Shutting down a water storage reservoir for renovation in a large urban water district is not a simple matter. The outage can interrupt service to thousands of homes, businesses and industrial facilities and play havoc with normal daily activities. In order to avoid major disruptions to water service, reservoir shutdown must be well-planned, and temporary alternative supplies must be in place to ensure uninterrupted flow and an adequate water supply to fight fires. When the planned renovation involves not one, but all 61 of the district's reservoirs, it is absolutely essential to have an orderly, efficient and flexible plan in place before the work begins.
This was the case with the East Bay Municipal Utility District, located on the eastern shore of San Francisco Bay. The District's water storage facilities, located above or near several major earthquake fault zones, were due for seismic retrofit. In 1994, the District Board of Directors approved a rate payer surcharge to fund the Seismic Improvement Program (SIP) to address an extensive list of seismic improvement projects. The funds will be applied over a 10-year period to strengthen critical water distribution system facilities so that they can withstand the expected impacts of a major earthquake in the District's service area. The District hired Kennedy/Jenks Consultants to help plan an orderly shutdown of the reservoirs during renovation.
Working with District staff, Kennedy/Jenks analyzed each pressure zone, then developed a comprehensive pressure zone report that defined key assumptions and modeling criteria. A reservoir outage plan was developed for each individual reservoir that was scheduled to come out of service. The consultant also analyzed the District's water distribution system, determined the impact that reservoir outage would have on water service and fire protection, and recommended temporary mitigation measures for taking reservoirs out of service.
Seismic Improvement Program
Kennedy/Jenks prepared reservoir outage plans for 26 of the District's 61 treated water storage reservoirs. This plan was to minimize the impact of reservoir shutdowns on customers and ensure an adequate water supply for fire protection during renovation. The Reservoir Outage Planning Project provides the District with information on the effects that emptying and removing each reservoir from service for seismic upgrades would have on the District's customers.
Hydraulic modeling was a team effort. Since the reservoirs slated for seismic upgrades are located in 46 different pressure zones, hydraulic modeling was required for most of the zones. An approach was designed with Kennedy/Jenks performing hydraulic analysis on 22 of the 46 pressure zones, and District staff completing the remaining 24 zones.
The project's approach for hydraulic modeling was designed by four teams of experienced computer modeling and design professionals. Each team was responsible for five to six separate pressure zones. In order to ensure that each team used identical procedures to complete the investigations and designs, all the project teams worked together to analyze an initial pressure zone and develop a prototype to be used for the remaining zones.
This prototype analysis served as a system shakedown that allowed the project teams to identify data gaps and establish the form and format for the remaining pressure zone investigations. From the prototype of the pressure zone, a protocol was developed for the modeling procedures that would be used for the analyses. The teams then proceeded to carry out each stage of the project (from model development through calibration and analysis of recommended outage plan alternatives.) This protocol was later presented to the District for use in their modeling of the remaining pressure zones.
As the project developed, a three-step process was used for the efficient exchange of information and to clarify requirements while building consensus with the District.
Step 1: Develop Hydraulic Model
Step 2: Analyze System
Step 3: Develop Recommended Plan
This simple three-step approach allowed the project team to organize the decision-making and planning into manageable units so that each team could focus their efforts and produce a final document for each reservoir outage.
Developing the Hydraulic Model
The first step consisted of creating, calibrating and verifying the model. Each of the District's pressure zones involved the exchange of a large amount of information, including base maps, the District's water distribution maps, demand information, pump information and operational data. With so much data exchange required to develop the hydraulic model, some quick and easy mechanisms were needed for this exchange. It was decided that the Internet would be used to transfer the District's water distribution system base maps to the Kennedy/ Jenks project team. By using the Internet, the project team could begin developing the models while the District was still gathering information.
To reduce the time required to enter data into the EPANET modeling program, the project team input pipe and node data into a KYPIPE-based computer program. The KYPIPE program allowed them to digitize the node locations in a CAD-based environment. By using the District's distribution maps in both CAD and hard copy format, the project team was able to place the District's CAD files on-screen as a background image to help locate and digitize pipelines and related facilities. After transferring the digitized node map (now with X and Y coordinates) into EPANET, they input the pump stations, storage reservoirs and other facilities.
This approach reduced the amount of time needed to create the computer model and facilitated the use of the graphical capabilities available with EPANET. (See Figure 1.) Since some of the District's pressure zones are very large, this modeling was restricted to the pipelines that would affect the validity of the analysis. These included pipes that carried significant flow (e.g., transmission lines), provided the sole feed to an area or completed a loop.
Kennedy/Jenks worked with District staff to identify representative locations such as hydrants at the top of the pressure zone, services and hydrants in remote areas and major service connections including large private fire services, pumping plants and pressure regulating valves (PRV) supplying water to other pressure zones. These rates were used to define the baseline conditions in each pressure zone.
Before proceeding with the calibration and verification of each of the 22 pressure zones, Kennedy/ Jenks met with District staff to verify that the pipelines, storage tanks, pump stations and fire hydrants that had been selected met all of the District's criteria.
Water Demands: Most of the SIP reservoir outages are scheduled between the months of November and May when water demands are at their lowest. To determine average-maximum daytime demand during this period, the District evaluated its Water Consumption Information System data from the past six years (see Table 1). The month with the highest historical water demand was selected to determine the impacts of reservoir outages.
The District also had data on the daily water demand for Winter 1995 and 1994. This data, separated according to pressure zone, was used to create a diurnal curve for the extended period simulation (EPS). (Table 2)
Calibration: Effective calibration is vital to create a model that accurately represents the District's water system operations. The following field data were required to properly calibrate the models:
- Pressure and amount of flow into the pressure zone from all sources.
- Elevations of water storage reservoirs.
- Flows and pressures out of the pressure zone at each pumping plant.
- Pressure gauge readings during fire hydrant tests in each pressure zone.
Although the District originally supplied Kennedy/Jenks with estimated "C"-factor (pipe friction headloss factor) information on all of their pipeline types, every pressure zone is unique. For each pressure zone, the project team developed hydrant tests to confirm the District's general C-factors for each pipeline type and to adjust those to reflect Kennedy/Jenks' field observations. In the process, the project team also found areas where the District had some closed valves and/or construction within the pipelines.
Model Verification: The models had been calibrated to static conditions using the hydrant flow tests, but it was still necessary to confirm that the models were predicting the correct pressures and flows over an extended period of time. Model verification was important to predict how the system would react once a reservoir was removed from service. Verification was provided by comparing the calibrated models with information from the District's Operations Network System Control Improvements (OSCI) system.
By selecting a two-day period, the project team was able to compare the reservoir level and pumping trends reported by the OSCI data to that predicted by the model. The closer the models came to simulating the OSCI trends, the stronger the verification. Once each of the models was calibrated and verified, the next step was to perform the baseline system analysis.
Analyzing the System
It was essential to maintain the existing level of service for the daily operation of the water distribution system during the planned outage of each reservoir. The availability of fire flow was also a critical element. Two baseline simulations were modeled to evaluate whether removal of the specified reservoirs from service would affect existing pressure and flow conditions during normal daily operation of the system:
- Baseline EPS: computer simulation over a 48-hour period with and without the reservoir in service.
- Baseline Fire Flow: static computer model simulation at a residual pressure of 20 psi with and without the reservoir in service.
To determine baseline conditions, the project team selected representative locations in the service area of each reservoir and evaluated the minimum and maximum pressures that would occur at these representative locations over a 48-hour period with and without the reservoir in service. They used the EPANET software to assess whether mitigation measures were required to maintain the existing level of service (Figure 2). In order to determine the existing availability of fire flow in the pressure zones, the study team also evaluated fire hydrants in areas where the pressure zone would be most vulnerable when the reservoir was taken out of service (such as the hydrants closest to the reservoir being removed from service).
A static computer model simulation was conducted to determine the available fire flow at a residual pressure of 20 psi. After evaluating the results of the computer simulations, mitigation measures were developed for areas where fire flows were less than 95 percent of the baseline analysis when the reservoir was not in service.
Developing the Recommended Plan
To develop a reservoir outage plan that would provide the District with the best mitigation alternative each time, the project team, working with District staff, developed a checklist of mitigation alternatives to evaluate those that were feasible. Of these plans, two alternatives would be later analyzed. Since reservoir outage is a temporary condition, the solutions selected had to provide a reliable water supply to the pressure zones during the water reservoir outage while keeping costs as low as possible.
To assess the feasibility of each alternative, Kennedy/Jenks put together a team of experts in tanks, pumping stations, pipelines and system operations. The Design Team conducted site visits to each pressure zone and the locations where mitigation alternatives would be implemented. After the site visits, the Team evaluated alternatives and summarized the results in the form of a mitigation Alternatives Checklist Table (Table 3). The checklist reflected the criteria that were considered significant in selecting the most appropriate alternatives. With the concurrence of District staff, two alternatives were chosen for further evaluation and hydraulic analyses was performed to see if the proposed mitigations would work hydraulically in that particular zone. Based on the analysis confirming the two recommended alternatives, the Design Team then selected a preferred alternative.
For each reservoir outage scenario, a plan was developed with a detailed description of how each preferred temporary mitigation would be implemented for supplying water during the reservoir outage. These recommendations provide District staff with the data they will need to proceed with any detailed design required for the implementation. The District was also provided with a summary matrix of recommended equipment and an Equipment Summary Document to provide overall guidance in equipment selection, purchase and implementation.
The Reservoir Outage Planning Project has provided EBMUD with a comprehensive and useful tool. As funds become available over the next ten years, the District can implement seismic retrofit of its reservoirs according to an orderly plan.
About the Author: Charles T. Duncan is with Kennedy/Jenks Consultants, San Francisco, Calif.