Section 4 - Hydraulic Model Development and Calibration. Section 4

Transcription

1 Section 4 - Hydraulic Model Development and Calibration Section 4

2 SECTION 4 HYDRAULIC MODEL DEVELOPMENT AND CALIBRATION This section describes the processes utilized to develop and calibrate the hydraulic model of EVWD s potable water system. First, the development of the model distribution network from EVWD s geographical information system (GIS) is described. Subsequently, the allocation of pressure zones, ground elevations, and water demands are discussed. This section concludes with a discussion of the model calibration process, which is performed to verify the model results with field measurements. In preparation for model calibration, a technical memorandum (MWH, 2013) was presented to EVWD which outlined the fire hydrant testing procedures, equipment list, and hydrant maps needed to perform hydrant tests at 20 locations throughout the system. The technical memorandum is attached to this report as Appendix B. The calibrated model will be used to perform system analyses of the system under existing demand conditions and future demand conditions. 4.1 Hydraulic Model Development The hydraulic model of the EVWD potable water system is created using Innovyze s InfoWater software, which is based on ESRI s ArcGIS platform. The ArcGIS files for the distribution system network provided by EVWD are used as a base for creating the hydraulic model for this Water System Master Plan (WSMP). The hydraulic model contains all pipelines and facilities (booster pumps, storage tanks, wells, and pressure reducing valves) present in the ArcGIS geodatabase provided by EVWD Data Collection Data used for the development of the hydraulic model is obtained from a variety of sources. Key information includes: GIS database of all water mains, laterals, and water facilities Hydraulic water system schematic Dimensions for storage reservoirs Pump curves and performance tests for booster pumps Pump controls and settings of pressure regulating valves Water production records ( ) Customer usage records ( ) Supervisory Control and Data Acquisition (SCADA) data General Plan and land use information Ground elevation contour lines Street centerline data Aerial photography coverage Imported and emergency water connections, sizes, and capacity Summary of projects currently under construction or scheduled for construction in near future Project Number: February 2014

3 Section Model Construction All pipelines and facilities included in the hydraulic model are obtained from the GIS information provided by EVWD. The spatial representation of the hydraulic model in NAD83 datum, California State Plane Zone V coordinate system is consistent with the coordinate system in which EVWD s GIS data are projected Pipelines All pipelines and facilities in the model are checked for accuracy and some pipelines and facilities are redrawn to resolve model connectivity issues. Model attributes for pipelines include the pipe number, pipeline length, diameter, material, roughness, and pressure zone. While most of these attributes are provided by EVWD, the roughness attribute is based on the age and material of the pipeline as shown in Table 4-1. Table 4-1 Pipeline Roughness Material Hazen Williams C-Factor (1) Asbestos Cement 130 Cast-Iron (3) 64 Cement Mortar Lined Steel 125 Copper 125 Dipped and Wrapped Steel (3) 100 Ductile Iron 130 Steel (2) 135 Unknown 100 (1) C Factors are estimated based on the age and the material of the pipeline. (2) Assumed to be cement lined based on the typical age of installation. (3) Pipelines older than 40 years are assigned a low C-factor of Valves and Junctions Pressure regulating valves (PRVs) are modeled with information such as valve diameter, pressure zones, valve settings, and minor loss coefficients. Pressure settings provided by EVWD are used for each active valve. Zone isolation valves are modeled where the geodatabase indicates the presence of normally closed valves. The zone isolation valves are modeled with an initial status set to CLOSED. Junctions are defined as the intersections of two or more pipelines, or at the location where any pipeline changes diameter or material. Attribute information for junctions include elevation, demand, and pressure zone. Fire hydrants are modeled as junctions and the fire flow demands are recorded in the model at these junctions. Project Number: February 2014

4 4.1.5 Storage Tanks Section 4 Storage tanks are modeled as cylindrical tanks. Their locations and pressure zones are determined from the system map provided by EVWD. Attributes such as elevation, diameter, tank height, and installation year are included based on a tank summary report provided by EVWD. For model calibration, the initial water level of each tank is set to the recorded water depth. The initial water level represents the water depth at the beginning of a hydraulic simulation (midnight). Hydropneumatic tanks are also included in the model but they are not modeled as tanks. Instead, they are represented by a pump set to the ON position flowing directly to a PRV. The valve is set to the pressure level observed in the field for the zone. Pressures experienced in a hydropneumatic zone can be satisfactorily simulated by using this modeling technique Pumps and Wells The pump database in the model is populated with a plant number and pump curve for each pump or well in the system. Manufacturer s pump curve information is provided by EVWD for the following pumps: Plant 39 booster #2 Plant 39 well Plant 125 well Plant 143 well Each pump is modeled as an adjusted multi-point curve based on the manufacturer s pump curves in conjunction with the most recent Southern California Edison Company (SCE) test data. Where pump curve data are not available, total dynamic head and corresponding flow information obtained from the most recent SCE pump test results are used in the model. The model creates design point curves, which allow the pumps to produce head up to 133 percent of the recorded head and flow up to two times the recorded flow. It is recommended that as new pumps are installed throughout the system, the model be updated with the manufacturer s pump curve adjusted for SCE test data. Each well is modeled as a reservoir and a pump, where the reservoir represents the groundwater aquifer and the pump represents the well pump. The reservoirs are modeled as fixed head (i.e. unlimited volume of water at a specified elevation) reservoirs with a water elevation equal to the static groundwater level minus drawdown Surface Water Treatment Plant The surface water treatment facility at Plant 134 is modeled with its associated booster pumps and tank supplying the Upper, Foothill, and Canal zones. The treatment facility is modeled as a fixed head reservoir connected to a flow control valve ensuring a steady flow of water into the system. The flow from the plant is adjusted based on the average flow observed for each calibration period. Project Number: February 2014

5 4.1.8 Facility Nomenclature Section 4 The identification scheme used in the existing system model is based on type of facility. Tanks begin with the letter T, booster pumps with the letter PMP, well pumps with the letters WELL, and pressure reducing stations with the letters PRS. This prefix is followed by the number of the plant and lastly a sequential number if there are multiple facilities at the site. For example, T_134 is the tank at Plant 134 while PMP_134_4 is pump number 4 at the same plant. This nomenclature makes model navigation easier for the user Facility Elevation Data Elevations for the model are derived from contour data (two feet intervals) provided by EVWD. Using the contour data, ground elevations are extracted and assigned to all junctions and facilities (with the exception of storage reservoirs) in the model. Elevations for storage reservoirs are assigned based on information contained in the summary sheets provided by EVWD Water Losses The water demands for existing conditions are based on customer usage information (billing data) provided by EVWD. The billing data covers the water usage for nearly 22,796 accounts for the period of November 2002 to December The data includes information on service IDs, street addresses, and monthly consumption. The average daily demand used in the model is based on four years of data ( ) to reduce errors that may be caused by unusual consumption patterns in a single high or low water year. Based on these four years of data, the average water consumption for the entire water system is 12,200 gpm (17.6 MGD). Table 4-2 contains a comparison of the consumption and production data in the system over the same fouryear period. On average, EVWD loses around eight percent of the water produced through leaks, firefighting, maintenance tasks, and other unaccounted for use, which is typical and acceptable in potable water systems. Table 4-2 Water Losses Year Billed Water Losses Produced Water Water Losses Consumption (% of (gpm) (gpm) (gpm) production) ,100 13,100 1, ,800 11,500 1, ,800 11,800 1, ,500 12,500 1,000 8 Average 13,300 12,200 1,100 8 Project Number: February 2014

6 Section Geocoding The process of geographically locating each billing record is known as geocoding. Each billing record is geographically located using the street addresses in the billing data and street centerline GIS coverage. The geocoding process electronically places the location of each service connection on a map. The demands are then scaled up to account for the water losses in the system. Demands are allocated to demand junctions based on proximity to the geocoded consumption data. The existing system model is comprised of nearly 21,000 pipelines and 20,000 junctions. To incorporate the demands into the hydraulic model, demand nodes are selected that represent a small area of multiple accounts. Demand junctions are selected based on pressure zone boundaries and proximity to meter locations. All junctions associated with water facilities or transmission pipes are excluded from the demand allocation process. Upon completion of the demand allocation process, the locations of the top 25 largest customers in the service area are manually verified, as these large customers can impact system hydraulics significantly. Future demands are allocated geographically based on the location of vacant parcels in the existing land use GIS coverage. Information regarding the locations of proposed developments (described in Section 3) is considered. The total demand for each parcel (or group of parcels) is calculated based on the size of the parcel, land use classification, and the water duty factor. Once the future demands are determined, the demands are assigned to the closest existing demand node in the hydraulic model Diurnal Curve A diurnal curve represents the average hourly demand fluctuation in a water system. The diurnal curve for EVWD s potable distribution system is created based on hourly production and tank level information from the Supervisory Control and Data Acquisition (SCADA) system. Where flows at wells and booster pump stations are not recorded in SCADA, pump ON/OFF times are used along with the flow rates obtained from the SCE test data to estimate the volume of water produced at the pumping facilities. Total system inflow data is based on the production data provided by EVWD. The calculated diurnal curve is presented on Figure 4-1. This curve represents the average hourly demand fluctuation of all pressure zones on August 9, 2012 and is representative of a hot summer day for the year The diurnal curve is created by preparing an hourly mass balance using well production, imported water supplies, and change in storage. As shown on Figure 4-1, the peak hour occurs around noon, which has a demand of 1.5 times the average demand of that day. Individual diurnal curves for each pressure zone could not be created due to data limitations such as the lack of flow meters to record inter-zonal transfers at pressure reducing stations and pumping stations. The diurnal curve shows a unique demand pattern as there is little to no peaking in usage commonly seen in the evening in most systems that are predominantly residential. This pattern was confirmed with EVWD operations staff. Project Number: February 2014

7 Section 4 Water Use (hourly/average) Hour Figure 4-1 System Wide Diurnal Curve 4.2 Model Calibration The hydraulic model with the existing system configuration and demands is calibrated to enhance the accuracy of the model results and provide a planning tool that can be used to identify system deficiencies and recommend pipelines and facilities to address system deficiencies. Model calibration is the process of comparing model results with field results and making model modifications where appropriate to simulate the field results as closely as possible. Typical adjustments include adjustments to system connectivity, operational controls, facility configurations, diurnal patterns, elevations, etc. Several indicators are utilized to determine if the model accurately simulates field conditions: water levels in storage tanks, the run times for pumps, and static and residual pressures from the fire flow tests, and roughness coefficients for pipelines. This also acts as the debugging phase for the hydraulic model where any modeling discrepancies or data input errors are discovered and corrected. The hydraulic model is calibrated for two scenarios: Steady State Calibration: Simulating fire hydrant flow tests to match field results (May 14 th and 16 th, 2013) 24-hour Extended Period Simulation (EPS) Calibration: Modifying the model until it mimics the field operations on the day of calibration (August 9, 2012) Steady State Calibration The objective of the steady state calibration is to validate the assumed pipeline roughness coefficients (C-factors) in the hydraulic model and make modifications, where appropriate. To Project Number: February 2014

8 Section 4 facilitate this exercise, fire hydrant tests are conducted at 20 locations throughout the distribution system. Appendix B provides a detailed description of all data gathered for the steady state and EPS calibration efforts. Each test consists of opening a fire hydrant (indicated as flowing hydrant) and flowing the open hydrant until the residual pressure at an adjacent hydrant (indicated as the gauging hydrant) stabilizes at least 5 pounds per square inch (psi) lower than the static pressure recorded at the gauging hydrant. The flow measured at the hydrant is then input to the hydraulic model as an additional demand and the pressures at the node that represents the gauging hydrant location with and without this fire flow demand is then compared with the field results. The locations of the 20 fire hydrant tests are shown in Figure 4-2. Table 4-3 presents information on hydrant location, hydrant number, static and residual pressure, and actual flow. The results of the fire flow test calibration are also summarized in Table 4-3. The static and residual pressures in the field are compared with the residual and static pressures predicted with the hydraulic model. As shown in Table 4-3 and Figure 4-3, nearly all of the model results are within 10 feet of head (4.5 psi) of the observed field data as promulgated by AWWA s Computer Modeling Manual M32. Four tests (Location Numbers 17, 18, 19, and 20) are outside the acceptable limits. Each of these four tests is performed in zones regulated by hydropneumatic tanks. In general, hydropneumatic zones have at least one small pump to meet demands under normal operations and one large pump to meet fire or other emergency conditions. The large pump remains OFF under normal operations. When running fire flow tests with a steady state model, the emergency pumps are turned on immediately to respond to the drop in system pressures. In reality, the emergency pumps turn on after some pressure is already lost from the system. Based on discussions with EVWD staff, it was agreed that calibrating the model for static pressures is sufficient for fire flow tests conducted within hydropneumatic zones. Two locations required further evaluation during steady state calibration. A 90 psi pressure drop was observed during the hydrant test at Location 12 in the Canal Zone. However, the hydraulic model simulated a 21 psi drop at this location. It appeared that this could be caused by a partially closed valve in the vicinity of the test location. This information was presented to EVWD staff and field investigations later revealed the presence of a closed valve in the system consistent with the model results. The hydrant at Location 16 is located in the Highland Upper zone where flow is controlled solely by PRVs. Based on discussions with EVWD staff, it was found that one of the valves at the station was closed because of leaks in the system. Adjusting the pressure setting at this PRV resolved the pressure discrepancy between the observed data and the simulated results Extended Period Simulation A model calibrated for a steady-state scenario provides an instantaneous snapshot of a water distribution system. As steady state modeling does not involve time-steps, the behavior of a water distribution system over time cannot be analyzed. An extended period simulation (EPS) model provides a better understanding of the operations of a water distribution system than a Project Number: February 2014

9 Section 4 steady state model. The goal of the EPS calibration is to estimate the accuracy with which the model simulates the field operations over a 24-hour period. The EPS calibration is performed for the 24-hour period between midnight August 8, 2012 and midnight August 9, 2012, approximating operations on a peak summer day. The total water production on this day was calculated to be 17,700 gpm (25.5 MGD). This is equal to 1.33 times the Average Day Demand (ADD) for the time period. The model results are compared with the field data to determine if the model reflects the actual system operating conditions over a 24-hour period. The modeled versus field data for the storage tanks, booster stations, and groundwater wells on calibration day are presented in Appendix C. Project Number: February 2014

10 INTENTIONALLY LEFT BLANK

11 Section 4 Location Number Zone Date (2013) Time Address of gaging hydrant (direction to flowing hydrants) Table 4-3 Steady State Comparison Table Average Flow Rate Observed/ Modeled (gpm) Static Pressure Observed (psi) Static Pressure Simulated (psi) Change in Static Pressure Over Observed Residual Pressure Observed (psi) Residual Pressure Simulated (psi) Change in Residual Pressure Over Observed Observed Pressure Drop Comparison (psi) Simulated Pressure Drop Comparison (psi) 1 Lower May14 8:15 AM Pine St./ E. 4th St. (West) % % Lower May14 9:04 AM 7114 Elmwood Rd. (South) 1, % % Intermediate May14 11:52 AM Hibiscus St./ Central Ave. (East) 1, % % Intermediate May14 8:38 AM Base Line St./ McKinley St. (South) 2, % % Intermediate May14 10:00 AM 2355 Osbun Rd. (South) 1, % % Upper May14 10:15 AM Val Mar Dr./ Newcomb St. (South) 1, % % Upper May14 12:10 PM Fisher St./ Center St. (East) % % Upper May16 9:54 AM Canyon Oak Dr./ Streater Ave. (East) 1, % % Foothill May16 9:42 AM Lochinvar Ct./ Lochinvar Rd. (North) 1, % % Foothill May16 8:45 AM 21st St./ Rainbow Ln. (West) 1, % % Foothill May14 10:35 AM Edgemont Dr./ Arden Ave. (East) 1, % % Canal May14 10:52 AM Juniper Dr./Manzanita Dr. (Northeast) % 44 44* 0% Canal May16 7:54 AM Lynwood Dr./ Palm Ave. (Southeast) 1, % % Canal May16 9:30 AM Havenwood Ln./Westwood Ln. (East) 1, % % Mountain May16 9:20 AM Horner Ln./ Bernard Ln. (West) 1, % % Highland Upper May14 11:33 AM Fisher St./ Orange St. (West) % % Hydro 34 May14 9:36 AM 6547 Monte Vista Dr. (East and North) % % Hydro 59 May14 11:10 AM 3712 N. Hemlock Dr. (Southwest) % % N Small Canyon Dr (South and East 19 Hydro 101 May16 8:25 AM on Mountain Top Dr.) % % Hydro 149 May16 9:02 AM 6456 Emmerton Ln. (Southwest) 1, % % *Original model result showed a residual pressure of 113 psi. The value in the table resembles the model after a partially closed valve was simulated. EVWD has since found the partially closed valve and opened it. Project Number: February 2014

12

13

14

15 Pressure (psi) Static Pressure Section Hydrant Location Number Residual Pressure Observed Simulated 120 Observed Simulated 100 Pressure (psi) Hydrant Location Number Pressure Drop Observed Simulated Pressure (psi) Hydrant Location Number Figure 4-3 Hydrant Testing Pressure Comparisons DRAFT 4-11 February 2014 Project Number:

16

17 Section 4 In order to achieve a balanced and calibrated model, the following adjustments are made in the model: Based on the SCE test point and the standard pump curves in the model, five wells were producing nearly two times the flow observed in the field. These wells are modeled with flow control valves to better mimic the field conditions. Adjust facility controls for pumps. For example, based on discussions with EVWD operations staff, the reservoir levels at which certain pumps turned ON was adjusted. Change many pump controls from tank level control to timer control based on discussions with EVWD operations staff. 4.3 Calibration Conclusions The American Water Works Association (AWWA) Manual of Water Supply Practices M32 provides guidelines for computer modeling of water distribution systems. These guidelines include Hydraulic Grade Line (HGL) predictions and water level fluctuation predictions. HGL predictions by the model should be within 5 to 10 feet of those recorded in the field which is equivalent of 2.2 to 4.3 psi. The tank water level fluctuations predicted by the model should be within 3 to 6 feet of those recorded in the field. The lower accuracy range in these guidelines can typically be applied to models used for design and operational evaluations while the higher accuracy guideline (4.3 psi) is typically applied to models used for long range or master planning. Consistent with the above mentioned guidelines, it can be concluded that the results from the hydraulic model are satisfactory for the purposes of long term planning. While this model can be used for long term planning, it is important to understand the inherent errors in the model are due to the input data used to develop the model. The following list gives possible causes for the discrepancies between the model and field data. Temporal variation in demand between EPS and steady state calibration days. The diurnal curve created for the calibration day is also used to determine demand at each hour for the fire flow tests. However, customer demands change from day to day and hour to hour resulting in different diurnal curves on different days. Demand variance in different pressure zones. A lack of sufficient flow meter data for each pressure zone of the system results in the use of a generalized diurnal curve for the entire system. With individual pressure zone diurnal curves, a more accurate demand can be captured as some zones have little to no irrigation demand and others have high irrigation demand. Inaccuracies in elevation data. Elevations used throughout the system for junctions, pump stations, and valves are based on ground elevation. Inaccuracies in observed pump flow. Because a majority of the flows calculated for the pump stations is based on on/off times and flow rates from SCE tests, the actual flow from any of these devices could vary. Project Number: February 2014

18 Section 4 Inaccuracies in pump curves: EVWD has limited information on pump curves and therefore, the model creates a generic pump curve based on a single design point. This can drastically change the flow versus head relationship for each pump station resulting in flow or head variances from field conditions. The lack of SCADA data to record flows at pump stations compounds these inaccuracies. Unknown groundwater level: Changes in depth to groundwater are not accounted for in the model. Groundwater levels vary throughout the year and from year to year. The groundwater elevations used throughout the system are based on the depth of water during the most recent SCE tests provided by EVWD. However, groundwater drawdown can vary significantly depending on the pumping rate and the static groundwater level conditions. These factors introduce additional inaccuracies in the model. Based on the findings from the steady state and the EPS calibration, the following items are recommended to improve and refine the predictive capability of the model in the future: Installation flow meters at pumping stations and PRVs between zones. Flows at these meters should be relayed to EVWD s SCADA system. Installation of pressure loggers to capture pressures at key points in the system such as the suction and discharge pressures at pump stations. Pressures at these loggers should be relayed to EVWD s SCADA system. Utilizing manufacturer s pump curves adjusted for SCE test data rather than design point curves in the hydraulic model. Project Number: February 2014