Field Data Collection Activities and
Water Distribution System Modeling
Three reasons exist for initiating a synoptic, system-wide collection of hydraulic and operational data. First, after conducting preliminary simulations using an equivalent network representation of the water-distribution system, results indicated higher than expected pressures, exceeding 125 psi in some locations (in the southernmost areas of Dover Township including the borough of South Toms River and Berkeley Township). Measured data were not available to either confirm or negate these initial simulation results. Second, to understand the present-day distribution of water from points of entry (sources) to locations throughout the distribution system, a calibrated model of the distribution system needed to be developed; however, a database of spatially and temporally varying data by which to characterize the distribution system was not available. Third, to reconstruct historical characteristics of the water-distribution system, a synoptic, system-wide characterization of the water-distribution system, based on measured data, was required. Neither present-day nor historical system-wide pressure measurements were available for the water-distribution system being investigated. Hence, ATSDR investigators decided to obtain present-day measurements to accurately characterize the water-distribution system.
ATSDR, in coordination with NJDHSS and the water utility, developed a protocol to collect pressure data and operational information during winter- and peak-demand periods of the year March and August 1998, respectively. Details of the protocol are provided in a report by Maslia and Sautner (1998a) and are briefly described below.
Hydrant Selection
Twenty-five hydrants (out of a system total of 2,127 in 1997) were
initially selected as test hydrants (designated as H-1, H-2, etc.) on which
continuous pressure recording-equipment (described below) would be installed.
The number and location of the proposed test hydrants (H-1, H-2, etc.,
shown in Plate 4) were selected based on the following:
could not be used to monitor pressure during the test periods (see Plate 5 for location of alternate test hydrants). During installation of the data-gathering equipment on the hydrants, it was determined that 5 of the designated test hydrants (H-5, H-6, H-14, H-19, and H-22; Plate 4) were not suitable for use as measuring points. Therefore, associated alternate hydrants (AH-5, AH-6, AH-14, AH-19, and AH-22; Plate 5) were used instead. The 5 hydrants (H-5, H-6, H-14, H-19, and H-22) were not used because when they were identified and turned on, investigators noticed excessive leakage flowing from the bolts joining the hydrant base to the underlying pipeline. Investigators felt that the excessive leakage would worsen with a pressure gauge or pressure recording device attached to the hydrant, resulting in erroneous pressure measurements. The final set of test hydrants used for both the March and August 1998 tests is shown in Plate 6. These 25 hydrants are connected to network pipelines that: (1) were installed between 1963 and 1996, (2) are constructed of asbestos cement (14 hydrants), plastic (PVC, 10 hydrants), and ductile iron (1 hydrant) materials, and (3) range in diameter size from 6 in. to 12 in.
Table 2. Identification, coordinates, and
location of test hydrants used for the March and August 1998 pressure tests,
Dover Township area, New Jersey
ATSDR
Hydrant ID1 |
2UWTR
Hydrant ID |
2Pipe
Diameter (inches) |
2Year
Installed |
Hydrant Location | 5Land
Surface
Elevation (feet) |
||||
3Geodetic
Coordinates
NAD 1927 |
New Jersey State Plane Coordinates, NAD 1927 | Street Identification4 | |||||||
Latitude
(decimal degrees) |
Longitude
(decimal degrees) |
Northing
(feet) |
Easting
(feet) |
||||||
H-1 | 6-075 | 8 | 1993 | 39.953265 | -74.286958 | 408142.264 | 2106457.294 | Millbrook Dr. & Westbrook Dr., NW corner | 49.42 |
H-2 | 6-014 | 8 | 1983 | 39.977381 | -74.283213 | 416931.503 | 2107469.462 | Costa Mesa Dr. & Pine Valley Dr., SE corner | 55.77 |
H-3 | 1-863 | 12 | 1989 | 39.989508 | -74.265614 | 421370.722 | 2112381.845 | Route 37, W. Floracraft @ no. 1600 | 59.91 |
H-4 | 6-173 | 8 | 1974 | 39.964666 | -74.245314 | 412347.723 | 2118112.826 | Pembroke Ln. & Fort de France Ave., SE corner | 29.92 |
AH-5 | 6-227 | 12 | 1988 | 39.945245 | -74.240738 | 405279.324 | 2119429.296 | Prince Charles Dr. & Davenport Rd, NE corner | 51.60 |
AH-6 | 1-156 | 6 | 1960 | 39.971423 | -74.221675 | 414841.283 | 2124726.787 | Oakside Dr. & Shady Nook Dr., NW corner | 44.14 |
H-7 | 1-304 | 12 | 1966 | 39.957666 | -74.212299 | 409843.256 | 2127380.239 | Lakehurst Rd. & Edgewood Dr., SW corner | 32.86 |
H-8 | 5-012 | 10 | 1988 | 39.950151 | -74.199608 | 407124.136 | 2130952.345 | South Main St. & Flint Rd., SE corner | 6.92 |
H-9 | 1-061 | 6 | 1991 | 39.952620 | -74.193779 | 408032.128 | 2132581.850 | Washington St. & Hooper Ave., NE corner | 32.68 |
H-10 | 1-078 | 12 | 1976 | 39.953508 | -74.170042 | 408391.758 | 2139234.937 | Washington St., across from Pine St. | 35.68 |
H-11 | 1-731 | 8 | 1988 | 39.947713 | -74.145932 | 406319.347 | 2146006.636 | Elizabeth Ave. & Berkeley Ave., NW corner | 13.91 |
H-12 | 1-702 | 8 | 1988 | 39.944769 | -74.121910 | 405287.147 | 2152748.462 | Minturn Rd. & Bay Shore Dr., NW corner | 4.26 |
H-13 | 1-665 | 12 | 1988 | 39.952430 | -74.118217 | 408084.170 | 2153766.796 | Marshall Rd. across from Maritime Dr. | 6.56 |
AH-14 | 1-762 | 8 | 1985 | 39.958156 | -74.146045 | 410123.244 | 2145952.755 | Windsor Ave. & Huckleberry Ln., SW corner | 28.27 |
H-15 | 1-591 | 12 | 1986 | 39.968498 | -74.122447 | 413930.025 | 2152545.164 | Bay Ave. & Bermuda Dr., SE corner | 6.18 |
H-16 | 1-973 | 8 | 1996 | 39.973560 | -74.206719 | 415640.998 | 2128914.673 | S. Dakota Ave. & N. Carolina Ave., NE corner | 38.66 |
H-17 | 2-332 | 6 | 1987 | 39.992107 | -74.206319 | 422397.734 | 2128991.902 | Indian Head Rd. & Hill Grass Ct., NW corner | 89.30 |
H-18 | 2-124 | 12 | 1984 | 39.999978 | -74.228648 | 425233.402 | 2122721.279 | Whitesville Rd. & Clayton Ave., N corner | 46.76 |
AH-19 | 2-348 | 12 | 1987 | 40.020870 | -74.220673 | 432854.892 | 2124917.511 | Route 9 Hwy. & Riverwood Dr., NW corner | 71.78 |
H-20 | 2-720 | 12 | 1996 | 40.050572 | -74.250934 | 443633.608 | 2116391.508 | Whitesville Rd. & Clear Lake Blvd., NW corner | 82.68 |
H-21 | 2-171 | 12 | 1974 | 40.009504 | -74.215593 | 428721.714 | 2126361.277 | Church Rd. & Rte 9 , NE corner | 71.24 |
AH-22 | 2-025 | 8 | 1965 | 40.018729 | -74.154675 | 432174.358 | 2143406.735 | Hovsons Blvd. & Adirondack Pl., NE corner | 14.17 |
H-23 | 2-093 | 12 | 1963 | 40.003062 | -74.154607 | 426467.378 | 2143458.575 | Fisher Blvd. & Hooper Ave., SE corner | 10.50 |
H-24 | 1-400 | 12 | 1971 | 39.992985 | -74.173007 | 422767.509 | 2138324.083 | Indian Hill Rd. & 300 ft South of Hooper Ave. | 88.89 |
H-25 | 1-308 | 8 | 1988 | 39.975514 | -74.176163 | 416398.421 | 2137474.800 | Brookside Dr. & Bay Ave., SW corner | 42.81 |
1Refer to
Plate 6 for hydrant locations.
2Data obtained from United
Water Toms River, Inc., November 1997.
3Geodetic
data obtained by ATSDR staff using global positioning system (GPS) equipment,
January 1998; refer to Sautner, et al. (1998).
4Refer
to Appendix A for photographs showing hydrant
location referenced to street corners.
5Land
surface elevation, referenced to sea level datum, determined by use of
GPS equipment. For hydrants H-2, AH-6, H-11, H-12, and H-18, GPS data were
determined to be in error. Therefore, elevations were determined from U.S.
Geological Survey digital elevation model (DEM), 7 ½-minute quadrangles
for the Dover Township area.
The location of the final set of test hydrants, described by: (1) geodetic coordinates, (2) New Jersey state plane coordinates, and (3) street location, is listed in Table 2. Photographs showing the location of the test hydrants with reference to street corners are provided in Appendix A. The geodetic locations were determined using global positioning system (GPS) equipment(3). Details pertaining to determining hydrant locations are provided in Sautner, et al. (1998). Land surface elevation at the hydrant locations was also determined using GPS equipment by placing the top of the receiving antenna on the ground next to the hydrant (see photograph of hydrant H-8 in Appendix A), and subtracting the length of the antenna (1.83 feet [ft]) from the recorded GPS elevation. Because elevations of the measuring points will be used in the calibration process to assess the reliability of the model (see section on "Hydraulic Model Calibration"), elevations determined by use of the GPS equipment were further verified by using land surface elevations obtained from 7½-minute U.S. Geological Survey (USGS) digital elevation model (DEM) quadrangles for the Dover Township area. For six hydrants (H-2, AH-6, H-11, H-12, and H-18; Table 2) the GPS elevations were found to be in error, although the cause of the error was unknown. Therefore, elevations for these hydrants were derived using the DEM data instead. (For these hydrants, field verification of land surface elevations indicated that the GPS-determined elevations were in error.)
To record system pressures continuously over the 48-hour duration of the tests, each test hydrant was equipped with a RADCOM Lolog LL™ continuous recording pressure data logger that sampled pressures at one-minute intervals. Components that made up the hydrant pressure measurement configuration were (Figures 2 and 3): (a) a hydrant adapter kit, (b) a brass lever handle shut-off valve, (c) a coiled pressure hose with a quick release coupling, and (d) the RADCOM Lolog LL™ pressure data logger. The pressure data recording equipment was chosen because of:
Figure 2. Components of hydrant pressure
measurement configuration.
(A) Dickinson A7983 Hydrant Adapter Kit | (B) Dickson brass lever handle shut-off valve, 1/4 NPTF x 1/4 NPTF, shown in "off" position |
(C) Coiled Pressure Hose with a quick release coupling | (D) RADCOMM Lolog LL™ pressure data logger |
Figure 3: Assembly of hydrant pressure measurement configuration: (A) hydrant adapter kit, (B) brass shut-off valve in "on" position, (C) coiled pressure hose with quick release coupling, and (D) RADCOM LoLog LL™ pressure data logger.
ATSDR established a test protocol or workplan (Maslia and Sautner 1998a) to ensure that successful data-gathering would occur during the test and to establish quality-assurance procedures that would be followed for each hydrant monitored. Because a large number of hydrants were to be monitored on a continuous basis, two installation teams were used. As part of the test protocol and quality-assurance procedure, each logger was identified by a hydrant number (H-1, H-2, etc.) and a logger serial number. Table 3 lists hydrant identifications and the associated pressure data logger serial numbers used for the March and August 1998 tests. Before conducting the tests, the loggers were factory calibrated to 150 psi, and, therefore, did not need field calibration (a letter from the manufacturer indicating the loggers were calibrated is provided in Appendix B). However, as part of the quality-assurance procedure, logger pressure during installation was verified with a manual pressure gauge, as described below. Installation teams were composed of ATSDR, NJDHSS, and water-utility staff. The following 7-step procedure was used to install and quality-assure the pressure data logger installation on each of the 25 hydrants:
Table 3. Hydrant identification and corresponding
data logger serial numbers used for the March and August 1998 pressure
tests, Dover Township area, New Jersey
Hydrant
Identification |
Data Logger
Serial Number |
Hydrant
Identification |
Data Logger
Serial Number |
H-1 | 294 | AH-14 | 307 |
H-2 | 295 | H-15 | 308 |
H-3 | 296 | H-16 | 309 |
H-4 | 297 | H-17 | 310 |
AH-5 | 298 | H-18 | 311 |
AH-6 | 299 | AH-19 | 312 |
H-7 | 300 | H-20 | 313 |
H-8 | 301 | H-21 | 1314, 222 |
H-9 | 302 | AH-22 | 315 |
H-10 | 303 | H-23 | 316 |
H-11 | 304 | H-24 | 317 |
H-12 | 305 | H-25 | 318 |
H-13 | 306 |
1 Logger 314 was determined to be defective
during installation of the March 1998 test. The manufacturer supplied a
replacement logger (number 222) that was installed on hydrant H-21at approximately
10:34 hours on March 24, 1998 and used for the remainder of the March 1998
test and for the August 1998 test.
Figure 6. (A) hydrant adapter kit, brass lever handle shut-off valve in "off" position, and Ashcroft Duralife industrial pressure gauge, and (B)
Figure 8. ATSDR staff querying pressure data logger attached to test hydrant H-11 for parameters of date, time, and pressure using the Psion HC-120™ hand-held computer.
Figure 9. ATSDR staff covering test hydrant H-11 with a heavy-duty (0.003 in.) plastic bag to indicate "out-of-service" condition.
Figure 10. Data being downloaded from the Psion HC-120™ hand-held computer to a laptop computer using the RS232 ports on both computers.
Review of installation pressure data indicated that all data logger and gauge pressures were within 3 psi for both the March and August 1998 tests (e.g., Appendix C) with the exception of the logger installed on hydrant H-21 (Plate 6; Table 3) during the March 1998 test. Because of the quality-assurance procedures in place, the logger at this test hydrant location, was determined to be defective and a new logger was flown in overnight and installed the next morning. Therefore, the first 10 hours of data for the March 1998 test were not recorded for hydrant H-21 (see pressure graph for test hydrant H-21 in Appendix D). For the duration of the March and August 1998 tests, ATSDR and NJDHSS staff routinely checked each test hydrant and pressure data logger and, using the Psion HC-120™ hand-held computer to observe instantaneous pressure readings at each hydrant, assured that the loggers were functioning properly (e.g., Appendix C). Additionally, routine checks of each test hydrant were made to assure that no significant water leaks, potentially affecting the pressure measurements, had developed. After the tests were concluded, loggers were removed in a process that was the reverse of the one described in steps 1-6 above. The recorded data in the data loggers were downloaded to and stored in the Psion HC-120™ hand-held computer (Figure 4) for post-test analysis. After all data loggers were removed from the test hydrants, data in the Psion HC-120™ hand-held computer were downloaded, as an additional quality assurance step, to a laptop computer in the manner shown in Figure 10.
Because data from the tests were gathered at one-minute sampling intervals, the pressure graphs show transients or "spikes" in the data (Appendix D and E). To use the measured pressure data as a reference for calibrating a model, the data should be averaged over one-hour time periods--the smallest time step for which the water-distribution system model is valid (see section on "Requirements for Model Input"). Therefore, the measured pressure data from the tests, shown in Appendix D for the March 1998 test and in Appendix E for the August 1998 test, were averaged over one-hour time periods for use with model calibration. From this point in the report, all pressure data will be referenced to the one-hour average values of the measured pressure data. Graphs of the one-hour average of measured pressure values along with graphs of measured storage tank levels and well and pump flows will be presented and discussed later in the report in the section on "Hydraulic Model Calibration."
Analysis of pressure data from both tests indicates that pressures throughout the water-distribution system generally range from a minimum of approximately 40 psi (test hydrant H-24, Table 4) to a maximum of about 108 psi (test hydrant H-12, Table 4). Because of higher demand conditions existing in August 1998 (Figure 1), maximum one-hour averages of measured pressures for the August test are lower than that those for the March test. For example, at test hydrant H-12, the maximum one-hour average measured pressure of 93.0 psi for the August test is more than 15 psi lower than the maximum one-hour average measured pressure of 108.2 psi for the March test. In the southern Dover Township, Berkeley Township, and the borough of South Toms River areas, where initial model simulation indicated areas of questionably high pressure (some in excess of 125 psi--see section on "Initial Model Simulation"), test results indicate that for March 1998, hourly averages of measured pressures ranged from about 50 psi to 104 psi (test hydrants H-1 - H-10, Table 4 and Plate 6) and for August 1998, pressures ranged from 48 psi to 93 psi (for test hydrants H-1 - H-10). Specific details for the tests and results for each test are provided in reports by Sautner and Maslia (1998) and Maslia and Sautner (1998b).
Table 4. Comparison of maximum and minimum
pressures obtained from the March and August 1998 pressure tests, Dover
Township area, New Jersey
Hydrant
Identification1 |
2Pressure, in Pounds Per Square Inch | |||
March 1998 Test | August 1998 Test | |||
Maximum | Minimum | Maximum | Minimum | |
H-1 | 92.1 | 55.8 | 79.1 | 48.1 |
H-2 | 90.7 | 50.0 | 80.0 | 50.4 |
H-3 | 87.3 | 51.8 | 76.5 | 47.6 |
H-4 | 103.4 | 66.0 | 93.1 | 63.1 |
AH-5 | 98.6 | 58.7 | 87.8 | 55.4 |
AH-6 | 97.8 | 61.6 | 80.1 | 55.8 |
H-7 | 102.0 | 66.4 | 89.8 | 66.5 |
H-8 | 103.8 | 74.4 | 92.4 | 76.2 |
H-9 | 99.0 | 65.4 | 82.2 | 63.4 |
H-10 | 90.5 | 59.2 | 78.7 | 62.9 |
H-11 | 101.8 | 68.9 | 88.2 | 70.6 |
H-12 | 108.2 | 71.3 | 93.0 | 72.8 |
H-13 | 107.6 | 71.6 | 91.4 | 73.7 |
AH-14 | 94.3 | 62.4 | 82.9 | 66.8 |
H-15 | 103.1 | 70.9 | 90.0 | 74.0 |
H-16 | 91.9 | 62.7 | 78.9 | 62.1 |
H-17 | 72.0 | 45.6 | 61.6 | 44.3 |
H-18 | 87.9 | 62.8 | 76.3 | 59.5 |
AH-19 | 71.6 | 52.7 | 61.7 | 46.2 |
H-20 | 61.9 | 50.4 | 57.3 | 42.8 |
H-21 | 72.8 | 50.2 | 63.6 | 47.2 |
AH-22 | 97.4 | 68.9 | 80.1 | 63.0 |
H-23 | 102.5 | 71.1 | 84.3 | 70.5 |
H-24 | 68.9 | 39.6 | 53.8 | 39.8 |
H-25 | 89.0 | 58.5 | 74.1 | 59.1 |
1See Plate
6 for hydrant locations.
2Pressure values are hourly
averages derived from one-minute sampling measurements.
Mathematical modeling has been used for more than 60 years to analyze flow in water-distribution system networks since the concept was proposed by Cross (1936). Using computers for conducting analyses of flow in pipe networks originated in the early 1960s and was greatly expanded during the ensuing decade of the 1970s with the advent of enhanced solution algorithms (Epp and Fowler 1970, Wood and Charles 1972) and the implementation of modeling techniques for devices such as pumps and valves (Jeppson and Davis 1976). In the late 1970s, single-time-period simulations were advanced to extended period simulations with techniques developed by Rao and Bree (1977). Hydraulic models can be used to analyze systems where demand and operating conditions are static or are time varying. The former type of model is a 'steady-state' model, and the latter is referred to as an 'extended period simulation' or EPS model.
Modeling the spatial distribution of water quality in pipelines first began with a steady-state modeling approach as suggested by Wood (1980) who studied slurry flow. Other researchers developing steady-state water-quality models in the 1980s and early 1990s include Chun and Selznick (1985), Metzger (1985), Males et al. (1985), Clark et al. (1988), Grayman et al. (1988a), Wood and Ormsbee (1989), and Clark (1993). The representation of temporally varying conditions for contaminant movement in a distribution system or 'dynamic' water-quality models began to be used in the mid-1980s. Investigators developing such models include Clark et al. (1986), Liou and Kroon (1986), Grayman et al. (1988b), and Hart (1991). With the widespread use and relatively low cost of personal computers and desktop workstations during the mid-1980s and 1990s, many models, both proprietary and public domain, can now be used to conduct hydraulic and water-quality analyses. Two such models in use today are the proprietary model Piccolo (SAFEGE Consulting Engineers 1994) and the public domain model, EPANET (Rossman 1994, Rossman et al. 1994) developed by the U.S. Environmental Protection Agency. The reader is referred to Rossman (1999) and Clark (1999) for a thorough discussion on the evolution and development of hydraulic and water-quality models.
Hydraulic modeling of water-distribution systems can be conducted by
solving mathematical equations that characterize the pipe network of the
distribution system. The EPANET water-distribution system model was chosen
to conduct an extended period simulation of the hydraulic behavior within
the water-distribution system. EPANET solves the following set of equations
for each storage node s (tank or reservoir) in the system (Rossman
1994, Rossman et al. 1994):
(1) | |
(2) | |
(3) | |
and the following equations for each link (between nodes i and j) and each node k: | |
(4) | |
(5) | |
where the unknown quantities are:
ys = height of water stored at node s, (L); |
|
the known constants are:
As = cross-sectional area of storage node s, (L2); |
Equation (1) expresses conservation of water volume at a storage node while equations (3) and (5) do the same for pipe junctions. Equation (4) represents the energy loss or gain due to flow within a link. For known initial storage node levels ys at time zero, Equations (4) and (5) are solved for all flows qij and heads hi using equation (2) as a boundary condition. The system of equations is solved using a technique known as the gradient method, and the reader is referred to Todini and Pilati (1987), Salgado et al. (1988), and Rossman (1994) for details.
Water-Quality Simulation Model
The fate of a dissolved substance flowing through a distribution network over time is tracked by EPANET's dynamic water-quality simulator. To model water quality of a distribution system, EPANET uses flow information computed from the hydraulic simulation as input to the water-quality model. The water-quality model uses the computed flows to solve the equation for conservation of mass for a substance within each link connecting nodes i and j, such that:
(6) |
where:
cij = concentration of a substance in link i, j as a function of distance and time (i.e., cij = cij(xij,t)), (ML-3); |
Equation 6 must be solved with a known initial condition at time zero
and the following boundary condition at the beginning of a link (i.e, at
node i) where xij = 0:
(7) |
where:
Cki lkivZt = concentration at end node of link k, i, of length Lki, (ML-3); |
The summations are made over all links k, i that have flow into the head node i, of link i, j. Note, the boundary condition for link i, j depends on the end node concentrations of all links k, i that deliver flow to link i, j. Hence, Equations (6) and (7) form a coupled set of differential and algebraic equations over all the links in the network. These equations are solved using a numerical method called the Discrete Volume-Element Method; for details the reader is referred to Rossman et al. (1993) and Rossman (1994).
The EPANET water-quality simulator provides a mechanism to account for the gain or loss of a substance by considering its reaction as it travels through the distribution system ((cij) in Equation (6)). For the intended use of the present study, however, the transport of conservative constituents (no reaction, transformation, or decay) will be analyzed. Therefore, the reaction rate, (cij), is set equal to zero for all simulations.
Identifying the source of delivered water in a distribution system has become a necessity when trying to determine the location of a source that may supply water that exceeds a given level of a chemical or biologic constituent. Wood and Ormsbee (1989) developed an explicit method to calculate the percentage of flow, under steady flow conditions, originating at various source points at a specific location in a distribution system. EPANET also has the ability to track the percentage of water reaching any point in the distribution network over time from a specified location (source) in the network (i.e., the "proportionate contribution" of water from a specified source). In this case, the value of cij in Equation (7) is set at 100 percent for the source location and the value of cij in Equation (6) becomes the percentage of flow the source location has contributed to the location of interest. Given the multiple number of points of entry to the water-distribution system serving the residents of Dover Township, the ability to track the percentage of water originating from a point of entry becomes a very useful analysis tool for this investigation.