United States
Environmental Protection
Agency
Municipal Environmental Research
Laboratory
Cincinnati OH 45268
Research and Development
EPA-600/S2-83-049 Aug. 1983
SER& Project Summary
Combined Sewer Overflow
Characteristics from Treatment
Plant Data
James A Mueller and Dominic M. DiToro
Research was undertaken to eval-
uate the adequacy of using a mass
balance technique with daily munici-
pal wastewater treatment plant data to
determine combined sewer runoff and
overflow characteristics.
An hourly simulator was used to
generate known runoff and overflow
concentrations as well as plant con-
centrations, similar to raw wastewater
data. The daily balance technique was
used to analyze the treatment plant
data that compared the calculated with
the known runoff and overflow con-
centrations.
The bias and variability associated
with the mass balance technique are
presented together with a theoretical
analysis of the effects of plant meas-
urement error. Also given are the unit
loads and average concentrations from
the New York City 26th Ward Treat-
ment Plant area and the effects of
rainfall characteristics on combined
sewer runoff concentrations.
This Project Summary was developed
by EPA's Municipal Environmental Re-
search Laboratory, Cincinnati, OH, to
announce key findings of the research
project that is fully documented in a
separate report of the same title (see
Project Report ordering information at
back).
Introduction
Assessing the magnitude and charac-
teristics of urban runoff loads for specific
regions is a difficult task because of the
random nature of storm events. Tech-
niques typically used to evaluate urban
runoff inputs include (1) direct sampling
of storm overflow concentrations and
flows, and (2) use of a stormwater quality
model based on land use and rainfall
characteristics. Because of the highly
variable nature of rainfall and associated
runoff phenomena, an extensive sampling
program is generally needed for the first
technique to provide accurate estimates of
overflow loads. But such sampling would
be costly and time-consuming. Further-
more, the first technique may lead to
significant errors If it is based on the
default values incorporated into the models.
To obtain reliability in the latter approach,
the models must be calibrated for specific
areas, normally by direct sampling of
stormwater overflows. Or existing data
bases (namely, treatment plant influent
data) can be used to determine combined
sewer overflow loads. These bases should
provide municipalities with an alternative
method to assess the importance of their
combined sewer overflows rapidly and
economically when formulating water
quality management plans.
Approach
The objective of this research was to
evaluate the adequacy of a mass balance
technique using treatment plant influent
data to determine the magnitude of com-
bined sewer runoff and overflow loads.
The initial concept of using treatment
plant data to obtain these loads was
developed to evaluate the relative impor-
tance of urban runoff inputs to New York
Bight The mass balance method is a
mathematical framework consisting of
mass and flow balances for the sewer
system and regulators over the total drain-
age area served by a treatment plant
I nputs to the sewer system include the dry
weather sewage flow, runoff into the com-
-------
bined sewer system during storm events,
and tidegate leakage. Outputs from the
system include the wastewater flow to the
treatment plant and the combined sewer
overflow from the regulators to the re-
ceiving waters. During the overflow event,
the quality of the overflow from the regu-
lators is assumed .to be equal to that of the
treatment plant influent Since daily treat-
ment plant data are normally available,
hourly mass and flow balance equations
are integrated over the sampling day to
provide estimates of the temporal and
areal average daily overflow and runoff
concentrations. The runoff concentration
includes the contribution from both sur-
face runoff and interceptors. The initial
study used data from the 26th Ward Plant
in New York City. A large degree of vari-
ability in the daily runoff and overflow
concentrations resulted. Values over a
number of years were used to characterize
loads from the drainage area. Partial
verification was obtained by comparisons
with existing combined sewer sampling
data from portions of the area.
The present study was conducted to
determine the bias and variability associ-
ated with the technique and to evaluate
modifications required to provide max-
imum accuracy for the available data base.
The approach taken was to develop an
hourly simulator in which all influent char-
acteristics (both dry weather sewage and
runoff) were known. The daily composite
simulator output was analyzed by the
mass balance technique and compared
with the known inputs. This comparison
served as the basis for modifying the
computational technique. Two modifica-
tions were developed: One employing
equal volume plant sampling similar to the
New York City sampling technique, and
the other using real time, which allows
rainfall events to be correlated with dry
weather sewage diurnal variability. The
study evaluated the effects of errors in (1)
estimating dry weather sewage charac-
teristics and runoff volumes, and (2)
measuring plant concentration. The tech-
nique was examined for its ability to ex-
tract the effects of rainfall characteristics,
the interval between storms, and the
storm duration from runoff loads. Both the
New York City sampling routine (every 4
hr, skipping the 2 a.m. sample) and an
hourly sampling routine were studied in
this regard.
The modified computational techniques
were used on the existing 26th Ward data
from New York City to evaluate the impact
of the improved methodology on the run-
off and overflow load estimates. A litera-
ture review and letter survey were also
conducted to evaluate the nationwide
applicability of the methodology.
Discussion of Findings
A mass balance technique was eval-
uated for its ability to use treatment plant
influent data to determine accurately the
overflow loads and runoff characteristics
from combined sewers. An hourly simu-
lator was used to generate known runoff,
overflow, and plant influent concentra-
tions. The plant influent data generated by
the simulator were analyzed by the daily
mass balance technique to determine con-
centrations, which were compared with
the true values. This comparison provided
the basis for analyzing the bias and vari-
ability associated with the technique.
The initial results showed that a sig-
nificant bias existed when interceptor
capacity was greater than dry weather
flow if a flow-weighted analysis of influent
data was used on plant composite sam-
ples collected in equal volumes. The bias
was removed by modifying the technique
to an equal-volume analysis of the plant
composite samples. Variability resulting
from the averaging technique was mini-
mized by using the hourly dry weather
concentrations coinciding with the time of
the storm.
The variability in calculated runoff and
overflow concentrations resulting from
plant measurement error is significant A
theoretical analysis of the error structure
indicated that the variability of the runoff
estimates was greater than that for the
overflow estimates. The variability in the
individual concentrations could be reduced
by deleting low-average storm intensities
(<0.03 inyhr) and low storm durations
that provided only one wet sample at the
plant. But excluding lower duration storms
from the mass balance analysis reduced
the capability of extracting first flush
effects from the data Random variability
in hourly dry weather sewage concen-
trations (using standard deviations of 10
and 20 percent on the hourly values) was
significant but somewhat lower than that
resulting from measurement error. A
summation of the variance of each of the
individual errors provided an excellent
estimate of the total variance of the esti-
mated runoff and overflow concentrations.
For the New York City sampling mode,
linear regression analysis was used to
evaluate the ability of the mass balance
technique to analyze the effect of rainfall
characteristics on runoff concentrations
when both averaging and measurement
errors were present The actual effects
that both interval and duration had on the
average storm runoff concentrations (pro-
vided by the simulator) were successfully
obtained from an analysis of the daily plant
data. Approximately 1 50 to 200 days of
data are required to ensure that the con-
fidence limits on the interval effect (as
measured by the slope of the regression
curve) are above zero when the runoff
concentrations are significantly affected
by a first flush. The correlation coefficients
obtained from these regressions are low,
explaining only 3 to 14 percent of the
observed variability. The remaining vari-
ability results from the averaging and
measurement errors inherent in the anal-
ysis and not from the random variability of
runoff concentrations. Thus the mass
balance technique can accurately predict
effects of duration and interval on storm-
weighted average runoff concentrations.
In the simulated runoff data, the first
flush effect was limited to the first hours of
the storm events, with background levels
attained after 3 to 4 hr. So when short
storms were neglected in the analysis,
lower runoff concentrations resulted, with
regression parameters similarly reduced.
Thus to properly evaluate the first flush
effects on runoff characteristics, short-
duration storms had to be included in the
analysis.
Collecting samples every hour (instead 4
of the New York City sampling routine of ™
every 4 hr, skipping the 2 am. sample)
caused a higher degree of variability in the
results, especially when short-duration
storms were analyzed. The reason is that
durations of 1 or 2 hr result in fewer than
10 percent of the collected samples re-
flecting wet-weather conditions Analyzing
durations that are only equal to or greater
than 4 hr for the hourly sampling routine
provided results similar to analyzing all
plant data sampled by the New York City
routine, as long as a runoff event occurred
during a plant sampling time. Thus a plant
sampling routine based on hourly sam-
pling reduces the capability of evaluating
runoff and overflow characteristics from
plant data
The actual daily data from the 26th
Ward Plant in New York City were then
analyzed using the hourly mass balance
technique. Unit loads were similar to those
for the previous flow-weighted daily mass
balance analysis, with the exception of the
soluble BOD5 data, which were signifi-
cantly lower than previously estimated.
For these estimates, the hourly variability
in dry weather concentrations for all four
parameters (SS, VSS, BOD., soluble BOD5)
was taken from the BOD5 variability.
Interval and duration significantly affected .
runoff concentrations. For the 26th Ward \
data, similar first flush effects were ob-
-------
tained when both 1- and 2-hr-minimum-
duration storms were analyzed, with a
higher correlation coefficient for the latter.
Plant data analysis using a mimimum
storm duration of 2 hr and minimum
average intensity of 0.03 in./hr provided
the best estimates of average runoff and
overflow concentrations as well as of the
effects of storm characteristics on runoff
concentration.
Conclusions
The following conclusions have been
drawn from the study:
• Average annual runoff and overflow
loads and concentrations can be ob
tained by using a mass balance anal-
ysis of long-term influent data from
treatment plants with combined
sewer systems.
• To remove bias from the analysis,
the original flow-weighted mass
balance technique must be modified
to reflect the type of composite sam-
pling being conducted at the plant.
For the New York City plant sampling
routine, an equation based on equal
sample volumes is required for the
measured plant concentrations.
• Individual estimates of daily runoff
and overflow concentrations have a
high degree of variability because of
subtractions inherent in the mass
balance technique. Thus, lone data
bases are required to provide good
estimates of average loads.
• An hourly mass balance technique
using dry weather hourly sewage
concentrations and flows with hour-
ly rainfall intensities reduces averag-
ing errors inherent in the daily anal-
ysis. But for evaluation of longterm
BOD5, suspended solids, and volatile
suspended solids for the 26th Ward
Plant in New York City, the extra
complexity of the hourly analysis
was not justified because average
loads were not significantly different
Such was not the case for soluble
BOD5, which was significantly lower
for the hourly analysis. In the hourly
analysis of the 26th Ward data, the
diurnal variability of all dry weather
constituents was assumed to be
similar to that for BOD5, for which
data were available.
• Measurement error associated with
plant concentrations causes a major
portion of the variability in estimated
runoff and overflow concentrations
using the hourly analysis. Other
causes of this variability are fluc-
tuations in hourly dry weather sewage
concentrations and within-storm
hourly runoff concentrations.
Runoff concentrations can be reliably
related to rainfall characteristics if a
sufficient length of record is anal-
yzed (with the hourly analysis pro-
viding greater reliability than the
daily analysis). The manner of sample
collection and compositing signifi-
cantly affects the length of record
required. For example, hourly sam-
pling for the daily plant composite
requires about 400 days of data.
whereas sampling at 4-hr intervals
would require approximately 150
days of data.
• Use of the mass balance technique
to obtain drainage area integrated
runoff and overflow concentrations
from plant influent data should pro-
vide significant cost savings when
laboratory analytical costs are high,
as in the case of the toxics.
The full report was submitted in fulfill-
ment of Grant No. R806519 by Manhattan
College under the sponsorship of the U.S.
Environmental Protection Agency.
James A. Mueller and Dominic M. Di Toro are with Manhattan College, Bronx, NY.
Douglas C. Ammon is the EPA Project Officer (see below).
The complete report, entitled "Combined Sewer Overflow Characteristics from
Treatment Plant Data," (Order No. PB 83-224 543; Cost: $13.00, subject to
change) will be available only from:
National Technical Information Service
5285 Port Royal Road
Springfield, VA22J61
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Municipal Environmental Research Laboratory
U.S. Environmental Protection Agency
Cincinnati, OH 45268
ftUS GOVERNMENT PRINTING OFFICE' 1983-659-017/7164
-------
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
Postage and
Fees Paid
Environmental
Protection
Agency
EPA 335
Official Business
Penalty for Private Use $300
LUU rt TlLLtY
HEGiON v hPA
LItiKARIAfo
230 S UEARbUKN
CHILAGU IL
ST
------- |