rst unit in the series:
QQ APETI
11
in
()
(17.4)
where
area of tank number 1, m
flow ra
1
2
1
A
Q
tte out of tank 1, m /d
3
Flows are thus computed sequentially, from inlet to outlet,
for the number of tanks chosen (PTIS).
The input data requirements for water mass balances are
1. Inow (Q
in
)
2. Rain (P)
3. Evapotranspiration (ET)
4. Inltration (I)
5. Area (A)
6. Apparent number of tanks in series (P-value)
An example is detailed in Table17.1. This hypothetical
example is congured to include small rainfall, considerable
ET, and some inltration; in other words, a leaky arid region
system. The net loss of incoming water is 41%, of which 44%
is ET and 56% is inltration.
POLLUTANT MASS BALANCES
The TIS model is then carried forward via a sequential cal-
culation of pollutant concentrations for each “tank” in the
chosen hydraulic model (Figure 17.1).
A rst-order areal model with rate constant k is selected
with necessary wetland background concentration C*. A
volumetric rst-order model may also be chosen for which
k Ehk
V
. The pollutant mass balance for the rst of the wet-
land segments, designated by subscript “1” for steady-state,
FIGURE 17.1 Conceptual TIS model for pollutant reduction.
Tank 1
PET
Q
1
C
1
Q
n–1
C
n–1
Q
1
C
1
kAC
1
Removal
Conversion
Burial
Infiltration
I
0.5T
kAC*
Return
Flux
Tank 2
Tank N
© 2009 by Taylor & Francis Group, LLC
Sizing of FWS Wetlands 629
nonuniform ow is
QC Q C I AC ET AC
kA C
11 11 11
11
()( )( )
(
in in
A
CC*)
(17.5)
In this simple version, rainfall has been assumed to have zero
pollutant concentration, but it is easy to add an atmospheric
input of the pollutant if it exists. Inltration is assumed to
occur at the outlet concentration. Transpiration ow of the
contaminant has been included. Combining Equation 17.4
with Equation 17.5 gives the concentration exiting the hypo-
thetical segment number one:
C
QC k A C
QETAIAkA
1
1
1111
iin
(*)
() ()(A
))
(17.6)
Or
C
qC kC
qETIk
1
1
in in
(*)
()A
(17.7)
Note that the hydraulic loading rates in Equation 17.7 are the
individual tank loading rates, not the overall system load-
ing rates. This computation is then repeated sequentially
for the remaining segments, using, in each case, the outlet
concentrations and ows from the preceding unit. The wet-
land outlet concentration is that exiting the nal hypotheti-
cal segment.
Outgoing pollutant loads are calculated as the product of
the volumetric outow (m
3
/d) and the outow concentration
(kg/m
3
).
The additional input data requirements for the pollutant
mass balances are
7. Input concentration (C
in
)
8. Background concentration (C*)
9. Rate coefcient (k)
10. Transpiration fraction (A)
The hypothetical example presented for the water budget
is extended to illustrate these calculations, continuing the
choice of P 3 TIS. Phosphorus is chosen as the pollutant of
interest, entering the wetland at 2.00 mg/L. The value of C*
is chosen to be low, 0.01 mg/L. The rate constant is selected
to be the median shown for phosphorus in FWS systems in
Table 10.11, k 10 m/yr. Because the transpiration ux is
presumed to draw phosphorus into the root zone, the frac-
tion of ET, that is, transpiration, must be selected, and in the
example it is picked as A 0.5.
The computed phosphorus concentration in the surface
outow water is C
o
0.62 mg/L, or a concentration reduction
of 69% (Table 17.2). However, the existence of ET losses and
inltration creates a different result for mass removal: 82%
of the phosphorus entering does not leave with surface water.
If inltration water departs vertically downward without fur-
ther treatment, then 15% of the phosphorus mass removal
is due to inltration. Thus, it is seen that load reduction and
concentration reduction are two different goals, which may
lead to different designs. Both these types of design sizing
require mass balance computations for water and rate calcu-
lations for pollutants.
TABLE 17.1
Water Budgets for a FWS Design Example
Parameters
Flow rate 5,000 m
3
/d
Rain 0.05 cm/d
ET 0.40 cm/d
Inltration 0.50
PTIS 3
Area 24 ha
Area/tank 8 ha
Depth 0.3 m
Porosity 0.95
Volume/tank 22,800 m
3
Inflow Tank 1 Tank 2
Tank 3 Outflo
w
Total
Flow rate (m
3
/d) 5,000 4,320 3,640 2,960 —
Rain (m
3
/d) — 40 40 40 120
ET (m
3
/d) — 320 320 320 960
Inltration (m
3
/d) — 400 400 400 1,200
Nominal detention (d) 13.7 5.3 6.3 7.7 19.2
HLR (cm/d) 2.08 5.40 4.55 3.70 —
Note: Sequential tank-to-tank calculations are based on Equation 17.4; this example has a com-
bined detention time of 19.2 days, whereas averaging the ows leads to 17.2 days; the required
input data are shown in bold.
© 2009 by Taylor & Francis Group, LLC
630 Treatment Wetlands
INTERCONNECTED POLLUTANTS:THE CASE OF NITROGEN
Nitrogen species interconvert, thereby linking the mass bal-
ances for organic, ammonia, and oxidized nitrogen. It is some-
times possible to disconnect these species, as, for instance, in
the case of wetlands that receive nitrate but little or no organic
or ammonia nitrogen. However, in many cases, it is neces-
sary to account for the (internal) production of ammonia from
organic sources, i.e., from either the incoming water or the
decomposition of wetland necromass, and the internal produc-
tion of oxidized nitrogen. A simple, presumed chemistry is
ORG-NNH-NNO-NN
4x2
lll
(17.8)
In a simplied version of analysis, uptake and return from
biomass is not included. The effects of the biogeochemical
cycle on nitrogen will be explored in a subsequent part of the
design process. The three mass balances then become linked,
and the tank equations are:
QC Q C I AC ETAC k A
11 11 11 1OinO,in O OO
()( )( )(A CCC
OO
*
1
)
(17.9)
QC Q C I AC ETAC
kA
A11 11 11
1
AinA,in A
A
()( )( )
(
A
CCC kACC
AAOOO111
**
)( )
(17.10)
QC Q C I AC ETAC
kA
11 11 11
1
NinN,in N N
N
()( )( )
(
A
CCC kACC
NNAAA
*
111
*
)( )
(17.11)
where
organic N concentration, mg/L
amm
O
A
C
C
oonia N concentration, mg/L
oxidized N co
N
C nncentration, mg/L
organic N rate coeffic
O
k iient, m/d
ammonia N rate coefficient, m/
A
k dd
oxidized N rate coefficient, m/d
N
k
and the subscript “in” denotes parameters associated with the
inuent ow for each nitrogen form.
These may be rearranged to solve the outlet concentra-
tions for each tank:
C
QC kAC
QETAIAkA
O
in O,in O
*
O
1
1
1111
¤
¦
¥
³
µ
()A
´´
(17.12)
C
QC k AC k C C
QETA
A
in A,in A A
*
OO1 O
1
1
11
()
()
*
A IIA k A
11
¤
¦
¥
³
µ
´
A
(17.13)
C
QC k AC k C C
QETA
N
in N,in N N A A A
*
1
11
11
*
()
()A
IIA k A
11
¤
¦
¥
³
µ
´
N
(17.14)
Equation 17.12 is a direct analog of Equation 17.7 for an
unspecied generic pollutant. Equations 17.13 and 17.14
contain extra production terms from ammonication in the
ammonia balance—and nitrication in the oxidized nitrogen
balance. The three must be solved sequentially—Equation
17.12 followed by Equations 17.13 and 17.14.
DESIGN PARAMETERS:SOURCES OF INFORMATION
The P-k-C* design model is the basis for this sizing analysis;
it is therefore necessary to select values of these three param-
eters for all pollutants of concern.
Background Concentrations
Wetland systems are dominated by plants (autotrophs), which
act as primary producers of biomass. However, wetlands also
include communities of microbes (heterotrophs) and higher
animals, which act as grazers and reduce plant biomass. Most
wetlands support more producers than consumers, resulting
in a net surplus of plant biomass. This excess material is
TABLE 17.2
Phosphorus Budgets for a FWS Design Example
Parameters
C
i
2.00 mg/L
C* 0.01 mg/L
k 10 m/yr
k 0.0274 m/d
Area/tank 80,000 m
2
Theta factor 0.5
Inflow Tank 1 Tank 2
Tank 3
Outflo
w
Removed Reduction
Concentration (mg/L) 2.00 1.42 0.96 0.62 — 69%
Load in surface water (kgP/yr) 3,650 — — 666 2,984 82%
Load inltrated (kgP/yr) — 207 140 90 437 15%
Load stored (kgP/yr) — — — — 2,547 70%
Note:
This example is based on the water budget shown in Table 1
7.1; sequential tank-to-tank calculations are based
on Equation 17.6; the required input data are shown in bold; the shaded cells represent potential design targets.
© 2009 by Taylor & Francis Group, LLC
Sizing of FWS Wetlands 631
typically buried as peat or exported from the wetland (Mitsch
and Gosselink, 1993). The net export results in an internal
release of particulate and dissolved biomass to the water col-
umn, which is measured as nonzero levels of biochemical
oxygen demand (BOD), total suspended solids (TSS), total
nitrogen (TN), and TP. These wetland background con-
centrations are typically denoted by the term C*. Enriched
wetland ecosystems (such as those treating wastewater) are
likely to produce higher background concentrations than oli-
gotrophic wetlands. These elevated background concentra-
tions are largely due to increased biomass cycling resulting
from the higher levels of nutrients and organic carbon in the
wastewater. Even land-locked wetland basins, which only
receive water inputs through precipitation, will have nonzero
background concentrations.
Consequently, many pollutants are not reduced to zero
in treatment wetlands—including BOD, TSS, organic nitro-
gen, and phosphorus. However, it is important to distinguish
between artifacts of data tting and real wetland processes.
Short-circuiting can lead to high values of data-t C* for
heavily loaded systems, but these high C* can be dealt with
by improving the hydraulics. Independent of hydraulics, the
wetland can manufacture water-phase organics and solids,
and cycle nutrients into and out of the water body. The chap-
ters of Part I contain estimates of the C*-values for the com-
mon pollutants, and many literature sources provide ranges of
background concentrations (U.S. EPA, 1999; IWA Specialist
Group on Use of Macrophytes in Water Pollution Control,
2000; U.S. EPA, 2000a; Wallace and Knight, 2006; Crites
et al.
, 2006). A summary is given in Table 17.3.
S
ome individual exotic chemicals are foreign to typi-
cal wetland environments and are not expected to exhibit
background concentrations. Examples include halogenated
hydrocarbons and pesticides.
Number of Tanks to Be Used in the Model
The performance of the wetland depends on the number
of tanks (TIS) selected—very strongly if the design is to
approach wetland background concentrations or high degrees
of removal (see Chapter 6). If a high degree of removal is
required, it will necessitate a very large wetland with poor
hydraulics, or a smaller wetland with good hydraulics.
Figure 17.2 illustrates this high degree of sensitivity in the
region of low P for large reductions. This chart quanties the
fact that a tiny fraction of the water following a fast short-
circuit will carry enough unreacted material to the outlet to
make 99% reduction almost impossible to achieve. Low
P numbers represent ow patterns that, by virtue of either
fast forward mixing or velocity proles with high-speed ele-
ments, carry fractions of unreacted material directly to the
outlet.
The P-value is somewhat at the discretion of the designer.
More cells and greater length-to-width ratios can increase the
P-value. As an illustration of the design decision to be made,
consider a hypothetical case relative to Figure 17.2. Suppose
the wetland is to achieve a 90% reduction. It is possible to
consider a one-cell wetland, with a presumptive P 3 that
needs 84 m
2
/(m
3
/d). Or, the designer can opt for two cells in a
series, each with a presumptive P 3 that needs 68 m
2
/(m
3
/d).
The question might be resolved based on the economics, i.e.,
does the cost savings of 20% area reduction outweigh the
added cost of the divider berm and structures? This illus-
tration carries a zero background concentration, but the
concepts are applicable to any pollutant, provided the reduc-
tion fraction to background is used instead of percentage
removal.
It should be remembered that the designer can control
the N-value for the wetland (inert tracer tanks in series) but
cannot entirely control the P-value. Pollutants that are mix-
tures, which may undergo weathering in the wetland, act to
reduce the applicable P-value. Table 6.3 provides some guid-
ance on the apparent P-values to be selected relative to tracer
N-values.
Rate Coefficients
In this initial, annualized analysis, the appropriate rate coef-
cients are those (shown in the chapters of Part I) as the
result of the tting of annual data from existing wetlands.
Those are variable across wetlands and years, thus producing
frequency distributions of the tted k-values. These are to be
found as follows:
BOD
5
Chapter 8, Table 8.2
Organic N Chapter 9, Table 9.11
Ammonia N Chapter 9, Tables 9.17 and 9.20
Total Kjeldahl
nitrogen (TKN) Chapter 9, Table 9.12
Oxidized N Chapter 9, Table 9.23
Total N Chapter 9, Table 9.14
Total P Chapter 10, Table 10.11
Fecal coliforms Chapter 12, Table 12.3
Conspicuously absent from this list is the common constitu-
ent TSS. Some individual system data were presented in
•
•
•
•
•
•
•
•
TABLE 17.3
Summary of Background Concentrations for FWS
Wetlands
Parameter
Lightly
Loaded
Heavily
Loaded
BOD
5
(mg/L) 2 10
TSS (mg/L) 2 15
Organic N (mg/L) 1 3
Ammonia N (mg/L) <0.1 <0.1
Oxidized N (mg/L) <0.1 <0.1
Total phosphorus (mg/L) <0.01 0.04
Fecal coliforms (CFU/100 mL) 10–50 100–500
Note: In the case of fecal coliforms, lightly/heavily loaded refers
to animal use.
© 2009 by Taylor & Francis Group, LLC
632 Treatment Wetlands
Chapter 7 and analyzed for rate coefcients. However, incom-
ing TSS is often reduced rather quickly, and wetland-efuent
TSS results from a balance of generation and resuspension in
the FWS wetland. The removal rate coefcients that pertain
to the inlet region of the wetland are often quite high, ≈10 m/d
(3,650 m/yr), as shown in Figure 7.8. Colloidal materials are
an exception to this generality. The design recommendation
suggested here is the use of a high-rate coefcient for TSS
(perhaps 200 m/yr), unless the design-limiting bottleneck
happens to be TSS. In that event, it is strongly suggested that
settling tests be conducted on the candidate wetland inuent
waters.
The frequency distributions of the reference tables are
generally quite broad. It is up to the designer to narrow the
selection for a given application, either by choosing the pre-
ferred degree of risk (blind to modifying factors), or by delv-
ing further into the details of existing wetland data sets, and
to narrow the selection to wetlands closest to the intended
application in terms of operating conditions. Regrettably,
there is no modern, published, all-inclusive database to which
to turn. The reader is cautioned that older databases such as
the NADB (Knight et al., 1993) and the 1994 Danish data-
base (in Kadlec and Knight, 1996) have been superseded.
Also, old databases are uneven in the quality and quantity of
the data presented. There are numerous examples of misin-
terpretation in such databases, and it is concluded that their
use as a sole source of narrowing the information eld is dan-
gerous. Because there are now so many treatment wetlands, it
is not feasible to provide detailed data. In this book, the com-
promise is to provide analysis, references, and the generic list
of sources used herein (see Appendix A).
DESIGN SIZING GOALS:LOAD REDUCTION
VERSUS
CONCENTRATION REDUCTION
A key feature of treatment wetlands is the ability to design
or manage the system for either concentration reduction or
for mass removal, but only one at the expense of the other
(Trepel and Palmeri, 2002). Wetland performance, as mod-
eled previously, follows the rule of mass action: the removal
rate of a pollutant is greater at higher pollutant concentra-
tions in the water. The rst-order model assumes a (nearly)
direct proportionality: doubling the concentration doubles the
removal rate. As a result of this observed behavior, removal
rates decrease as water passes through the treatment wet-
land, and pollutant concentrations are reduced (Figure 17.3).
However, the actual mass of the pollutant that is removed
increases with increasing hydraulic loading. Thus, increas-
ing hydraulic loads result in more kilograms removed, but at
the expense of higher efuent concentrations. This trade-off
between removal efciency and load reduction is a key feature
of wetland design for nutrient control. These may be quanti-
ed via the rst-order model. A simple version for C* 0
and no water loss is
Concentration reduction
io
i
¤
¦
¥
³
µ
´
CC
C
k
P
11
P
¤
¦
¥
³
µ
´
(17.15)
Load reduction = Input load r Concentration reduction
¤
¦
¥
³
µ
´
¤
¦
¥
¥
³
µ
´
´
qC
k
Pq
P
i
11
(17.16)
The maximum mass of pollutant that can be removed in a given
footprint area results from a hydraulic load so high that little
or no concentration reduction is achieved. Under these condi-
tions, pollutant concentration is at a maximum everywhere in
the wetland, thus maximizing the mass removal rate.
If the output load of a pollutant is to be held below some
regulatory limit, then that has the same general effect as
specifying an outlet concentration. However, because the
load of a pollutant in the wetland outow is usually taken to
FIGURE 17.2 The effect of the number of tanks on the area required for different degrees of removal. The calculations are for k 15 m/yr
and C* 0.
10
100
1,000
10,000
100
,
000
1 10 100
P-Value
Area (m
2
/(m
3
/d))
99.9%
99%
90%
70%
50%
© 2009 by Taylor & Francis Group, LLC
Sizing of FWS Wetlands 633
be that in the surface water discharge, any reduction in ow
through the wetland assists in providing lower output loads
as well as higher load reductions.
All three of the potential design goals (concentration out,
load out, load reduction) may be sensible, depending on the
pollutant in question. These are marked as shaded cells in
the example in Table 17.2. For instance, a low required outlet
concentration can be of use in preventing ammonia toxicity
in receiving waters. But if there is a load allocation, as might
be the case for a discharge contributing to the total maxi-
mum daily load (TMDL) for an impaired water body, then
the mass of the contaminant is of more direct interest. Lastly,
if there is a desire to maximize the benets of a given wet-
land footprint, then the goal should be to dissipate or retain
the maximum amount of pollutant in the wetland.
Because it is easy to confuse these potentially conicting
goals, it is recommended that the designer clearly identify
and state the purpose of the design.
17.2 AREA COMPUTATIONS
At this point in the development of constructed wetland tech-
nology, it would be disingenuous to provide overly simplistic
design-calculation procedures. But it is also a mistake to think
that adding more factors that may play a signicant role in perfor-
mance lends more accuracy or precision to design predictions.
GOAL SEEKING:DETERMINATION
OF THE
REQUIRED WETLAND AREA
The fundamental and straightforward technique for deter-
mining the area is to adjust that area until the specied crite-
rion is met. That is easy if the calculations have been set up
on a spreadsheet; the area can then be sequentially and man-
ually changed by the user until the criterion is met, or auto-
matic searches may be invoked, such as the Solver
™
routine
in Microsoft Excel
™
. The hypothetical phosphorus example
is continued to illustrate this process.
Concentration Criterion
A common criterion for phosphorus in the United States is
for monthly means to be less than 1.0 mg/L. Realistically,
the project owner would not want to encounter exceed-
ances very often. For the example, choose the compliance
frequency to be 90%. From Table 10.13, the multiplier to
contain exceedance at that frequency is 1.94. Therefore, the
design target is adjusted downward to 1.00/1.94 0.52 mg/L,
which becomes the value to be achieved as the wetland area
is varied.
In just three manual iterations, the starting guess of 24
ha (C
o
0.62 mg/L, Table 17.2) is changed to 27.6 ha (C
o
0.52 mg/L). The same result may be obtained using Solver
™
,
which provides the answer in an essentially instantaneous
search from any plausible starting condition.
The exceedance containment multipliers for common
constituents may be found as follows:
BOD
5
Chapter 8, Table 8.6
Organic N Chapter 9, Table 9.11
Ammonia N Chapter 9, Table 9.21
Oxidized N Chapter 9, Table 9.25
Total N Chapter 9, Table 9.16
Total P Chapter 10, Table 10.13
Fecal coliforms Chapter 12, Table 12.4
(TKN is not in this list because it is not normally
regulated.)
Maximum Load Criterion
The same philosophy of regulation might lead to an annual
load limitation on the efuent from the system. An outlet con-
centration of 1.0 mg/L at a ow rate of 5,000 m
3
/d (inow rate)
implies a maximum annual load of 1,825 kgP/yr leaving the
wetland (50% load reduction). A search for the wetland area
leads to a value of 10.1 ha. This is very much lower than that
for the concentration goal for two reasons. First, the exceed-
ance factor is not in play because of the annual character
•
•
•
•
•
•
•
•
0
20
40
60
80
100
120
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95
Hydraulic Loading, HLR (m/yr)
Concentration Reduction (%)
0
50
100
150
200
250
300
Load Removal (gN/m
2
· yr)
Concentration Reduction
Load Removal
FIGURE 17.3 Concentration reduction and load reduction as a function of hydraulic load for a hypothetical nitrate treatment wetland.
Parameters: C
i
10 mg/L, k 35 m/yr, C* 0, P 4.
© 2009 by Taylor & Francis Group, LLC
634 Treatment Wetlands
of the load limit. Secondly, credit is built into the calculation
for the loss of water to ET and for wetland inltration.
MINIMUM LOAD REDUCTION CRITERION
A slightly different regulatory philosophy requires a minimum
load reduction. For instance, some wetland systems in South
Florida require a minimum of 75% phosphorus load reduction.
A search for the wetland area leads to a value of 19.8 ha.
MULTIPLE COMPOUNDS OF CONCERN
The projected outlet concentrations of all constituents of
interest are calculated via the preceding Equation 17.7 (or
Equations 17.12–17.14 in the case of nitrogen). All computed
outgoing concentrations and loads vary with the selected
wetland area. Area is adjusted until the most stringent cri-
terion is met. Criteria often include outlet concentration
specications, each with an associated allowable frequency
of exceedance. But in other cases, outlet pollutant loads may
be specied (or load reduction). Loads are easily calculated
using the outlet concentrations and ows from the last unit.
If the contaminants are considered singly, then a last step
remains: all performances are calculated on the controlling
(maximum required) wetland area.
As an example of multiple contaminants, the previous
phosphorus illustration, along with the additional require-
ments to also reduce BOD and total nitrogen, is continued.
BOD enters the wetland at 30 mg/L, and total nitrogen, at 30
mg/L. The value of C* is chosen to be 2 mg/L for BOD, and
P 1. The value of C* is chosen to be 1.5 mg/L for TN, and
P 3. The rate coefcients are selected to be the medians
shown for BOD and TN in FWS systems in Tables 8.2 and
9.14, i.e., k
BOD
33 m/yr and k
TN
13 m/yr. It is assumed
that exceedances must be contained at the 90th percentile.
The multipliers are 1.56 for BOD
5
(Table 8.6) and 1.55 for
TN (Table 9.16).
The calculations of the required areas for each of the
three contaminants are shown in Table 17.4. The largest area
is needed for the reduction of total nitrogen—40 ha for the
5,000-m
3
/d ow. When that area is applied to the calculation
of performance for BOD
5
and TP, those pollutants are reduced
more than necessary (Table 17.5).
This concludes the preliminary calculation of the
required wetland area. It may seem that there are no further
steps needed, but in fact there is no guarantee that these pre-
liminary calculations t into reasonable, known patterns of
wetland behavior. It is critical that the bounds of plausible
biogeochemical cycles are not exceeded, that the required
ancillary chemicals (oxygen, carbon) are present in ample
supply, and that the forecasted results are reasonable com-
pared to known performance data. Further, the seasonality of
the system has yet to be investigated via forecasting.
17.3 CHECKING THE BIOGEOCHEMICAL
CYCLES
In this phase of design, vegetation in the prospective wet-
land and its role in nutrient processing is brought into the
picture. The rate coefcient-based analysis has provided rst
estimates of the quantities of materials entering and leaving
the system, but there has been no allocation of the removals
to the various processes that comprise the entire ecosystem
function. It is not feasible to go too far in breaking down
processes because the knowledge base does not support great
detail. Here, the processing of carbon, nitrogen, and phos-
phorus are examined via the mass balances used in ecosys-
tem analysis in Part I (see Chapter 9, Figure 9.14).
In the analysis that follows, two points are important.
First, the analysis is based on estimates of what the eco-
system will be like. Such estimates cannot be precise, and,
consequently, the analysis is “order-of-magnitude” only. The
intent is to gain some insight into the relative importance of
wetland processes that will likely be operating if the proj-
ect is built. Second, the analysis is for an annual period, and
hence there remains the issue of seasonality. Seasonal analy-
sis proceeds more readily if we can rst establish whether
the wetland should be viewed as an agronomic system or a
microbial system.
TABLE 17.4
Areas Needed for a Three-Parameter Wetland Sizing
BOD
5
TN TP
Inlet C mg/L 30 20 2.00
Regulatory C mg/L 10.0 5.0 1.00
Containment multiplier mg/L 1.56 1.55 1.94
Target C mg/L 6.41 3.23 0.52
Background C* mg/L 2.0 1.5 0.01
PTIS — 1 3 3
Annual k m/yr 33 13 10.0
Area ha 30.3 40.0 27.6
Hydraulic loading cm/d 1.65 1.25 1.76
Outlet C mg/L 6.41 3.23 0.52
TABLE 17.5
Performance for a Three-Parameter Wetland Sized
for TN Reduction
BOD
5
TN TP
Inlet C mg/L 30 20 2.00
Inlet loading
g/m
2
·yr
137 91 9.1
Regulatory C mg/L 10.0 5.0 1.00
Containment multiplier mg/L 1.56 1.55 2.00
Target C mg/L 6.41 3.23 0.50
Background C* mg/L 2.0 1.5 0.01
PTIS — 1 3 3
Annual k m/yr 33 13 10.0
Area ha 40.0 40.0 40.0
Hydraulic loading cm/d 1.25 1.25 1.25
Outlet C mg/L 5.48 3.23 0.29
© 2009 by Taylor & Francis Group, LLC
Sizing of FWS Wetlands 635
C, N, AND P CYCLES
The pollutant removals simplistically described in the pre-
ceding section take no account of wetland functions; they are
based on purely empirical k-values and outlet concentration
data. It is therefore prudent to examine the empirical fore-
casts and to ascertain whether they comply with an estimated
set of wetland processes. Of particular interest are the car-
bon, nitrogen, and phosphorus cycles.
The carbon cycle involves the growth, death, and decom-
position of biological materials, including animals, plants,
algae, and microbes. For many wetlands, the standing crop
of these organic materials is relatively constant throughout
the year, but the proportions of living and dead may vary
considerably, as may the physical location (i.e., standing dead
or litter). Further, the speed of cycling varies with the nature
of the material and the time of the year (see Chapter 3). Fine
detritus from microbes cycles rapidly, whereas the decompo-
sition of some woody plant parts may take years. However,
the important feature of the carbon cycle is the amount of
material that does not decompose, for it is this residual that
accretes in the ecosystem and forms storage for many pollut-
ants, including phosphorus. This storage is relatively perma-
nent under appropriate hydroperiod conditions.
An approximate assessment of the implied impacts of
the carbon cycle can be made on the basis of two numeri-
cal characteristics: the speed of the cycle in grams of dry
material per square meter per year, and the fraction of the
cycled material that does not decompose. The speed of the
cycle may also be characterized by the standing crop, in
g/m
2
, plus a turnover rate, in number of times per year. The
recycled organic fraction contains carbon that is a source of
support for denitrication and other heterotrophic processes.
The burial fraction leads to sediment buildup and storage of
nitrogen, phosphorus, and other trace contaminants that par-
tition to organics. Typical values for the vegetative standing
crop, turnover time, speed, and burial fraction are given in
Table 17.6.
In general, the algal and microbial standing phytomass
is small compared to the vegetative phytomass. However,
turnover times are also small, so the annualized rate of
uptake and burial for this component of the phytomass may
be estimated to be 50–100% of the vegetative annual rates.
Simply stated, the amounts of N and P processed by the big
green biomass are not very much larger than those processed
by the nearly invisible microbes and algae. This nondisparity
has been characterized as the “buckets and teacups” analogy
set forth by Richardson et al. (1986).
The term nutrient poor would reect very low nutrient
status, with TP < 20 µg/L and NH
4
-N 0.2 mg/L. Nutrient-
moderate wetlands would have TP < 200 µg/L, with NH
4
-
N < 1.0 mg/L. Nutrient-rich wetlands would have TP ≈ 1.0
mg/L, with NH
4
-N ≈ 5.0 mg/L. Very-nutrient-rich systems
would have TP 5.0 mg/L, with NH
4
-N 10.0 mg/L. These
denitions do not correspond to the equivalents for aquatic
water bodies, where the dominant vegetation is plankton. In
the treatment wetland context, bacterial and algal materials
are of comparable importance with above- and belowground
macrophytic vegetation.
The magnitude of the biogeochemical cycle increases
with increasing nutrient availability—up to some presumptive
limit enforced by availability of space and sunlight. This fer-
tilizer response is not well-quantied for treatment wetlands
and, therefore, cannot be used directly in design. However,
rough estimates are of value in assessing the potential impor-
tance of the biogeochemical cycle—particularly to check that
the empirical design calculations do not imply unreasonable
ecosystem functions.
One technique for making such checks is to graphically
link a presumed cycle to the mass balance calculations avail-
able from the preliminary sizing step. The biogeochemistry
check, thus, is implemented via linked ecosystem mass bal-
ances, one each for C, N, and P. These do not replace, nor
can they substitute, the water column mass balances used in
k-rate removal calculations. In an annualized mass balance
storage calculation, the following equations may be used:
GDB
(17.17)
GX DX BX
GDB
(17.18)
BX GX
BG
B
(17.19)
where
biomass burial, g/m d
biomass deco
2
B
D
mmposition, g/m d
biomass growth, g/m d
2
2
G
X fraction of the pollutant in each biomass compartment
fraction of pollutant uptakeB that is stored
The Carbon Cycle
The nomenclature for biomass processes in treatment wet-
lands is a bit confusing. The growth, death, and decomposi-
tion processes are referred to as part of the wetland carbon
cycle, but more than carbon is involved. However, most veg-
etation and other wetland organisms are about 40% carbon;
so, either dry biomass or carbon serves to track the amount
of the material involved. Carbon itself is withdrawn from
atmospheric sources as carbon dioxide for photosynthesis.
Likewise, it is returned to the atmosphere as methane from
anaerobic mechanisms, or carbon dioxide from oxidative
processes (respiration included).
The ability to estimate nutrient cycling rests upon our
knowledge of the biomass pools in the wetlands and their
changes. Guidelines are shown in Table 17.6. According to
Equations 17.17–17.19, on an annual basis, the important esti-
mation quantities are
Annual growth rate, g/m
2
·yr (standing crop phytomass
necromass biomass times turnover per year)
Annual burial fraction (undecomposable residual
fraction)
© 2009 by Taylor & Francis Group, LLC
636 Treatment Wetlands
An assumption is made that the nutrients taken up, but not
buried as accretion, are returned to the water column of the
FWS wetland. For nitrogen, this is the maximum estimate, as
microbial processes in abovewater tissues can transfer nitro-
gen to the atmosphere without entering the water. These phy-
tomass quantities, together with phytomass nutrient content
(percentage or mg/kg), allow checks on the empirical removal
calculations.
For purposes of design, for order-of-magnitude checks
on the calculated nutrient removals, the total growth rate and
burial fraction are assumed based on the strength of the water
to be treated. This provides a rough annual estimate of the
biomass cycle (Figure 17.4). Note that the magnitude of this
cycle and the nutrient contents are a function of the degree of
fertilization of the wetland (see Chapter 3).
The wetland carbon cycle is also critical to observe per-
formance as it relates to sediment oxygen demand and to the
carbon supply for denitrication. The implied supply con-
straints of this carbon cycle are examined in the constraint
check section of this chapter.
The Phosphorus Cycle
For phosphorus, the calculated removal is represented as a
large uptake (GX
G
)—in major part balanced by the return
of soluble phosphorus from tissue decomposition (DX
D
). For
TABLE 17.6
Estimates of Wetland Nutrient Cycling in Biomass
Nutrient Poor Nutrient Moderate Nutrient Rich Very Nutrient Rich
Organic matter
Standing crop gDW/m
2
150 500 3,000 10,000
Turnover time times/year 2 4 3 2
Cycling rate (G)gDW/m
2
·yr 300 2,000 9,000 20,000
Burial fraction (b)% 2 10 20 25
Accretion (B)gDW/m
2
·yr 6 200 1,800 5,000
Phosphorus
Tissue content mgP/kg 1,000 2,000 3,000 4,000
Uptake gP/m
2
·yr 0.30 4.0 27.0 80
Return gP/m
2
·yr 0.29 3.6 21.6 60
Accretion gP/m
2
·yr 0.01 0.4 5.4 20
Nitrogen
Tissue content %dw 1.0 1.5 2.0 2.5
Uptake gN/m
2
·yr 3.00 30.0 180 500
Return gN/m
2
·yr 2.94 27.0 144 375
Accretion gN/m
2
·yr 0.06 3.0 36 125
Note: Bacterial and algal cycling are included in these values.
Source: For rst two categories, data from Davis (1994) In Everglades: The Ecosystem and Its Restoration. Davis
and Ogden (Eds.), St. Lucie Press, Delray Beach, Florida, pp. 357–378. For the second two, data from Kadlec (1997a)
Ecological Engineering 8(2): 145–172.
FIGURE 17.4 Estimated annual biomass cycle in a FWS treatment wetland for a rich nutrient condition. Note that the standing stock and
turnover refers to above- and belowground material, and to macrophytes, algae, invertebrates, and microbes.
Water
Biomass
9,000 g/m
2
· yr
9,000 g/m
2
· yr
2,250 g/m
2
· yr
6,750 g/m
2
· yr
Standing stock: 3,000 g/m
2
· yr
Turnovers per year: 3.0
Turnover rate: 9,000 g/m
2
· yr
Burial fraction: 0.250
Bulk density: 0.100 g/cm
3
Accretion: 2.25 cm/yr
Necromass
Soil
Air
© 2009 by Taylor & Francis Group, LLC
Sizing of FWS Wetlands 637
a stable ecosystem past startup, the net phosphorus removal
associated with the k-rate calculation is assigned to accretion.
Sorption and the building of additional biomass are no lon-
ger sinks for phosphorus (Kadlec, 1997a). Because the k-rate
calculation is independent of the biomass cycle calculation,
there is one degree of freedom, which is taken to be the com-
puted accretion fraction. It is known through extensive data
from treatment wetlands that this percentage ranges from
≈500 to 5,000 mgP/dry kg, or from 0.05–0.5% dry weight.
A somewhat narrower range is shown in Table 17.6 because
extremely nutrient-poor wetlands are uncommon in treat-
ment applications. It is unlikely that higher removals can be
sustained via accretion. The corresponding accretion rates
are 0.01–20 gP/m
2
·yr. The values in Table 17.6 are provided
as an order-of-magnitude check point. If a design is forecast
to exceed these approximate cycling, removal, and storage
values, then more detailed mass balances should be invoked.
If it appears that the phosphorus removal calculated by a k-
rate exceeds the ecosystem capacity to store, then the k-value
should be adjusted downward.
As a graphic illustration, the design example is car-
ried forward (Figure 17.5). Because the nutrient concentra-
tions are high (2 mg/L P and 20 mg/L TN), it is anticipated
that the nutrient-rich condition will prevail, as illustrated in
Figure 17.4. The biogeochemical cycle removes 27 gP/m
2
·yr
from the water column (9,000 gdw/m
2
·yr × 0.3% P), far more
than the loading to the wetland of 9.1 gP/m
2
·yr. The k-rate cal-
culations indicate that 7.4 gP/m
2
·yr are removed, and, hence,
it is deduced that (27 − 7.4) 19.6 gP/m
2
·yr are returned to
the water from decomposition and leaching of the biomass.
The removal to accretion is thus 7.4/27 28% of the biomass
uptake. These uptakes and returns involve aboveground plant
parts (≈50%), belowground plant parts (≈15%), and microbes
and algae (≈35%). The 28% burial rate for phosphorus is
somewhat higher than the comparison value in Table 17.6 of
20%. If phosphorus were the controlling substance for siz-
ing, it would be advisable to adjust the rate coefcient down-
ward from the value of 10 m/yr used in the design forecast.
But, the wetland is overdesigned for phosphorus reduction,
and, hence, it is probably not necessary to revise the k-rate
calculation.
The Nitrogen Cycle
For nitrogen, a more complex situation occurs as a result of the
multiple speciation of water column nitrogen. The details of
nitrogen mass balancing have been presented in Part I, Chap-
ter 9. The water column contains organic, ammonium, and
nitrate nitrogen. Their interconversions are computed from
the empirical k-rates. The biogeochemical cycle is linked in
a manner analogous to phosphorus, but with abstraction from
both the ammonium and nitrate pools in the water, and return
from both the ammonium and organic pools in the water. This
allocation recognizes a split of plant uptake between ammo-
nium and nitrate (Martin and Reddy, 1997) and the fact that
decomposition processes produce organic nitrogen. The nitro-
gen content in accreting sediments is known from extensive
data from treatment wetlands to range from ≈1.0–2.5% dry
we
ight (Table 17.6). A lower value would be associated with
nutrient-poor wetlands, a higher with nutrient-rich systems.
Again, because the k-rate calculations are independent of the
cycle calculation, there is one degree of freedom, which for
nitrogen is taken to be an assumed percentage of the cycled
nitrogen that is buried in accreting sediments. A nitrogen
decit may be assumed to be supplied by xation, a process
known to occur in nitrogen-decient wetland environments.
FIGURE 17.5 Estimated annual phosphorus cycle in a FWS treatment wetland for a nutrient-rich condition. Flows, concentrations, and
loadings leaving the wetland are from k-rate forecasts. The cycle turnover and tissue-phosphorus concentrations are presumptive. The frac-
tion of phosphorus buried as accretion is calculated by mass balance.
Total Phosphorus
in Water
Litter
Live
Soil
Influent
g/m
yr
Effluent
g/m
yr
Death
Burial
Releas!
Infiltration
&).'
Uptake
Influent Characteristics
'/*
&).'
'"
'/*
&).'
'"
E
ffluent Characteristics
*!#
-,*$!(,*$#)( $,$)(+
$)'++" .'
/*
$++-!'"%" .
-*$!
Site Conditions
© 2009 by Taylor & Francis Group, LLC
638 Treatment Wetlands
An illustration of the use of the nitrogen cycle and inter-
conversions to conrm the design k-rate calculations is shown
in Figure 17.6. The previous illustrative example is continued,
but for simplicity, the details of the k-rate calculations for the
various nitrogen species are not shown here. As indicated in
Figure 17.6, the incoming nitrogen is presumed to be mostly
organic (9 mg/L) and ammonia (10 mg/L), comprising nearly
all of the 20 mg/L of TN. The nitrogen loading to the wetland
is 91 gN/m
2
·yr, which is a low loading that places the system
in the category of an agronomic system (see Chapter 9). It is
therefore expected that the biogeochemical cycling of nitro-
gen will play an important role in the overall reduction, but
not to the exclusion of microbial processes.
Once the size of the biomass cycle has been selected,
together with the phytomass nutrient content, mass balances
x all the nitrogen uxes for the specied inows and out-
ows. For this proposed design example, the cycling of bio-
mass nitrogen is very important. The required nitrogen to
build the annually cycled biomass is 180 gN/m
2
·yr, which
is just about double the amount of nitrogen supplied in the
wastewater (Figure 17.6). However, there is no concern that
the plants will starve, because there is a correspondingly
large return ux of nitrogen from leaching and decomposi-
tion of necromass. It is again necessary to keep in mind that
these uptakes and returns involve aboveground and below-
ground plant parts as well as microbes and algae.
The ultimate fate of removed nitrogen in the example
is apportioned to accretion (56%), denitrication (21%), and
seepage (23%). Interestingly, this does not mean that there are
only small amounts of nitrication occurring, because about
two thirds of the incoming nitrogen load winds up being
nitried (61.7 out of 91.3 gN/m
2
yr, Figure 17.6) because of
leakage and a small amount of biomass uptake. The question
then arises as to whether there is sufcient oxygen supply to
support the nitrication implied by this mass balance. This
implied supply is evaluated in the next section.
17.4 CHEMICAL SUPPLY CONSTRAINTS
Traditional chemistry assumptions indicate a requirement for
carbon to support heterotrophic denitrication and oxygen to
support nitrication.
Table 17.7 lists some of the more important chemical and
biological constraints on wetland processes. The top block
of information shows that the carbon balance, and, hence,
the implied biochemical oxygen demand (BOD), is due to
removal of incoming BOD and to the carbon formed by the
biogeochemical cycle. Denitrication requires 4.0 g of chem-
ical oxygen demand (COD) to reduce 1 g of nitrate nitrogen
(U.S. EPA, 1993b; Crites and Tchobanoglous, 1998). Decom-
posing biomass is far more important than added BOD in the
water in many instances. However, some moderate fraction
of the biomass decomposition takes place in air, thus reduc-
ing the available carbon for denitrication.
The second block concerns nitrication, for which there
are associated oxygen and alkalinity requirements. An oxy-
gen equivalent is retrieved as the result of denitrication
(third block, Table17.7), as well as some of the required
FIGURE 17.6 Estimated annual nitrogen cycle in a FWS treatment wetland for a nutrient-rich condition. Flows, concentrations, and load-
ings leaving the wetland are from k-rate forecasts. The cycle turnover and tissue–N concentrations are presumptive. The fraction of biomass
nitrogen buried as accretion is assumed to be 25%.
Denitrification
Uptake
Uptake
Death
Nitrification
Ammonification
Release
Burial
gN/m
yr
gN/m
yr
!#&/5
%,$/,#
$!
6$/,#
$%
4
6$/,#
'$/,#
'"$/,#
!#&/5
%,$/,#
$!
6$/,#
$%
4
6$/,#
'$/,#
'"$/,#
+-
$2+)-)*1
/11,*3/
5
'112+$*3
)$!
6$(0.+
$2+*
© 2009 by Taylor & Francis Group, LLC
Sizing of FWS Wetlands 639
alkalinity for nitrication. The carbon produced in the wet-
land supplies some or all that is necessary for denitrication
(fourth block, Table 17.7). The accretion process implies a
rate of sediment buildup determined from the sediment bulk
density. Finally, the possibility of sulfur-driven denitrica-
tion carries a demand for sulde-sulfur.
These constraints are critical to design because wetland
chemical processes cannot proceed without the necessary
ancillary compounds. Nowhere is this more apparent than in
the need for oxygen to drive ammonia removal routes. There-
fore, if inadequate supplies of ancillary chemicals are fore-
cast, rate constants and loadings must be reduced until the
constraints are met. Alternatively, additional supplies may be
introduced into the wetland. This may require costly supple-
ments, such as the addition of methanol to fuel denitrication
(Gersberg et al., 1984). However, sometimes rearrangements
of ows can resolve the supply problem, such as the feed-
forward of high BOD unpretreated inuent for a carbon sup-
plement for denitrication (Burgoon, 2001).
OXYGEN SUPPLY
The various oxygen requirements for the example design are
calculated (estimated) in Table 17.8. Removal of BOD (0.35
gO/m
2
·d) and nitrication (0.73 gO/m
2
·d) appears to exert
only a small demand, which should easily be supplied by
atmospheric reaeration. The decomposition of the generated
necromass would consume something like 10 gO/m
2
·d, but
that decomposition is in part due to anaerobic processes and,
in part, occurs above the water as standing dead materials
oxidize. Another part of oxidation may involve root respira-
tion, which comprises plant oxygen transfer. Therefore, it is
difcult to attach great signicance to the apparent need for
large amounts of oxygen for necromass oxidation.
CARBON SUPPLY
The carbon produced by biomass in the example wetland (4.93
gC/m
2
·d) far exceeds that necessary for denitrication (0.05 gC/
m
2
·d), and in fact could be supplied easily by the reduction of
BOD (0.13 gC/m
2
·d). Thus, despite the fact that only a fraction of
the biomass-produced carbon is available in the water, there is no
apparent shortage of carbon for denitrication in this example.
Because there is a large phytomass cycle, the accretion
of solid residuals is an important factor. The implied rate of
sediment buildup is 2.3 cm/yr, which is comparable to the
rates observed in operating wetlands (Kadlec, 1997a).
INTERSYSTEM PERFORMANCE CHECKS
The question addressed next is, how does the proposed design
compare with results from existing wetlands for which there
is operating data? The annual performance period is retained
for this phase of investigation.
The intersystem loading chart for phosphorus (Fig-
ure 17.7) shows that the overall scatter follows an increasing
trend of outlet concentrations with increasing loading. How-
ever, in design, the loading variable is usually restricted to
a xed inlet concentration, whereas the hydraulic loading is
variable in response to different choices of the wetland area.
Therefore, in any particular design, only a subset of the larger
loading group is relevant. For phosphorus and most common
pollutants, a particular inlet concentration subset shows a
more modest increasing trend than that of the whole set.
A generic loading chart looks similar to the concep-
tual “cloud” indicated in Figure 17.8. The larger data group
encompasses all systems for all inlet concentrations, whereas
each subgroup at a specic range of inlet concentrations
occupies a smaller group. The left side of either data cloud
represents less efcient systems, and hence these are proto-
types of low risk for the design. The right side represents
more efcient systems, a riskier assumption. Any perfor-
mance calculation is represented by a single point on this
plot and, therefore, may be compared to the intersystem data
scatter on this chart.
The loading graphs that allow intersystem comparisons
for common constituents may be found as follows:
TSS Chapter 7, Figure 7.19
BOD
5
Chapter 8, Figure 8.9
Organic N Chapter 9, Figure 9.17
Ammonia N Chapter 9, Figure 9.37
TKN Chapter 9, Figure 9.23
Oxidized N Chapter 9, Figure 9.51
Total N Chapter 9, Figure 9.29
Total P Chapter 10, Figure 10.17
Fecal coliforms Chapter 12, Figure 12.11
•
•
•
•
•
•
•
•
•
TABLE 17.7
Supply and Demand Constraints
Process Units Basis
BOD removed gO/m
2
·d k-rate load reduction
Nitrate-N equivalent of BOD
removed
gN/m
2
·d 0.25 Mass ratio to
COD (BOD)
SOD equivalent of decomposing
biomass
gO/m
2
·d 0.1–0.5 Mass ratio
Max oxygen required for BOD
and SOD
gO/m
2
·d Sum of BOD and SOD
Reaeration gO/m
2
·d 0–4
Oxygen required for traditional
nitrication
gO/m
2
·d 4.57 Mass ratio to N
Oxygen required for anammox gO/m
2
·d 1.72 Mass ratio to N
Alkalinity required for nitrication g/m
2
·d 7 Mass ratio to N
Oxygen from denitrication gO/m
2
·d 2.86 Mass ratio to N
Carbon required for denitrication gC/m
2
·d 1.1 Mass ratio to N
Alkalinity produced by
denitrication
g/m
2
·d 3 Mass ratio to N
Biomass produced g/m
2
·d 2–50
Biomass accreted g/m
2
·d 2–25% of production
Usable carbon generated gC/m
2
·d 10% of biomass
Carbon accreted gC/m
2
·d 2–25% of production
Carbon burned gC/m
2
·d 75–98% of production
Sediment generated cm/yr 0.05–0.2 bulk density
Sulfur for denitrication gS/m
2
·d 1.69 Mass ratio to N
© 2009 by Taylor & Francis Group, LLC
640 Treatment Wetlands
After the necessary wetland area has been calculated based
on the limiting pollutant and the performance for the other
contaminants of concern recalculated for that area, the posi-
tion of the design may be plotted on each loading chart of
concern. For the example design that has been carried for-
ward, these are shown in Figures 17.9 (BOD), 17.10 (TN), and
17.11 (TP). It is seen that in all cases, the design point is close
to the central tendency of the intersystem data. That is not
an accident, because the rate coefcients used in the example
were selected to be at the 50th percentile of the intersystem
values. In general, there is no design requirement to choose
that percentile, and the loading charts can assist in interpre-
tation of the consequences of picking higher- or lower-rate
coefcients.
As should be expected, the wetland area is sensitive
to the choice of rate coefcients. The two modes of design
calculations are the area search and performance forecast
that follows once the limiting (largest) area has been found.
Figure 17.12 shows the effect of changing rate coefcients for
phosphorus during the area-seeking phase of design. Recall
that the goal was to achieve an outlet concentration of 0.52
mg/L. For a rate coefcient of 10 m/yr, this required 27.5 ha.
Picking a lower rate coefcient (2.8 m/yr) results in a higher
area (67 ha) and, thus, in a lower phosphorus loading (5.4 gP/
m
2
·yr) (Figure 17.12). Picking a higher rate coefcient (25 m/
yr) results in a lower area (12 ha) and, thus, in a higher phos-
phorus loading (30.4 gP/m
2
·yr). It is seen that the effect of
changing the estimate of the k-value moves the design point
from the left to the right on the loading chart. Note that at
low k-values for phosphorus, the area exceeds that needed for
nitrogen reduction, and, thus, the limiting constituent would
change from nitrogen to phosphorus.
In the forecast mode, the effect of changing the esti-
mate of the k-value moves the design point up and down on
TABLE 17.8
Supply and Demand Constraints for the Example
Annual Flux Daily Flux Basis
BOD removed 129 gO/m
2
·yr 0.35 gO/m
2
·d —
SOD equivalent of decomposed biomass 3,600 gO/m
2
·yr 9.9 gO/m
2
·d 32/12 Mass ratio
Max oxygen required for BOD and SOD 3,729 gO/m
2
·yr 10.2 gO/m
2
·d —
Oxygen required for nitrication 265 gO/m
2
·yr 0.73 gO/m
2
·d —
Alkalinity required for nitrication 120 g/m
2
·yr 0.33 g/m
2
·d 7 Mass ratio
Oxygen from denitrication 59 gO/m
2
·yr 0.16 gO/m
2
·d 48/14 Mass ratio
Carbon required for denitrication 19 gO/m
2
·yr 0.05 gC/m
2
·d 1.1 Mass ratio
Carbon equivalent of BOD removed 48 gO/m
2
·yr 0.13 gC/m
2
·d 12/32 Mass ratio
Alkalinity produced by denitrication 51 g/m
2
·yr 0.14 g/m
2
·d 3 Mass ratio
Biomass produced 9,000 g/m
2
·yr 24.7 g/m
2
·d —
Biomass accreted 2,250 g/m
2
·yr 6.16 g/m
2
·d 25% of Production
Carbon generated 1,800 gC/m
2
·yr 4.93 gC/m
2
·d 20% Content
Carbon accreted 450 gC/m
2
·yr 1.23 gC/m
2
·d 25% of Production
Carbon burned 1,350 gC/m
2
·yr 3.70 gC/m
2
·d 75% of Production
Sediment generated 2.3 cm/yr — 0.1 g/cc Bulk density
y = 0.954x – 3.51
R
2
= 0.778
y = 0.240x – 2.57
R
2
= 0.411
–6
–4
–2
0
2
4
6
–202468
ln (PLI, g/m
2
yr)
ln (TP Out, mg/L)
All
0.12–0.54
Linear (All)
Linear (0.12–0.54)
FIGURE 17.7 The phosphorus loading chart for FWS wetlands replotted from Figure 10.17. Log-linear regressions are shown for the entire
data set and for the inlet concentration range 0.12–0.54 mg/L.
© 2009 by Taylor & Francis Group, LLC
Sizing of FWS Wetlands 641
the loading chart. The system area is xed at that needed
for nitrogen reduction, 40 ha in the example. As seen in
Figure 17.13, if the phosphorus k-value is too low, the outlet
phosphorus concentration exceeds the maximum allowable
design value of 0.52 mg/L, and, thus, the limiting constituent
would change from nitrogen to phosphorus.
Clearly, there are a number of choices open to the
designer for selection of wetland performance. The proce-
dure set forth here advocates the use of rate coefcients,
complemented by the use of loading charts as a cross-check
on intersystem performance.
17.5 ADJUSTMENTS FOR SEASONALITY
The methods of analysis, up to this point, have utilized annual
average performance measures. However, some wastewater
constituents display seasonal removal behavior in FWS treat-
ment wetlands, as has been discussed on a case-by-case basis
in Part I. Seasonal patterns are particularly evident for nitrogen
species, driven either by temperature effects on microbial pro-
cesses or seasonal plant growth patterns, or both. Consequently,
the magnitudes of seasonal swings in pollutant reduction are
FIGURE 17.9 Location of the example design on the BOD-load-
ing graph. The upper and lower lines are the approximate bounds
of data shown in Figure 8.9. The central line represents the central
tendency of annual data.
FIGURE 17.10 Location of the example design on the TN-load-
ing graph. The upper and lower lines are the approximate bounds
of data shown in Figure 9.29. The central line represents the cen-
tral tendency of annual data.
FIGURE 17.11 Location of the example design on the TP-loading
graph. The upper and lower lines are the approximate bounds of
data shown in Figure 10.17. The central line represents the central
tendency of annual data.
0.01
0.1
1
10
100
0.1 1 10 100 1,000
TP Load (gP/m
2
· yr)
TP Out (mg/L)
FIGURE 17.8 Concepts of risk on the pollutant-loading chart. The
overall data set of comparison systems occupies a band of increas-
ing concentrations with increasing load. However, the data set for a
particular inlet concentration occupies a smaller band, oriented at a
lesser slope.
1
10
100
1,000
10,000
0.1 1 10 100 1,000
Load (g/m
2
· yr)
Outlet Concentration (mg/L)
Low risk
High risk
© 2009 by Taylor & Francis Group, LLC
642 Treatment Wetlands
sensitive to the magnitude of the annual temperature cycle or
the growth-senescence patterns of a particular climatic region.
There are three fundamentally different ways of dealing with
these effects: temperature coefcients (theta factors), monthly
rate coefcients, and outlet trend amplitudes. If theta factors
are appropriate, then the water temperature must be forecast.
WETLAND WATER TEMPERATURE
A rst approximation for long detention wetlands has been
shown to be the mean daily air temperature during the
unfrozen season (see Part I, Chapter 4). The next level of
approximation involves estimating longitudinal proles of
wetland water temperature via the use of accommodation
coefcients. The water temperature may depend on wet-
land size due to lack of time to reach the balance condition.
In that case, design iteration may become necessary. The
prediction of wetland water temperatures is briey summa-
rized here.
Bal
ance-Point Temperature
A treatment wetland may contain one or two thermal
regions, depending on water loading (detention time). For
long detention times, there is an inlet region in which water
temperatures adjust to the prevailing meteorological condi-
tions and an outlet region in which that adjustment is com-
plete. The value reached is determined by the balance of
energy ows and is termed the balance temperature. For
short detention times, near the wetland inlet, the adjust-
ment may not be completed. To a rough approximation,
wetland water balance temperatures are equal to air tem-
peratures during the unfrozen season. More accurate fore-
casts are possible utilizing the energy balance, as described
in Chapter 4.
We
tland Water Temperature Cycles
The annual cycle of wetland water temperatures in mild to
warm climates follows a sinusoidal pattern, with a summer
maximum and a winter minimum. In northern climates, the
onset of frozen conditions typically is accompanied by under-
ice water temperatures of 1–2°C. The sinusoidal model, trun-
cated for frozen conditions, is as follows:
For the unfrozen season (t
1
< t < t
2
),
TT A tt
wavg
§
©
¶
¸
1cos(-)
max
W
(4.27)
For the frozen season (t
2
< t < t
1
),
TT
wo
(4.28)
where
fractional amplitude of the sinusoidA ,, dimensionless
yearly cycle frequency =W 22 /365 = 0.0172 d
time, Julian day
ic
1
1
P
t
t
ee-out time, Julian day
freeze-up time, J
2
t uulian day
time of annual maximum tempe
max
t rrature, Julian day
water temperature, °C
w
T
TT
T
avg
annual average water temperature, °C
oo
under-ice water temperature, °C
Table 4.6 lists the parameters for this sinusoidal model for a
variety of FWS wetlands at different latitudes. The determin-
istic trend represents the central tendency of water tempera-
tures, but there are also stochastic variations. For instance,
the standard deviation of the monthly means for Columbia,
Missouri, is 1.6°C.
FIGURE 17.12 Location of the example design on the TP-loading
graph. The upper and lower lines are the approximate bounds of
da
ta shown in Figure 10.17. The central line represents the central
tendency of annual data.
k!
A!
k!
A!
k!
A!
FIGURE 17.13 Location of the example design on the TP-loading
graph. The upper and lower lines are the approximate bounds of
data shown in Figure 10.17. The central line represents the central
tendency of annual data.
k
k
k
© 2009 by Taylor & Francis Group, LLC
Sizing of FWS Wetlands 643
The Accommodation Zone
The inlet zone of a treatment wetland exhibits temperature
changes as the water approaches the balance temperature.
For short detention times (typically less than three days for
FWS), the adjustment may not be completed, and the balance
temperature is not reached (Kadlec, 2006c).
An empirical exponential model may be easily cali-
brated and used to describe the approach to the balance
temperature:
TTTT
t
ch
TTT
wb wib
p
bwib
¤
¦
¥
³
µ
´
()exp ()e
H
R
xxp
¤
¦
¥
³
µ
´
t
T
A
(4.40)
where
heat capacity of water, 4.182 10
p
6
c rJ/kg·°C
water depth, m
wetland water t
w
h
T
eemperature, °C
wetland balance temperatu
b
T rre, °C
inlet water temperature, °C
acc
wi
T
H oommodation coefficient, MJ/m ·d·°C
volu
2
p
Rc mmetric heat capacity of water, MJ/m ·°C
n
3
t oominal detention time to an internal point,, d
The quantity T
A
Rc
p
h/H represents characteristic accommo-
dation time for the wetland water on its travel through the
system, during which 63.2% of the change from inlet to bal-
ance temperature is achieved. At 3T
A
, 95.0% of the change is
accomplished. Data for free water surface wetlands indicate
that adaptation takes a detention time on the order of one to
three days (see Table 4.9).
TEMPERATURE COEFFICIENTS
Temperature effects on k are usually summarized using the
modied Arrhenius equation
kk
T
T
20
20
Q
()
(17.20)
where k
T
is the rate constant at temperature T T°C and
k
20
is the rate constant at 20°C. Values of the temperature
correction factor (Q) have been estimated for data sets with
adequate operational temperature data.
Frequency distributions for theta factors and correspond-
ing k
20
values for common constituents may be found as
follows:
BOD
5
Chapter 8, Table 8.4
Ammonia N Chapter 9, Table 9.29
TKN Chapter 9, Table 9.16
Oxidized N Chapter 9, Table 9.40
Total N Chapter 9, Table 9.20
Total P Chapter 10, Table 10.12
Fecal coliforms Chapter 12, Table 12.5
Note: Data are presently not adequate to determine
a temperature coefcient for either TSS reduction or
ammonication.
•
•
•
•
•
•
•
•
RATE-CONSTANT ADJUSTMENT
The goal of design calculation is the selection of the wetland
area that will provide the necessary treatment in all seasons
of the year. The annualized calculations that have gone before
are presumed to represent conditions roughly corresponding
to the average annual water temperature. However, it is now
necessary to adjust these rate coefcients downward to win-
ter conditions, which range downward to about 2°C.
The controlling pollutant in the example is total nitro-
gen. It is also shown that this is an agronomic wetland, with
considerable inuence of the biomass cycle. It is probably not
appropriate to use theta-factor corrections under these con-
ditions, but, nonetheless, it shall be done here for illustra-
tion purposes. From Table 9.20, the median k
20
21.5 m/yr
and the median Q 1.056. Suppose it is forecast that the low
winter water temperature is going to be 4°C. (This would
apply to the midsection of the eastern United States, e.g.,
Kentucky/Tennessee). The corresponding k
4
9.0 m/yr. The
annual calculation used is k 13 m/yr, so a larger wetland
area is needed, which is found to be 52 ha—up from 40 ha.
Note that if the temperature coefcient is large and the
winter temperature is low, e.g., 1–2°C, then the area may
become quite large, and winter storage may be warranted.
The median theta factors for some pollutants are very
close to, or even less than, 1.000, implying very little tem-
perature dependence. BOD removal in FWS wetlands has a
median of Q0.985 (Table 8.4), and fecal coliform reduc-
tion has a median of Q 0.963 (Table 12.5). Of course, sus-
tainable, postsorption phosphorus removal mechanisms rely
heavily on the balance of vegetative uptake over return, and,
therefore, agronomic models should apply. That is reected in
median Q1.005 (Table 10.12). It is various nitrogen species
that exhibit theta factors, which implies a need for tempera-
ture correction in design.
MONTHLY RATE COEFFICIENTS
As discussed in Chapter 9, temperature coefcients apply to
microbial controlled systems but not necessarily to agronomic
systems. It would be desirable to have monthly rate coef-
cients for the various sequential nitrogen conversions under
agronomic conditions, but these are generally not available
at this time. What is clear is that agronomic wetlands display
a peak of ammonia removal during the spring growth period
and that this does not correspond to the maximum of water
temperature (see Chapter 9, Figure 9.28).
The mean ammonia behavior of 11 cold-climate wet-
lands under partial or total agronomic control is shown in
Table 17.9. Note that the rate coefcient is highest in May,
which is typically the maximum growth month. Summer
is, in general, a time of greater ammonia reduction, per-
haps reecting microbial activity as well as growth in some
instances. The average total nitrogen loading to these wet-
lands was 130 gN/m
2
·yr, just over the putative limit of 120
for agronomic wetlands; the ammonia loading was a bit
lower at 90 gN/m
2
·yr. It is noteworthy that January remains
the “bottleneck” month, with a rate coefcient that is about
© 2009 by Taylor & Francis Group, LLC
644 Treatment Wetlands
four times lower than the annual average and about ten times
lower than the growing season maximum.
TREND AMPLITUDES
When contaminant concentrations are well above wetland
background levels, the seasonal variations in wetland efu-
ent concentrations may be described equally well by a P-k-
C* model or an empirical sinusoidal model. Both employ
the three constants with which the data is t. Under such
conditions, either may form the trend basis from which
the exceedance frequencies are determined. The absence
of a signicant temperature coefcient for the rate coef-
cient (different from unity) does not necessarily imply the
absence of an annual cycle in wetland efuent concentra-
tions. If the wetland is large enough to reach approximately
background concentrations, then there may be a seasonal
pattern to the outlet concentration resulting from a seasonal
pattern in C*.
However, when the wetland efuent concentration is close
to background, the value of C* prevails and the rate constant,
whether agronomic or microbial, does not control behavior.
This situation is known to occur with fair frequency for TSS
and BOD. Performance forecasting then depends on under-
standing the seasonality of C*.
It is possible to use an Arrhenius factor for C* (Kadlec and
Knight, 1996; Stein et al., 2006b). Values ranging from 0.935
< Q < 1.029 were reported for COD, depending upon vegeta-
tion type (Stein et al., 2006b). In general, C* for COD was
found to decrease with increasing temperature (Q < 1.000).
A central tendency for a TSS of Q 1.065 was reported by
Kadlec and Knight (1996), which reects an increase in C*
with increasing temperature.
The parameters for seasonal trends in TSS for several
FW
S wetlands are shown in Table 7.7, with an example cycle
shown in Figure 7.16. On an average, the magnitude of the
cycle is 39% of the mean, with a maximum in midsummer on
yearday 166. The parameters for seasonal trends in BOD for
several FWS wetlands are shown in Table 8.5, with example
cycles shown in Figure 8.15. On average, the magnitude of
the cycle is 27% of the mean, with a maximum in midsum-
mer on yearday 129.
A hypothetical annual TSS C* cycle is illustrated in
Figure 17.14. If the winter to summer temperature swing is
15°C, the corresponding swing Q 1.056 for C*. But note that
the TSS is highest in summer, and, therefore, it is the summer
peak and the associated scatter that are of concern in design.
There is no design control over the TSS background.
TABLE 17.9
Apparent First-Order Rate Constants
for Ammonia Removal in FWS
Treatment Systems in Cold Climates
Month
Rate Coefficient
(m/yr)
January 2.4
February 4.1
March 7.1
April 14.7
May 20.3
June 13.1
July 11.5
August 12.4
September 11.7
October 8.0
November 5.8
December 3.0
Annual Average 9.5
Note:
Model
parameters are P = 3 and C* = 0;
data from 11 wetlands; the mean nitrogen load-
ing was 130 gN/m
2
·yr (N = 11).
0
2
4
6
8
10
12
0 30 60 90 120 150 180 210 240 270 300 330 360
Yearday
TSS (mg/L)
FIGURE 17.14 Hypothetical scatter of TSS around a seasonal trendline. The annual mean is set at 5 mg/L, and a ± 40% scatter is indicated.
The peak TSS occurs in midsummer.
© 2009 by Taylor & Francis Group, LLC
Sizing of FWS Wetlands 645
The annual values of BOD and TSS may not reach as
low as C*, in which case the annual cycle parameters are
used to adjust the design target. The amplitude of the sinusoi-
dal trend can serve to show what “extra” concentrations may
occur seasonally. A procedure for accounting the seasonality
is to add the seasonal effect to the scatter effect. The trend
multipliers tabulated for use with the P-k-C* model can be
extended to include the seasonality. For instance, the scatter
in Figure 17.14 is 40%, and can be contained in a design using
a trend multiplier of 1.40. However, the summer peak in the
trend, 7 mg/L, is also 40% higher than the annual mean of
5 mg/L. The overall multiplier is thereby increased to 1.96,
which is used to contain both the random scatter about the
mean trend and the seasonal trend. This is the procedure
recommended in Kadlec and Knight (1996), who reported a
100th percentile total containment (trend plus scatter) multi-
plier of 1.9 based on data for 49 wetlands.
WINTER STORAGE VERSUS WINTER OPERATION
TSS and BOD reductions in treatment wetlands are approxi-
mately independent of season, and the decision for winter
storage involves only hydraulic concerns with freezing con-
ditions. But, winter rate constants are about one third and
one tenth the summer values for phosphorus and ammonia,
respectively (Kadlec and Reddy, 2001). Thus, for nutrient
removal, the winter removal-process slow-down adds to the
hydraulic concerns. There are several options for dealing
with winter slow-down, including the following:
1. Provide no winter storage, and increase the wet-
land size to meet desired winter performance.
2. Provide winter storage by adding storage lagoon
volume, but operate the wetland year-round and at
a reduced (bleed) ow in winter.
3. Provide winter storage by adding storage lagoon
volume, do not attempt to operate the wetland
in winter, and control the warm-season releases
according to water-temperature patterns.
4. Provide winter storage by adding storage lagoon
volume, do not attempt to operate the wetland in
winter, and release water from the pond and wet-
land at a uniform rate throughout the warmest
season.
Options 3 and 4 are the choices for climates so severe that
the hydraulic operation is questionable because of freezing.
In more moderate climates, where ice formation is absent, or
hydraulic operation of the wetland can continue with minor
ice formation, all options may be considered.
Winter operation is hydraulically possible for lagoons,
even for fairly severe cold, where ice thicknesses may range
up to a meter (Heaven and Banks, 2005). Indeed, most lagoon
systems receive water year-round. Winter operation is also
possible for FWS wetlands, even in moderately cold condi-
tions, because water can be managed to ow under ice. Com-
plete freeze-up leads to over-ice water ows and ice buildup.
Cold water temperatures lead to large reductions in micro-
bial processes, and the vegetation cycle is dormant in frozen
conditions. Oxygen transfer to under-ice water is blocked by
ice and snow. These individual process effects combine in
complex ways to produce the overall observed slow-down of
treatment.
Performance modeling may be used to resolve the require-
ments of the four options for pond volumes and wetland areas.
Typically, the storage options will require the least land (small-
est footprint) and the most earthmoving, whereas Option 1 (no
winter storage) will require the most land and the least earth-
moving. Controlled discharge (Option 3) is operationally the
least convenient, because it requires operator attention to set
the ows on a periodic schedule throughout most of the year.
For North American conditions, although Option 3 often leads
to capital investment economics that favor controlled dis-
charge, operational considerations almost invariably lead to a
uniform warm season release (Option 2).
An
Il
lustration
A hypothetical example of some of the design and perfor-
ma
nce factors for a large-scale system is given in Table 17.10.
The ow to the system is 4,000 m
3
/d in all seasons, and the
effects of seepage, rain, and ET are assumed negligible for
simplicity. The design target for this example is to meet a
monthly ammonia efuent concentration of 3 mg/L on an
average during all the months of the year, and this value
includes the allowance for random variability. The degree of
storage pond treatment is as predicted by procedures in the
lagoon literature (U.S. EPA, 1983c; Crites et al., 2006), and
the treatment in the wetland is computed using the P-k-C*
model with P 3, C* 0, and a temperature dependent k
value, and with k
20
0.05 m/d and Q 1.09. The pretreat-
ment facilities, upstream of wetlands and storage lagoons,
are presumed to produce 20 mg/L. Under these conditions,
the design bottleneck is winter for the no-storage opera-
tion. Storage moves the bottleneck to early spring and late
autumn, which are the times of lowest temperature during
the unfrozen discharge season. Option 2, employing a winter
bleed ow through the wetland, has a dual requirement: nei-
ther winter nor summer efuent concentrations must exceed
the target. Therefore, the wetland area is determined by the
spring and fall bottlenecks, and the winter bleed ow is set to
meet the target in winter. Option 1, year-round wetland oper-
ation, uses a wetland size that meets winter requirements.
The ow and storage time series for the various options are
shown in Figures 17.15 and 17.16. Associated efuent ammo-
nia concentrations are summarized in Figure 17.17.
The ranking of alternatives indicates an increasing foot-
print (total area) as storage alternatives move from Options 2
to 3 to 4 (Table 17.10). Option 2, use of a winter bleed ow,
has the smallest footprint, 58 ha, just a little more than half
the year-round operation footprint of 100 ha. This system
includes 87 days of lagoon storage. If the choice is to avoid
winter discharges altogether, then the best option is to utilize
a controlled release strategy. Small releases are used in early
© 2009 by Taylor & Francis Group, LLC
646 Treatment Wetlands
TABLE 17.10
Summary of Comparison of Winter Flow and Storage Options for a FWS Ammonia Reduction Example
No Storage Winter Bleed Controlled Release Uniform Release
Parameter Unit Wetland Only Pond Wetland Pond Wetland Pond Wetland
Flow m
3
/d 4,000 4,000 2,000–7,800 4,000 2,600–9,400 4,000 8,000
Inlet C mg/L 20.0 20.0 11.3 20.0 9.7 20.0 11.3
Max temperature °C 17.0 17.0 17.0 17.0 17.0 17.0 17.0
Min temperature °C 2.0 2.0 2.0 2.0 2.0 2.0 2.0
Max k m/yr 14.9 9.6 14.9 9.6 14.9 9.6 14.9
Min k m/yr 3.9 3.7 3.9 3.7 3.9 3.7 3.9
Storage days of ow 0 87 — 138 — 184 —
Working depth m 0.25 2.0 0.25 2.0 0.25 2.0 0.25
Max month outlet C mg/L 3.0 13.8 3.0 11.7 3.0 11.6 3.0
Mean outlet C mg/L 1.7 11.9 2.3 9.8 2.2 9.7 1.7
Area ha 100 18 40 28 39 38 51
Total area ha 100 — 58 — 67 — 89
Volume 1,000 m
3
250 450 100 700 98 950 128
Total volume 1,000 m
3
250 — 550 — 798 — 1,078
0
100,000
200,000
300,000
400,000
500,000
600,000
700,000
800,000
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Month
Water Volume (m
3
)
Controlled Release
Winter Bleed
Uniform Release
FIGURE 17.16 Water volumes in storage for a seasonal storage and release example.
FIGURE 17.15 Wetland ows for various strategies for the seasonal storage and release example.
0
1,000
2,000
3,000
4,000
5,000
6,000
7,000
8,000
9,000
10
,
000
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Mean Monthly Flow (m
3
/d)
No Storage
Controlled Release
Winter Bleed
Uniform Release
© 2009 by Taylor & Francis Group, LLC
Sizing of FWS Wetlands 647
spring and late autumn, and maximum release takes place
in midsummer, the time of warmest water temperature. An
area of 67 ha, containing 138 days of storage, is forecast to
achieve the required ammonia standard. However, the conve-
nience of a steady summer discharge might lead to the choice
of the 89 ha system, which has 184 days of storage. The extra
two months of storage are needed to avoid the (high) uniform
ows during early spring and late autumn, or else the wet-
land will become nearly as large as that for no storage. Land
acquisition, and perhaps liners, are area-specic costs, which
form a large proportion of the total system capital cost and,
hence, Options 2 and 3 are favored for lowering those costs.
Earth moving is another major cost in natural systems,
and there are large differences among the three alternatives
in the illustration. Year-round wetlands require the least
earth moving, and a complete winter shutdown requires the
most. Table 17.10 shows that total basin volumes (lagoon plus
wetland) of the alternatives vary widely for the example, with
the winter operation of the wetland needing only 25% of the
earth moving that a uniform summer discharge system would
require. Thus, the presence or absence of liner requirements,
or the type and cost thereof, can shift the economic advan-
tage from one option to another. This example suggests that
wetlands should be used in winter if hydraulically practi-
cal, because land and liner costs typically will not outweigh
earth-moving costs.
Wetland Systems with Storage
There are considerable numbers of treatment wetlands that
op
erate seasonally, with off-season storage (Table 17.11).
Many of these are preexisting lagoon systems with wetlands
added, so storage was already available. A common mode of
operation of lagoons and facultative ponds is pulse discharge
during spring or autumn, thus taking advantage of the maxi-
mum dilution potential while receiving surface waters. Some
or all of the needed storage capacity may, in fact, be present
prior to wetland add-ons.
17.6 STORMWATER WETLANDS
Stormwater wetlands are nearly always FWS systems because
of the need to hydraulically accommodate large-event ows
and because there is no perceived health hazard associated
with the water. Design follows one of the four following
tracks:
1. Percentage of the contributing watershed
2. Design storm detention
3. Annual average performance
4. Detailed dynamic modeling
The rst two of these are technology-based because they pro-
vide prescriptions that imply a general level of treatment rather
than specic numerical concentration or loading goals. Tech-
nology-based volume or area specications are attractively
simple because they do not require any information about
the incoming water chemistry. They suffer from an inabil-
ity to be adjusted for removal targets other than the origi-
nal presumptions. The second two are performance-based
methods, which depend on forecasts of wetland reductions of
target pollutants. Performance-based sizing calculations may
be either on an average basis, usually annual, or on a short
time scale to capture the full dynamics, usually daily. For
example, annualized performance can be related to hydrau-
lic loading and inlet TSS by using Equation 14.8 as a design
prediction tool. Exploration of this formulation shows that
it is consistent with the technology-based procedures in the
75–80% removal range.
All of the methods require knowledge of the runoff
volume characteristics of the contributing watershed; the
performance-based methods also require the incoming con-
taminant concentrations.
Associated with stormwater wetlands is the concept of
bypassing. Bypassing is necessitated when site constraints
prevent the treatment of all imaginable ows. The constraint
may occur because of limited land availability, which creates
$ " " " ! #
%"!
! *! &+
'! +(- ), (-
FIGURE 17.17 Calculated wetland efuent ammonia concentrations for different storage strategies for a seasonal storage and release example.
© 2009 by Taylor & Francis Group, LLC
648 Treatment Wetlands
a hydraulic conveyance constraint on how much water can
be passed through the system. For instance, the intense rains
of subtropical regions can occasionally create runoff ows
that simply cannot be routed through a reasonably sized wet-
land. The second constraint arises owing to the availability
of maintenance water. If the wetland is sized large enough
to accommodate major events, then the demand for mainte-
nance water during interevent periods may be too large.
Agricultural stormwater systems might differ somewhat
from urban stormwater systems because the target may be sed-
iment or nutrient control and load or concentration reduction.
Chapter 14 has provided the experience base for the
design of stormwater wetlands. Here, the design sizing pro-
cedures are outlined. These rely on source characteriza-
tion, just as for the continuous ow wetlands discussed in
Chapters 14 and 16.
INFLOW ESTIMATION
The amounts of water to be expected to reach the wetland
depend on the size and character of the watershed. Rainfall
records provide an accessible and useful means to determine
the amount of water received by the watershed, and runoff
fractions provide the resultant ows to the wetland. Run-
off coefcients vary with land use; they are typically high
(80–90%) for impervious surfaces but lower (20–30%) for
more absorbent conditions in the watershed (Table 14.1). An
example of the determination of the runoff coefcient for
particular applications is the New York calculation (Center
for Watershed Protection, 2001):
Runoff coefficient
V
V
q
RI
R
0 05 0 009
020
.
(17.21)
where
percent imperviousI
However, regional considerations can easily preempt generic
information such as the impervious cover estimates summa-
rized in Table 17.12. For instance, runoff coefcients for the
Everglades agricultural area are about 50%, not because of
imperviousness but because of extremely high water tables
TABLE 17.11
Characteristics of Pond Wetland Systems by Latitude
System Name (Years of Operation) Type
Area
(ha) Latitude
Winter
Operation
HLR
(cm/d)
TSS
(mg/L)
BOD
(mg/L)
NH
4
N
(mg/L)
TP
(mg/L)
FC
(#/100 mL)
Houghton lake, Michigan (29) P–W 200 44 N in 0.3 30 17 6.1 3.15 89
Nat (warm) out — 10 4 0.08 0.04 20
Minot, North Dakota (12) P–W 51 48 N in 3 32 16 4.5 — 381
Con (warm) out — 14 10 2.1 — 219
Cannon beach, Oregon (17) P–W 7 45 N in 25 39 31 — 12.2 19
Nat (wet) out — 9 8 — 4.6 53
Lake Nebagamon, Wisconsin (1) P–W–I 0.8 46 N in 2.7 30 10 4 — —
Con (warm) out — 10 4 0.05 — —
Bellaire, Michigan (12) P–W 7 44 N in 1.0 — — 8.54 2.78 —
Nat (warm) out — — — 0.73 0.30 —
Drummond, Wisconsin (24) P–W 6 46 N in 0.14 — — 0.13 2.36 —
Nat (warm) out — — — 0.17 0.94 —
Musselwhite, Ontario (6) P–W 2.5 54 N in 52 — 6 12.8 0.018 —
Nat (warm) out — — 5 3.9 0.012 —
Estevan, Saskatchewan (11) P–W 11 49 N in 3.0 33 9 2.82 2.49 —
Con (warm) out — 24 5 0.35 2.18 —
Saginaw, Michigan (5) M–W 8 43 N in 0.30 — — 457 — —
Con (warm) out — — — 17 — —
Isanti-Chisago, Minnesota (6) M–W 0.6 46 N in 3.5 45 — 8.9 — —
Con (warm) out — 4 — 0.76 — —
Vermontville, Michigan (35) P–W 4.6 43 N in 1.00 — — 6.5 1.8 —
Con (warm) out — — — 5 0.64 —
Mt. Angel, Oregon (11) P–W 3.6 45 N in 13.3 11 25 2.6 1.5 —
Con (wet) out — 8 14 2.2 1.4 —
Note:
Years operational period; operation is yearlong (Y), warm, wet, or dry season; P pond, AP aerated pond, W wetland, I inltration.
Source: Adapted from Kadlec (2003d) Water Science and Technology 48(5): 1–8.
© 2009 by Taylor & Francis Group, LLC
Sizing of FWS Wetlands 649
and high rainfall. Accordingly, runoff should be considered
highly site specic.
A design storm is specied according to regulatory or
negotiated considerations. For instance, the 90th percentile
of the event number distribution is used in New York (90% of
all rain events are less than this event). The rainfall amount
varies with geographical location but is approximately 3 cm
in the northeastern United States. Given the area of the water-
shed and the runoff coefcient, the volume of water that the
design event delivers to the wetland, V
design
, may be deter-
mined. This amount may then be converted to the annual
volume reaching the wetland, via the ratio of the annual total
rainfall to the amount of the 90th percentile. More detailed
estimates may be made on the basis of the variation of R
V
with
the rainfall amount, whereby very small rains are assumed to
produce no runoff. Further, by implication, larger events are
bypassed, either in total or partially.
HYDRAULIC CONSIDERATIONS
Interpretation of performance for event-driven wetlands
depends on the amount of a given event that is retained in the
system and by the length of the interevent period, rather than
by detention time or the average hydraulic loading. Deten-
tion and loading may, nonetheless, be computed if desired.
As an example, consider a wetland of volume V m
3
that
quickly receives the same volume of water as an inow. If
the wetland were in plug ow, then the new water would just
displace the antecedent wetland water. That efuent water
would have been subjected to batch treatment over the period
since the preceding event. Thus, the efuent concentration
would be determined by the concentration that entered with
the preceding event and by the detention during the inter-
event period, and would have nothing to do with the concen-
tration of the water that displaced it.
Stormwater wetlands receive pulses of widely varying
volume and frequency and are not usually anywhere close to
plug ow. The TIS model may be used to explore how much of
the water of events of different sizes will stay within the wet-
land (see Chapter 6 for details of the TIS model). In general,
the fraction of the event that remains in the wetland increases
as the number of tanks increases, to the limit of plug ow
(see Figures 14.5 and 14.6). Some existing design methods
specify the particular return frequency event for which the
runoff is to be nominally contained in the wetland (Schueler,
1992). This design event is characterized by the value 1.0 in
Figure 14.5. At this event size, the actual fraction of the
event that is retained is strongly dependent on N, the number
of TIS. It is therefore desirable to maximize this number
in the design. Walker (1998) suggested that computational
uid mechanics could be used to investigate the geometri-
cal design factors that contribute to the retention that is to
be expected. This procedure utilizes a dispersion coef-
cient to explore deviations from plug ow, and it leads to
Figure 14.6.
PERCENTAGE OF THE CONTRIBUTING WATERSHED
The amount of water and pollutants that reach stormwater
treatment wetlands from contributing watersheds are not
typically known in advance. The number and duration of the
events that produce the input, together with the interevent
spacing, are thought to affect the efciency of the stormwa-
ter wetland. Because of this variability of actual ow rates,
detention time and hydraulic loading are difcult to dene,
and other single-number sizing rules have evolved. One such
method of presumptive sizing is a specication of the wet-
land-to-watershed area ratio (WWAR). This rule of thumb
states that the wetland size should be a specied fraction of
the contributing watershed, usually in the range of 1.0–5.0%.
A little arithmetic shows that this is equivalent to the range
of hydraulic loading rates cited earlier for continuous ow
treatment wetlands. In a moderate climate,
Annual rainfall 60 cm
Average annual rain rate 60/365 0.164 cm/d
Watershed runoff coefcient 0.75
Runoff 0.75 × 0.164 0.123 cm/d
Watershed/wetland area ratio:
WWAR 1/.025 40 (for 2.5% of the watershed)
Average annual wetland HLR 40 × 0.123 4.9 cm/d
Because this average annual HLR is close to the median of
the distribution of HLRs for FWS wetlands (e.g., 4.0 cm/d for
TSS systems and 4.8 cm/d for TP systems), it is reasonable to
expect that stormwater wetlands designed in this way would
perform somewhere near the average for the emergent marsh
database set. For instance, the mean reduction for phospho-
rus in 284 FWS wetlands was 38% at a median HLR 4.8
cm/d; the mean reduction for several constructed stormwa-
ter marshes with an average WWAR 4.3% was also 57%
(Strecker et al., 1992).
TABLE 17.12
Imperviousness of Various Land Uses
Land Use Category
Mean Impervious
Cover
Agriculture 2%
Open urban land (parks, golf
courses, cemeteries)
9%
One-acre lot residential 14%
Quarter-acre lot residential 28%
Multifamily residential 44%
Institutional (schools, churches) 28–41%
Light industrial 48–59%
Commercial 68–76%
Source: Data from Center for Watershed Protection
(2001) New York State Stormwater Management Design
Manual. Report by the Center for Watershed Protection
(CWP) for the New York State Department of Environ-
mental Conservation, Albany, New York.
© 2009 by Taylor & Francis Group, LLC
650 Treatment Wetlands
Ratios of 2% for marshes alone and 1% for pond–marsh
combinations have been advanced (Schueler, 1992), but some
authorities require more. Ontario and Alberta target 5%.
However, such generalizations are dangerous as they do not
extend to other event-driven wetland applications. For exam-
ple, Braskerud (2001a) found 50–60% TSS removal in four
wetlands over a six-year period, with WWAR between 0.03
and 0.07%.
DESIGN STORM DETENTION
A stormwater wetland may also be sized to contain a specic
volume of water, usually the volume associated with a rain
event of a specied return frequency or probability of occur-
rence. For instance, Schueler (1992) suggests that wetlands
have sufcient volume to fully contain any rain event up to
the 90th percentile of the rainstorm quantity distribution.
Again, this may be shown to match the loading and detention
design ranges used for point-source wetland systems.
In the vicinity of Washington, D.C., there is 104 cm
annual rainfall, and the 90th percentile storm is 3.18 cm:
Watershed area 40 ha 400,000 m
2
Watershed runoff coefcient 0.75
Design storm runoff volume 0.75 × 0.0318400,000
9,540 m
3
Wetland area @ 0.3 m depth 9,540/.3 31,800 m
2
3.18 ha
WWAR 100 × (3.18/40) 8%
Annual ow 0.75 r 1.04 r 400,000 312,000 m
3
Average annual detention time 9,540/312,000
0.03 years 11 days
Average annual wetland HLR 312,000/31,800
9.8 m/yr 2.7 cm/d
This single-number design technique has the advantage of
allowing a variable percentage of watershed, depending on
the annual rainfall pattern and annual rainfall total. As in the
case of WWAR design, the loading and detention times cor-
respond to the median values for point-source treatment wet-
lands. It is, therefore, not surprising that Schueler (1992) lists
pollutant reductions that are in the mid-range for other treat-
ment wetlands. For instance, the total phosphorus removal
is projected to be 45%, compared to the 38% mean for FWS
systems in general.
The selection of volume or area also corresponds to a
projected level of TSS removal. For example, Schueler (1992)
suggests that 75% TSS reduction can be reliably achieved by
stormwater wetlands that capture the 90th percentile of rain
events. Although conrmation of that assumption remains
elusive, the criterion has been adopted in New York (Cen-
ter for Watershed Protection, 2001). Other states, such as
Wisconsin, have adopted similar criteria, which recommends
a design rain of 3.8 cm and suggests that this corresponds to
an 80% reduction (Prey et al., 1995).
ANNUAL AVERAGE PERFORMANCE
Another level of detail may be added via annual average
performance calculations. The average hydraulic loading is
computed from individual events of the year, utilizing pre-
cipitation records and the denition of the maximum event
to be treated. Incoming concentrations are determined from
the watershed data or from estimates. The ow-weighted
concentrations are used. Both regression equations and rate
coefcient formulations have been proposed.
Regression Equations
Duncan (1998), as referenced and reported by Wong et al.
(1999), suggested the use of regression equations for purposes
of sizing. He presents relations based on the analysis of 76
stormwater wetland systems, presumably including pumped
systems. The proposed relations are
TSS percent remaining R
i
2
78 0 80
033 049
qC
.
(14.8)
TP percent remaining R
2
12 0 71
044
q
.
.
(14.9)
TN percent remaining R
2
14 0 78
043
q
.
.
(14.10)
where
inlet concentration, g/m
hydrauli
i
3
C
q
cc loading rate, m/yr
Carleton et al. (2001) also suggest that the WWAR may be
used as a predictor of performance. Their regression equa-
tions are developed for small data sets, including natural
wetland systems, and have low correlation coefcients (R
2
≈
0.15–0.45). Within the range 0 < WWAR < 0.1, the data scat-
ter was particularly large. Regressions of the data for TSS,
TP, and nitrogen compounds in Tables 14.5–14.10 are gen-
erally unsatisfactory. For TSS, for example, the remaining
fractions increase slightly with WWAR.
Annualized Rate Coefficients
It has been suggested that the design techniques from con-
tinuous ow wetland performance may be transferred to
event-driven systems by using rate coefcients to forecast
average performance (Wong and Geiger, 1997; Carleton
et a
l., 2001). However, the rate coefcients are not nec-
essarily the same as those found for the continuous ow
systems. Stormwater data sets analyzed by Strecker et al.
(1992) and extended by Carleton et al. (2001) include about
half of the natural wetland systems. Strecker et al., in fact,
found that natural wetlands performed somewhat better than
constructed wetlands, but the use of natural wetlands is not
encouraged owing to real and perceived negative impacts
(U.S. EPA, 1993e).
The median annual areal rate constants reported by Car-
leton et al. (2001) for FWS-constructed gravity systems (N
9) are as follows:
© 2009 by Taylor & Francis Group, LLC
Sizing of FWS Wetlands 651
Total phosphorus: 8.3 m/yr
Ammonia: 5.0 m/yr
Nitrate: 6.7 m/yr
These are very low compared to the k-values reported for con-
tinuous ow systems in Part I. Analysis of the few detailed
data sets that exist indicates that higher rate coefcients than
these apply to individual events. For instance, Kadlec (2001a)
reported k 100 m/yr for phosphorus, and Werner and Kadlec
(2000b) measured 125 m/yr for nitrate, both based on pumped
event dynamics.
The designs of the 16,000 ha of constructed stormwater
treatment areas (STA wetlands) of South Florida, which target
phosphorus removal, were essentially based on annualized
rate coefcients. The prototype was an event-driven natural
wetland that produced k 10 m/yr for C* 2 µg/L and P
∞ (Walker, 1995). Subsequent analyses have shown that the
constructed systems have annualized k-values of the same
general magnitude. However, the dynamic k-values are higher
(Walker and Kadlec, 2005). Further, the k-values are sensitive
to the characteristics of the event sequences. It is likely that
more accuracy can be gained only by dynamic simulations.
DETAILED DYNAMIC SIMULATIONS
The most detailed design calculations involve dynamic simu-
lations of wetland performance, accounting for nonideal ow
patterns and the time series of ows and concentrations that
enter the wetland. This procedure has been calibrated for
some few wetlands (Wong et al., 2006), but the input param-
eters are of unknown transferability to other wetlands and
sediment types. For instance, data from stormwater wetlands
at Elbow Valley, Alberta (Amell, 2004), indicate a rst-order
removal rate coefcient for TSS on the order of 0.6–0.9 m/d,
whereas the rate coefcient found by Wong et al. (2006) were
16–21 m/d. The magnitude of this difference is not surpris-
ing, given the extreme variability in the TSS settling rates,
shown in Figure 7.8. There are some suggested procedures
that account for the size distribution of the incoming TSS
(Lawrence and Breen, 1998), but such measurements require
signicant data acquisition for the source water. In terms of
model nomenclature, the parameters k, C*, and N are not
known in advance. Wong et al. (2006) found it necessary
to calibrate the dynamic TIS model for the number of TISs.
Treatment wetland technology is not yet capable of deter-
mining that number a priori from design and layout details.
Computational uid mechanics would offer a way to do this
(Walker, 1998), but there is not yet a verication of the accu-
racy of this procedure using wetland eld data.
At present, the only generally available dynamic simu-
lation tool developed solely for the design of event-driven
wetlands is for phosphorus removal in warm climates. The
dynamic model for stormwater treatment areas (DMSTA)
computer routine is given at Web site alker.
net/dmsta/. This model has seen extensive use in the design
of constructed wetlands in South Florida, including eight
large systems of the order of 1,000 ha each. There are numer-
ous other computer codes that have varying capabilities to
deal with time-variable stormwater ows and concentra-
tions, together with some form of constituent tracking. Some
of these have been listed in Chapters 9 and 10. However,
calibrations to wetland data are typically lacking, except for
DMSTA, which has been calibrated to more than 100 wet-
land data sets, ranging from mesocosms to large full-scale
systems.
A key feature of the dynamic simulation of event-driven
wetlands is the inclusion of the storage of contaminants.
Nutrients are stored in vegetation and sediments and cycled
on a preset time scale. Movements of pollutants into and out
of storage have important effects on the long-term reductions
in these pollutants.
Detailed dynamic simulations are very data hungry.
Rather extensive time series of inows, concentrations, and
meteorology are needed. The probable rainfall sequence may
be constructed from historical records, but the inows and
concentrations will typically have to be estimated. As a con-
sequence, full dynamic simulations are likely to be restricted
to situations involving large and expensive wetlands.
FLOW EQUALIZATION VERSUS PULSE OPERATION
Event ows create large instantaneous loads on a treatment
wetlands that, if not contained, will be the cause of low
degrees of treatment. The peaking factor (PF) may be dened
in a manner similar to that used to describe the uctuations
in continuous ows, namely, as the ratio of the peak ow to
the mean ow. To illustrate the effect of ow equalization,
consider a highly idealized example in which an average of
2,000 m
3
/d at 100 mg/L is to be treated in 8 ha. The P-k-C*
values are selected to be P 3, k 10 m/yr, and C* 0. For
continuous, uniform ow, the calculated exit concentration
is 39 mg/L. Suppose the peaking factor is 10, i.e., the ow
all occurs at 20,000 m
3
/d, but only for 10% of the time. The
calculated exit concentration is then 90 mg/L. The reduction
has dropped from 61% to 10%. This example is unrealistic
for several reasons; for example, there are changes in storage
in the water body, which have not been accounted.
Next, we broaden this illustration to include storage
and (batch) treatment during the interevent periods. The
event sequences will be limited to a time series of ten-day
events, separated by periods that increase as the peaking fac-
tor increases. The average ow is maintained at 1,000 m
3
/d.
Treatment continues during the interevent period, but there
is no ow. Water from the next ow event mixes with the
antecedent water and displaces it.
It is supposed that the design should achieve 50% reduc-
tion; two alternatives are to be compared: a wetland only, and
a storage pond plus wetland. The pond is sized to fully equal-
ize the ow sequence and thus generate a uniform continuous
ow. The pond performs no treatment.
© 2009 by Taylor & Francis Group, LLC
