North Jersey Develops Computerized Reservoir Management Program
By Donald Distante, Lawler Matusky & Skelly Engineers, William Goble, North Jersey District Water Supply Commission, and Pen Tao, United Water New Jersey
The North Jersey District Water Supply Commission (NJDWSC) operates two reservoirs, as well as two pump stations, that are used to ensure sufficient supplies of water to customers in northern New Jersey. The Wanaque and Monksville reservoirs have storage capacities of 29.6 and 7 billion gallons, respectively. The Ramapo pump station has four 37.5-MGD pumps that can deliver up to 150 MGD, and the Two Bridges pump station uses up to five 50-MGD pumps that can deliver up to 250 MGD. An inter- watershed connection exists to supply water from these systems in the Passaic watershed to the Oradell reservoir, which is located in the Hackensack River watershed. United Water New Jersey (UWNJ) operates four reservoirs in the Hackensack River watershed and uses this interconnection to supplement its supplies.
Both NJDWSC and UWNJ saw the need to develop a computerized management tool to help minimize the costs of operating the two pump stations, project future supplies during drought conditions, and provide a means to evaluate safe yield given the various operating scenarios for pumping. They envisioned a user-friendly computer program that could provide graphic results that could be presented to regulatory agencies, the public and in-house staff to show and discuss projected supplies, anticipated pumping costs and the effects of various pumping alternatives on cost and supply.
To meet this need, NJDWSC and UWNJ retained Lawler, Matusky & Skelly Engineers (LMS) to develop the required model. LMS developed the model using the database and programming software Paradox for Windows and named it the Wanaque South Management Program, or WSMP. The following are summaries of WSMP`s primary features.
Data Management: The program provides a simple means to access, update and evaluate historical data. Data incorporated into the WSMP includes the daily reconstituted flows (i.e., natural flows) from 1919 to 1993, monthly rainfall data, reservoir stage-volume curves, reservoir rule curves (i.e., monthly reservoir storage objectives below which pumping may occur), system water use demands and electrical energy costs. Daily reconstituted flows from 1919 to 1993 form the basis for performing reservoir and pump station simulations. Daily data were used because minimum reservoir release requirements and minimum passing flow requirements at the two pump stations are specified by the New Jersey Department of Environmental Protection (NJDEP) as daily flows. A statistical distribution that fits the distribution of the observed data is used so that drought recurrences beyond those represented by the data can be used to simulate future supplies. An extreme value distribution commonly known as the Gumbel distribution was used.
Drought Prediction: By entering current monthly rainfall and runoff data the user can ascertain the drought recurrence interval up to the present period by considering single or cumulative periods (e.g., November or June through November). A rainfall/runoff forecast module is also included that uses historical trends to estimate the next month`s rainfall and runoff.
Reservoir Storage Simulation: Reservoir supplies can be simulated for up to 12 months ahead of the current month. The user chooses daily flows based on either an anticipated recurrence interval (e.g., 1/100 years) or a historical period (e.g., 1963) as direct inputs to the reservoirs and as flows at the two pump stations. System drafts, initial storage, sewage treatment plant contributions and pumping alternatives are chosen and the simulation model executed.
Pumping Costs: Two levels of electrical costs are included: a rough estimate based on historical costs per million gallons pumped for each pump station and a refined costing module that incorporates the hourly changes in electrical charges (e.g., off-peak, intermediate peak and on-peak), as well as the pump rating curves, which are a function of the reservoir elevation and the interaction between pumps from the two pump stations, which share a common connection.
As of the end of September 1995, the Northeast was in a severe drought and NJDEP had declared a drought emergency. NJDWSC and UWNJ used WSMP to simulate possible future supplies. Runoff data from August 1995 back through January 1995 indicated that the area could be headed for the worst drought on record. The runoff was the lowest on record.
Based on the assessment it was considered reasonable to simulate future supplies using a 12-month, 1/100-year drought running from Sept. 1 through Aug. 31. Using all available pumping and drafts equal to the average draft over the last five years, a simulation of possible supplies was run. This simulation took about one minute to set up and about 30 seconds to run on a 100-MHz Pentium PC. As can be seen, water supplies under the 1/100-year assumption are shown to be above the drought warning line. Planning on the part of NJDWSC and the availability of pumping to replenish supplies resulted in the Wanaque Reservoir being almost full at the beginning of June. In July and August, pumping is not allowed due to water quality and ecological considerations, but the September starting reservoir volume was a result of this planning. As a result, even with the projection of a serious drought, NJDWSC projected that it could maintain the reservoir supply above the drought warning line, assuming that a 1/100-year drought occurred for the subsequent 12-month period.
Recently, there has been substantial rainfall in northern New Jersey and the drought emergency has been lifted. The ability of WSMP to simulate storage under these new conditions resulted in a cost-effective decision to reduce pumping.
In the months of September and October 1995, rainfall and runoff increased substantially (12.8 inches of rain fell at the Wanaque reservoir gauge during the period). WSMP was used to evaluate how much pumping would be necessary to ensure that the reservoirs are full in spring 1996 (as a general rule, NJDWSC tries to have full reservoirs going into the summer and early fall dry seasons) assuming that future flows represent a 1/10-year recurrence. Several pumping alternatives were tested by simulating reservoir storage from Nov. 1, 1995 to Oct. 31, 1996. The following table summarizes some of the key results of these model runs:
As indicated, alternative 3, Only Ramapo Pumps Available, meets the storage objective and would save approximately $206,000 compared to alternative 1, Pumping From Both Stations, which also meets the objective. These estimated costs are based on average costs per MG pumped and are used for general planning. WSMP also includes more refined projected pumping cost calculations which are intended for one-month-ahead cost planning. This example demonstrates how WSMP is used to minimize the cost of pumping and still meet storage objectives. This process of using WSMP to evaluate future pumping needs is continually updated to reflect the previous month`s rainfall and runoff patterns as well as actual reservoir storage. If actual rainfall is lower than anticipated and storage falls below the reservoir rule curves, then a re-run of WSMP would indicate the increased pumpage required to meet storage objectives.
LMS has made several real-time presentations of WSMP by projecting the computer screen images to a larger projection screen. The program has been very well received and one of its key assets is that it can immediately involve the audience. For example, the audience can ask such “what-if” type questions as, “What happens if only three pumps are operated at the Two Bridges Pump Station or if the system drafts are increased?” With a couple clicks of the mouse, the answers are displayed to the audience. This provides a powerful way to present results and to make persuasive cases.
The development of WSMP has provided LMS with a general framework that can readily be applied to other watersheds. A very important element in the development of WSMP was the input received from NJDWSC and UWNJ, who have the knowledge essential to successfully operate their systems. Their practical input was incorporated into the WSMP code, and this type of input is considered crucial for application to any watershed.