United States
Environmental Protection
Agency
Risk Reduction Engineering
Laboratory
Cincinnati OH 45268
Research and Development
EPA/600/S2-90/030 Aug. 1990
&EPA Project Summary
Operation of Water Distribution
Systems to Improve Water
Quality
Robert M. Clark, Judith A. Coyle, Richard M. Males, and Walter M. Grayman
The quality of drinking water can
change dramatically between the
point of discharge from the treatment
plant and the point of consumption.
To study these changes in a
systematic manner, EPA's Drinking
Water Research Division in
conjunction with the North Penn
Water Authority (NPWA) in Lansdale,
PA, developed and field tested
contaminate propagation models
over a 3-1/2 yr period.
Temporal and spatial variations in
water quality were found to be much
greater than expected. Steady-state
predictive modeling of water quality
provided insight into overall water
quality variations and patterns within
the distribution system, which
received water from multiple sources.
Point prediction of water quality was,
however, limited. Dynamic modeling,
though more difficult, provided better
insights into system behavior and
more accurately reflected changes in
water quality.
This Project Summary was
developed by EPA's Risk Reduction
Engineering 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
The research summarized here
explored the possibility of using
operational modifications of the water
distribution system (i.e., changes in
pumping, .valving, etc.), rather than
additional treatment, to improve
delivered, quality water to the consumer.
The NPWA uses multiple water sources,
including a purchased, treated, surface
water source and numerous wells. The
purchased water has a significantly
different chemical profile than does the
well water, and water quality also varies
among the wells. Water from different
sources mixes and blends within the
distribution system, changing with time.
All of this results in variations in the
quality of water delivered to the
consumer.
This U.S. Environmental Protection
Agency's Drinking Water Research
Division has developed the Water Supply
Simulation Model (WSSM), computer
programs designed to create and
maintain a data base of information
concerning a water distribution system,
so that a variety of models could be
applied to that data base. The component
of the WSSM known as the "Solver"
algorithm was developed to predict
mixing of water in a distribution system,
based on steady-state assumptions and
known hydraulic flow patterns. The
WSSM had been applied on a theoretical
basis", but had not been tested against
field data. This study was designed to
apply the predictive model of mixing to
the NPWA system, field test the
predictions, propose operational changes
to modify the flow and mixing patterns,
and test these modifications. During the
study, after a short-term sampling
-------
program indicated the dynamic nature of
quality variations, a dynamic water quality
model (DWQM) was developed. Some
operational modifications have been
made in the NPWA system based on the
data developed in this project.
The report details the significant use of
computer modeling and the development'
of a number of generally applicable
models. It also details the testing of two
automated samplers that are capable of
preserving volatile organics. -
The North Perm Water Authority
The NPWA serves 14,500 customers in
19 municipalities in Montgomery County,
north of Philadelphia. Water sources for
the 5-mgd system include 1 mgd of
treated surface water purchased from the
Keystone Water Company and 4 mgd
from 40 wells operated by NPWA. The
system is largely contiguous; a few
unconnected satellite systems were not
modeled in this study. The system has
five storage tanks and two pumping
stations. Figure 1, a schematic
representation of the 225 miles of pipe in
the contiguous NPWA distribution
system, shows the location of wells, the
Keystone "tie-in," and the three pressure
zones (Souderton Zone, Lansdale Low
Zone, and Hillcrest Zone): Pipe sizes
range from 3 to 24 inches but the
majority of the pipe is cement-lined
ductile iron or unlined cast iron.
Surface water enters the NPWA
system at the Keystone tie-in. The rate of
flow into the system is determined by the
elevation of the tank in the Keystone
system and by a throttling valve at the
tie-in. Flow, which is monitored
continuously, is relatively constant; the
throttling valve is adjusted seasonally.
Water flows into the Lansdale low
pressure zone and from there enters the
Lawn~"Avenue tank, from which it is
pumped into the Souderton Zone.
Additional water, solely derived from
wells in the Hillcrest Pressure Zone,
enters the Lansdale system at the Office
Hillcrest transfer point. Except for unusual
and extreme circumstances, such as fire
or main breaks, water does not flow from
the Souderton Zone in the Lansdale Low
Zone, nor from the Lansdale Low Zone
into the Hillcrest Zone. This study
emphasized modeling water flow in the
Lansdale Low Zone, which is the largest
portion of the distribution system.
The well pumps operate on time
clocks. Although some wells pump
continuously, some are not pumped
during the late evening and early morning
hours. At these times, Keystone water
moves further into the distribution
system.
The chemical characteristics of
Keystone water differ from those of the
well waters. Keystone water contains total
trihalomethanes (TTHM) at significantly
higher levels than does the well water.
Certain wells show the occurrence of
trichloroethylene (TCE), or cis-1,2-
dichloroethylene (cis-1-2-DCE), or both.
Inorganic chemicals also vary from well
to well and between the wells and
Keystone water.
The NPWA distribution system is well
instrumented, with detailed continuous
records available on well pumpage, tank
heights, and flow at various locations with
the system. In addition, NPWA maintains
a detailed water quality sampling
program, both at sources and in the
distribution system; thus, there is
significant historical data available on
system water quality.
Description of the Study
Steady-State Prediction of
Mixing
With the use of WSSM, a steady-state
prediction of mixing within the distribution
system was obtained. This effort required
the following steps: after establishing a
WSSM data base for the NPWA
distribution system and developing and
parameterizing a hydraulic model of the
NPWA distribution system as well as
hydraulic "scenarios," hydraulic
information obtained from selected
scenarios was stored within the WSSM
data base. Running the Solver module of
the WSSM provided steady-state
prediction of water mixing.
The steady-state algorithm as applied
in this case did not take into account the
concentration of Keystone water resident
in the tanks. According to the late night
scenario, at least some Keystone water
will enter the tanks and then be
discharged into the distribution system as
another source of Keystone water. This is
a more complex analysis, but, in the
majority of the hydraulic scenarios
selected in this study, tanks were filling,
rather than discharging to the system.
The steady-state analysis inherently
cannot take into account water already in
the pipes of the distribution system. For
example, the late night scenarios, in
- which wells are off, results in Keystone
water flowing throughout the system. In
fact, much of the water in the pipes at
that time is well water; the pipes act as a
reservoir for this water, even though the
wells are off. Obviously, blending takes
place, rather than water being either
well water or fully Keystone water.
volume of water in the pipes of iw
distribution system (as represented in the
WSSM, i.e., neglecting smaller pipes)
was calculated to be 2.8 million gallons,
or over half of the average daily supply of
5 million gallons for the NPWA
distribution system. The reservoir effect,
which results when the wells are shut off
at night and Keystone water enters pipes
partially filled with well water, is likely to
be significant and cannot be taken into
account in traditional steady-state
modeling. This further confirms the
desirability of dynamic quality modeling.
Comparison of Steady-State
Prediction with Historical Data
Solver module predictions were
compared with water quality sampling
data to assess the spatial variations in
quality. Average values for specific water
quality variables, at defined locations,
were calculated. Because the TTHM
formation of the purchased surface water
was assumed to have achieved a steady-
state relationship and the TTHM
production for the other sources in the
system was assumed to be zero, TTHMs
were used as a predictor of mixing and
blending of surface with well water,
Solver predictions in general agree
the analyzed historical data. "Point-
prediction" capacity (the ability of the
model to accurately estimate the level of
a pollutant at a specific point in the
distribution system) was, however,
limited.
Short- Term Sampling Program
A short-term, pilot sampling program
was undertaken to characterize more
accurately the spatial (space) and
temporal (time) variations of quality in the
water distribution systems. Six sites
throughout the distribution system were
sampled at 4-hr intervals for a 36-hr
period. This sampling program showed
significant variations in TTHMs at the
Keystone surface water source and
obvious blending and mixing of the well
and surface water in the distribution
system and, as predicted by the steady-
state modeling analysis, one location
alternately fed by well water and by
surface water.
The pilot sampling program clearly
showed the dynamic nature of quality
variation within the system and led to
methods to predict dynamic quality. In
addition, the difficulty in obtaining manual
samples led to development of
automated samplers.
2
-------
Dynamic Quality Modeling
•forts
Sequential Steady-State
Modeling
The initial attempt to model the
dynamic quality variations found in the
sampling study was based on
successively applying the steady-state
Solver technique to different demand
patterns and boundary conditions. This
technique was implemented in
developing the quality model by
designing different steady-state hydraulic
scenarios, which were representative, of
different time periods within the sampling
study, and successively predicting
hydraulics and the associated water
quality for each period. This technique,
called sequential steady-state modeling,
again yielded reasonable results in terms
of overall patterns. Theoretically,
however, this approach fails to take into
account the water that is resident in the
pipes of the distribution system which is
obviously in a blended/mixed state, and
is extremely cumbersome to handle
logistically.
Development of Dynamic Water
Quality Model/Flow Tracing
Algorithm
A dynamic water quality model
(DWQM) was developed that would better
represent the variations observed in the
water quality. DWQM is based on a
routing technique that continually
accounts for the quality of water in pipes
within the distribution system. As with the
steady-state model, the DWQM relies on
externally available, detailed hydraulic
information relating to flows in pipes and
at nodes available for each time step in
the quality simulation. The DWQM is
based on flow tracing rather than on the
simultaneous equation solution used in
the Solver algorithm. When the DWQM
was tested against the conditions of the
sampling study, an extended-period
hydraulic simulation was developed that
showed good agreement between
predicted and actual tank levels and the
pressures in the system. This was then
used to model the quality variations
which again showed good agreement
over the 36-hr sampling run.
Conclusions and
Recommendations
Steady-State Predictive
Modeling
Based on the results to date, steady-
state predictive modeling appears to be a
reasonable first step to characterize the
distribution of water quality in multisource
systems. Although point prediction
capability is probably not accurate, and
probably highly sensitive to hydraulic
assumptions, general trends can be
established, and then verified, through
field sampling.
Steady-state prediction might be better
suited to systems that are less dynamic
in terms of operation, with fewer sources,
than is the NPWA distribution system.
Steady-state assumptions might be more
consistent with a less dynamic system.
Because of the manner in which the
NPWA system was developed, by
connecting of a number of separate
systems, it is somewhat disjointed, a
number of portions of the system are
connected by a single pipe. Once the
hydraulic solution establishes flow
direction, the system is effectively
partitioned, at least in certain areas, into
zones of uniform concentration.
Dynamic Water Quality Model
Results of the pilot sampling study
clearly showed the spatial and temporal
variability of water quality in the NPWA
distribution system. Because little
information was available on short-term
spatial/temporal variability for distribution
systems (most monitoring strategies
involve taking samples at daily, monthly,
or quarterly intervals, at a selected
number of sites) within-day quality
variations were seldom noted.
Although steady-state prediction of
quality does provide certain insights into
system behavior, dynamic water quality
models, though more demanding of input
data describing the system, more
accurately reflect changes in water
quality.
The DWQM developed in this study is
relatively simple to use and can easily be
extended to more complex situations of
nonconservative constituents. As applied
to the NPWA pilot run, good agreement
between predicted and actual results
were found.
When using DWQMs, two major
problems arise: the need to manage the
large quantity of data that the models use
and the need for improved methods of
data analysis and display.
Dynamic water quality models both
generate and use large quantities of data.
Detailed descriptions of dynamic
hydraulic behavior are required, and the
models can generate extensive
information on within-pipe and nodal
water quality over time. Improvements in
data handling, and development of
techniques for analyzing and displaying
data reflecting spatially varying, dynamic
water quality, would contribute greatly to
ease of using dynamic water quality
models.
Monitoring Needs
A clear result from this research is the
need to obtain more representative
monitoring results than are normally
acquired from distribution system
sampling. Contaminant values can vary
greatly over a relatively short time at a
given point. There are also no doubt
weekly and yearly cycles that, when
combined with hydraulic and mixing
variations, will have great effect on the
contaminant levels at a given point in a
distribution system. Existing compliance
monitoring strategies probably fail to truly
reflect population exposure because they
are based on assumptions about quality
behavior in distribution systems that
imply essentially steady-state
characteristics.
The use of automatic samplers, such
as those developed and tested in this
study, should prove to be of significant
value in rapidly characterizing the degree
of temporal/spatial variability of water
quality in distribution systems. It is
expected these samplers can, with minor
modifications, be useful field instruments.
Extension of the approach to bacterio-
logical sampling would also prove very
useful.
Having the tools to predict time-of-
travel between points in a system and to
estimate the quality of water provided to
any point from any source would allow for
realistic water quality monitoring
strategies.
The full report was submitted in
fulfillment of Cooperative Agreement No.
811011 by North Penn Water Authority
under the sponsorship of the U.S.
Environmental Protection Agency.
-------
The EPA author, Robert M. Clark (also the Project Officer, see below), is with Risk
Reduction Engineering Laboratory, Cincinnati, OH 45268; Judith A. Coyle is
with North Penn Water Authority, Lansdale, PA 19446; Richard M. Males is with
RMM Technical Services, Inc., Cincinnati, OH 45208; and Walter M. Grayman is
with Walter M. Grayman Consulting Engineers, Cincinnati, OH 45229.
The complete report, entitled "Operation of Water Distribution Systems to Improve
Water Quality," (Order No. PB90-246 539/AS; Cost: $17.00, subject to change)
will be available only from:
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Risk Reduction Engineering Laboratory
U.S. Environmental Protection Agency
Cincinnati, OH 45268
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
BULK RATE
POSTAGE & FEES PAID
EPA
PERMIT No. G-35
Official Business
Penalty for Private Use $300
EPA/600/S2-90/030
------- |