Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
TASK 500
TECHNICAL MEMORANDUM NO. 501
COLLECTION SYSTEM MODELING
FINAL DRAFT
December 2010
2700 YGNACIO VALLEY ROAD • SUITE 300 • WALNUT CREEK, CALIFORNIA 94598 • (925) 932-1710 • FAX (925) 930-0208
pw://Carollo/Documents/Client/CA/SFPUC/7240A00/Final Draft PM-TM/500 Collection System/Task500TM501_CollectionSystemModeling.doc (FinalDraft_rev2)
CITY AND COUNTY OF SAN FRANCISCO
2030 SEWER SYSTEM MASTER PLAN
TASK 500
TECHNICAL MEMORANDUM
NO. 501
COLLECTION SYSTEM MODELING
TABLE OF CONTENTS
Page No.
LIST OF FIGURES
Figure 1 Model Network for the SSMP, DDMP, and SSIP ...................................... 501-2
Figure 2 Focus Areas of DDMP Analysis (2007)..................................................... 501-3
After completion of the SSMP, SFPUC through amendments and additional contracts, have
continued to update the collection system model and further refine baseline sewer system
performance estimates. Reports from these studies and projects are described below (by
reference) in this cover memo to provide the reviewer with additional resources to examine
understand the current baseline configuration of the model. The extent of the model
network for the SSMP, DDMP, and SSIP are shown in Figure 1.
The seven focus areas for the study have been selected with input from Bureau of
Engineering (BOE) staff based on the potential and susceptibility to flooding issues.
The types, causes, and degree of potential for flooding vary across the seven areas.
The study will address the potential flooding issues, recommend mitigation criteria
and alternatives, and provide insight for further future analyses.
The San Francisco Sewer System Master Plan (SSMP) model was developed using
the InfoWorks software package to simulate dry-weather and wet weather flows in
pipes 30-inches and larger. It was developed primarily to predict the frequency and
volume of combined sewer discharges (CSD) to the Ocean and San Francisco Bay.
Once calibrated, this tool was then used to analyze the impact of alternatives
designed to reduce flooding and combined sewer overflows to San Francisco Bay
and the Pacific Ocean. The current planning level model does not provide the level
of detail necessary to conduct detailed drainage and flooding assessments. As
such, it will need to be extended, refined and updated to improve drainage and
flooding analysis capability. The modifications will server to better simulate existing
conditions as well as the measures and alternatives that may be considered to
Sewer System Improvement Program Report, DRAFT Report for SFPUC Commission
Review Included by reference only. In support of the SFPUC, the DDMP model was used
to complete a series of additional analyses to address flooding and conveyance concerns
outlined in the memo. Substantative updates to the model were not made from the DDMP
configuration, however, minor updates were made to local sewer networks to better reflect
the objective of the configurations explored as part of this program. The model network
developed under this project represents the most up-to-date representation of San
Francisco’s sewer collection system.
3.0 SUMMARY
In general, the consultants and SFPUC staff were confident of model representations and
sewer system performance estimates and believe the baselines used for the different
studies were an appropriate starting point for considering control options. The additional
documents, included by reference only, provide a reader with information that can be
reviewed in order to understand the evolution and development of the collection system
model and baseline sewer system performance estimates used for master planning efforts.
Model Development,
Validation and Baseline
Report
Prepared for
October, 2007
Contents
Contents .................................................................................................................. i
Acronyms and Abbreviations..................................................................................... v
1.0 Introduction .............................................................................................. 1-1
1.1 Project Description and History......................................................... 1-1
1.2 Description of Collection System ....................................................... 1-3
2.0 Model Development ................................................................................... 2-1
2.1 Software Description ....................................................................... 2-1
2.2 Hydrology....................................................................................... 2-1
2.2.1 Sub-basin Delineation ........................................................... 2-3
2.2.2 Flow Monitoring and Rainfall Gauging Program ....................... 2-5
2.2.3 Flow Monitoring Analysis ....................................................... 2-8
2.2.4 Parameter Estimation.......................................................... 2-13
2.3 Model Network.............................................................................. 2-16
2.3.1 Collection System ............................................................... 2-16
2.3.2 Facilities Characterization .................................................... 2-22
2.3.3 Tidal Data.......................................................................... 2-34
3.0 Model Calibration and Validation.................................................................. 3-1
3.1 Calibration Process .......................................................................... 3-1
3.1.1 Dry Weather Flow (DWF) Calibration ...................................... 3-1
3.1.2 Wet Weather Flow (WWF) Calibration .................................... 3-2
3.2 Calibration Storms......................................................................... 3-11
3.3 Calibration Criteria......................................................................... 3-11
3.4 Calibration Results......................................................................... 3-12
3.4.1 Model Calibration................................................................ 3-12
3.4.2 Meter Calibration ................................................................ 3-14
3.4.3 DCS Calibration .................................................................. 3-15
3.5 Validation Results.......................................................................... 3-15
3.5.1 Flooding ............................................................................ 3-16
3.5.2 Combined Sewer Discharges................................................ 3-16
4.0 Baseline Analysis........................................................................................ 4-1
4.1 Baseline Model................................................................................ 4-1
4.2 Design Rainfall Conditions................................................................ 4-2
4.2.1 Rainfall Events ..................................................................... 4-2
4.3 Model Results ................................................................................. 4-5
4.3.1 Capacity and Flooding........................................................... 4-5
4.3.2 Combined Sewer Discharges.................................................. 4-9
4.3.3 Wastewater Treatment Flows ................................................ 4-9
5.0 Conclusions ............................................................................................... 5-1
i
Tables
Table 1-1 Highlights of CCSF Sewer System History................................................ 1-2
Table 1-2 Age of Sewers ....................................................................................... 1-6
Table 2-1 Basin Number for Major Drainage Basins.................................................. 2-2
Table 2-2 Summary of Sub-basins in each Major Drainage Basins ........................... 2-4
Table 2-3 SFPUC Rain Gauge Network and Location ................................................ 2-5
Table 2-4 - Rain Gauge Data substitutions .............................................................. 2-9
Table 2-5 Zone and Percent Impervious .............................................................. 2-15
Table 2-6 System Types ...................................................................................... 2-21
Table 2-7 Major Offshore Outfalls in the Wastewater System.................................. 2-31
Table 2-8 Parameters Used to Adjust Bayside Tidal Data........................................ 2-35
Table 3-1 Manning’s Roughness Coefficient............................................................ 3-9
Table 3-2 Horton Infiltration Model ..................................................................... 3-10
Table 3-3 Summary of Calibration Storms ............................................................ 3-11
Table 3-4 Summary of Meter Calibration Results.................................................... 3-12
Table 3-5 Model Predicted CSD Activations Compared to Reported CSD Activations for
October 19, 2004- April 3, 2004 .................................................................... 3-17
TABLE 4-1 COMPARISON OF TYPICAL PERIOD TO ORIGINAL PERIOD AND 30 YEAR
AVERAGE ...................................................................................................... 4-3
TABLE 4-2 SUMMARY OF DESIGN STORMS USED FOR BASELINE ANALYSIS.............. 4-4
Table 4-3 Model Predicted CSD Activations for 5-Year Typical Period......................... 4-9
Figures
ii
Figure 2-18 Plan view of IWCS Network of Lincoln Way Overflow............................ 2-33
Figure 2-19 Plan View of IWCS Network of Westside Pump Station ......................... 2-34
Figure 3-1 A Sample Plot of Dry Weather Flow Calibration........................................ 3-2
Figure 3-2 Rain Gauge Boundaries ......................................................................... 3-3
Figure 3-3 Subcatchment Table with Rainfall Profile Highlighted ............................... 3-3
Figure 3-4 Land Use and Runoff Surface Tables ...................................................... 3-4
Figure 3-5 Hydraulic Simulation Scheduling Interface............................................... 3-5
Figure 3-6 Parameter Selection for Runoff Hydrographs ........................................... 3-6
Figure 3-7 Sample Runoff Hydrographs Plotted in InfoWorks.................................... 3-6
Figure 3-8 Wet Weather Flow Calibration Plots ........................................................ 3-7
Figure 3-9 Example of Excluding Park from T Area .................................................. 3-8
Figure 3-10 DWF Model Calibration Results ........................................................... 3-13
Figure 3-11 WWF Model Calibration Results .......................................................... 3-14
Appendices
A Flow Meter Locations
A-1 2004-2005 Monitoring Period
A-2. 2005-2006 Monitoring Period
B Calibration Plots of Meters
C Calibration Plots of DCS Data
D Validation Plots
E Capital Improment Projects
F Rainfall Analysis Technical Memorandum
G SFHHDAR Rainfall Nework
H Baseline Results
iii
Acronyms and Abbreviations
v
ACRONYMS AND ABBREVIATIONS
vi
1.0 Introduction
Faced with an aging wastewater collection system and treatment facilities, and
an increase in operational problems associated with aged and undersized
infrastructure (such as combined sewer discharge (CSD) and flooding), the San Deleted: ng
This report presents a summary of the work activities that were undertaken by
the Scope C Master Planning Team (Metcalf & Eddy, Olivia Chen Consultants,
HydroConsult, and the San Francisco Bureau of Engineering) to build and
validate the computer model.
The model is viewed as tool that is continuously being refined and adjusted. The
discussion and the results presented here provide an overview of the initial
master planning model. During subsequent phases of the Master Plan project,
the model will be further refined and more detail will be added.
The construction of San Francisco’s collection system first started during the gold
rush era in the 1850s. As the City continued to develop, three Wastewater
master plans have been developed and implemented to accommodate the
increased flow, the need for environmental improvements, and various
regulatory requirements.
The 1899 and 1935 Wastewater Master Plans were developed to provide
direction for control of the dry weather pollution problems. The last Master Plan
1-1
1.0 INTRODUCTION
was adopted in 1973 to meet requirements of the “Clean Water Act” of 1972 and
was meant to protect the beneficial uses of coastal waters. The 1973 Master
Plan included the upgrade of treatment plants to secondary treatment and
control of the quantity and volume of the CSOs. To reduce the number of
overflows and meet the current State and Federal discharge regulations, a series
of large underground structures known as Transport/Storage boxes (T/S) were
built to intercept, temporarily store, and transport the wet weather flow to
upgraded treatment facilities before discharge to the San Francisco Bay and
Pacific Ocean. The highlights of the San Francisco Sewer system history are
listed in Table 1-1.
1850-99 City growth resulted in the construction of ~ 150 miles of combined sewers
First facilities plan adopted and guided development of ~700 miles of sewer though
1899
1935
The second facilities plan recommended 3 sewage treatment plants with deep-
1935
water outfalls
1938 The City’s first water pollution control plant was put into operation
1971 The City updates its Master Plan to comply with the Porter-Cologne Act
1972 Federal Water Pollution Control Act adopted by Congress and signed into law
1974 The City adopts the Master Plan after the EPA and the City completed the EIR
1982 Bayside core facilities put into operation (excluding Islais Creek System)
1986 Oceanside core facilities put into operation (excluding Lake Merced and Richmond)
1-2
1.0 INTRODUCTION
The City’s collection system currently consists of approximately 950+ miles of Deleted: 900
sewers, 24,600 manholes, 15,000 catch basins, and 200,000 laterals that
connect to the sewer mains. In addition to these facilities, there are
transport/storage facilities and several sewer tunnels in the City. Most of the Deleted: Over 99%
sewered area of the City is served by a combined sanitary-storm system with the
remaining area served by a separate sanitary system. As shown in Figure 1-1,in
a combined sewer system, runoff from streets and roofs are directed to the same
pipe as the wastewater from homes, businesses, and industry.
Conversely, in a separate system, the stormwater runoff from the streets and
roofs is directed to a stormwater pipe, while wastewater from homes,
businesses, and industry is directed to a separate sanitary sewer (Figure 1-2).
1-3
1.0 INTRODUCTION
In San Francisco’s combined sewer system, there are 10 major drainage basins
from where the wastewater and stormwater runoff is collected and conveyed to
treatment plants through various trunk sewers and transport structures. These
basins and major collection system components are illustrated in Figure 1-3.
1-4
1.0 INTRODUCTION
1-5
1.0 INTRODUCTION
NO INFO
19%
OTHER
1% VCP
45%
CONCRETE
11%
BRICK
7%
IRONSTONE
17%
Many of these sewer pipes are old, with more than 26% of the system built prior
to 1892. The City’s sewers range in age from newly installed to over 100 years.
Table 1-2 shows the percent of City sewers in relation to the year built.
1-6
2.0 Model Development
2.2 Hydrology
The CCSF, excluding Treasure Island, is naturally divided into two drainage
areas: Oceanside and Bayside. Each of these drainage areas can be further
divided into major drainage basins. Oceanside consists of 3 major basins –
Richmond, Sunset and Lake Merced. Bayside consists of 5 major basins – North
2-1
2.0 MODEL DEVELOPMENT
Shore, Channel, Islais Creek, Yosemite and Sunnydale. Each of these major
basins can be further divided into minor basins (depicted alphabetically). Table
2-1 lists the basin number assigned to each of these major drainage basins and
the number of minor drainage basins in each major drainage basin. Figure 2-1
shows the major and minor drainage basins for the CCSF.
20 Richmond 4 A, B, C, D
Oceanside 40 Sunset 13 A, B, C, E, F, G, H, J, K, L, M, N
60 Lake 5 A, B, C, D, E
Merced
10 North 9 A, B, C, D, E, F, G, H, J
Shore
30 Channel 13 A, B, C, D, E, F, G, H, J, K, L, M, N
Bayside
50 Islais Creek 15 A, B, C, D, E, F, G, H, J, K, L, M, N, P, Q
70 Yosemite 5 A, B, C, E, H
70 Sunnydale 2 A, B
2-2
2.0 MODEL DEVELOPMENT
Each of the major and minor drainage basins were further divided into sub-
basins. The purpose of sub-basin delineation was 1) to capture dry weather flow
(DWF) from the tributary pipes and assign it to manholes in the model’s network,
and 2) to capture the likely wet weather flow (WWF) drainage patterns within
the sub-basins.
The approach taken to create sub-basins was to begin at the most upstream
portion of the model network and work downstream, using the following criteria
as guidelines:
2-3
2.0 MODEL DEVELOPMENT
Large parcels of land such as parks, golf courses and freeways that are not
connected to the collection system should be excluded from the sub-basins
for the purposes of collection system modeling.
Sub-basin delineations should not cross over combined or sanitary pipes and
should always end at a manhole.
Sub-basins were delineated based on the location of the smaller sewers in the
tributary areas, aerial photographs, city lot maps and contour information.
AutoCAD and ArcMap software programs were utilized to create these sub-
basins. Table 2-2 gives a summary of the number of sub-basins and the min,
max and average areas for each of the major basins. Figure 2-2 illustrates the
sub-basin delineation.
2-4
2.0 MODEL DEVELOPMENT
Rainfall and flow monitoring were conducted throughout the CCSF from
November 19, 2004 through April 9, 2005. Rainfall data are routinely collected
from 21 locations from the SFPUC rain gauge network (see Figure 2-3 and Figure
2-4). Table 2-3 lists the rain gauge locations.
2-5
2.0 MODEL DEVELOPMENT
15 Herbert Hoover Middle School 2290 14th Ave. (14th Ave. & Santiago)
San Francisco State University - Thornton 19th Ave. (between Holloway &
16
Hall Buckingham)
26 Sunshine High School 2730 Bryant (between 25th & 26th St.)
33 Dr. William Cobb School 2725 California St. (Scott & California)
ADS Environmental Services installed 109 flow meters in the CCSF combined
sewer system. These meters were installed at most pipe exits for minor
drainage basins and in areas of known flooding. In addition, nine flow meters
were placed in combined sewers in the SOMA area and were monitored from
March 19, 2005 through April 9, 2005.
Figure 2-3 show the locations of the rain gauges and flow meters. The general
naming convention of the meters is by major and minor drainages. A more
detailed description of each meter placement can be found in Appendix A-1.
2-6
2.0 MODEL DEVELOPMENT
Figure 2-3 Location of rain gauges and flow meters for the 2004-05 monitoring
period
An additional thirty-two flow meters were installed by ADS and monitored from
December 6, 2005 through April 4, 2006. These flow meters were installed in
locations where there was a lack of information from the previous monitoring
period or a desire for more information in particular areas. These flow meters
will be used for validation of the calibrated model. Figure 2-4 shows the location Deleted: at a future time
2-7
2.0 MODEL DEVELOPMENT
Figure 2-4 Location of rain gauges and flow meters for the 2005-06 monitoring
period
Rainfall data was collected at 5-minute intervals. The rainfall data was examined
for data gaps or unusual data. On the occasion that data was missing due to
malfunctioning, data from an adjacent gauge was used. Table 2-4 shows the
times and location where data was substituted for these particular rain gauges.
2-8
2.0 MODEL DEVELOPMENT
15 – Herbert Hoover Middle 1/18/05 00:00 through 1/18/05 16 – San Francisco State
School 23:55 University
2/13/05 01:40 through 3/19/05
00:00
34 – Palace of Fine Arts 1/18/05 00:00 through 2/04/05 33 – Dr. William Cobb School
15:25
Flow meters recorded depth and velocity at 5-minute intervals. Flow was
computed by multiplying the velocity by the cross-sectional area of the pipe.
Therefore, three sets of data (depth, flow and velocity) were obtained for each
the flow meter.
2-9
2.0 MODEL DEVELOPMENT
As a preliminary evaluation of the meter data, flow was plotted over time for the
entire monitoring period to identify missing and/or unrealistic data and to
observe the meter’s response to rain. Figure 2-5 illustrates an example of a
meter’s response.
5.00 0
Flow (mgd)
4.50 Baseflow 0.05
Rainfall (inches) Wet Weather
4.00 0.1
Flow (WWF)
3.50 0.15
3.00 0.2
Flow (MGD)
Rainfall (in)
2.50 0.25
1.00 0.4
0.50 0.45
0.00 0.5
1/25/05 1/25/05 1/26/05 1/26/05 1/27/05 1/27/05 1/28/05
12:00 AM 12:00 PM 12:00 AM 12:00 PM 12:00 AM 12:00 PM 12:00 AM
Date/Time
2-10
2.0 MODEL DEVELOPMENT
Dry weather flow (DWF) is defined as periods with no direct wet weather
influence, such as rainfall or snow melt. A typical DWF period was 3 days after a
storm event through the beginning of the next storm event. This was based on
the assumption that 3 days is adequate for flow to return to base flow depth
after the end of a storm.
DWF was divided into a sanitary flow component and an infiltration component,
as shown in the wastewater hydrograph (Figure 2-6). Infiltration, generally from
groundwater, enters the sewer through service connections and through
defective pipes, pipe joints, connections or manhole walls. Infiltration is
assumed to be approximately 88% of the minimum nighttime DWF.
5.00 0
Flow (mgd)
4.50 Baseflow 0.05
Rainfall (inches)
4.00 0.1
Rainfall (in)
Flow (MGD)
2.50 0.25
1.00 0.4
0.50 0.45
0.00 0.5
1/25/05 1/25/05 1/26/05 1/26/05 1/27/05 1/27/05 1/28/05
12:00 AM 12:00 PM 12:00 AM 12:00 PM 12:00 AM 12:00 PM 12:00 AM
Date/Time
An estimate of the average DWF for each meter was determined by taking the
average of the daily sanitary and infiltration component of the DWF for the entire
monitoring period. The average DWF was used as a QA/QC tool to determine
theflow-balance between meters. In the event that the downstream meter has a
lower flow rate than the upstream meter or sum of upstream meters, further
investigation of these meters was undertaken.
2-11
2.0 MODEL DEVELOPMENT
Diurnal Curves
Wet weather flow (WWF) usually refers to the stormwater or snowmelt runoff
that contributes directly to the combined sewer system. This is also known as
the inflow, as shown in the wastewater hydrograph in Figure 2-6. To assess the
wet weather response at each meter, storm events that were at least 2 days
apart were used in the analysis. Each wet weather flow period begins with the
first drop of rain and ends when flow returns to the DWF depth.
The inflow volume was plotted against rainfall depth for different storm events.
The slope of the line-of-best-fit is a function of the percentage of rainfall entering
the sewer recorded by that meter (see Figure 2-7). Contributions from
impervious areas cause the slope to increase for larger storms, while system
capacity limitations cause the slope to decrease. Percentage of rainfall captured
can be obtained by dividing the slope with the total drainage area contributing to
the meter. The x-intercept of the line-of-best-fit can be considered as
depression storage.
2-12
2.0 MODEL DEVELOPMENT
30.00
y = 3.2974x + 0.0399
2
R = 0.9865
25.00
Runoff Volume (MG)
20.00
15.00
10.00
Slope is a
function of
5.00 percent capture
0.00
0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00
Rainfall (in)
InfoWorks CS includes several runoff models. The one selected for this
application is the SWMM RUNOFF model because it is widely used in the US and
has been shown to perform well for urban settings. In this model, the
catchment is divided into impervious and pervious fractions. In the impervious
fraction, runoff is calculated using a non-linear reservoir model based on
Manning’s equation applied to overland flow. The flow depth is determined from
a mass balance over the basin with input from rainfall and output from runoff.
In Manning’s equation, the flow depth is reduced by the depression storage. The
same approach is used in the pervious fraction, with the addition of infiltration as
an outflow in the mass balance. Several equations are available to simulate
infiltration, and among those the Horton equation was selected for this project.
Overall, the parameters needed to simulate runoff are the catchment area,
percent impervious, Manning’s n, width, slope and depression storage, as well as
the Horton infiltration parameters.
For each sub-basin, the following parameters were estimated as initial values for
calibration purposes –unit wastewater rate, population, baseflow due to
infiltration, sub-basin width, slope, and percent impervious and pervious area.
2-13
2.0 MODEL DEVELOPMENT
Data from the SFPUC indicated that water consumption for residential use in San
Francisco was approximately 63 gallons per person per day. Based on this
information, the unit wastewater rate was estimated to be 50 to 60 gallons per
person per day.
2.2.4.2 Population
Population per sub-basin was estimated using GIS geoprocessing of the 2000
United States Census Bureau data based on the sub-basin boundaries.
The infiltration flow rate, determined from the DWF processing of each of the
meters, was distributed to the upstream sub-basins in proportion to the pipe
lengths or area of the sub-basin as an initial estimate of the baseflow for each
sub-basin. For sub-basins located between flow meters, the infiltration flow of
the upstream meter was subtracted from the downstream meter before the
infiltration flow was allocated to those sub-basins. Sub-basins that are
predominately parkland, such as the Presidio, were excluded in the allocation of
infiltration flow rate.
The sub-basin width was estimated by dividing the sub-basin area by the sub-
basin length. The length of the sub-basin is determined by measuring the
longest path of overland flow that it would take to reach the exit manhole.
Sub-basin slope was estimated using a spatial GIS analysis of contours from a
digital elevation map. The slope in each sub-basin was calculated by taking the
difference in the maximum and minimum elevation and dividing it by the sub-
basin length.
The percent of impervious area in each of the sub-basins was estimated using
GIS geoprocessing of a San Francisco simplified zoning map (see Figure 2-8)
based on sub-basin boundaries. A percent impervious value was assigned to
each zone as indicated in Table 2-5. The percent impervious for each sub-basin
was calculated by taking the sum weighted average of the zoning areas in that
2-14
2.0 MODEL DEVELOPMENT
C or N Commercial or Business 85
M Industrial 70
S Streets 98
2-15
2.0 MODEL DEVELOPMENT
In order to assure that the model would closely correspond to the City GIS
database for future updates and modeling use, a software program was selected
that would allow for multiple users to perform quality assurance and quality
control (QA/QC) work on the system data prior to the initial import into
2-16
2.0 MODEL DEVELOPMENT
InfoWorks CS. To accomplish this task, the City converted the GIS database
(SQL server) into an Oracle spatial database so that the Munsys software could
be used to interface with the data tables by staff who are more familiar with
AutoDesk drafting program tools and functionality.
The aim of this initial data population effort was to assure that major trunk lines
intended to be part of the Wastewater Master plan (WWMP) model had size and
shape data, as well as rim and invert elevations at the manhole node locations.
Data gaps in the Oracle spatial database were filled using as-builts via the
Munsys software. After information was input from available as-builts, the
Digital Terrain Model (DTM) was used to populate missing manhole rim
elevations. Missing invert elevations at nodes were populated with the
elevations obtained after subtracting a fixed depth value (average depth from
dataset that had values) from the rim surface elevation.
The City Sewer GIS database represented many of the various hydraulic
structures including weirs, orifices, sluice gates, and pumps. However, the
actual data required by the model for importing these various objects was not
known during the time that this City GIS was under development and
modification. Instead of inputting this data into Oracle via the Munsys interface,
it was decided to implement InfoNet (IN), an asset management database
software program companion to InfoWorks developed by Wallingford.
The sewer links and nodes were exported from Oracle spatial database into ESRI
GIS shapefiles for import into IN so that special structures could be coded and
for the continuation of QA/QC work on the existing data. Two access database
look-up tables were created and used in the export process so that the sewer
shapes and material types would have all data required by the IN software
program. Once data was inputted, modified, or coded within IN, it can be
exported back to ESRI GIS shapefiles and incorporated back into the City Oracle
spatial database for future use.
The sewer links and nodes were initially imported into IN on two systems types
2-17
2.0 MODEL DEVELOPMENT
so that a color scheme could assist in quickly identifying which sewers would be
included in the model. IN also allowed for some user defined fields to be
imported into the dataset from the City GIS database. Even though this
information is not required to run the hydraulic model, having as-built document
numbers, and surface elevations at nodes from the DTM aided QA/QC and
coding work.
Aside from collection system network data, subcatchments were also imported
into Infoworks CS from ESRI GIS shapefiles. After delineation work was
performed in AutoCAD, the boundaries were converted to ESRI GIS shapefiles so
that ESRI’s ArcMAP software could be used to clean up boundary gaps and
overlaps. Lastly, the subcatchment shape file was populated with preliminary
data required for running the Infoworks CS model.
Other data imported into the Infoworks CS model included the various diurnal
pattern profiles developed for each meter location, rain gage data, outfall tide
levels, and real time control (RTC) rules at various pump station locations.
The model is intended to cover only the main interceptors and sewers. The initial
criterion was that conduits 30 inches in diameter (or equivalent, if not circular)
and larger were included in the model, however, some smaller and larger sewers
were included and excluded from the model, respectively.
Sewer that were smaller than 30” diameter were included in the model when it
was determined that they were essential to the modeling effort and included
various auxiliary sewers, sewers on steeper grades with a large amount of
upstream area contributing to large flows, and even dry weather flow diversion
lines. Larger sewers were excluded from the model when it was determined that
the sewer (ie.3’x5’ brick) was plugged at its upstream end and no longer carried
significant flows, when larger brick wings were located at intersections in
upstream parts of the smaller sewer system network, and when there was a
small enough subcatchment contributing runoff to the upstream end of the pipe.
All manholes on the sewers and interceptors were included in the model. There
were cases where the City GIS nodes represented junctions of two pipes of a
different size and were not actually manholes. When identified, these were
noted appropriately. All manholes were included in order to provide a one-to-
one relationship between the model, the GIS, and the actual sewer system.
2-18
2.0 MODEL DEVELOPMENT
basins that are largely independent, although they do interact as they reach the
transport/storage (T/S) boxes or the plants. Therefore, model development
proceeded on a basin-by-basin basis with different modelers working on their
respective basins. Regardless, the individual basin models were combined into
two overall models for analysis of the Bayside and Oceanside systems.
The IN software was used for coding special structures as well as performing
QA/QC work on the existing network. Since IN allows for ESRI GIS shapefiles to
be imported, a boundary showing the location of these special structures was
exported from the City Oracle database and used in IN to identify their locations.
By writing queries in IN, it was possible to select sewers in the system network
that had adverse slopes so that they could be targeted for profile review. Linear
interpolation was used to correct sewer invert elevations as well as manhole
chamber depths.
2-19
2.0 MODEL DEVELOPMENT
Upon review of the network, it was evident where assumed data was used by
the City before the GIS was imported into IN. By creating and using customized
IN queries, locations where the rim elevation had the same value as the DTM
surface or where the chamber depth value was the assumed average depth were
easily located. If there was data missing upon review of a sewer profile, then
available as-builts were reviewed for required information. If no data was
available, then IN inference tools were used to populate missing data. Once a
sewer link and node was reviewed and checked, its system type was changed as
well as the flag. Figure 2-10 indicates the flags that were used to identify where
data originated and why it was changed.
Table 2-6 provides a list of the system types used throughout the QA/QC
process.
2-20
2.0 MODEL DEVELOPMENT
In the original import into IN, the user defined field that held the cross section
area data was used to put links and nodes connected to those links on either a
foul or unspecified system type. Also, links that had zero for size data were
assigned a system type of foul to assure that they were reviewed during the
QA/QC process.
In order to determine the model extents, obvious links for foul system types
were changed to an unspecified (smaller upstream links that had zero data) or
3’x 5’ brink wings. There were also cases where unspecified sewers needed to
be changed to a foul system type for further review. Once model extents were
established, all links and nodes on the foul system type were examined using
query tools and profile reviews. If data was changed for a sewer link and node,
it was flagged and the system type was changed to combined, surface, or
transition. The model was separated into major basins with a different modeler
assigned to a unique basin. This was done to expedite the validation and QA/QC
process. IN also has a built in conflict resolution flag if different users modified
the same piece of data.
Once network data was imported into Infoworks CS from IN, a few more QA/QC
checks were performed. Not all special structures data required to run the model
could be input into IN. For example, a large box chamber could be identified in
IN, but the top and side cross sectional areas required input within Infoworks CS.
Further, custom cross sections required to more accurately model a non-
standard sewer shape were identified as bizarre system types within IN and
addressed within Infoworks CS. Since further modifications to the network were
anticipated during calibration work, the data flags assisted the modelers in
determining whether data was either assumed, interpolated, inferred, or from an
as-built drawing.
2-21
2.0 MODEL DEVELOPMENT
In order to assure that this model will be an effective tool during the alternative
analysis phase of the Wastewater Master Plan Project, major collection system
facilities were included in the network. Major existing collection system facilities
are illustrated in Figure 2-11. The transport/storage (T/S) boxes and tunnels
ring the City like a moat around San Francisco’s shoreline. The Bayside T/S
collects wastewater from the east side of the City and ultimately routes flow to
the SEP during dry and wet weather seasons and to the NPF during wet weather
periods. The oceanside system conveys water from the western side of the City
to the OSP.
The three main functions of the T/S boxes are to capture combined storm water
from the gravity collection system and prevent overflows into the coastal waters,
store for later treatment, and provide a level of treatment by settling out grit
prior to discharge into the coastal water when the system’s storage capacity is
exceeded. Figure 2-12 shows the location of the various T/S boxes around the
periphery of the City.
2-22
2.0 MODEL DEVELOPMENT
2-23
2.0 MODEL DEVELOPMENT
The SEP has a wet weather treatment capacity of 250 MGD, a maximum
secondary treatment peak flow of 150 MGD and a design secondary average flow
of 85 MGD. The NPF is a wet weather only treatment facility with a wet weather
2-24
2.0 MODEL DEVELOPMENT
treatment capacity of 150 MGD. Dry and wet weather flow diagrams for the
Bayside Drainage Basins are shown in Figure 2-13 and Figure 2-14, respectively.
The Northern Bayside facilities include a seven-mile system of underground T/S
sewers that are 20 feet in
width and 45 feet in
depth, and feed two
pumping stations (North
Shore and Channel Pump
Station). The Southeast
Bayside facilities were
designed to store and
transport flow to the SEP
through 3 miles of
concrete T/S structures,
enlarged pipe sewers, and
four small pump stations.
This system is comprised
of five distinct facilities
listed below:
Mariposa Facilities
Islais Creek Outfall
Consolidation and
Flynn Pump
Station,
Hunters Point
Facilities
Yosemite/Fitch
Outfall
Consolidation and
Griffith Pump
Station
Sunnydale Outfall
Consolidation and Pump Station.
2-25
2.0 MODEL DEVELOPMENT
facilities and T/S structures are exceeded, the excess flow is discharged into the
Bay via the 29-shoreline overflow structures.
The Northern Bayside Facilities begin with the Jackson and Marina T/S systems
that receive flow from Golden Gate National Recreational Area (GGNRA) and the
northeastern sector of the City. The T/S structures plus associated sewer
storage capacity is 24 million gallons. In dry weather, flows are conveyed to the
North Shore Pump Station, which pumps 15-34 MGD to the Channel transport
system. In wet weather, it is capable of pumping 150 MGD of combined flows to
the North Point Wet Weather Facility (its maximum treatment capacity). The
NPF provides screening, primary treatment, chlorination and dechlorination for
wet weather flows, and discharges these flows offshore through the North Point
Bay Outfalls.
The Channel T/S structure collects and transports flows from the South of Market
and northern Mission Districts and has a total storage volume of 38 million
gallons.
The Channel System handles sewage pumped from the North Shore Pump
Station during dry weather and gravity flow from the Channel drainage area
sewers. The Channel Pump Station pumps all flows through a 66-inch force
main directly to the Southeast Water Pollution Control Plant (SEP). Dry weather
flows from the Channel and North Shore areas average 40 MGD. In wet
weather, the North Shore flows are diverted to the NPF, and the Channel Pump
Station is capable of pumping 103 MGD of combined flows from the local
drainage area to the SEP.
The Mariposa Pump Station (MPS) and the 20th Street Pump Station (TWS)
pump dry and wet weather flows to the SEP and were constructed in 1992. The
T/S structure has a capacity of 0.6 million-gallon volume and was designed to
collect/transport wastewater from the Potrero Hill and 20th Street areas. The
Mariposa pump station capacity is 3 MGD during dry weather and 13 MGD during
wet weather. Discharge from the Mariposa Pump Station is routed to a gravity
sewer at 21st and Illinois Streets for transport to the SEP for treatment.
2-26
2.0 MODEL DEVELOPMENT
The 20th Street transport and pump station facilities are included within the
Mariposa Facilities and flows are discharged to the same location. It has a
pumping capacity of 0.78 MGD and the transport storage volume is 0.2 MG.
Yosemite Facilities
Sewage and storm water runoff from the Bayview/Hunters Point and Candlestick
areas are intercepted and transported by the Yosemite Facilities, which were
completed in 1989. The combined volume of T/S structure plus associated
sewer storage capacity is 11.5 million gallons. Dry weather flow from the upper
Yosemite area flows to the SEP via the Hunters Point Tunnel. Flows generated
from the lower Yosemite areas combine with flows from Sunnydale and are
pumped to the SEP by the Griffith Pump Station.
The Griffith Pump Station capacity during dry and wet weather is 10 MGD and
120 MGD, respectively. Flows from the Griffith Pump Station are discharged to
the southeast sewer system and flow by gravity to the SEP.
Sunnydale Facilities
The Sunnydale T/S and associated facilities were completed in 1991 and
designed to collect and transport sewage and runoff from the Sunnydale
drainage area. The total storage volume of this structure and associated sewers
is 6.2 million gallons.
During normal dry weather operation, flow bypasses the Sunnydale Facilities,
flows toward the Yosemite system, and the Griffith pump station and pumps it to
the SEP. Combined flows are diverted from the gravity system to the T/S
structure, and then conveyed to the Sunnydale pump station. This station
2-27
2.0 MODEL DEVELOPMENT
pumps a maximum rate of 50 MGD toward the Candlestick tunnel sewer which
then flows towards the Yosemite Facilities.
The Westside System for which construction was completed in 1994, consists of
the Richmond (RMT), Westside (WST), and Lake Merced (LMT) transports, the
Westside Pump Station
(WSS), the Oceanside
Water Pollution Control
Plant (OSP), the Southwest
Ocean Outfall (SWOO),
and seven CSO near-shore
discharge points.
2-28
2.0 MODEL DEVELOPMENT
Dry and wet weather flow diagrams for the Bayside Drainage Basins are shown
in Figure 2-15 and Figure 2-16, respectively.
Richmond
Transport/Storage (T/S)
System
A 2100-foot long
segment of 10-foot
diameter pipe that
connects to the
upstream end of the
RMT
A 2000-foot long
segment of 42-inch
diameter pipe
connecting the RMT to
the WST
The Mile Rock outfall that provides hydraulic relief by allowing discharges
near Land’s End
Figure 2-16 Current Westside Wet Weather
Flow Diagram
Flows from the northwest section of the City are controlled and conveyed by
RMT. Combined sewage is intercepted at 17th, 22nd, 24th Avenues and the
2-29
2.0 MODEL DEVELOPMENT
Presidio, and is eventually delivered to the WST at Great Highway and Fulton
Street intersection. Because of the elevation of the Mile Rock Overflow structure,
segments of the Richmond Facilities are pressurized during wet weather events.
Richmond District drainage area, Sunset District and Lake Merced flows all
converge in the WST and are pumped by the Westside Pump Station (WSS) to
Oceanside Plant (OSP) for treatment. Flows that exceed the OSP capacity pass
through a decanting process in the WST. This process removes grit and debris
before it is pumped out via the SWOO (95 -110 MGD).
WST is a 45-foot deep and 25-foot wide reinforced concrete box structure
running north-to-south under Great Highway from Fulton Street to Sloat
Boulevard and has a useable storage volume of 49.3 million gallons.
All combined storm water and sanitary sewage flows leaving WST receive the
equivalent of primary treatment before being discharged out SWOO or, during
extreme storm conditions, from Mile Rock, Lincoln, Vicente, or Lake Merced
Outfalls. Nearly all WST influent arrives via gravity. Flows from low areas are
pumped by Pine Lake Pump Station (PLS) in the west end of Stern Grove and by
the Zoo Wet Weather Pump Station (ZWS) northeast of the intersection of Sloat
and the Great Highway.
The LMT system was constructed with the intent to capture and control flow
generated from the southwestern portion of the City. Major components consist
of the discharge structure, outfall line, transport line, and transition structure.
Combined sewage flows from the southwest of the City are conveyed to OSP via
the LMT, which discharges to the WST east box cunette. All Westside dry
weather flows (as well as combined flows prior to decanting to the west
chamber) mix at this point. Flows entering the WSS east inlet are pumped to the
OSP. A transition structure at the north end of the LMT directs flows under the
Great Highway to the east box. Flows enter the LMT System through a 3-
compartment sewer adjoining the transport structure on John Muir Drive. This
structure directs both dry and wet weather flows to the LMT and excess storm
flows to an outfall line and with a capacity of 10 MG.
Offshore Outfalls
2-30
2.0 MODEL DEVELOPMENT
There are four offshore outfalls in the wastewater system which discharge
treated effluent from the three treatment facilities. In addition to these outfalls,
a series of onshore outfalls exist in the wastewater system. These onshore
outfalls operate infrequently during high flow events and allow for discharge
from the transport/storage (T/S) system. Table 2-7 provides a summary of the
offshore outfalls in the City’s combined sewer system.
The IWCS model has the capability of modeling the operation of the various
collection system facilities and components. However, the model is limited to
collection system components and the operation of the treatment plant facilities
and discharge of treated flow through the various outfalls.
Before these complicated facilities and structures could be coded into the
modeling software, available as-builts and other in-house information was
reviewed so that schematics could be developed. Field inspections were
performed to obtain information that was not available upon review of in-house
records. The schematic in Figure 2-17 illustrates a few critical components of the
WST facility that were included in the model.
2-31
2.0 MODEL DEVELOPMENT
Figure 2-17 Isometric of Lincoln Way Overflow and Westside Pump Station
The pump station operations and modeling was achieved by coding the various
pumps, screens, gates, sump chambers, and force mains into the network as
illustrated in Figure 2-19. Data from the station pump curves were input into
model as well as real time control rules guiding how they operate during the
time-step simulations within the model. Many of the initial assumptions on the
operation of these facilities were modified during calibration phase of the project.
2-32
2.0 MODEL DEVELOPMENT
Manho
Lincoln Way
Overflow
Sewer
Weir
s Gat
2-33
2.0 MODEL DEVELOPMENT
Sloat Blvd
WS
Wei
Sluic
e
Screens
Pum
Sewer
Manhol Link
Tidal data was obtained from the National Oceanic & Atmospheric
Administration’s (NOAA) San Francisco gage. The gage is located on the
Northeast side of the National Parks Service Wharf on the western portion of
Crissy Field in the Presidio. The coordinates of the gage are Latidude: 37º 48.4’,
Longitude: 122º 27.9’.
For the 2004-2005 wet season, 12-minute tide data obtained from the NOAA
gage at Chrissy Field was adjusted to account for changes in high water levels
2-34
2.0 MODEL DEVELOPMENT
and shifts in time for four other locations in the bay, Pier 41, Pier 22 ½, Islais
Creek and Hunter’s Point. High water levels at these four locations were
increased by a value (as described in Table 2-8) and water levels for times within
3 hours of the high water level were scaled accordingly. The data was then
shifted in time by an appropriate value. Table 2-8 provides a summary of the
factors used to adjust Bayside tidal data at the various locations. The five tidal
data points were used to define the boundary conditions of the 36 different
outfalls in the InfoWorks model. The outfalls were assigned the tidal data that
corresponded to the data point located closest to the outfall. Oceanside outfalls
were assigned tidal data equivalent to the Chrissy Field Gage.
For the 1977-1981 5 year typical period, a similar procedure as described for the
2004-2005 wet season tidal data was used, with the exception that 1-hour tidal
data was used. Because the time scaling factors used to adjust the points along
the bay were all half an hour or less and a 1-hour interval was used, it was not
necessary to time shift the tidal data.
2-35
3.0 Model Calibration and
Validation
The hydraulic model of San Francisco’s system was calibrated using a three step
process. First, the dry weather flow in the system was calibrated to meter data
for a dry weather period during flow monitoring. Second, a wet weather
calibration was performed for a three-week period starting in late December,
2004. This period contained several moderate sized storms and a few dry
weather days. At this stage, model parameters were adjusted to match wet
weather flows and major discrepancies were resolved. Third, two other several
day periods were simulated.
Dry weather flow calibration involved determining the unit flow rates and 24-
hour diurnal flow profiles in the model that resulted in the best match of
modeled flows to monitored flows for a typical dry period. Two flow components
were calibrated during dry weather: the minimum dry weather base flow and the
domestic sanitary flow. The dry weather base flow (minimum flow in the sewer)
was determined from metering data. The measured base flow at each meter
was distributed to upstream sub-basins by area in gallons per acre per day
(gpad). Per-capita flows in gallons per day per capita (gpcd) were calibrated for
both domestic and employee populations. Equivalent populations were used to
account for flows from commercial, institutional and industrial uses. Calibrated
domestic per-capita rates ranged from 50 to 60 gpcd and were based on water
consumption rate data. Weekday and weekend diurnal profiles were adjusted to
better match hourly flows. Over 100 diurnal profiles were developed to match
each meter’s flow and sub-basin characteristics. Figure 3-1 shows an example
of dry weather calibration for meter 40G4. Overall, the DWF calibration results
are excellent for depth, flow and velocity, with different diurnal patterns for
weekdays and weekends apparent.
3-1
3.0 MODEL CALIBRATION AND VALIDATION
----------- Meter
Weekend Weekday ----------- Model
Wet weather flow calibration involved adjusting the model parameters for
groundwater infiltration, and for the volume and timing of I/I entering the sewer
system to match monitored flows during particular rainfall events. The rainfall
events from November 2004 to April 2005 were used for calibration and
verification.
Rainfall data were imported into InfoWorks and assigned to rain gauge polygons.
All subcatchments located within the rain gauge boundary used that gauge’s
data. When the need arose to use data from another guage, the centroid of the
subcatchment was modified to make it fall within the boundary of the other
gauge.. The rain gauge boundary setup overrode the rainfall profile number
specified in the subcatchment table (Figure 3-3). Figure 3-2 shows the
boundaries of the 21 rain gauges used in the model
3-2
3.0 MODEL CALIBRATION AND VALIDATION
Wet weather flow parameters were entered in the subcatchment table (Figure 3-
3) including total and contributing area, % impervious, catchment width, and
3-3
3.0 MODEL CALIBRATION AND VALIDATION
ground surface slope. Figure 3-4 shows other parameters used in the land use
and runoff surface tables. The parameters in these tables had the most influence
on the wet weather flows in the model. The land use table is used to globally
control population derived flows and runoff flows. In the current setup, there is
only one land use type in the Oceanside model, which is named as “West Side”.
If no population data and wastewater profiles are specified in the subcatchment
table, the model defaults to a population density of 10 person/acre and the first
wastewater profile specified in the model (for dry weather diurnal variation) to
generate dry weather flows for that subcatchment. For example, in figure 3-4,
runoff surfaces 10 and 20 are linked to the “Runoff Surface” table, where surface
10 is defined as “West Side Impervious” with overland surface Manning’s
n=0.011 using the surface runoff routing model SWMM1. Surface 20 is defined
as “West Side Pervious”. It also uses SWMM for the surface runoff routing with
overland surface Manning’s n=0.15. Infiltration is simulated using the Horton
equation. Infiltration parameters listed in Figure 3-7 are examples of the input
parameters.
Because the model relied so heavily on flow meter data accuracy, the meter data
was checked again prior to wet weather flow calibration using the following
factors:
1 Runoff routing models determine how quickly the rainfall enters the drainage system from the catchment. SWMM is one
of 5 routing models available in the Infoworks CS software. In Infoworks CS, one can assign different routing models to
each surface type but all the surfaces on a particular land use must have the same routing model. SWMM is a non-linear
reservoir model developed in the USA.
3-4
3.0 MODEL CALIBRATION AND VALIDATION
If the velocity data was questionable, but the depth data seemed reasonable, the
model was calibrated with depth data, or vice versa. In the absence of
reasonable data, the model was not calibrated with flow monitoring data in that
location.
Two approaches were employed to quickly see the effect of parameter changes.
First, a shorter period was run, and second, a runoff only simulation was
conducted using the option shown in Figure 3-5. This option allowed a quick
view of how the model responded to changes made to the following wet weather
flow parameters: contributing area, % impervious, catchment width, and runoff
surface Manning’s roughness.
A sample runoff hydrograph showing the results from simulating runoff only is
shown in Figure 3-7. If the volume of rainfall is less than the volume of
infiltration, there will be no runoff generated from the pervious area, i.e. surface
02 in figure 3-6. Once the wet weather runoff parameters have been refined, the
“Simulate Runoff Only” option was unchecked so that the real impact on pipe
flows could be seen.
3-5
3.0 MODEL CALIBRATION AND VALIDATION
3-6
3.0 MODEL CALIBRATION AND VALIDATION
results from a full simulation for one of the meters in Basin 40. The model under
predicts flow at the beginning of the storm but over-predicts at the end. Velocity
was under-predicted during dry weather. Volume comparison shows about 4 MG
over-prediction. In this case, decreasing the percent impervious area resulted in
a decrease in flow, but the catchment width was increased in order to maintain
the hydrograph peak and shape. The proper adjustment of these parameters
helped match the hydrograph shape (i.e. time to peak and recessions) and peak
flow rates.
In the subcatchment table, there are two columns for area, one for “Total Area”
and another for “Contributing Area”. By default, the contributing area is the
same as the total area. However, areas like parks and golf courses need to be
excluded from the contributing area since they do not usually connect to the
collection system. These areas were subtracted from the contributing area
acreages (Figure 3-9). For example, subcatchment 17613 covers a large area in
the Golden Gate Park, which was taken out of the contributing area for wet
weather flow calculations. Comment [b1]: How was this
done? How did you determine what
area came out. Surely some areas of
the park must drain towards a
manhole or combined sewer line. If
so, was GIS used?
----------- Meter
----------- Model
3-7
3.0 MODEL CALIBRATION AND VALIDATION
3-8
3.0 MODEL CALIBRATION AND VALIDATION
the sewers. The majority of the flow in a combined sewer system comes from
impervious areas. Impervious areas usually show a shorter recession limb than
pervious areas. Catchment width can change the hydrograph shape significantly.
Decreasing catchment width results in a lower peak and longer duration
hydrograph. Adjustments are usually done on a trial and error basis.
Increasing the Manning’s surface roughness coefficient attenuates the peak flow.
Table 3-1 presents Manning’s roughness coefficients from InfoWorks. The
catchment width can vary from basin to basin, but only one Manning’s roughness
coefficient can be assigned to each runoff surface type.
3.1.2.6 Infiltration
Factors that influence the process of infiltration into the soil include soil property,
soil surface condition, soil moisture content at the time of the storm and
compaction. When the rate of infiltration is greater than the rainfall rate, all
rainfall is lost to infiltration. If the rainfall rate is greater than the infiltration
capacity of the soil, surface ponding and/or surface runoff will occur. The Horton
equation was chosen to model infiltration, which requires the minimum and
maximum infiltration rates and decay factors to be defined. Different values for
Horton's infiltration model are available from the literature (see table below).
Minimum infiltration factors as determined by infiltrometer studies are highly
variable and vary by orders of magnitude on seemingly similar soil types. The
factors in Table 3-2 were adjusted if necessary during wet weather flow
calibration.
3-9
3.0 MODEL CALIBRATION AND VALIDATION
For example, if the contributing area is correct and the model is over-predicting
the flow, the percentage of impervious area was decreased with a corresponding
increase of pervious area. Adjusting percent imperviousness and infiltration
characteristics were done using several iterations. Once the total volume is
matched, then the depth and velocity were occasionally modified by changing
pipe slope (only if the inverts were estimated) and roughness.
3-10
3.0 MODEL CALIBRATION AND VALIDATION
The flow monitoring period of November 19, 2004 through April 9, 2005 included
a number of rainfall events. Some of these were selected to calibrate the model.
After reviewing flow data and rain gauge data, the main period selected for
calibration was from 12/26/04 to 1/14/05. This period was chosen based on
the sufficient length of time, the total rain depth, and peak intensities of the
rains. Two other periods, 1/25/05 to 1/30/05 for 5 days and 2/14/05 to
2/23/05 for 9 days, were used to verify the calibrated model. Table 3-3
summarizes the selected rainfall periods.
Typical calibration goals are indicated below for both dry-weather and wet-
weather conditions. Those goals recognize that the monitoring data themselves
have a certain degree of error. Flow measurements in sewer systems are
considered accurate to ±10% when conditions are good.
3-11
3.0 MODEL CALIBRATION AND VALIDATION
3-12
3.0 MODEL CALIBRATION AND VALIDATION
3-13
3.0 MODEL CALIBRATION AND VALIDATION
The collected meter data was verified both in-field and in-office.
The in-field data verification was performed by ADS technicians who manually verified
depth measurements to ensure their equipment was recording correctly. Whenever a
disparity arose between the electronic equipment and the manual equipment,
technicians would remedy the situation, either by recalibrating or replacing the
equipment. Any questionable data is flagged by the field technician to be further
analyzed by office technicians.
In-office data verification was performed by both ADS technicians and master plan
personnel. Although both groups did independent data verification, the same processes
were used to locate possible problems with the data. These processes include:
3-14
3.0 MODEL CALIBRATION AND VALIDATION
DCS flow data were used to assist with the calibration of the North Shore and
Channel pump stations.
DCS flow data was available at the downstream end of both the DWF and WWF
pumps. Since the pumps were manually operated, it was not possible to
calibrate these pumps to the exact operating conditions as captured by the DCS
data. However, crucial operating conditions, such as the DWF pumps are turned
off when the WWF pumps are on, were captured in the calibration.
Furthermore, the model was calibrated to the general average flow rates of both
the DWF and WWF pumps.
DCS flow data was available for each of the four pumps at the Channel Pump
Station, as well as at the downstream pipe of these pumps. Although there are
four pumps, a maximum of three pumps are operating at one time (one pump is
for standby). However, there is no one designated standby pump. This role of
standby is cycled between the four pumps, so it was difficult to match the DCS
data for each of the pumps. Therefore, for modeling purposes, Channel Pump
#4 was designated as the standby pump at all times and the remaining pumps
were calibrated to match the average flow rates.
After meter calibration, the network was put through a series of validation
procedures to complete the verification of the network. The model was used to
simulate various design storms and periods for comparison to known system
characteristics. Model results were compared to such parameters as flooding
and CSD activations.
3-15
3.0 MODEL CALIBRATION AND VALIDATION
3.5.1 Flooding
The model was used to compare simulated flooding locations with reported
flooding locations throughout the city. The 2 and 5-year design storms were
simulated and results analyzed to determine the water levels and surcharge
characteristics of the manholes. Manholes were characterized based on the
maximum surcharge level in relation to the manhole rim elevation during the
given storm. The surcharge characteristics of the manholes were then mapped
to show locations of flooding throughout the city. This map was analyzed to
compare locations of known flooding with those predicted by the model.
During the verification stage of the InfoWorks model, the calibrated model was
used to simulate the 2004-2005 wet weather season. Rainfall and flow
monitoring were conducted throughout the CCSF from November 19, 2004
through April 9, 2005. Rainfall data are routinely collected from 21 locations
from the SFPUC rain gauge network (see Figure 2-3 and Figure 2-4). Table 2-3
lists the rain gauge locations. The number and duration of reported CSDs were
compared to the modeled CSDs for the 2004-2005 wet weather season. The
modeled CSDs were based on actual rainfall that fell during the 2004-2005
calibration period over the SFPUC rain guage network discussed in section 2.2.2.
The CSDs reported from Bayside system were obtained from the NPDES
monitoring report sent to the California Regional Water Quality Control Board on
February 3, 2006. The CSDs reported from Oceanside system were obtained
from a Wastewater Enterprise Division Inter-Office Memo sent on August 11,
2005 with the subject “CSOs for 2004-2005.”
In order to remain consistent with San Francisco’s NPDES Permit, overflows were
considered multiple events if they occurred greater than six hours apart.
Overflows which occurred within six hours of each other were considered the
same overflow event.
It was concluded that for the entire wet weather season, the model accurately
predicted the reported number of overflow occurrences at various locations.
However, the model did not always match the reported number of individual
overflows. Many of these discrepancies are attributed to differences in pump
system operations, as discussed in Section 3.4. It should also be noted that
there are individual outfalls, particularly in the Islais Creek region, for which
there are no reported overflows. It is likely that at the time of the overflows,
monitoring equipment was not present at all of the outfalls. Therefore, this
report has grouped overflows by region instead of by individual outfall. Table 3-
5 includes a summary of the reported and modeled overflows for the 2004-2005
3-16
3.0 MODEL CALIBRATION AND VALIDATION
wet weather season. Refer to Appendix D for a list of modeled versus reported
overflows for the individual outfalls.
3-17
4.0 Baseline Analysis
The goal of the baseline analysis is to apply the model of the existing system
under various design conditions to identify limitations in the system, such as
CSOs and flooding. Two different types of design conditions were developed for
baseline analysis. First, a typical long term period of 5 years was developed
from historical data to evaluate master planning control alternatives. Second, a
series of short, intense, discrete storms of various return periods were developed
to evaluate system capacity. Once system conditions are evaluated, the baseline
model will serve as the basis from which different alternative solutions can be
analyzed.
5-1
4.0 BASELINE ANALYSIS
Two sets of design rainfall conditions were created for evaluating the San
Francisco sewer system.
Typical five-year and one-year rainfall periods for evaluating CSO control
alternatives.
Design storms of various return periods for evaluation of the system during
discrete storm events.
4-2
4.0 BASELINE ANALYSIS
Long-term typical periods lasting one and five years were developed by selecting
a period which closely matched the average record from the National Weather
Service (NWS) rain gauge for the past 30 years and adding and removing storm
events to further improve the match. The period that was selected was 1977-
1981. For typical period simulations, rainfall from the more detailed SFHHDAR
citywide rain gage network will be used to take into account spatial variation in
storms. The selected five year period was tailored to match rainfall averages for
the past 30 years instead of the overall period of record to take into account
factors such as global warming and El Niño (See Appendix F). A summary of the
30 year average as well as the five year typical period can be found in Table 4-1.
The rainfall analysis used the long-term historical data from the NWS gage. For
the InfoWorks model, however, the data from SFHHDAR network was utilized so
that spatial variation would be taken into account. The SFHHDAR network was
in operation from 1972-1985, and consisted of 27 gages. For the actual
locations of all the gages, refer to Appendix G.
During analysis of the SFHHDAR network data it was discovered that all gages
were missing rainfall during instances where the NWS gage reported storms. For
these time periods, the InfoWorks rain gage profiles were populated by spatially
varying the data from the NWS gage data. The NWS data was spatially varied
according to the parameters set forth in the Hydrometerorological Report
(Hydroconsult, 1981). Refer to Appendix G for a list of time periods when the
4-3
4.0 BASELINE ANALYSIS
SFHHDAR network was down as well as a list of the scaling factors used for
spatial variation.
After importing the complete rain gage data of the five year typical period into
InfoWorks, Thiessen Polygons were created around the rain gage network. The
rainfall profile for each subcatchment was assigned the value of the rain gage
whose Thiessen Polygon contains the subcatchment.
To develop design storms, an analysis was conducted based on the past 30 years
of rainfall recorded at the NWS rain gauge. Intensity-Duration-Frequency (IDF)
curves for the rain gauge were developed based on information provided by the
State Climatologist’s Office, and extrapolated to cover the 3-month, 6-month,
and 1-year return periods. The resulting design storms were compared to the
storms currently being used in the City of San Francisco. The currently used IDF
curves have a higher peak intensity for very short times (5-10 minutes), but a
lower overall volume compared the curves developed as part of this analysis.
There are significant implications to switching from currently used design storms
that affect issues outside of the Wastewater Master Plan, such as recommended
sewer design sizes. Therefore, it was decided to use the current design storms
for the design analysis. Table 4-2 is a summary of the characteristics of the
design storm events.
These design storms were multiplied by the regional factors described in the
Hydrometerorological Report (Hydroconsult, 1981). A list of the scaling
factors used for spatial variation can be found in Appendix G.
4-4
4.0 BASELINE ANALYSIS
The capacity analysis was performed by applying the calibrated hydraulic model
with added CIP improvements to compute flows and levels in each modeled pipe
under 2-year and 5-year design storm conditions. The results of these model
runs may indicate where additional capacity will be required to continue to
provide the desired level of protection against flooding and overflows. The 2-year
and 5-year design storms were chosen for the following reasons: the City’s
design criteria used a 5 year storm, the storms provided a valid characterization
of how the system might respond, and the storms gave the best means of
comparison to the City’s known areas of flooding. Figures 4-2 and 4-3 show the
relative surcharging and flooding in the system due to the 2-year and 5-year
storms. Figure 4-4 shows the severity of flooding related problems from the
February 25, 2005 storm (100-year storm).
4-5
4.0 BASELINE ANALYSIS
Figure 4.3.1 Baseline Model Capacity and Flooding using 2-year, 24 hour Design
Storm
4-6
4.0 BASELINE ANALYSIS
Figure 4.3.2 Baseline Model Capacity and Flooding using 5-year, 24 hour Design
Storm
4-7
4.0 BASELINE ANALYSIS
The 5-year design storm shows much more flooding than the 2-year design
storm for medium sized flooding incidents, or those between .0024 MG and .24
MG. There were only two manholes from the 5-year analysis that flooded more
than the largest single incident from the 2-year storm. In general, the areas with
the most flooding from the 5-year modeled storm match the observed areas
pretty well. There are a few areas where the model predicts large amounts of
flooding but the February 25, 2005 storm does not and vice versa. The model
shows large flooding in the northwest reach of the 10th Street zone but the
actual response is minimal. Actual responses are heavy in the North Point, Down
Town, and So Market zones, but the model shows very little response there.
These issues are important and can not be clarified with the model at this time.
However, a more detailed model will be built during the next phase of
4-8
4.0 BASELINE ANALYSIS
the project and should be able to provide better insight into the
capacity and flooding issues.
The baseline model was used to simulate the 5-year typical period. The results
of the 5-year typical period simulation were evaluated to determine the
frequency of CSD activations during the simulation. Using a similar method as
used during the model verification stage, as described in Section 3.5, the
occurrence of CSDs were compiled for the various locations and outfalls
throughout the city. CSDs which occurred within 6 hours of another overflow
were considered the same discharge event; dicharges separated by greater than
6 hours were considered separate events. To determine the effect of
alternatives on CSD activation frequency, the results from this baseline analysis
will be compared to the results from the analysis of system alternative
simulations. Table 4-3 includes a summary of the reported versus modeled
overflows for the 5-year typical period. Refer to Appendix H for a list of modeled
versus reported overflows for the individual outfalls.
The following table shows the modeled average and peak DWF an d WWF into
the City’s major treatment facilities for baseline conditions. The model
overpredicts flow at SEP by 40% for a typical dry day.
4-9
5.0 Conclusions
A detailed dynamic model has been developed and calibrated for the San Francisco
Public Utilities Commission. The steps taken to build, calibrate, and validate the model
have been discussed. Initial calibration of the model during both dry and wet weather
conditions was performed using a series of flow meters deployed across the system for a
5-month month period in 2004. The initial calibration was successful and will serve as
the foundation for more refinements later.
The capacity analysis was performed by applying the calibrated hydraulic model with
added CIP improvements to compute flows and levels in each modeled pipe under 2-
year and 5-year design storm conditions. The results of these model runs indicate
where additional capacity will be required to continue to provide the desired level of
protection against flooding and overflows.
Following initial calibration and baseline validation, the hydraulic model is being used to
evaluate the impacts to the collection system (both positive and negative) resulting from
the alternatives developed by the Master Plan team.
5-1
APPENDIX B
Appendix A
FLOW MONITORING LOCATIONS
12
APPENDIX B
Appendix B
FLOW METER CALIBRATION RESULTS
13
APPENDIX B
In the following summary tables, the DWF calibration period was from December
15, 2004 through December 22, 2004. The WWF calibration period was from
December 26, 2004 through January 14, 2005. If a meter met calibration criteria
it is marked YES, and if it did not, it is marked NO. If predicted results were
within +/- 15% of observed results at a given meter, it was successfully
calibrated.
The above criteria should be met for the verification period, unless circumstances
at the monitoring locations (1) cannot be modeled and are determined to be
unimportant, (2) are not detrimental to the model, or (3) are due to infiltration
and can be accounted for in subsequent use of the model. Currently, USEPA’s
Combined Sewer Overflow: Guidance for Monitoring and Modeling provides only
vague non-numerical criteria for calibration assessment.
14
TABLE B-1 CALIBRATION SUMMARY
5-1
APPENDIX B
40G3 24127.1 NO NO NO NO NO YES 40G1, 40G3 and 40G4: these three meters are in the same vicinity. 40G3 and 40G4 are upstream of 40G1 and installed in parallel pipes. 40G1 collects flows from
40G3, 40G4, and 40E1. Both 40G1 and 40E1 were fairly calibrated. The flow balancing at 40G3 and 40G4 very likely contributes to the discrepancies at these
locations.
40G4 24128.2 YES YES YES YES NO NO 40G1, 40G3 and 40G4: these three meters are in the same vicinity. 40G3 and 40G4 are upstream of 40G1 and installed in parallel pipes. 40G1 collects flows from
40G3, 40G4, and 40E1. Both 40G1 and 40E1 were fairly calibrated. The flow balancing at 40G3 and 40G4 very likely contributes to the discrepancies at these
locations.
40H1 21317.1 YES YES YES YES YES YES
40J1 21242.1 NO NO YES NO YES YES
40J2 18764.1 NO NO YES YES YES NO
40J3 85357.1 NO NO NO YES YES NO
40J4 17548.1 YES NO YES YES YES YES
40K1 86079.2 NO NO NO NO NO NO 40K1: This meter was installed in the West Side box. Most of the time, the meter was not functional due to the debris/sand coming from upstream.
40K2 300069.1 YES NO NO YES YES NO
40M1 100006.1 YES YES NO YES NO NO
40M2 24055.1 NO NO NO YES YES YES
Basin 50
50A1 28428.1 NO NO NO YES NO NO Depth is less than 12 feet so velocity and depth are not accurate. Minimal flow – upstream meter <0.67 MGD
50AK1 86277.1 YES NO NO YES NO NO Depth is less than 12 feet so velocity and depth are not accurate
16
APPENDIX B
* SOMA meters – use DWF and WWF calibration period from March 10 through March 20, 2005.
** Issue with meter 60B1 and 60E1. 60B1 installed upstream of weir and dry weather flow dropout that was found to be plugged. ADS reported a range of sediment depth at any one time of 0“, 2.5”, 10” and 22.5”. ADS also reported that results show 24” of standing water even during
periods of low flows. Downstream meter 60E1 recorded less flow than upstream meter 60B1. ADS corrected meter 60E1 by adding 0.66” to depth results. Due to the weir condition and since meter 60B1 is located mid reach of pipe ID 29264.1 with a relatively step slope, 1.01 feet was
added to ADS depth data so that depths could be compared at the downstream end or pipe ID 29264.1. DWF calibration of catchments upstream of 60B1 was based on data from 60E1 due to 60B1 metering issues. WWF calibration of 60B1 was only for depth due to meter location,
chunette, irregular section and possible unknown amount of sediment. Based on 60E1 dry weather flow calibration parameters values seem low suggesting unreliable data.
17
5-1
APPENDIX B
Appendix C
CALIBRATION PLOTS OF DCS DATA
20
5-1
Appendix D
VALIDATION RESULTS
22
Model Predicted CSD Activations Compared to Reported CSD
Activations
for October 19, 2004- April 3, 2005
23
24
25
Appendix E
CAPITAL IMPROVEMENTS PROJECTS
26
Baseline Capital Improve Projects (CIPs)
1.0 Oceanside
1.1 Constructed
27
1.2 Under Construction
2.0 Bayside
2.1 Constructed
28
o Replace existing 2’x3’ sewer on Third Street from Gilman to Salinas
Avenues.
29
The above list was reviewed against available information to ensure that the
correct and complete set of CIP projects was implemented. Changes to the
model were identified by using the network compare function in Infoworks, and
were compared against the above list.
30
31
Appendix F
Rainfall Analysis Technical Memorandum
32
Insert Sharon’s Rainfall Tech memo.
33
Appendix G
SFHHDAR Network
34
Dates of Storm Events During Which the SFHHDAR Network was Down
Duration
Start of Event (hours)
1/1/1977 14:00 44
5/3/1977 11:00 2
8/24/1977 7:00 3
11/20/1973 0:00 3
11/20/1977 18:00 30
2/4/1978 20:00 62
9/9/1978 11:00 8
2/18/1979 6:00 21
7/21/1979 8:00 2
5/9/1980 9:00 8
3/18/1981 8:00 70
3/25/1981 3:00 20
4/1/1981 0:00 2
4/18/1981 16:00 12
4/25/1981 14:00 2
5/17/1981 22:00 11
10/27/1981 8:00 40
11/21/1981 7:00 14
1/27/1982 23:00 14
12/30/1983 2:00 12
12/4/1985 20:00 8
35
24 1.12
25 1
26 1.18
27 1.16
28 1.1
29 1.12
41 1.17
42 0.94
43 0.95
44 0.95
46 1.03
47 1.05
36
37
Appendix H
Baseline Results
38
Modeled CSD Activation for the 5 Year Typical Period
39