EPA/600/R-20/165 | May 2020 | www.epa.gov
United States
Environmental Protection
Agency
XV EPA
National Water Infrastructure
Adaptation Assessment
Part II, Smart Urban Design (SUD) and
Application Case Studies
Office of Research and Development
Center for Environmental Solutions and Emergency Response
-------
EPA/600/R-20/165
May 2020
(Intentionally left blank)
-------
National Water Infrastructure Adaptation
Assessment: Part II, Smart Urban Design (SUD) and
Application in Case Studies
Y. Jeffrey Yang1, Heng Wei2, Xinhao Wang3, Steven Buchberger2, Marissa S. Liang4
Ni-bin Chang5, Britta Bierwagen6, Susan Julius6, Zhiwei Li7, Dominic L. Boccelli2,
Robert M. Clark8, Hou Liu2, and Jill Neal1
(1) U.S. EPA, ORD, Center for Environmental Solutions and Emergency Response, Water
Infrastructure Division, Cincinnati, OH 45268
(2) University of Cincinnati, College of Engineering and Applied Science, Cincinnati, OH
45221
(3) University of Cincinnati, School of Planning, Cincinnati, OH 45221
(4) U.S. EPA, ORD, Center for Public Health and Environmental Assessment, ORISE
participant, Cincinnati, OH 45268
(5) University of Central Florida, Department of Civil, Environmental, and Construction
Engineering, Orlando, FL 32816
(6) U.S. EPA, ORD, Center for Public Health and Environmental Assessment,
Washington, DC 20460
(7) Carbon Sequestration Technologies, LLC., Pittsburgh, PA 15241
(8) Environmental and Health Consultant, Cincinnati, OH 45268
-------
DISCLAIMER
The U.S. Environmental Protection Agency, through its Office of Research and
Development, conducted, funded, and managed the research described herein. The report,
National Water Infrastructure Adaptation Assessment: Part II, Smart Urban Design (SUD) and
Application Case Studies, EPA/600/R-20/165, has been subjected to the Agency's peer and
administrative review and has been approved for external publication. Any opinions expressed in
this paper are those of the authors and do not necessarily reflect the views of the Agency;
therefore, no official endorsement should be inferred. Any mention of trade names or
commercial products does not constitute endorsement or recommendation for use.
1
-------
FOREWORD
The U.S. Environmental Protection Agency (EPA) is charged by Congress with
protecting the Nation's land, air, and water resources. Under a mandate of national environmental
laws, the Agency strives to formulate and implement actions leading to a compatible balance
between human activities and the ability of natural systems to support and nurture life. To meet
this mandate, EPA's research program is providing data and technical support for solving
environmental problems today and building a science knowledge base necessary to manage our
ecological resources wisely, understand how pollutants affect our health, and prevent or reduce
environmental risks in the future.
The Center for Environmental Solutions and Emergency Response (CESER) within the
Office of Research and Development (ORD) conducts applied, stakeholder-driven research and
provides responsive technical support to help solve the Nation's environmental challenges. The
Center's research focuses on innovative approaches to address environmental challenges
associated with the built environment. We develop technologies and decision-support tools to
help safeguard public water systems and groundwater, guide sustainable materials management,
remediate sites from traditional contamination sources and emerging environmental stressors,
and address potential threats from terrorism and natural disasters. CESER collaborates with both
public and private sector partners to foster technologies that improve the effectiveness and
reduce the cost of compliance, while anticipating emerging problems. We provide technical
support to EPA regions and programs, states, tribal nations, and federal partners, and serve as the
interagency liaison for EPA in homeland security research and technology. The Center is a
leader in providing scientific solutions to protect human health and the environment.
This publication has been produced as part of the EPA ORD's Air-Energy (A-E) research
program. The research report is published and made available by ORD to assist the user
community and to link researchers with their clients.
Gregory Sayles, Ph.D.
Director, Center for Environmental Solutions and
Emergency Response
11
-------
PREFACE
Water is essential to life. Uneven distribution of population and water resources in the
world results in more than 1.1 billion people with a lack of access to clean drinking water and 2.6
billion people deprived of adequate water sanitation. Today fresh water is being consumed at an
alarming rate, almost doubling every 20 years. Global changes further exacerbate this already
stressed situation. It can be said that water availability is not only a problem for developing
countries, but one facing developed nations that are saddled with aging water infrastructure.
Pressed by challenges, however, civilizations always have found innovative solutions to meet
water resource needs and adapt to evolving social and environmental conditions. This spirit of
adaptation continues to date and will continue.
Today, one of the most complex challenges facing our nation revolves around water
supply sustainability, often framed in the name of water-environment-energy nexus. The
challenge is acute considering occurring and future changes in land use and hydroclimatic
conditions and, thus, requires a holistic water management approach. For the purpose,
interdisciplinary research and development to supplement and improve water management and
engineering practice are often the first steps of the effort.
The national adaptation assessment reports synthesize the results of multidisciplinary
research and development in the past 8 years. This report presents an assessment of our nation's
water resource infrastructure, characterizes hydroclimatic provinces, and future hydroclimatic
and land use conditions. It further introduces planning and engineering means to develop the
quantitative scientific basis for adapting water infrastructure and, in general, for urban
development. The systematic adaptation approach is structured at multiple levels from integrated
watershed management to urban-scale planning and individual water system engineering.
In considering water infrastructure adaptation needs, a suite of tools ranging from those
in master planning and systems engineering to those in watershed modeling and drinking water
plant simulations has been developed or adopted. These adaptation techniques for different levels
of planning and engineering are described in this report and accompanying publications and are
illustrated by case studies. The focus is to develop actionable science and engineering bases for
adapting to the likely future environmental stressors at local scales and, by doing so, to support
water resource managers and technical stakeholders who face the technical complexity. Although
this report provides a wealth of technical data and information, it only marks the beginning of a
long march toward the goal of the sustainable water resource and resilient infrastructure in a time
of accelerating global changes.
in
-------
ACKNOWLEDGMENTS
The U.S. Environmental Protection Agency (EPA), through its Office of Research and
Development (ORD), funded, managed, and conducted the intramural and extramural research
described herein. The research was a part of the ORD Air and Energy (A-E) research program. It
was implemented by the EPA Water Resources Adaptation Program in ORD's Water Systems
Division by Pegasus Technical Services, Inc., through EPA Contract EP-C-05-056, and Cadmus,
Inc., through EPA Contract EP-C-06-100.
Programmatic guidance from ORD's former Aging Water Infrastructure Program is
acknowledged. Special thanks are due to Jeff Peterson, Karen Metchis, Elizabeth Corr, Robert
Cantilli (now with ORD), Rachael Novak, Elana Goldstein, and Curt Baranowski of the EPA
Office of Water for their efforts to bring together experts and practitioners from around the
country and for their involvement in this research. Additionally, the EPA Office of Air and
Radiation Climate Division and the Office of Transportation Air Quality were engaged with
interests in this research through the A-E research program.
The project team would like to acknowledge numerous technical staff and participants
from EPA and contracting research organizations. This investigation of both a wide breadth and
a substantial depth was accomplished only with their participation and contribution, including
the administrative and contract support from Dr. Michael Moeykens, Michelle Latham, Steve
Harmon, and Stephen Wright. Finally, technical data and collaboration efforts by the Manatee
County [Florida] Water Department, the Greater Cincinnati Water Works, the Las Vegas Valley
Water District, and local governments on the Massachusetts coast are acknowledged.
The national adaptation assessment reports were a result of continuing research efforts
over the past several years. These reports initially were prepared in 2011 and reviewed by
individuals inside and outside of EPA. Per review comments, additional technical contents were
added with new research, especially in the areas of adaptation tools and methods. This
development led to rewriting and reorganization of the entire reports. In the process, three rounds
of internal and external peer reviews were conducted. After these peer reviews, Part I of the
national water infrastructure assessment report was published in 2015. This current Part II report
also has been subjected to administrative review and is now approved for publication.
The EPA project team includes Y. Jeffrey Yang (ORD/ Center for Environmental
Solutions and Emergency Response [CESER]), Jill Neal (CESER), Chelsea Neal (former
CESER ORISE participant, now with Department of Energy's Sandia National Laboratories),
Marissa Liang (ORD/Center for Public Health and Environmental Assessment [CPHEA] ORISE
participant), Britta Bierwagen (CPHEA), Susan Julius (CPHEA), and Jeri Weiss (EPA Region
1), among many others involved in the past several years. Principal research participants from
contracting organizations are listed below.
University of Cincinnati:
iv
-------
Dr. Steven Buchberger, P.E.; Dr. Zhiwei Li (now with Carbon Capture Scientific,
LLC.); Dr. Heng Wei; Dr. Xinhao Wang; the late Dr. Timothy C. Keener, P.E.;
Dr. Dominic L. Boccelli (now at the University of Arizona); Dr. Hou Liu; Dr. Zhuo
Yao (now at the State of California Air Resources Board); Ting Zuo; Dr. Yu Sun
(now at the University of North Georgia); Amy Burguess; Heng Yang; Jie He; Patcha
Huntra; Dr. Pamela Heckle, P.E. (now at American Red Cross); Dr. Thushara
Ranatunga (now at the Houston-Galveston Area Council, Texas)
Consultants:
Dr. Robert M. Clark, Dr. Walter Grayman, P.E.
University of Central Florida:
Dr. Ni-Bin Chang, P.E.; Dr. Ammarin Makkeasorn; Dr. Sanez Imen; Lee Mullon
Pegasus Technical Services:
Dr. Karen Koran
Cadmus, Inc.:
Chi Ho Sham (now at Eastern Research Group), Jaime Rooke, Brent Ranalli,
Laurie Potter, Laura Blake, Julie Blue, Donna Jensen, Patricia Hertzler
The following reviewers are acknowledged for peer reviews of this report and its
previous versions.
Dr. Weiwei Mo, University of New Hampshire
Dr. Frederick Bloetscher, Florida Atlantic University
Dr. James Goodrich, ORD/National Homeland Security Research Center, EPA
Karen Kleier Schrantz, NRMRL
Dr. Mark Rodgers, NRMRL
Dr. Vahid Alavian, World Bank
Mr. Jeff Adams, ORD
Dr. Nancy Beller-Simms, National Oceanic and Atmospheric Administration
Dr. E.P.H. Best, ORD
Dr. Pratim Biswas, P.E, Washington University
Dr. Levi Brekke, Bureau of Reclamation
Mr. Mao Fang, P.E., Las Vegas Valley Water District
Mr. Gary Hudiburgh, EPA Office of Water, American Indian Environmental Office
Dr. Timothy C. Keener, P.E., University of Cincinnati
Dr. Paul Kirshen, University of New Hampshire
Dr. Julie King, U.S. Geological Survey
Dr. Thomas Johnson, ORD
Mr. Craig Patterson, P.E., ORD
Dr. Joo-Youp Lee, University of Cincinnati
Dr. Steven McCutcheon, P.E., D.WRE, ORD
Mr. Ken Moraff, EPA Region 1
Ms. Angela Restivo, EPA Region 6
Dr. Neil Stiber, EPA Office of Science Advisor
v
-------
Mr. Michael J. Wallis, East Bay Municipal Utility District, CA
Dr. Xinhao Wang, University of Cincinnati
Cadmus, Inc.: Glen Boyd, Dr. Rudd Coffey, Laura Dufresne, Charles A. Hernick,
Ken Klewicki, Dr. Jonathan Koplos, William M. Jones, Frank Letkiewicz,
Richard Kro, Vanessa M. Leiby, Jeff Maxted, G. Tracy Mehan III, Tom
Mulcahy, Karen Sklenar, Mary Ellen Tuccillo
Dr. Audrey Levine, formally at the National Science Foundation
-------
ABSTRACT
This report, National Water Infrastructure Adaptation Assessment: Part II, Smart Urban
Design (SUD) and Case Studies, is a part of the research effort led by the EPA Office of
Research and Development (ORD) Water Systems Division. The multiyear research, organized
by ORD's Air and Energy (A-E) research program, has generated data, models, and methods to
assess the water infrastructure vulnerability and develop sustainable planning and designs for
urban infrastructure. The research results are summarized in separate EPA reports.
The first report, published in 2015, contains a preliminary regulatory and technical
analysis of the U.S. water infrastructure and its relationship to hydroclimatic and socioeconomic
changes. This second report presents SUD tools and methods for urban planning and
infrastructure adaptation design. The report and its content aim to assist water practitioners and
urban planners in developing resilient water supply systems and water management programs.
The tools and methods also help users with an understanding of the interconnectedness among
urban growth, transportation, and water systems. In sequence, the report first outlines adaptation
objectives and the SUD framework. Next, it evaluates the unique environmental properties
associated with urban growth and describes current planning practices that facilitate such growth.
In Sections 3.0 through 7.0, the core SUD components in urban planning and water system
engineering are described with case studies for illustration. In Section 8.0, the SUD applications
in coastal areas are presented to illustrate adaptation consideration and approaches against
complex and interconnected hydroclimatic impacts.
Water infrastructure adaptation may take place at three different levels: (1) urban-wide
planning and adaptation, (2) system-scale water master planning, and (3) local-scale adaptive
engineering and design of infrastructure components. The urban-wide adaptive planning relies on
integrated analysis and scenario-based simulation of future land use, socioeconomics, and
transportation and water infrastructure. The aim is to improve urban efficiency and achieve
adaptation co-benefits in economics and systems resilience. At the system scale, SUD adaptation
tools enable systems evaluation (e.g., trade-off analysis) for master planning options. At the local
scale, the newly developed SUD tools and methods enable users to model, evaluate, and
optimize water treatment, distribution, storage, and energy conservation. Together with
monitoring and forecasting techniques to be described in other assessment documents, these
SUD methods and tools form a suite of adaptation techniques designed for water infrastructure
planning and system improvement.
The SUD adaptation methods were examined in case studies in the U.S. inland and
coastal regions. These applications include studies of urban infrastructure and water systems in
Cincinnati, Ohio; Manatee County, Florida; Las Vegas Valley Water District, Nevada; and
Mattapoisett, Massachusetts. They are presented to highlight the adaptation considerations and
the use of SUD tools and methods.
vii
-------
Table of Contents
DISCLAIMER i
FOREWORD ii
PREFACE iii
ACKNOWLEDGMENTS iv
ABSTRACT vii
Definitions and Abbreviations xviii
Part Two: Smart Urban Design (SUD) and Application Case Studies 1
1. Sustainable Development of Urban Water Systems 1
1.1. Adaptation considerations 2
1.1.1. Defining adaptation obj ective 3
1.1.2. Understanding urban adaptation constraints 6
1.1.3. Revi sing or redefining pi anning and engineering focus 10
1.1.4. Selecting adaptation evaluation matrix 13
1.2. Three levels of water infrastructure adaptation 15
1.3. Smart Urban Design (SUD) for systems analysis 15
1.3.1 Integrated Watershed Modeling (IWM) 15
1.3.2 Adaptive Urban Planning and Engineering Tool (AUP&ET) 20
1.3.3 SmartWater for water supply 22
1.3.4. Source-to-tap water supply in a systems approach 24
2. Adaptive Urban Planning in Urban Scales 24
2.1 Physical infrastructure and urban forms in current practice 25
2.1.1. Land use encouraging urban sprawl 26
2.1.2. Transportation and energy performance 28
2.1.3. Water planning and engineering 29
2.2 Transformation toward smart growth 30
2.3. Monitoring and re-evaluation 31
3. SUD Methods and Tool in Adaptive Urban Planning 32
3.1 AUP&ET principles and utilities 33
3.1.1. Land use projection - CA-Markov model and ICLUS 33
3.1.2. Calibration and validation of the land use simulation model 35
3.1.3. AIR-SUSTAIN system for transportation simulation 36
3.1.4. The linkage to water infrastructure simulations 38
3.2 The AIR-SUSTAIN simulation tool for transportation 38
3.2.1. Basic functions and interfaces of AIR-SUSTAIN 39
viii
-------
3.2.2. Travel demand forecasting - VISUM 41
3.2.3. Assistance in traffic congestion identification 42
3.2.4. Microscopic simulation using VISSIM 43
3.2.5. Emission estimation using MOVES 43
4. Adaptive Urban Planning in SUD Case Studies 45
4.1 Urban form and urban infrastructure 46
4.1.1. Urban form and land use patterns 46
4.1.2. Transportation and traffic distribution 48
4.1.3. Urban form and air quality 50
4.1.4. Thermal inversion and mixing height 53
4.1.5. Urban and exurban differences 54
4.1.6. Urban form effects on urban heat island and air quality 55
4.2. Adaptive urban planning modeling and analysis in Cincinnati 64
4.2.1 Three development scenarios 64
4.2.2 Transportation and emission analysis using AIR-SUSTAIN tool 65
4.3 Adaptation analysis for water master planning in Manatee County, Florida 75
4.3.1 Water supply assessment 76
4.3.2 Expansion scenario analysis 82
4.3.3. Quantitative modeling and systems analysis 85
4.3.4. Adaptation analysis results on cost and carbon/energy footprint 89
5. System-Scale Adaptation for Existing Urban Water Infrastructure 93
5.1. Basic considerations in adaptation engineering 95
5.1.1. Adaptation engineering for water infrastructure 95
5.1.2. Adaptation attributes of three types of water infrastructure 97
5.1.3. The capacity reserve concept and climate resilience 99
5.1.4. CR and engineering practice 101
5.2. Water infrastructure capacity reserve and resilience 102
5.2.1. Stormwater infrastructure functions and design tolerances 102
5.2.2. Drinking water infrastructure functions and design tolerances 107
5.2.3. Wastewater infrastructure functions and design tolerances Ill
5.3. Water infrastructure vulnerability analysis for adaptation 115
5.3.1. The resilience assessment and two approaches 115
5.3.2. Water resilience evaluation and resilience tool - CREAT 117
5.3.3. From vulnerability analysis to adaptation engineering 120
6. SUD Methods and Tools for Drinking Water Treatment 120
6.1. Principle, models and algorithms in WTP-cam 121
6.1.1. Conventional treatment unit processes 121
ix
-------
6.1.2. Customization of GAC unit process 122
6.1.3. Adaptation cost and economics 124
6.1.4. Other unit processes for adaptation 126
6.2. Adaptation Analysis using WTP-cam 126
6.2.1. Treatment process and compliance targets 126
6.2.2. Monte Carlo methods in modeling source water quality 128
6.2.3. Advanced unit process and adaptation cost 132
6.3. GCWW Richard Miller Treatment Plant case study using WTP-cam 133
6.3.1. Miller water treatment plant operation and performance 134
6.3.2. WTP-cam simulation of hydroclimatic change impacts 137
6.3.3. Engineering analysis for water treatment adaptation 143
7. Adaptation Engineering for Drinking Water Distribution 146
7.1. Water age and water quality changes in distribution: The need for adaptation 147
7.1.1. EPANET-based risk assessment on DBP formation 147
7.1.2. Water age variations, modeling and adaptive control 150
7.2. In-network water treatment as adaptation measure 152
7.3. Water conservation, storage and reuse through adaptive planning 155
8. SUD Applications in Coastal Regions: Water Infrastructure and Emergency Planning ..155
8.1. Water infrastructure vulnerability in coastal regions 155
8.2. Wastewater vulnerability and adaptation in storm surge 159
8.3. Emergency evacuation and water supplies 160
9. Summary and Recommendations 162
10. References 165
Appendix A AIR-SUSTAIN Program Input and Output Structures 184
Appendix B Water Treatment Plant - Climate Adaptation Model (WTP-cam) User's Manual
248
x
-------
List of Tables
Table 2-1 Adaptation attributes for common objectives 14
Table 2-2 General advantage and challenges of three-level adaptation actions 16
Table 2-3 Selected urban functions impacted by hydroclimatic conditions 25
Table 2-4 Four daily periods of traffic compositions on the highway in Cincinnati, OH 49
Table 2-5 Locations and traffic flow in 2009 for selected locations in the Cincinnati road
network 52
Table 2-6 Statistics of daily temperature measurements at NQAAS Station 17-061-00040....
53
Table 2-7 Temperature differences between the reference station and other stations
abstracted from the >10-year daily temperature measurements 59
Table 2-8 Trip generation results in the number of trips per day by the target year 2030.... 72
Table 2-9 Trip distribution results for the number of trips per day originated from and
attracted to centers 72
Table 2-10 Average queue length, average wait time, total delay, and average delay during
morning peak hours (7:00 -9:00 am) 73
Table 2-11 Water demand in 2006 and projections for wholesale customers in annual
average 78
Table 2-12 Water demand projections for retail and significant users in annual average 78
Table 2-13 Twenty alternatives for water supply expansion in the county master planning.. 79
Table 2-14 Maximum water credit and unit cost of the twenty water supply alternatives 82
Table 2-15 Life cycle analysis of carbon footprint for the twenty water supply alternatives 84
Table 2-16 Optimal solutions of the multi-objective model 90
Table 2-17 The Pareto optimal expansion strategies (n = 1) 90
Table 2-18 The Pareto optimal expansion strategies for the best and worst cases (n = 1) 91
Table 2-19 Water infrastructure design and engineering domains, and their attributes 103
Table 2-20 Important engineering attributes for stormwater infrastructure adaptation 105
Table 2-21 Important engineering attributes and likely vulnerability in drinking water
treatment and distribution systems for community water supplies 109
Table 2-22 Important engineering attributes and potential vulnerability of wastewater
infrastructure 112
Table 2-23 Types and approaches of eight water utilities in climate vulnerability assessment
116
Table 2-24 GAC contactor cost estimate parameters 124
XI
-------
Table 2-25 GAC reactivation cost 125
Table 2-26 Illustration of calculating running annual average for finished water TOC 128
Table 2-27 Options for Monte Carlo analysis 131
Table 2-28 Miller WTP unit process design parameters 134
Table 2-29 Inflow and chemical feed levels for the Miller WTP 135
Table 2-30 Statistics of full-scale field measurements 135
Table 2-31 Source water inputs for the Miller water treatment plant in 1998 (Baseline) 138
Table 2-32 Correlation matrix for source water quality parameters (for Ohio River from July
1997 to December 1998) 140
Table 2-33 Projected raw water quality parameters for the Miller WTP in 2050 141
Table 2-34 Comparison of sampled and modeled water quality results 142
Table 2-35 Selected hydrological impacts and adaptation variables in coastal area 157
Table 2-36 SLOSH modeling parameters for storm surge modeling at Mattapoisett, MA.. 159
Table 2-37 Population affected for evacuation under four categories of hurricane 161
xii
-------
List of Figures
Figure 2-1 Schematic process diagram of iterative monitoring-adaptation framework for
water infrastructure adaptation 3
Figure 2-2 Typical spatial relationships of water infrastructure in an urban environment with
illustration of infrastructure adaptation scales and general process 4
Figure 2-3 Schematic diagram showing three major types of urban forms and their typical
properties 7
Figure 2-4 General process of the current urban master planning and its relations to
transportation and water infrastructure engineering 8
Figure 2-5 Three major types of urban sprawl expanding the urban footprints into exurban
areas 10
Figure 2-6 General process of adaptive urban planning and engineering 12
Figure 2-7 Schematic diagram of the Smart Urban Design (SUD) structure for scenario-based
urban development planning and engineering 17
Figure 2-8a Process flow diagram of the Integrated Water Management (IWM) modeling for
watershed simulations 18
Figure 2-8b Process flow diagram of the Integrated Water Management (IWM) modeling for
urban catchment using EPA's National Stormwater Calculator (U.S. EPA, 2014).
19
Figure 2-9 Process flow diagram of the scenario-based Adaptive Urban Planning and
Engineering Tool (AUP&ET) for urban planning and engineering 21
Figure 2-10 Schematic diagram of conceptual modeling framework for WTP3.0 as a major
SUD element 22
Figure 2-11 Schematic diagram of water supply and major system variables 23
Figure 2-12 Urban expansion and urban form transformation for Atlanta (upper) and Phoenix
(lower) metropolitan regions between 1970 and 1992 27
Figure 2-13 Transportation efficiency (annual delay, travel index, excess fuel use, and annual
cost) in year 2007 as a function of urban population in the U.S. urban centers... 28
Figure 2-14 Simulation block diagram for CA-Markov based urban land projections 34
Figure 2-15 AIR-SUSTAIN modeling framework for transportation analysis of efficiency and
carbon dioxide emission in urban infrastructure adaptation 37
Figure 2-16 AIR-SUSTAIN graphic interface for scenario modeling and analysis 39
Figure 2-17 The Results Comparison module interface in the AIR-SUSTAIN tool 40
Figure 2-18 General relationship between model precision and data requirements in traffic
modeling 44
Xlll
-------
Figure 2-19 Modeling framework for emission estimation using both the macroscale VISUM
and microscopic VISSUM traffic simulation models 45
Figure 2-20 Major transportation traffic routes and the urban physical footprints of the
Cincinnati metropolitan region 47
Figure 2-21 Different land use patterns in areas among the 12 EPA's NAAQS monitoring
stations 48
Figure 2-22 A schematic diagram of three-dimensional model for the urban form, traffic and
atmospheric structure in the Cincinnati metropolitan region 49
Figure 2-23 Truck and passenger car traffic volume distribution in the Cincinnati metropolitan
region 51
Figure 2-24 Representative temperature profiles showing the boundary inversion and capping
inversion 54
Figure 2-25 Temporal Lr and HU1V variations showing diurnal thermal inversion in the urban
boundary layer in October 2011 55
Figure 2-26 Tmin, AT, and PM2 5 variations with time at NAAQS monitoring station 061-0040.
56
Figure 2-27 Temporal change of ambient temperature Tmin and AT at station 061-0040 in the
Cincinnati urban core 57
Figure 2-28 Tmin at 17-061-0040 station is linearly correlated with those of other stations in the
year 2005 measurements 58
Figure 2-29 Spatial variations of temperature difference for mean and maximum Tavg and
PM2.5 in cross section O-O' 60
Figure 2-30 Co-variation of minutely average wind speed at the 1-75 site with the lapse rate in
the boundary layer during the roadside black carbon dispersion experiments .... 61
Figure 2-31 Schematic diagram showing major types of microclimate conditions in the surface
roughness layer (SRL) equivalent to the urban canopy layer (UCL) 63
Figure 2-32 The base-year distribution maps for the Cincinnati metropolitan area in 2009: A)
population; B) household; and C) employment 65
Figure 2-33 Three development scenarios for the Cincinnati metropolitan area in the target
year 2030 66
Figure 2-34 Setup of a new scenario in AIR-SUSTAIN 67
Figure 2-35 Program interface for A) importing the Base Year data; B) assigning population
change; and C) assigning employment changes at TAZ levels 67
Figure 2-36 Simulation module of demographic analysis for a development scenario 68
Figure 2-37 Transportation congestion areas identified for typical peak-hour traffic for 2009 in
Hamilton County, showing concentrations along 1-71,1-75,1-275N, and Ronald
Reagan Highway 69
xiv
-------
Figure 2-38 Population changes by 2030 for the three developmental scenarios (SI, S2, and S3
in the inserts) in comparison to the distribution of base year 2010 (background) ...
70
Figure 2-39 Changes in number of household by 2030 for the three developmental scenarios
(SI, S2, and S3 in the inserts) in comparison to the distribution of base year 2010
(background) 71
Figure 2-40 Employment changes by 2030 for the three developmental scenarios (SI, S2, and
S3 in the inserts) in comparison to the distribution of base year 2010
(background) 71
Figure 2-41 Simulated peak hour (7:00-9:00 am) traffic volume distribution over the
Cincinnati road network for the base year (2010) and under three development
scenarios in the target year (2030) 73
Figure 2-42 Comparison of three development scenarios (SI, S2, and S3) in peak hour (7:00-
9:00) vehicular CO2 emission and energy (fuel) consumption 74
Figure 2-43 Location of the Manatee County water supply system along the upper Manatee
River in Florida 76
Figure 2-44 Locations of WTP, ASR, well fields (ECWF-1, MPWF), and the twenty potential
water supply alternatives A1-A20 77
Figure 2-45 The life-cycle system analysis flow diagram for determining carbon footprint in
water infrastructure expansion alternatives 83
Figure 2-46 Pareto solution fronts for the best compromised solutions to meet the projected
future water demand (base case) and the demand with 10% uncertainties (best
case and worst case) 92
Figure 2-47 Suggested optimal facility expansion strategies in each of five-year periods based
on the optimization modeling of water infrastructure expansion options for
Manatee County, Florida 94
Figure 2-48 Assessment-adaptation process for water infrastructure planning and engineering.
96
Figure 2-49 Process schematic diagrams for typical centralized drinking water, wastewater,
and stormwater infrastructure in an urban watershed 98
Figure 2-50 Four types of infrastructure vulnerability under the threat of external impact event
(e.g., storm surge) 99
Figure 2-51 Maximum percentage increase (AQ%) in hydraulic capacity of stormwater
conveyance using commercial concrete pipes of discrete nominal diameters (D)
106
Figure 2-52 BOD removal efficiency of a wastewater activated aeration tank as a function of
flow, BOD mass loading, and aeration capacity 113
xv
-------
Figure 2-53 Relative magnitude of infrastructure CR installed in current engineering practice
(left) in comparison with the relative precipitation change (solid bar) and
uncertainty (pattern and solid line with whisker) by 2060 114
Figure 2-54 The process of climate vulnerability analysis using the EPA tool CREAT 118
Figure 2-55 Unit process, inputs and outputs in model simulation of WTP2.0 adopted in WTP-
cam program 122
Figure 2-56 Cost curve for annual cost of GAC unit, indicating the cost associated with a
given GAC reactivation period for the Miller WTP in Cincinnati 125
Figure 2-57 Schematic diagram for (A) treatment unit process at the GCWW Miller water
treatment plant, and (B) WTP-cam program flow in the example simulation.... 126
Figure 2-58 Original input data for the example process train at the Miller WTP 127
Figure 2-59 Program logic sequences in Monte Carlo simulation of future source water quality
variations using the correlation matrix method 129
Figure 2-60 Graphic user interface for inputs in Monte Carlo simulations of future water
qualities 130
Figure 2-61 Manual input window for influent water quality statistics 132
Figure 2-62 Temporal variations of influent and blended effluent TOC in the GAC unit .... 136
Figure 2-63 Temporal variations of inflow, mass inflow and active number of GAC contactors
136
Figure 2-64 Temporal variations of active contactors and blended effluent TOC concentration
137
Figure 2-65 Time series of plant inflow and EBCT variations 137
Figure 2-66 Normal probability plots for source water pH (107 samples) and TOC (93
samples) for Ohio River from the ICR database (July 1997-December 1998)... 138
Figure 2-67 Modeled treatment performance of the Miller WTP in baseline (1998) and future
(2050) scenarios 144
Figure 2-68 Accumulative probability of net annual adaptation cost for the source water
change scenario in year 2050 145
Figure 2-69 Schematic diagram showing the simultaneously occurring chlorine reactions in
bulk and wall demands, and mass exchange between the bulk water and pipe wall.
149
Figure 2-70 Water demand and computed Re variations in a two-week period for a single
home, 31 and 114 homes of a pipe dead-end section, showing significant
differences between the APAD model and the generalized water demand pattern.
151
Figure 2-71 Probability distribution and corresponding CDF of simulated water age for the
network 152
xvi
-------
Figure 2-72 Schematic views of in-network aeration in the LVVWD water distribution
network to remove volatile THM from drinking water in the alpha tank reservoir
153
Figure 2-73 EPANET simulation of flow and THM distribution in the Western Hill portion of
the LVVWD water distribution system 154
Figure 2-74 Schematic illustration of long-term climate and short-term meteorological and
disruptive storm surge events in atypical coastal zone 156
Figure 2-75 Schematic diagram showing wave action and storm surge height as a function of
storm surge, tidal cycle, and sea level rise 158
Figure 2-76 Location of the water infrastructure at the Town of Mattapoisett aside of the
Mattapoisett Harbor 160
Figure 2-77 A cartoon illustration of SLOSH modeling results on likely inundation risk for the
wastewater transfer station at Mattapoisett, MA 161
Figure 2-78 Hourly traffic map in the Mattapoisett region after evacuation order activated at
noontime 162
xvii
-------
ABBREVIATIONS AND NOTATIONS
Definitions and Abbreviations
AADT
annual average daily traffic
AERMOD
American Meteorological Society and U.S. Environmental Protection Agency
Regulatory Model
AIR-SUSTAIN
Air Impact Relating Scenario-Based Urban Setting and Transportation Asset
in Network
APAD
all-pipe and all-demand
ASR
aquifer storage and recovery
AUP&ET
Adaptive Urban Planning and Engineering Tool
AwwaRF
former name of Water Research Foundation
BASINS
U.S. Environmental Protection Agency's Better Assessment Science
Integrating Point and Non-Point Sources model
BOD
biological oxygen demand
CA-Markov
cellular automata-Markov
CAL3QHC
CALINE3-based monoxide model with queuing and hot spot calculations and
with a traffic model to calculate delays and queues
CBD
central business district
CDF
cumulative density function
CR
capacity reserve
CREAT
Climate Resilience Evaluation and Awareness Tool
CRWU
Climate Ready Water Utilities
CSO
combined sewer overflow
CSS
combined sewer system
CWA
Clean Water Act
D/DBP
di sinfectant/di sinfection by-product
DALR
dry adiabatic lapse rate
DBP
disinfection by-product
DOC
dissolved organ carbon
EBMUD
East Bay Municipal Utility District
GAC
elemental carbon in air emission
ECWF-1
East County Wellfield I
EPA
U.S. Environmental Protection Agency
FHWA
Federal Highway Administration
GAC
granular activated carbon
GCWW
Greater Cincinnati Water Works
GHG
greenhouse gas
GIS
geographical information system
GUI
graphical user interface
HAAs
haloacetic acids (nine individual species and the total of five (HAA5), six
(HAAe) and nine (HAA9) species)
xviii
-------
HDVC
hourly demand variation curve
HSPF
Hydrological Simulation Program - Fortran
ICLUS
U.S. Environmental Protection Agency's Integrated Climate and Land Use
Scenarios
ICR
information collection rule
IDF
precipitation intensity-duration-frequency
IPCC
United Nation's Intergovernmental Panel on Climate Change
IWM
integrated watershed modeling
LANDSAT
land remote-sensing satellite (system)
LCA
life cycle analysis
LID
low-impact development
LVVWD
Las Vegas Valley Water District
MARS
Manatee [County, Florida] Agricultural Reuse Supply
MCE
multiple criteria evaluation
MCL
maximum contaminant level
MCUD
Manatee County Utility Department
MS4s
Municipal separate storm sewer systems
MEOW
Maximum Envelope of Water
MGD
million gallons per day
MIA
most impacted area
MODIS
moderate resolution imaging spectroradiometer (for satellite)
MOM
Maximum of MEOW (Maximum Envelope of Water)
MOVES
U.S. Environmental Protection Agency's Motor Vehicle Emission Simulator
model
MPWF
Mosaic Phosphate Wellfield
MSX
EPANET-Multi Species Extension
NAAQS
U.S. Environmental Protection Agency's National Ambient Air Quality
Standards
NHSA
North Hudson Sewerage Authority
NO A A
National Oceanic and Atmospheric Administration
NOM
natural organic matter
NPDES
National Pollutant Discharge Elimination System
NRMRL
U.S. Environmental Protection Agency's National Risk Management
Research Laboratory
NTU
nephelometric turbidity unit
NYCDEP
New York City Department of Environmental Protection
O&M
operation and maintenance
OC
organic carbon in air emission
OKI
Ohio-Kentucky-Indiana Regional Council of Governments
ORD
U.S. Environmental Protection Agency's Office of Research and
Development
OTAQ
U.S. Environmental Protection Agency's Office of Transportation Air Quality
PR/MRWSA
Peace River Manasota Regional Water Supply Authority
PVC
polyvinyl chloride
RSSCT
rapid small-scale column test
XIX
-------
SAWS
San Antonio Water System
SBL
stable boundary layer
SCMs
Stormwater control measures
SDWA
Safe Drinking Water Act
SLOSH
Sea, Lake, and Overland Surges from Hurricanes
SRES
Special Report on Emissions Scenarios
SLID
Smart Urban Design
SWAT
Soil and Water Assessment Tool
SWFWMD
Southwest Florida Water Management District
SWMM
U.S. Environmental Protection Agency's Stormwater Management Model
TAZ
traffic analysis zone
TDF
travel demand forecasting
THM
trihalomethane
TOC
total organic carbon
TTHM
sum of four individual species of trihalomethanes
UHI
urban heat island
USGS
U.S. Geological Survey
UVA
ultraviolet absorbance at 254 nm
VIS SIM
a microscopic traffic model after "Verkehr In Stadten - SIMulationsmodeH'
VISUM
a macroscopic traffic model after "Verkehr In Stadten - SIMulationsmodeH'
VSP
vehicle specific power
WASP
Water Quality Analysis Program
WEAP
Water Evaluation and Planning
WTP
water treatment plant
WTP-cam
water treatment-climate adaptation model
WUP
water use permit
Notation and Symbols in Equations
a
level of significance
77bod
biological oxygen demand (BOD) removal rate
a vector of independent, normally distributed random variables with mean zero
and variance one
9
hydraulic residence time
0C
biomass cell age in aeration tank
»
defined by 3 = (l + kd6c)/dc
Mo
average of water quality for the baseline scenario
Ah
average of water quality for the future scenario in 2050
P
correlation coefficient
standard deviation of water quality for baseline scenario
o~i
standard deviation of water quality for future scenario in 2050
aenj
standard deviation of On/k
XX
-------
&o, standard deviation of Qiijk
defined by = x/y0
c0 initial BOD concentration
D GAC reactivation period
D a known correlation matrix for the nine raw water quality parameters
EBCT empty bed contact time
i sequence number of pixels of a quantitative component
/ sequence number of time period
kd BOD degradation constant
m0 mass loading
q flow rate
TOC effluent TOC concentration from GAC processing
roc input TOC concentration to GAC unit
usrt process design or operating variable
v aeration tank volume
w weighing factor
J microorganism concentration in the aeration tank in milligrams per liter; and TOC
increment over the compliance criterion, 2 mg/L
y capital of operation and maintenance cost for GAC processing
Y maximum yield coefficient in mg/mg for an aeration tank; and net annual cost of
GAC processing
z, vector of nine raw water quality parameters used in water treatment plant -climate
adaptation model (WTP/WTP-cam) modeling
xxi
-------
Part Two: Smart Urban Design (SUD) and Application Case Studies
The national adaptation report Part I (U.S. EPA, 2015a) described multiple environmental
and economic stressors facing our nation's water infrastructure. It further discussed the
adaptation need for improving infrastructure resilience and sustainability. This Part II report
investigates the relationship through scenario-based adaptation among the factors of
hydroclimatic and land use changes, urban growth, population shifts, transportation, energy, air
and water pollution, and water management. These factors can be shown to interact at the
watershed, urban, and system-specific scales. For example, urban development may lead to the
occurrence of an urban heat island (UHI), which increases energy use and water consumption,
but may reduce overall energy needs when smart growth policies are devised. Low-density
development leads to a lesser UHI effect but higher energy use in transportation, adding to air
pollution. Development can alter rainfall and runoff characteristics, which, subsequently, can
impact water quality and water supplies. The water quality and quantity changes may require
water plant processes to be altered, potentially increasing energy needs. Poorly planned patterns
in urban development also affect water demand distribution and sewer system operations with
respect to water age and quality, sanitary and storm sewer pipe network, and, hence, power
usage. These examples show that water infrastructure sustainability is a multidimensional issue
intrinsically related to watershed management, urban development pathways, and individual
water system engineering.
This Part II report presents the Smart Urban Design (SUD) tools and methods, their
principles for urban planning and infrastructure adaptation design. SUD development in this
research aims to define the interconnectivity and to assist water practitioners and urban planners
in developing more resilient and efficient water infrastructure. In sequence, the report first
outlines adaptation objectives and the SUD framework at three spatial scales. Next, it describes
unique environmental properties associated with urban growth and current planning practices
that facilitate such growth. In Sections 3.0 through 7.0, the core SUD components in urban
planning and water system engineering are described. Case studies are presented to provide
further insight into the function and utility of SUD tools and methods. For SUD application in
coastal areas, Section 8.0 briefly illustrates the complex factors of the hydroclimatic impacts in
adaptation planning. The summary and recommendations are presented in Section 9.0.
1. Sustainable Development of Urban Water Systems
Given that practitioners and water managers are risk-averse, a properly defined design
basis for hydrological impacts from climate and land use changes is the first and fundamental
step. How to develop the design basis can affect directly the infrastructure management
objectives in years to come. Because much of this infrastructure will remain in place for 50 to
100 years, there is significant uncertainty on how it relates to planning today and on the potential
for stranded capacity in the future, which is costly to tax and ratepayers. Failure to provide
adequate infrastructure has serious economic and social consequences. Similarly, the uncertainty
resulting from the capacity excess or deficit creates significant concern for local officials charged
with infrastructure management. Providing methods and tools to help reduce these uncertainties
and, thus, assist technical managers to construct a more resilient future is the focus of this
research report.
1
-------
1.1. Adaptation considerations
The projections of future hydroclimatic and land use conditions have uncertainties. Often
the degree of uncertainty can be too large for widely used planning and engineering practices
without undue economic cost in urban development. Inflexible water infrastructure is capital-
intensive, and the water industry is risk-averse. Incorporating uncertainty into the planning and
design process is essential. The adaptation process defined by this report is intended to manage
the uncertainties of the hydrological and land use projections (Figure 2-1). It aims to reduce the
uncertainty to the degree appropriate for infrastructure projects. The process incorporates a step
of reevaluating the adaptation design and objectives for the ability of further adaptation to the
changing circumstances. This adaptive practice is pertinent when significant hydroclimatic
impacts are realized in a local watershed.
Because of the uncertainty in future projections, water managers need to manage the risk
from inadequate, poorly planned, or delayed adaptation efforts. The consequences of inadequate
adaptation and adaptation limitations have been described in the literature (e.g., Felgenhauer and
Bruin, 2009; Felgenhauer and Webster, 2013). For this consideration, SUD tools and methods
are developed in systems approach with the aim to maximize adaptation co-benefits of the long-
life and capital-intensive water infrastructure investment by minimizing contributions to global
impacts (e.g., reduced emissions), providing for added economic efficiency (e.g., improved
transportation and water service), protecting public and private infrastructure investments, and
minimizing the need for future costly infrastructure retrofitting and even reconstruction.
Hydrological responses to climate and land use changes are realized over long durations,
while short-term disruptive meteorological events or climate impacts of large magnitudes can be
explicit and quantifiable. Examples of disruptive events include storm surge, urban flooding, and
salt water intrusion, all of which commonly are found in coastal areas. For these explicit impacts,
actionable technical information is usually available for the design and implementation of
adaptation actions. This scenario is marked as step 1 in Figure 2-1. For long-term hydrological
impacts, model simulations of land use and future climate frequently are used to project
hydrological conditions in a local watershed where the water infrastructure is located. These
projections often have a substantial degree of uncertainty. Corresponding adaptation actions are
marked as the Tier-I in Figure 2-1. The integrated model simulation and monitoring framework
to quantify the combined hydrological effects of future climate, land use, and population changes
will be presented in subsequent publications.
Processes A and B in Figure 2-1 involve real-time or near real-time monitoring and data
analysis as tools to validate and further refine climate and land use projections. This approach
can decrease the uncertainty in future projections. The land use and climate modeling updates,
marked as actions a and b in Figure 2-1, may improve the projections for less uncertainty. Then
the outputs and subsequent integrated hydrological modeling may help develop the technical
basis for adaptation planning and engineering design. This step is marked as step 2 in Figure 2-1.
In case the results have large certainty for engineering design and adaptation evaluation, the
iterative process continues in steps 3 and 4. For this objective-oriented monitoring-to-adaptation
process, the tools for integrated hydrological modeling and the near real-time monitoring are
available. Examples include the framework using MODIS and LANDSAT satellite imagery
(e.g., Chang et al., 2006, 2014a,b), the integrated hydrological modeling of climate and land use
changes in a local watershed (e.g., Tong et al., 2012).
2
-------
Tier-I
Climate modeling
revision/update
Tier-
Land use
projection
Watershed hydrology
projection
Remote-sensing based
short-term and near
real-time monitoring
Land use plan
update
Hydrologic variables
Water quantity, quality
Integrated watershed
hydrological modeling
Hyd rological
projection revision
for plan update
Response and action
implementation
©
J®
®-
Adaptati
planning
Adaptation action
implementation
c
3
Time
Figure 2-1 Schematic process diagram of iterative monitoring-adaptation framework for water
infrastructure adaptation. Symbols show steps in design basis development for processes
A/a and B/b (see text for details).
Water infrastructure and transportation infrastructure are two fundamental elements in
urban developments that provide vital urban services and support economic activities. Several
considerations in framing the infrastructure adaptation are important, including objective
definition; constraint assessment; adaptation feasibility analysis; adaptation option comparison;
and, finally, adaptation effectiveness evaluation. These considerations are specific to the physical
boundary of the service area or projected service area under consideration (Figure 2-2), whether
the adaptation is on the scale of a watershed, urban water systems, or unit operations (e.g.,
distribution pipe network). The adaptation effectiveness evaluation yields data and results to
compare with the urban development objectives. When necessary, the stakeholders and local
managers may take further urban adaptation actions, and even adjust the development goals and
objectives. This systematic approach can serve as a venue to better communicate the adaptation
options and their trade-offs to stakeholders.
1.1.1. Defining adaptation objective
Water infrastructure adaptation to future hydroclimatic and land use conditions is
effective when taken in the context of sustainable urban and socio-economic development, a
central objective for many stakeholders. This emphasis agrees with the objective "downscaling"
concept described by Brown et al. (2012). Although specific goals of adaptation may vary among
3
-------
Watershed-scale adaptation
Urban-scale adaptation
Figure 2-2 Typical spatial relationships of water infrastructure in an urban environment with illustration of infrastructure adaptation scales
and general process. (SW- stormwater, DW- drinking water, GW- groundwater, WW - wastewater)
GW recharge
System adaptation: DW
DW plant &
Adaptation techniques: storage
Commercial
(DW consumption,
WW and SW
generation)
Residential
(DW consumption,
WW and SW generation)
Adaptation techniques:
water loss reduction
(DW consumption,
WW and SW
generation)
WW treatment &
discharge
4
-------
stakeholders and local conditions, an overreaching and commonly shared objective can
be described as follows:
¦ To enhance water infrastructure resilience. The ultimate purpose is to provide
uninterrupted water supply and wastewater services and to provide stormwater
management and urban drainage for a projected socioeconomic growth under both
current and future climate conditions.
¦ To increase the technical ability to comply with the existing regulations and help the
implementation of urban development policies. The environmental regulations related to
hydrological impacts were reviewed in the National Water Infrastructure Adaptation
Assessment Part I (U.S. EPA, 2015a). For example, the case studies in Sections 6.0 and
7.0 show examples of adaptation actions against the projected increase of total organic
carbon (TOC) content in source water, the increase of water age in underutilized sections
of the drinking water distribution network, and, consequently, the increase in disinfection
by-product (DBP) formation in the water supply. Such a combination of natural and
developmental factors can be a prominent concern for water managers.
¦ To achieve the co-benefits of water infrastructure adaptation in environmental resilience
and sustainable urban growth. Water infrastructure construction and operation consume a
significant amount of energy and yield air emissions, water pollution, and negative
ecological impacts. Thus, the co-benefit in optimized transportation and water
infrastructure development and operation should be and can be maximized through
systematic analysis. This is important in the view of urban growth and future energy
needs (Yang and Goodrich, 2014; Yang, 2010; Dodder, 2014; Dodder et al., 2011).
¦ To minimize the systems' adaptation cost
There is a need to analyze water and transportation infrastructure together; both
infrastructure types are the traditional and fundamental focus of urban planning and
development. Transportation and water infrastructure are planned for a given urban
development. In turn, they also induce and can facilitate further expansion and shifting of
population and economic activities as the urban area grows. In addition to the traditional water
management functions, attention has been galvanized recently on water availability on the supply
side and water footprints on the consumption side. For water infrastructure, these fundamental
concepts can be expressed as water reuse or reclamation; water storage; water loss prevention;
water conservation; and, more importantly, water-energy nexus (PNNL, 2012; Yang and
Goodrich, 2014, and references therein).
One important attribute in adaptation is the time horizon for infrastructure planning and
urban development in general. In many parts of the United States, rapid urbanization and newly
improved or constructed infrastructure services are projected to concurrently occur with
significant changes in hydroclimatic conditions. Globally at the beginning of this decade, urban
centers only occupied about 2% of land area on Earth but accounted for 70% of global energy
consumption and air emissions (e.g., Parshall et al., 2010; ADB, 2012; IEA, 2013a,b). The urban
population is projected to increase further, from 3.4 billion in 2009 to 6.3 billion by 2050 (IPCC,
2014). In the future, the urban change likely will lead to an even greater contribution to global
energy, water, and food consumption, as well as air emissions. Adaptation action is necessary to
improve sustainability to meet the needs of projected urban growth.
5
-------
Because of the higher population density in urban centers, emission intensity, and water
consumption rates on a per capita basis are mostly lower than national averages (Dodman, 2009;
ADB, 2012). Therefore, the shifting of energy and water consumption into high-density urban
centers creates the location-specific socioeconomic dynamics that adaptation needs to address.
The conditions in the U.S. are similar. In this urbanization trend, adaptation offers opportunities
to reduce per capita emissions and water consumption, enabling meaningful changes in global
energy consumption growth (Dodman, 2009). Urban infrastructure development and
redevelopment have significant potentials to recognize these co-benefits (Yang and Goodrich,
2014). Further decreases in per capita emissions and water intensity are possible, depending on
the design and implementation of urban planning and adaptation actions that are both effective
and economically viable.
Effective infrastructure adaptation can be achieved in a systems approach. Many urban
sustainability issues and assessment matrices are described in relevant U.S. Environmental
Protection Agency (EPA) publications (e.g., U.S. EPA, 2007b,c; 2009b; 2012a). For example,
high-density developments, mixed-use zoning, walkable communities, and green development
all are specific sustainability measures. These measures have the potential to eliminate
unnecessary urban sprawl, thereby effectively adapting water and transportation infrastructure to
a changing environment. Such developments all are focused on urban performance and
efficiency in the form of energy and water footprints and their combination with economic
benefits (e.g., Chang et al., 2012; Yao et al., 2014). These considerations will be discussed
further in Sections 2.0 through 4.0.
1.1.2. Understanding urban adaptation constraints
The National Water Infrastructure Adaptation Assessment Part I (U.S. EPA, 2015a)
described the vast water infrastructure built in the nation over the past century, and discussed the
stressors on these water infrastructure systems (e.g., aging infrastructure, increasing demand) and
the implications of their ability to be adaptable to future changes while complying with
regulations. In the past, a significant national investment has been made to create this vast
physical urban infrastructure. Now it is being made continuously to improve the infrastructure's
reliability, resilience, and service. Thus, the physical footprints, planning guidelines, and existing
engineering practices, all define the premise on which the constraints must be understood and
managed for adaptation.
In urban development, water infrastructure is associated spatially with transportation
infrastructure: highways, roads, and mass transit. Both types of urban infrastructure form the
structural building blocks of urban communities. The resulting "urban form" defines the social
structure, population, and business distribution, and it is reinforced by local zoning laws and
ordinances. Common types of urban forms are monocentric, polycentric, and the combination of
two (Figure 2-3). Each of the urban forms defines how urban population and economic activity
are distributed in space. This reinforcing mechanism results in the so-called infrastructure
"locked-in condition" that limits the optimization potential in water infrastructure and other
urban systems. Consequently, change to how urban systems are planned is necessary. This
requires the ability to overcome the physical as well as socioeconomic barriers associated with
these locked-in infrastructure systems.
6
-------
' J
• •
\
Mass transit
Personal transport
%
CBD, exclusively zoned
o
CBD, mixed zoned
•
Communities
-> High
Compactness, self-contained development, density
Figure 2-3 Schematic diagram showing three major types of urban forms and their typical
properties: A - monocentric, B - mixed development with high reliance on personal
transport, and C - polycentric developments with mass transit. (GS - green space;
LDD - low-density development area; HDD - high-density development area; and
CBD - central business district).
The current urban development is oriented toward protecting public health and meeting
service demands, while being limited by economic considerations. The development mode has
resulted in an unprecedented urban sprawl that expands the urban footprint into exurban areas.
Figure 2-4 shows the process in the current practice of urban infrastructure planning and
engineering. It starts with stakeholder engagement to determine urban socioeconomic goals, the
projected or anticipated growth factors, and other socio-physical conditions. The subsequent
master urban planning guides the type and spatial distribution of urban land use, economic
activities, residential distribution, and environmental assets, including water resources, parks,
green space, and preservation of environmentally sensitive areas. The guidelines can be
implemented and enforced using the zoning ordinances and other local regulations. This
traditional practice in urban planning leads to a final urban form, in which population and urban
activities are distributed in the monocentric, polycentric forms, and variations between these two
end-members (Figure 2-3).
7
-------
General planning framework
policy
Growth factors
¦ Economic expansion
/ contraction
¦ Population change
¦ Life style change
Sociophvsical
constraints
¦ Climate, carbon
allocation
¦ Water and land
availability
¦ Economics and
policy preference
planning
Planning
Urban forms
Buildings -
Residential
Buildings -
Commercial
Buildings -
Industrial
Parks,
recreations
Green space,
eco. reservation
<=>
T ransformation
districts
Polycentric
Low density
^ High density
Mixing
Green
development
Urban sprawl
New
Development
0
Economic and policy adjustment
Evaluation /assessment
• Efficiency
• Economics
• Satisfaction
Evaluation
Engineering
Energy source;
Source water
Material flow
Construction Phase
Operation Phase
Building structure
Energy consumption
Water consumption
Waste generation
Transportation Infrastructure
¦ Subway, mass transit
¦ Urban roads
S Mass transit
J Individual transit
¦ Other roads (biking,
pedestrian, etc.)
- Telecommuting
Water Infrastructure
Centralized
S Water supply
•/ Storm water
S Wastewater
s Waste and solid waste
Figure 2-4 General process of the current urban master planning and its relations to transportation and water infrastructure engineering.
Less emphasis on urban form transformation and the evaluation criteria often results in urban sprawl. The element in gray text
in Planning step is not applicable; those in the engineering step are not discussed in this report.
8
-------
To the extent to which a specific urban form depends on topography and natural
environments, planning policies play a significant role. They either reinforce the monocentric
form or change it to a multicenter polycentric configuration. As shown in Figure 2-4, each of
these urban forms has distinct composition and configuration of land use patterns; population
distribution; and, thus, different characteristics of transportation and water infrastructure. The
subsequent phase of infrastructure planning and design follows various guidelines and economic
considerations. For example, the Department of Transportation has published a series of
guidelines on transportation mobility and infrastructure improvements (e.g., FHWA, 2002,
2012). The EPA's Office of Transportation Air Quality (OTAQ) has issued guidelines on the
emissions criteria, fuel standards, and transportation-vehicle technologies that affect urban air
quality (e.g., U.S. EPA, 2015a, 2011). Other technical models and tools are used widely to
evaluate and simulate the transportation needs, travel demand simulation (e.g., VISUM,
VISSIM), and air quality analysis (e.g., MOVES, AERMOD, CAL3QHC). These topics related
to urban adaptation will be discussed further in Section 3.0.
Water infrastructure is one principal element of urban infrastructure supporting and
shaping the urban form. The water services start with potable water supplies in distributing
drinking water to customers, followed by the collection and management of sewage and
stormwater to protect public health and property. Generally, master plans are developed for a
given set of land use and economic projections with the purpose to satisfy the current and future
water supply and water sanitation expectations. Many municipalities follow a well-defined
process in developing planning objectives and determining planning variables. Planning and
engineering tools are widely available, including EPANET and its commercial derivatives (e.g.,
WaterCAD, H20map) for drinking water supply, EPA's Stormwater Management Model
(SWMM) and related stormwater packages for stormwater management and urban drainage, and
engineering software platforms (e.g., SewerGems, H20Map/Sewer, HydraSewer). Overall, most
municipalities pay attention to the operation and management of existing infrastructure, which is
aging across the U.S. Some communities have expended efforts focusing on component
optimization, system improvement, and capacity expansion, but system-wide re-planning and
redesign rarely happen. These focus areas, for example, are identified in the nationwide
assessment (see Section 7.0 of U.S. EPA, 2015a).
Overall, the current practice promotes the expansion of the existing water system
infrastructure and its physical footprints. In the master planning process, municipalities and
stakeholders periodically assess urban infrastructure performance after construction and a period
of operation (Figure 2-4). This step aims to compare infrastructure performance against the intent
of the original master plans or new urban growth objectives. The performance evaluation serves
as a basis for master planning revision and modifications of existing urban infrastructure. This
master plan revision occurs periodically; a revision frequency every 5 to 10 years is common in
practice. Many county or municipal master planning time horizons are 5 to 30 years, depending
on the infrastructure types. The exact planning timeframe has more to do with the uncertainty
than with other factors; generally, public officials are reluctant to invest too far ahead,
particularly when the future is uncertain. A side effect of this current planning and engineering
process is the continuous urban sprawl into the exurban areas, as opposed to reevaluating the
underlying framework of urban systems (Figure 2-4). Flanders et al. (2014) described this type of
urban sprawl in an EPA internal report and further analyzed its implications on urban
infrastructure development.
9
-------
Radial sprawl, ribbon sprawl, and leapfrog sprawl, as shown schematically in Figure 2-5,
are the three common types of urban sprawl. In terms of water infrastructure, general planning
and engineering consists of three major steps: (1) land use and economic projection, (2) analysis
of spatial population distribution, and (3) projection of water demand and wastewater generation
in a planning timeframe. The economy of scale for operations favors a centralized water supply
system, which results from the initial monocentric urban form. The result is the single water,
wastewater, and stormwater management network, if the hydrographic conditions permit, as
commonly found in most U.S. cities.
Radial Sprawl Ribbon Sprawl Leapfrog Sprawl
Figure 2-5 Three major types of urban sprawl expanding the urban footprints into exurban areas.
Reproduced from Sudhira et al. (2005).
In principle, a centralized water supply system delivers water from a central treatment
plant through a vast distribution network at a lower treatment cost than a decentralized system.
But it may do so at the expense of increased energy consumption and greater risk of water
quality changes in downstream use areas. The same trade-off between the economy of scale and
energy expense occurs for sewer systems that collect wastewater from individual users and carry
it to a central location for treatment before discharge. Municipal separate storm sewer systems
(MS4s) drain an urban area and discharge overland runoff, often untreated, into local water
bodies. This arrangement is the most energy-efficient service within a monocentrically
distributed population.
Compounding the current planning and engineering practice is that all design guidelines
are based on the assumed climate stationarity. This assumption and implication on water
infrastructure were described in the adaptation report Part I (U.S. EPA, 2015a) and in the early
EPA's adaptation conference proceeding (U.S. EPA, 2009c). The stationarity issue is embedded
in planning guidelines and engineering codes but needs to be assessed; for example, the recently
revised American Society of Civil Engineers' code of ethics includes language on the need to
"comply with the principles of sustainable development in the performance of their professional
duties."
1.1.3. Revising or redefining planning and engineering focus
There is a critical need for a significant national investment to improve and renovate the
nation's water infrastructure. This new investment will provide a rare opportunity to reevaluate
the current urban development framework and, if needed, to break up the current urban sprawl
cycle and reorient the growth pathways toward sustainability (Yang and Goodrich, 2014).
Among many technical pathways, the master planning and revision process presents the most
practical opportunity to reduce unnecessary urban expansion and increase urban efficiency. This
~ High-Density Sprawl
~ Medium-Density Sprawl
~ Low-Density Sprawl
10
-------
point in the process also enables initiating and developing effective water infrastructure
adaptation to future hydroclimatic changes, in addition to the traditional land use considerations.
The urban planning process to incorporate global changes is schematically illustrated in
Figure 2-6. The revised process, called adaptive urban planning, contains two sets of adjustments
to overcome deficiencies of the current practice that are identified in performance evaluation.
The first pathway revolves around the urban and infrastructure adaptation through
adaptively realigning of the urban layouts and basic functions. These preplanned and adaptive
actions induce changes to urban forms for better sustainability attributes. Through adaptive
planning, for example, transformation districts and corresponding infrastructure (e.g., smart
transportation, water supply systems) may be able to induce existing urban transformation into
multicenter high-density configurations and to avoid low-density development to the extent
possible (Figure 2-6). The transformation district concept stems from spatial continuity in urban
development and stipulates that an urban initiative can attract additional development radiating
from the original location or district; some are called transformation corridors. A handy example
is the development of the 1-270 technology corridor in the Washington, DC, metropolitan region.
Many large cities, such as Philadelphia, PA; Glynn County, GA; and Ludwigsburg, Germany,
have transformation districts in their master planning, largely for sustainability consideration.
The principles, along with an example of urban transformation, will be illustrated in
Section 4.1 by analyzing the potential development scenarios for the Cincinnati metropolitan
area. In general, such scenario-based analysis in urban adaptation is important to compare the
cost and benefits among developed adaptation alternatives. The results provide a technical basis
to inform decision makers on the limitation of physical adaptation approaches and the likelihood
of success for water and other urban infrastructure systems. The results also can help understand
the feasibility and limitation of adaptation compared with other options, such as an infrastructure
rebuild (Felgenhauer and Webster, 2013). In the analysis, the future hydroclimatic and land use
changes need to be first assessed.
The second pathway is to change or adjust the urban developmental objectives or local
land use regulations (Figure 2-6). Using the measured urban sustainability and performance
attributes, developmental objectives can be adjusted in a way to change or modify the growth
against the hydrological and socio-physical constraints (i.e., the basic planning variables). The
process and its attributes such as environmental justice, capital flows, centralized versus
decentralized management, have been discussed in the literature (Small and Song, 1994; Ewing,
2008; Heikkila et al., 1989; U.S. EPA, 2006, 2007b; Baynes, 2009; Ostrom, 2010).
Adaptation option evaluation and analysis depend on how the impacts of future
hydroclimatic and land use changes are projected and defined at the scale of interest. Also
relevant is the relative magnitude between the projected future changes and those originally
assumed in the existing master planning. This important hydrological evaluation includes several
key elements, including those below.
¦ The degree of hydroclimatic change that affects the precipitation intensity-duration-
frequency (IDF) relationship in the watershed of interest. Because of its impacts on the
infrastructure's hydrological design basis, the long-term hydrological effects deserve
11
-------
Adaptive Planning Framework
Planning
Growth factors
¦ Economic expansion
/ contraction
¦ Population change
¦ Life style change
Sociophysical
constraints
¦ Clnoate, carbon
allocation
¦ Water and land
availability
¦ Economics and
policy preference
Buildings -
Residential
Buildings -
Commercial
Buildings -
Industrial
Parks,
recreations
Urban Adaptation
Economic and policy adjustment
Green space,
eco. reservation
Urban forms
Monocentnc
/ \
Transformation
districts
Polycentric
Low density
v-
[ High density
Mixing
Green
development
i>
Minimize New
Developments
Evaluation assessment
¦ Carbon footprint
(global & local)
¦ Water footprint
Efficiency
Economics
Satisfaction
Engineering
Energy source;
Source water
Material flow
Construction Phase
Operation Phase
r
Building structure
Energy consumption
Water consumption
Waste generation
Transportation Infrastructure
¦ Subway, mass transit
¦ Urban roads
S Mass transit
S Individual transit
¦ Other roads (biking,
pedestrian, etc.)
- Telecommuting
Water Infrastructure
¦ Centralized
S Water supply
S Storm water
S Waste water
S Waste and solid waste
Figure 2-6 General process of adaptive urban planning and engineering. Compared to traditional master planning in Figure 2-4, adaptive
planning promotes economic and policy adjustment on urban development goals and urban adaptation for high-density,
polycentric form through transformation and proper transportation and water infrastructure adjustment. The element in gray text
in the planning step is not applicable; those in the engineering step are not discussed in this report.
12
-------
careful reevaluation (Mailhot et al., 2007). The relevance of this important hydrological
impact has been extensively discussed in the literature (e.g., Wilby, 2007; Beck, 2005;
Semadeni-Davies et al., 2008; Ashley et al., 2007; Pielke et al., 2007).
¦ The degree of infrastructure capacity reserve in the current planning and engineering
practices; for example, the safety factors used in a design basis. A quantitative evaluation
of the capacity reserve (see Section 5.2 later) helps identify the vulnerability under future
conditions. The evaluation results can inform decision makers on the need for local
economic and developmental policy adjustments (see U.S. EPA, 2015a). This type of
"bottom-up" assessment is facilitated by the Climate Resilience Evaluation and
Awareness Tool (CREAT) available from the U.S. EPA's Office of Water1. It is
generally accepted that the precipitation and temperature changes in the future can
directly affect hydrological and water quality engineering. The impact can be exacerbated
through a complex interaction among hydrology, land use, and population growth in an
urban catchment. Furthermore, atmospheric feedback of land use change can be
significant to the urban microclimate, as described in Section 4.1. The effect on mass
conservation and energy momentum in the planetary boundary layer is known to create
precipitation variations in local and regional scales (Adegoke et al., 2007; Pielke et al.,
2007), and in changes of soil erosion and soil moisture (O'Neal et al., 2005; Miller et al.,
2007). For simplicity, this type of feedback-loop interaction is often neglected in
adaptation analysis. All these factors can affect the capacity reserve needed for successful
infrastructure adaptation.
¦ The adaptation co-benefits in energy efficiency when developing and evaluating the
adaptation options. These co-benefits are often neglected currently in the evaluation of
urban infrastructure performance, causing a major unrealized benefit to remain
unassessed. However, this co-benefit concept deserves attention in adaptive urban
planning as shown in Figure 2-6.
1.1.4. Selecting adaptation evaluation matrix
In adaptive urban planning, the urban performance evaluation and assessment step (see
Figure 2-6) requires a selection of the appropriate evaluation matrix. The matrix may include
criteria that address hydrologic adaptation impacts, describe the dependence between water and
carbon footprints, and consider the time of adaptation in terms of capital flow and adaptation
limitations in trade-off analysis. Some attributes in adaptation evaluation are listed in Table 2-1.
The adaptation co-benefits in energy and air emissions are an important and basic
attribute in urban infrastructure adaptation (Yang and Goodrich, 2014). Water infrastructure
contains significant energy footprints, yielding significant air emissions both during construction
and thereafter operations. Water infrastructure operation is often the largest energy user in most
communities. To evaluate the co-benefits and tradeoffs, one method relies on conjugate water
and energy/carbon footprints (PNNL, 2012; Yang, 2010). These two sustainability indices
unlock the dependence between energy usage and water availability, and therefore can provide a
useful criterion to find compromised solutions in the adaptation option analysis. For example,
adaptation solutions to address water availability in water-stressed regions often include water
1 http ://water.epa. gov/infrastructural/watersecurity/climate/creat. cfm
13
-------
Table 2-1 Adaptation attributes for common objectives
Attribute
Objective
1.
Adaptive urban planning
Urban form
Urban sprawl index
Population density
Housing density
Transforming district
Zoning
Sustainable land use and resilient infrastructure
Reducing exurban de\«lopment
Achieving compact de\«lopment
Achieving compact de\«lopment
Transitioning to polycentric form
Land use change for planned de\«lopments
2.
Urban transportation
Traffic delay
Trip generation
Fuel consumption
Emissions
Increasing transportation efficiency
Promoting walkable community and mass transit
Reducing fuel use in urban activities
Increasing mitigation co-benefit
3.
Urban water systems
Water availability
Water quality
Energy use
Energy and emission
Resilience
Adequate supply to meet demand
Compliance to SDWA and CWA regulations
Reducing energy cost in managing water systems
Reducing life-cycle emission and improving o\«rall
energy efficiency for the mitigation co-benefit
Ability to provide service function under natural and
man-made emergency and disrupts events
reclamation, reuse of treated wastewater for non-portable and even potable purposes,
desalination, and water storage (Oron et al., 2007; Yang et al., 2007, 2010). These adaptation
options have high energy intensities and generate air emissions when producing the new "virgin
water." Similarly, compromised solutions between energy/carbon footprint and economic cost
are relevant to water infrastructure planning (Chang et al., 2012). For illustration, the case study
on water infrastructure expansion alternatives in Manatee County, Florida is described in Section
4.2.
Evaluation matrix selection is always objective-dependent. Some commonly investigated
pairs include water and energy/carbon footprints, water availability and cost analysis, and water
footprints (see Table 2-1). Defining the evaluation matrix is often the first step in the adaptation
process that can affect adaptation pathways and outcomes (Figure 2-6), and often involves the
14
-------
extensive engagement of the public and stakeholders. Therefore, the evaluation matrix may be
highly location-specific and should be clearly described in the performance evaluation.
1.2. Three levels of water infrastructure adaptation
Physical adaptation to urban infrastructure can take place at three levels on a spatial scale
(Figure 2-2). Referring to Figure 2-6, adaptation may also occur at the different stages of the
planning-engineering-evaluation process, such as in the planning phase or the engineering of
specific adaptation measures against specific hydroclimatic impacts (e.g., floods and chronic
droughts). As the adaptation level changes from systems adaptation (e.g., storage, water
conservation, and water loss prevention) to urban-scale or watershed-scale adaptation, the
complexity increases for the systems analysis, and thus for adaptation planning and engineering
design. The remainder of this report describes the technical approach for adaptation in these
three levels and illustrates the considerations in selected case studies.
1.3. Smart Urban Design (SUD) for systems analysis
Water infrastructure adaptation in the watershed scale, the urban scale, and the water
system scale has own advantages and challenges (Table 2-2). To quantify specific adaptation
actions, one integrated modeling tool "Smart Urban Design (SUD)" has been developed from
this research. It consists of scenario-based modeling tools integrated as a platform to assist the
design of adaptation actions (Figure 2-7). The main SUD components are described below.
1.3.1 Integrated Watershed Modeling (IWM)
The IWM tool is built upon the EPA's Better Assessment Science Integrating Point and
Non-Point Sources (BASINS) program (U.S. EPA, 2019) with further integration of a land use
model under the future global conditions (Figure 2-8a,b).
The EPA's Integrated Climate and Land Use Scenarios (ICLUS) is a land use model that
provides an explicit projection of population, housing, and land use under future climate
scenarios. Future climate is specified in the four Intergovernmental Panel on Climate Change's
(IPCC) Special Report on Emissions Scenarios (SRES) scenarios. At present, the ICLUS
projections are made at the county-scale spatial resolution through the year 2100. This
development scenario tool is based on a pair of models: a demographic model for population
projection and a spatial allocation model to distribute the projected county population into
housing units at a 1-ha (1-hectare) pixel resolution. Population allocation from a county scale to
census tract resolution is technically challenging, because of the model assumptions for the
present, near-term and distant economic growth. For example, the spatially explicit regional
growth model (SERGoM) is used in population allocation to generate the projections at a
spatiotemporal resolution of 10 years and one (1) hectare. The associated uncertainty with low-
resolution population projection may be excessive for infrastructure planning. This potential risk
has not been assessed fully at the time of this reporting. More details on the methodology can be
found in U.S. EPA (2009b, 2010b). Projections covering the contiguous U.S. can be accessed at
https://www.epa.gOv/ICLUS/ICLUS-downloads#tab-l.
15
-------
Table 2-2 General advantages and challenges of three-level adaptation actions
Adaptation
Level
Methods/To
ols
Advantages
Challenges
Integrated
Watershed
Management
IWM, ICLUS,
CA-Markov,
HSPF,
SWAT
• Ability to protect source
water quality and assure
water availability
• High implementation
feasibility through the
CWA and SDWA
regulatory framework
• Data requirements for
watershed process
analysis
• Land use planning and
action often difficult to
implement
• Close interactions of land
use and urban catchment
hydrology
Adaptive
Urban
planning
AUP&ET
SWMM,
VISUM,
MOVES
• Large emission mitigation
potential
• Changes amendable to
urban development goals
• Increasing urban
resilience
• Potential to accommodate
multiple objectives (e.g.,
economic development)
• Complex, requiring
integrated planning
• The transformation
required in urban
development
• Cost and time for capital
investment payoff
• Public acceptance
Infrastructure
Systems
Adaptation
WTP-cam,
EPANET,
SWMM,
SWC
• Taken as a part of capital
improvement
• Well-defined actions for
decision making
• Increasing infrastructure
capacity for specific
needs
• Independent for quick
actions at a relatively
small capital cost
• Difficult to resolve urban-
wide performance issues
• Limited adaptation
potentials after years of
improvement
• Difficult to resolve urban-
wide performance issues
Urban land use changes are dynamic and often difficult to model. They occur at a spatial
resolution much finer than the county level resolution used in ICLUS. For this reason, the
cellular-automaton Markov (CA-Markov) land use model can be used available in the IWM
module (Figure 2-8a,b). The CA-Markov method for land use projection combines the stochastic
probability of future evolution that builds on the current situation (namely, the statistical state),
and the geographic association with current and projected land use. The latter is captured using
the cellular automaton modeling that depicts the probability of spatial association in state
changes. In combination, the CA-Markov modeling can estimate probable land changes in spatial
aggregation using the geographical information system (GIS) modeling capabilities (see Tong et
al., 2012; Sun et al., 2013). Nevertheless, population and land use projections in planning
scenarios are the most difficult and least quantifiable for urban areas, especially as they may
relate to large-scale hydroclimatic impacts like sea level rise that will alter the current landform.
16
-------
Watershed management
programs (e.g., TMDL)
Figure 2-7 Schematic diagram of the Smart Urban Design (SUD) structure for scenario-based urban development planning and engineering.
Program linkage between the three major modules of IWM - integrated watershed management models for hydrological changes,
SmartWater - water supply engineering design tool, and AUP&ET - adaptive urban planning & engineering tool for scenario
analysis. Colors indicate different blocks in the integrated simulation process.
17
-------
Surface water projections
(flow, water quality) at t12,...
IWM - watersheds
Figure 2-8a Process flow diagram of the Integrated Water Management (IWM) modeling for watershed
simulations. It consists of the climate-influenced hydrological model HSPF and land use
models either by ICLUS or high-resolution CA-Markov modeling. The program resides in
the EPA's BASINS framework. Colors indicate different blocks in the integrated simulation
process.
The problem is confounded for projections requiring high spatial resolutions, like in the
census tract levels. In addition, disruptive development decisions and events can make model
projections less accurate and not useful. These factors can lead to conditions inconsistent with
the spatial continuum assumptions embedded in the semi-empirical CA-Markov method. This
potential problem can result in erroneous model projections, a limitation that cannot be under-
estimated.
Figure 2-8a shows a general modeling framework for suburban and rural watersheds. For
urban catchments experiencing significant changes, the modeling framework is shown in Figure
2-8b. In Figure 2-8a, the hydrological parameters (e.g., streamflow and water quality) are
modeled for future time frames of interest (e.g., ti, I2, etc.) using EPA's BASINS program.
BASINS for assessment of water quality and flow variations in watershed runoff and surface
streams is documented in U.S. EPA (2013b) and application studies (Tong et al., 2012; Sun et
al., 2013). The newly released BASINS4.1 is a comprehensive platform, providing a choice of
multiple hydrological simulation engines. Available models include Hydrological Simulation
Program - Fortran (HSPF), the Soil and Water Assessment Tool (SWAT), the EPA's
Stormwater Management Model (SWMM), Generalized Watershed Loading Function model
extension (GWLF-E) MapShed, and the simple watershed model Pollutant Loading Estimator
(PLOAD), as well as two instream water quality models AQUATOX and the Water Quality
Analysis Program (WASP).
18
-------
Runoff projections for a given
return interval at t12
IWM - urban catchment
Figure 2-8b Process flow diagram of the integrated watershed management (IWM) modeling for urban
catchment using EPA's National Stormwater Calculator (U.S. EPA, 2014). Modeling
functions in the light blue box exist in the National Stormwater Calculator. Colors indicate
different blocks in the integrated simulation process.
There are some generalities with model inputs for the IWM tool. The parameters for
hydrological and land use are of greatest interest. First, land use projections can be made for the
desired future period, using past land use in digital format for model calibration and validation.
Additional land use constraints, including nature preservations, water bodies, historical
preservations, are specified as model constraints in the GIS land use simulation. Examples can be
found in Sun et al. (2013). Separate anticipated population changes can be directly downloaded
from the ICLUS model outputs. When higher spatial resolution than the county-scale projection
is needed, population change is often available from master plans created by local governments.
The model outputs of land use types and population distributions are the inputs for subsequent
BASINS hydrological modeling (Figure 2-8a).
Future climate parameters (i.e., precipitation, temperature, dew point, wind) are another
set of input parameters for hydrological modeling in BASINS. The IWM module obtains these
parameters from climate models. The model projections used in IWM are further revised for
post-bias corrections using the techniques from Liang and Julius (2017) and Yang et al. (2017).
The projected precipitation for given return intervals is used as one HSPF parameter in BASINS
simulation. Furthermore, the urban catchment is smaller in size than a rural watershed but may
contain more dynamic changes in land use and land cover, and built infrastructure. For urban
application, EPA's National Stormwater Calculator (U.S. EPA, 2014) is the main simulation
engine (Figure 2-8b); it is a simplified model based on EPA's SWMM on a GIS platform. The
Stormwater Calculator accepts future land use either by green infrastructure design (e.g.,
detention and retention ponds, swales, and other catchment areas) or directly from CA-Markov
19
-------
modeling of land use at census tract resolutions. The land use modeling techniques are described
in Tong et al. (2012), Sun et al. (2013), and Fu et al. (2018).
Precipitation is the other important modeling input for IWM-urban catchment. For this
purpose, downscaled climate modeling outputs may not be suitable for infrastructure design or
planning because of their coarse spatial resolution and large projection uncertainty. One principal
reason is that microclimate in urban centers can significantly differ from the regional climate of
natural land cover. One example is the UHI effect often discussed in the literature. The unique
nature of the urban microclimate is discussed in a case study described in Section 4.1.
For both watershed and urban catchment areas, the IWM helps project key hydrological
parameters at a future time for subsequent analysis in the Adaptive Urban Planning and
Engineering Tool (AUP&ET) discussed next. These projections include:
¦ Unit hydrographs for storms of a given return interval. Both peak flow and time of
concentration are specified. Often these parameters are given for specific storm return
intervals.
¦ Stream base flow. The model outputs can be analyzed for changes in stream base flow
under the future land use and climate conditions. Application examples can be found in
Johnson et al. (2015) and U.S. EPA (2013b).
¦ Surface water quality parameters such as total nitrogen, total phosphorus, turbidity, and
organic pollutants (e.g., Tong et al., 2012).
1.3.2 Adaptive Urban Planning and Engineering Tool (AUP&ET)
The schematic diagram in Figure 2-9 shows major modeling components and data flows
for the scenario-based Adaptive Urban Planning and Engineering Tool (AUP&ET). The tool
considers urban development scenarios for transportation, water supply, wastewater, and
stormwater systems. These infrastructure systems are the controlling factors over the basic urban
form, employment and economic activity, and population distributions. The scenario-based
feature in model simulation allows one to develop alternative development scenarios.
Urban infrastructure has a large physical footprint. The existing infrastructure systems are
capital-intensive and difficult to change once built. The nature of the infrastructure precludes the
potential to perform real-world experiments for optimal planning and design solutions. Thus
AUP&ET takes the technical approach in a scenario-based computer simulation. The tool relies
on two major inputs. First, the development objectives are defined, for which a set of
development options can be created for a given physical and environmental setting. Second, for
water infrastructure adaptation, water availability and hydrological parameters of surface streams
(e.g., peak and base flows, water quality) are basic variables for quantification in developing the
urban scenarios (Figure 2-9). Each urban scenario yields quantitative outputs of the future land
use and urban parameters, including urban form, community functions such as parks,
transportation, water sanitation, and supply services. These model variables are then incorporated
into the land use modeling and imported into GIS for spatial analysis (Figure 2-9).
Perrone et al. (2011) analyzed the roles of transportation and water infrastructure in
determining the physical form and efficiency of urban systems. Flander et al. (2014) further
investigated the attributes and inter-dependency of the infrastructure. Here in AUP&ET, the
inter-dependence and interactions are modeled for environmental attributes on an urban scale in
20
-------
development scenarios. An application example will be described in Section 4.1. Potential
analysis outcomes may include:
¦ Population distribution for specified development goals;
¦ Daily and peak traffic flow at road link levels;
¦ Urban-wide emissions, and traffic congestion identification;
¦ Drinking water supply needs and their spatial variations;
¦ Stormwater and wastewater generation rates and spatial variations;
¦ Energy consumption and cost comparisons in the transportation and water sector.
Figure 2-9 Process flow diagram of the scenario-based Adaptive Urban Planning and Engineering
Tool (AUP&ET) for urban planning and engineering. Each of the four program modules -
traffic, drinking water, wastewater and stormwater, is discussed separately in Sections
3.1 and 5.0. Colors indicate major elements in the integrated simulation process.
21
-------
1.3.3 Smart Water for water supply
Within AUP&ET, one engineering tool is the SmartWater module for water treatment
and distribution. Different from the other planning-centric AUP&ET modules, the SmartWater
tool is developed for system engineering, evaluation, and detailed unit process analysis in
adaptation. It consists of an updated Water Treatment Plant (WTP) model for water treatment
process engineering, and a sensor-based data-driven EPANET engine for water distribution
(Figure 2-10).
SmartWater's WTP3.0 consists of two separate modules that are linked by an overall
graphic user interface (GUI) (Figure 2-10). WTP2.0/2.2, originally developed in 1994 and
updated in 2004 (U.S. EPA, 2005), was intended for the national evaluation of water treatment
plant performance to support the promulgation of the Safe Drinking Water Act (SDWA) DBP
Stage II regulations. In the SUD, the water treatment plant - climate adaptation model (WTP-
cam) is developed from WTP2.2 for plant-specific adaptation analysis. Its application in a case
study at the Greater Cincinnati Water Works (GCWW) Richard Miller Water Treatment Plant
("Miller WTP") is described later in Section 6.3.
The SmartWater module in SUD treats the two processes (water treatment, and
distribution) as a single system (Figure 2-11). This approach aligns well with U.S. water utilities
starting to align water treatment and distribution operations under the same management, or to
Inputs from IWM results
Inputs from AUP&ET results
SUD
Requlatorv/national analysis
National plant
inputs
WTP2.0/2.2
L
Unit process modeling
using database
Model outputs: DBP, CI'
, TOC, UVA, I.R., etc.
Svstem-specific analysis
i»Raw water quality
¦ ~ TOC/UVA module
'»Alkalinity/pH module
1 ~ CI decay module
'~THM/HAA module
i» Other DBP module
i» Ct/lnactivation module
o New improvements
Smart Water modules
MC engine for
r.w. variability
Plant spec
Unit process modeling
using WTP2.2 or
mechanistic models
Process change and
optimization
Model outputs:
• Water quality: DBP, CI",
TOC, UVA, I.R., etc.
• Cost-curve
• Compliance evaluation
Figure 2-10 Schematic diagram of conceptual modeling framework for WTP3.0 as a major SUD
element. It consists of the system-specific analysis using WTP-cam (Water treatment
plant - Climate Adaptation Model) and the regional analysis in WTP2.0/2.2.
22
-------
Commercial
(DW consumption,
WW and SW
generation)
drinking water
source
{DW consumption,
WW and SW generation)
GW recharge
Industrial
(DW consumption,
WW and SW
generation)
WW treatment
and discharge
drinking water plant
and distribution
Water Distribution (EPANET)
mixing/
coagulation ^
settling/
sedimentation
l
settling/
pH sedimentation
adjustment ,
1 1
sand filtration
* ~ ~ "¦ | pH adjustment,
GAC adsorption | disinfection
1 i
¦>
11 r reservoir
Source Water ¦
coagulant addition
OnsiteGAC
regeneration
Water
Variables, Demand
Local Treatment
Source water
Variables,
Source Water
Water Treatment (WTP-cam)
Figure 2-11 Schematic diagram of water supply and major system variables. Water treatment and distribution are the two engineered systems
to meet variations in source water quality and water demand. The consideration of disinfection by-products (DBPs) is cited for
specific management and technical considerations. Other abbreviations: DW - drinking water; GAC - granular activated carbon;
SW - stormwater; and WW - wastewater.
23
-------
link those distributors with water producers in real-time data exchange. As shown in Figure 2-11,
this integrated approach allows one to manage and optimize the treatment and distribution
infrastructure in a timely manner to address changes in source water quality and to meet
changing water demands. This communication exchange has become feasible because of recent
technical advances in sensor-based monitoring, real-time data communication, and algorithm-
assisted system operation. More technical details will be presented in Section 6.0.
1.3.4. Source-to-tap water supply in a systems approach
The SUD methodology takes a systems approach toward efficient and resilient water
infrastructure. In current engineering practice, line process diagrams in infrastructure analysis
usually describe a water system and unit processes with simplistic consideration of the spatial
interactions with other urban components and surrounding watersheds. This traditional approach
is convenient in technical analysis. However, it may discount interactions between the highly-
dense socioeconomic activities and an ever-changing urban environment.
Taking water supply as an example, the "source-to-tap" systems approach (Figure 2-11)
is the basis of SUD. Currently, the SUD tool only has surface water as the source water in
drinking water production; modules for groundwater and reclaimed water will be considered for
addition at a later time. The rest of this Part II report discusses the four basic steps in the systems
approach, its tools, methods, and application examples:
¦ IWM and analysis for water quantity and quality variations in watersheds;
¦ AUP&ET modeling and analysis for urban development scenarios. The objective is to
define the water demand and its spatiotemporal variations at present and in the future.
Energy and economic efficiency of the urban scenarios for decision making also is
analyzed;
¦ SmartWater modeling and analysis to optimize water supply efficiency. First, the system
capacity and capacity reserve are defined for the water supply objectives and service
resilience against future hydroclimatic impacts. Second, for changing source water or
water demand, the potential system alteration/expansion/addition is evaluated;
¦ Upon evaluation of the infrastructure performance, a new round of system evaluations
may take place. The results are used to evaluate the necessary adaptation for improving
infrastructure's resilience and sustainability. This iterative re-evaluation and adaptation
process, as commonly practiced in periodic master planning, is shown in Figure 2-7.
2. Adaptive Urban Planning in Urban Scales
In the recent fifth IPCC climate assessment report (IPCC, 2014), the 3rd Work Group
investigated mitigation and adaptation in the urban environment. They concluded that urban
form transformation has, by far, the single largest potential to achieve meaningful carbon
emission reductions and urban efficiency improvement. Many publications (e.g., U.S. EPA,
2006, 2009b, 2012a, 2013a, and references therein) identified several common planning options
including infill, inner-city redevelopment, mixed land use, and employment centers. These
measures can introduce urban transformation to more desired and sustainable configurations.
24
-------
These smart growth practices and transformation measures have been applied in U.S.
cities (U.S. EPA, 2007b, 2013a). They are designed to slow down urban sprawl and achieve
adaptation-mitigation co-benefits, but often require changes in metropolitan transportation and
water services. Other urban forms alternative to the traditional monocentric configuration may
offer a smaller urban
physical footprint with a
higher population and
housing density. However,
this change is
accompanied with more
complicated transportation
and water infrastructure,
and thus greater
difficulties with planning
on how to incorporate
existing infrastructure
assets. To overcome these
planning and engineering
challenges, new
approaches are essential
for smart growth through
urban form
transformation. Attention
in this report is given to
the planning methods and
tools for transportation
and water infrastructure.
Because of the
high population density
and integrated urban
infrastructure, urban
centers also are vulnerable
to natural and man-made
disruptions including
significant hydroclimatic
changes in the future.
Three major categories of
impacts on urban
functions are listed in
Table 2-3.
2.1 Physical infrastructure and urban forms in current practice
The monocentric urban formation is common in the U.S., where the urban population is
distributed around a single central business district (CBD) of concentrated economic activities.
In this urban form, automobile-based mobility is a precondition to facilitate the urban-suburban-
Table 2-3 Selected urban functions impacted by hydroclimatic
conditions
Hydroclimate Factors
Urban Functions
Long-term drought and
large swings in
precipitation variation
• Water supply, landscape, local
agriculture
• Wastewater and stormwater NPDES
discharge to streams
• Urban heat island effects and heat
spells on population health
• Example: U.S. Southwest, Southeast,
Rocky Mountains
Heavy downpour,
disruptive
climate/meteorological
events (e.g., tornados
damaging winds, etc.)
• Transportation management and
roads operations
• Urban flooding and water service
systems operation
• Water pollution management from
nonpoint source
• Inundated sewer systems resulting in
sewer overflows and property
damage
Storm surge and sea
level rise
• Disruption to water supplies; changes
to hydraulic gradients affecting
stormwater drainage and wastewater
collection
• Disruption to transportation systems
• Inundations of roads and pipe
systems
Note: CSO - combined sewer overflow
25
-------
exurban arrangement (Theobald, 2005). The typical geometry of the urban form is schematically
shown in Figure 2-5. Examples include numerous, mostly middle-to-large sized, urban centers
such as Las Vegas, Cincinnati, Houston, and most urban centers of the Northeast and the
Midwest.
As a city grows into a very large metropolitan center, the population becomes more
dispersed and the monocentric form evolves into a polycentric arrangement of connected satellite
cities. This urban form is now characteristic of very large metropolitan regions, such as New
York City, Washington DC, San Francisco, and Los Angeles. The urban form transformation,
and its implications to the CBD formation, population distribution, and transportation service,
have been investigated in literature (e.g., Gordon et al., 1986; Small and Song, 1994; Heikkila et
al., 1989; Larson et al., 2012, Garcia-Lopez, 2012, and Zhou et al., 2013). The nature and
process of the transformation have significant implications for the feasibility of developing and
implementing adaptation options.
Polycentric urban form is marked by a multi-center urban configuration; for example, in
the leapfrog sprawl of Figure 2-5. The transition toward a polycentric form may take different
pathways. Continuous urban expansion toward a more dispersed polycentric form is a persistent
trend leading to unplanned uncontrolled urban sprawl. On the contrary, the transition can permit
high-density development, less personal travel, better use of mass transit and green space. This
requires a different configuration for fixed urban infrastructure assets. As population and urban
activities are redistributed, water infrastructure is accordingly transitioned in space for a new set
of operational requirements to meet new water service and management needs. The three typical
urban expansion configurations in Figure 2-5 are all linked to the transportation routes and other
infrastructure services, forming the mode of radial, ribbon and leapfrog sprawl (Sudhira et al.,
2005). For these different urban forms, the trade-off is under debate on urban efficiency and
infrastructure sustainability ranging from resource allocation and urban ecology to engineering
and operations.
Urban-developmental effects on water infrastructure have been widely recognized. For
example, the centralized operation and management in water services have allowed for better
control of water pollution and management toward meeting water regulations. It benefits from
the economy-of-scale. However, negative environmental consequences are found in energy use
and thus potentially higher indirect emissions, barriers to resource recovery, excessive water age
in distribution systems, and vulnerability to the impact of natural and man-made incidences. The
alternative form of urban development promotes more decentralized water systems. As urban
transforms into polycentric form, the centralized water system may become decentralized, and
the urban water cycle may become more localized (Hering et al., 2013; Luthy, 2013). This can
result in better service to localized, high-density population centers. However, the required
infrastructure transformation can be a difficult technical and engineering challenge. It requires
coordinated urban planning among land use, transportation, and water services.
2.1.1. Land use encouraging urban sprawl
The three types of sprawl modes (Figure 2-5) can be easily found in the historical
developments of U.S. cities. In Figure 2-12, the old urban centers of Atlanta and Phoenix
expanded radially toward exurban at a rapid rate in merely 22 years from 1970 to 1992. The
older urban centers, such as DeKalb County in Atlanta, further evolved into spatially continuous
26
-------
high-density development. Smaller development centers in exurban perimeters in 1970 were later
expanded in size, linking to the major urban centers through fill-in development. This leapfrog
pattern is very common; for example, in the Norcross, Marietta, and Douglas communities in the
Atlanta metropolitan area, and in the communities of Sun City, Mesa, and Chandler in the greater
Phoenix area (Figure 2-12). Furthermore, the ribbon sprawl (Figure 2-5) can be observed along
transportation roads, forming linear spreading of urbanized lands. This development pattern is
obvious along the roads of regular shapes around the Luke Air Force Base west of Phoenix (see
1992 map in Figure 2-12).
Urban population and land use are difficult to project. Future population and land use are
a function of urban economic conditions, political motives, and development initiatives; the last
can introduce sudden changes in spatial continuity of land use patterns, and thus poses a
modeling challenge in mathematic formulations. As an approximation, the CA-Markov
simulation in GIS can be used with model boundary conditions representing urban land
development restrictions. Wei et a I (2012, 2017), Tong et al. (2012), and Sun et al. (2013)
Figure 2-12 Urban expansion and urban form transformation for Atlanta (upper) and Phoenix (lower)
metropolitan regions between 1970 and 1992. Red color indicates developed urban
land use. Imagines obtained and modified from Auch et al. (2004).
27
-------
successfully projected future land use changes in the urban communities, suburban watersheds of
the Cincinnati and Las Vegas metropolitans. Their modeling methodologies incorporate
population and land use variables as a GIS model filter in the C A-Markov simulations. The
ICLUS tool and projections (U.S. EPA, 2010b) is an alternative to project future housing density
and land use categories. See associated discussions in Section 1.3.1.
2.1.2. Transportation and energy performance
The concept of urbanization along transportation routes is shared by most U.S. cities.
Such urban expansion, facilitated by current urban planning practices, has a set of characteristic
physical layouts for water and transportation infrastructure, which in turn defines urban functions
and affects infrastructure efficiency and adaptability.
Figure 2-13 shows the evolutionary trajectory of transportation efficiency as the U.S.
cities grow from medium to very large metropolitan areas. Plotted statistical data were obtained
from the Department of Transportation annual urban mobility reports prepared by the University
of Texas (Schrank and Lomax, 2009). In these plots, the efficiency variables (annual delay,
y=-10209+17673P
F?=0.84
10
Population (mil) Population (mil)
Figure 2-13 Transportation efficiency (annual delay, travel index, excess fuel use, and annual cost)
in year 2007 as a function of urban population in the U.S. urban centers. Data from the
2009 urban mobility report (Schrank and Lomax, 2009). The blue dash lines are 95%
upper and lower bounds of the regression of all data.
.1 1 10 100
Population (mil)
Population (mil)
28
-------
excess fuel usage, travel index, and annual cost) in 2007 are all based on a comparison between
peak hour traffic and free flow conditions in principal freeway and arterials. Travel index is a
ratio between time used in peak hour versus free flow at 60 and 35 miles per hour (mph) on the
freeway and arterials, respectively. The excess fuel usage is defined as fuel wasted at vehicles
moving at a slower speed than at free flow conditions. These measures quantify the
consequences of urban traffic, indicative of urban transportation efficiency.
The transportation efficiency is correlated to urban population size (Figure 2-13). The
correlation is the strongest (R2=0.84) on the excess fuel use. The correlation slope indicates the
change in transportation performance (e.g., delay or excess fuel use) as the population grows and
urban sprawls. The excess fuel use and cost curves indicate that as cities grow into very large
metropolitan centers, the slope becomes smaller and nearly a constant. For cities of population
<3 million, they tend to plot to the left side of the regression line indicating greater excess fuel
consumption (Figure 2-13). Efficiency appears to be attributable to the effect of mass transport
and high-density development in large cities (Schrank and Lomax, 2009). The similarities and
differences reveal the underlying principles that govern the efficiency of urban transportation
systems.
In all cases, the limitation in infrastructure adaptation under the current decades-long
urban planning practice is important. More meaningful improvement may come from the change
in urban form from the traditional monocentric to polycentric urban arrangements.
Transformation districts and adaptive planning are critical elements in the process (see Figure 2-
6). Such change is conceptually illustrated in Figure 2-5. How to facilitate the urban form
transformation for improved urban efficiency is a challenge to urban planners and infrastructure
engineers. An example of this transition has been examined in a detailed mechanistic study of the
transportation system in the Cincinnati metropolitan area. The results are described in the
subsequent sections to illustrate the likely benefits from urban form transformation using
computer simulation of adaptation planning scenarios.
2.1.3. Water planning and engineering
Water infrastructure planning and design follow the guidelines in urban development
master plans, and further details the needed assets and management required to provide water
services (see Figure 2-4 and 2-6). In general, the water infrastructure is scoped mostly during or
after transportation planning according to master plans or development policies.
In expansion, an urban form evolves and, sometimes, develops into polycentric
configurations. The transportation structure reinforces the changes. Such infrastructure-
facilitated change can alter the spatial distribution of population and economic activity, and in
return generates new water service needs. Often passively, the water service is compelled to
adapt and expand to meet the new water service demands (Figure 2-4). It is not uncommon that
the legacy of the centralized water system configuration remains intact even after cities are
transformed into a polycentric formation. Practical examples are numerous, such as the vast
centralized water service infrastructure in Cincinnati, Detroit, Cleveland, Los Angeles, and New
York City.
Existing water systems are mostly monocentric in the U.S.: centralized water treatment
and water distribution, centralized wastewater collection and discharge treatment, and, to a lesser
extent, centralized stormwater systems (mainly gravity-driven) with discharges to available
29
-------
waterways. Specific engineering of the water supply and water sanitation infrastructure is
described later in Section 5.0. Additionally, current water infrastructure planning and design are
often focused on component optimization, system improvement, and capacity expansion. This
tendency in development has the following notable attributes:
¦ Water infrastructure, most of which is buried, is planned and designed to meet water and
wastewater demands as defined in urban master plans. Once designed and built, the water
infrastructure and their functions create a "locked-in" condition whereby the
infrastructure framework can be difficult to change or modify in the future.
¦ For the most part, the treatment plants characteristic of centralized water and wastewater
systems are located away from urban centers. This was done to protect water supplies
from pollution by discharging treated wastewater downstream of the population to limit
the potential for waterborne disease. In addition, few people desire to live in the vicinity
of these plants. Compounding the issue, many older city centers have lost population and
industry to the extent to which having excess capacity to "sell." This current practice in
the development cycle results in a natural tendency to expand distribution and collection
pipe networks into the new areas of development because of relatively small capital cost
and leverage over the utilization efficiency of the centralized system. However, this
sprawling expansion occurs at the price of a potential increase in energy usage and a
decrease in environmental qualities. There is a limit to this expansion before the basic
system configuration and operational parameters require changes, often at a substantial
economic cost.
¦ Water infrastructure has as its primary service function to ensure compliance with
applicable SDWA and Clean Water Act (CWA) regulations. "Secondary" requirements
include providing adequate capacity and reliability to meet the urban service needs,
providing fire service, and controlling rates through managing capital and operational
costs. The system efficiency, energy consumption, and emissions are often lesser
priorities in master planning (U.S. EPA, 2015a).
¦ Although subject to the master development plans, urban water infrastructure is often
engineered independently from transportation infrastructure. The two may become
decoupled and uncoordinated. As a result, water infrastructure may not be adequate to
meet the service needs when transportation infrastructure and associated land use induce
further spatial shifts in population and business activities (Flanders et al., 2014). This
nature in planning may not only create conflicts with construction and service timing for
the two types of urban infrastructure, but also add greater complexity when changes and
adaption become necessary to support new urban functionalities in the future.
2.2 Transformation toward smart growth
Smart urban growth aims to achieve low-carbon and energy-efficient service-reliable
development through adaptive planning. Urban transformation is one approach to change the
existing urban form to a configuration of high urban density, walkable communities, and livable
environments (U.S. EPA, 2013a). This smart growth concept is now being incorporated by many
municipalities. In the national trends, smart growth often entails techniques such as infill, green
planning, and high-density residential developments, which has been increasingly applied
throughout the U.S. (U.S. EPA, 2013a, 2009b). Infill development and mixed transportation
30
-------
mode are demonstrated to improve system efficiencies and reduce transportation emissions. A
series of EPA reports have been published on smart growth applied to residential development
and its pertinent transportation and water infrastructure (e.g., U.S. EPA, 2006, 2009b, 2011,
2013a, and 2012a).
Transformation districts as a smart growth measure (Figure 2-6) can induce a transition
from a monocentric to a polycentric urban form that has higher efficiency. These districts are
planned with degrees of flexibility to evolve into polycentric, high-density, walkable
communities with ready access to mass transit. The planning process rests on the ability to
modify infrastructure to accommodate urban growth and population increase with the minimum
environmental impacts. For this purpose, the transformation districts are the necessary links for
the natural evolution toward very large polycentric urban centers. For example, multi-mode
transportation systems and multiple water supply or wastewater management districts are
common examples for very large urban centers. These urban features are characteristic of smart
urban growth principles. They can be initiated by adaptive urban planning as a part of the long-
term master planning process.
The transition from monocentric to polycentric forms, when realized in practice, has
significant implications for water infrastructure planning, engineering, and operation. In the
polycentric urban development, highly urbanized centers of impermeable surfaces are scattered
among and surrounded by undeveloped natural or low-impact developments (See Figure 2-6 and
related discussions in Section 1.1.3). Smart urban development, through measures such as
compact neighborhood design and the use of infill and green infrastructure, can reduce water
demand, improve water availability and water quality, and provide reliable water services at a
reduced cost (U.S. EPA, 2006). The use of green infrastructure is emphasized in the management
of combined sewer overflows (CSO) for many cities in the U.S. Midwest, East, and Northeast.
With adaptive planning, several possibilities are potentially achievable through the design and
implementation of a polycentric urban form. For example, the polycentric distribution of urban
populations and activities may allow for developing the decentralized or satellite systems for
water supply and wastewater management. Decentralized water management shortens the urban
water cycle, by increasing water recycling and onsite water infiltration, thus making it possible to
increase wastewater reclamation, nutrient recovery, and potential energy harvesting (see Luthy,
2013; Lee et al., 2013). In addition, the high-density housing development in the multiple centers
yields a smaller carbon footprint per capita (ADB, 2012), and may facilitate the development of
mass transit systems to connect the new urban centers. Examples include Washington DC, San
Francisco, and New York City. The resulting higher urban efficiency and lower carbon footprint
per capita in these urban centers can be observed from Figure 2-13 in Section 2.1.2.
2.3. Monitoring and re-evaluation
Adaptive planning helps examine viable urban development options against a set of
adaptation objectives. Such analysis aims to evaluate the capacity and efficiency of existing
transportation and water infrastructure, identify future improvement options, and compare their
benefits against a set of planning objectives, mostly through model simulations. Major planning
activities may include:
¦ Population and land use planning and future projections
¦ Transportation analysis and planning, including air quality analysis
31
-------
¦ Water infrastructure analysis and planning to either assist or limit transportation
development in urban development scenarios
Adaptive urban planning can be readily incorporated in the conventional master planning
process. The current urban planning (Figure 2-4) evaluates water and transportation
infrastructure conditions, and defines infrastructure improvements, mostly by increments,
between two adjacent master planning periods. Often the transportation and water infrastructure
development are uncoordinated, producing a condition that could potentially hamper future water
service optimization. To avoid this undesired consequence, adaptive planning uses an iterative
process and integrates the planning, engineering, outcome assessment, and re-planning through
scenario simulations (see Figure 2-6). It first evaluates urban efficiency against the evaluation
criteria, such as energy consumption, urban efficiency, and compliances. Then it gradually, and
systematically shifts the development paradigm toward those favoring smart growth.
The change in the development path by adaptation takes place at two endpoints. At one
end, the adaptation weighs into the readjustment of developmental goals, local water, and land
use policies. This adjustment varies among locations and individual cities, because of potentially
different constraints in local environmental and socioeconomic conditions. At the other end,
adaptive planning is focused on urban form and infrastructure itself. Urban growth is adaptively
planned to change the paradigm from urban sprawl to the high-density, low-carbon, and high-
efficiency urban form (Figure 2-6). One example of such adaptive planning is to expand
wastewater management service through a combination of gray and green water infrastructure,
often in decentralized management, for increased water harvesting and overland runoff
reduction. This report does not cover the adaptation approach through adjustment of the
development goals, but instead focuses on adaptive planning for physical systems.
In addition to the mandatory environmental standards, water and energy/carbon footprints
are two indices of urban efficiency that can be used in evaluating urban adaptations. The two
non-parameterized orthogonal indices can be used to quantify the water and energy tradeoffs at a
systems scale. This evaluation matrix can be used to compare developmental options. Published
studies are mostly based on simple water or energy usage and for analysis of a single industry or
single service sectors such as a municipal drinking water supply or a transportation system.
Nevertheless, these previous studies provided insight into the water-energy interactions in energy
production (Cooley et al., 2011; Rothausen and Conway, 2011; Zhou et al., 2013; Azadi et al.,
2013; Dodder, 2014; Dodder et al., 2011; Webster et al., 2013; Chang et al., 2012; and Ibrahim et
al., 2008) and in urban planning and operation (Perrone et al., 2011; Hering et al., 2013; Yang et
al., 2013; Wang et al., 2013; Kenworthy, 2006; Novotny, 2013; Lee et al., 2013). More attention
is now made toward both indices (water and energy/carbon footprints) and their relative
importance for a given system, infrastructure asset or industrial sector; for example, to the
planning of energy biomass and hydropower production in the water-stressed U.S. west (PNNL,
2012; Yang and Goodrich, 2014).
3. SUD Methods and Tool in Adaptive Urban Planning
Much of the discussion to this point has been focused on the water sector and how the
change in urban form might impact it. The limited discussion has concerned the fact that in urban
planning, energy usage and air emissions are intimately related to the construction and operation
of water infrastructure. This interrelationship can be dissected in many ways. For example, as
32
-------
noted in Section 1.1 (especially 1.1.2), transportation and water infrastructure are closely related
to the urban form and the potential development mode. In return, an urban form defines
population and economic activities, and thus can significantly affect the energy use of water
supply in both infrastructure construction and operation.
The other principle effect of urban development is found in spatiotemporal variations of
water demand, because of the demand distribution and the unique UHI effect. The concept of the
UHI is that urban gray infrastructure (such as pavements, buildings, and concrete structures)
creates a greater amount of reflective heat than undeveloped vegetated areas. Hence as an area
develops, the man-made gray structure absorbs and emits heat in a day cycle, causing higher
ambient temperature and stagnant air flow in urban centers especially during night time. The
degree of UHI effect and temperature variation depends on land use and land covers, local
topography, and ultimately the urban form. Such conclusions were made by several studies on a
detailed thermal mapping of the UHI effects (e.g., Liu et al., 2012; Buyadi et al., 2013; Weng et
al., 2004). In later Section 4.1, a case study in the Cincinnati metropolitan region shows how the
urban form, defined by highways and roads, can lead to the occurrence of UHI, its magnitude
and spatial distributions. Overall, the urban form has an impact on traffic emission, ambient air
quality, the distribution of population and business activities, water demand, and water services.
These combined effects are the basis for adaptive urban planning.
3.1 AUP&ET principles and utilities
The adaptive urban planning and engineering tool (AUP&ET) is developed to assist the
analysis of urban development options. This integrated simulation tool is intended for planners to
simulate urban transportation performance (e.g., travel delay, air emission) in adaptation
scenarios. The integrated modeling framework for AUP&ET is shown in Figure 2-9. Overall, the
AUP&ET tool consists of three major modules: land use projection, urban-scale transportation
modeling, and water infrastructure modeling. In this framework, the urban variables refer to
physical attributes such as topography, environmental conditions, and natural resources. The
urban developmental goals and growth factors, along with the impacts of climate variations, are
collectively represented in the scenario attributes. Future developmental and environmental
conditions can be defined in terms of the probabilistic occurrence.
In this section, the AUP&ET module for transportation infrastructure, its simulation
methods, functionality, and applications in the Cincinnati metropolitan area, are described.
3.1.1. Land use projection — CA-Markov model and ICL US
Land use projection is one of the primary bases for scenario-based planning (Figure 2-9).
In subsequent publications, methods and examples of future land use projections will be
discussed in detail for rural, suburban, and urban watersheds. As discussed in Section 1.3.1., the
ICLUS land use database can be used when the analysis is based on county-level spatial
resolutions. Many urban adaptive planning exercises, however, require land use and employment
projections in finer spatial resolutions, typically at census block levels. Thus, the CA-Markov
modeling technique is incorporated as the default in the AUP&ET module.
A land use model predicts target year land use according to the base year data (land use,
demographic, and socioeconomic factors) and develops viable land use scenarios involving
demographic and socioeconomic changes anticipated for the target year. In a CA-Markov
33
-------
analysis, the CA model is combined with Markov Chain analysis, incorporating the Multiple
Criteria Evaluation (MCE). The Markov model is based on the formation of the Markov random
process for the prediction and optimal control theory (Jiang et al., 2009). The calculation is a
multifaceted cross-tabulation between a pair of land use images from two times of different
historical observations. Future change probabilities are derived from observed change patterns
(Eastman, 2009). Markov modeling predicts each land use transition area for a future year using
the transition probability derived from two different historical land use data sets (Sang et al.,
2011; Eastman, 2009).
Geographic proximity, also known as spatial autocorrelation, assumes that adjacent areas
tend to be similar in land use in a gradual land use change. In a natural environment, similar soil
characteristics, terrain, weather, and vegetation are usually found within a defined region. The
impacts of all these factors are evaluated according to the factor's relative importance or weights
(Rao, 2005). MCE as a multi-attribute decision-making tool is incorporated in CA-Markov
modeling to provide land suitability analysis with the support of GIS (Fu et al., 2018). Overall,
the CA-Markov model allocates land use under the objective that was produced by Markov
Chain analysis according to terrestrial suitability produced by MCE. It takes original land use
and its neighborhood land suitability into consideration (Feng et al., 2011; Eastman, 2009).
These basic principles and applications can be found in land use projection literature (Tong et al.,
2012; Sun et al., 2013; U.S. EPA, 2010b, and references therein).
Figure 2-14 shows a framework of urban land use projection used in the AUP&ET. There
are three properties essential to calculate the transition probability: (1) past trends, (2) geographic
proximity, and (3) spatial dependency.
Past trends are land use changes observed during a previous period. They can be
measured by comparing land uses in the initial year and the base year. The elements of
Figure 2-14 Simulation block diagram for CA-Markov based urban land projections.
34
-------
multicriteria evaluation using different criteria weights are converted numerically into a
sustainability score (Figure 2-14) that can be analyzed spatially. Furthermore, geographic
proximity and local developmental drivers are necessary considerations in any land use analysis.
For example, population density and land value are similar within a defined geographic unit (i.e.
neighborhoods, cities), but significantly differ among such units. Spatial dependency may restrict
or promote future land use changes. Although spatial dependency factors may vary by location
and types of land use, they are derived from four major categories: (1) population density, (2)
accessibility, (3) administrative restrictions, and (4) physical limitations.
The land use projection in AUP&ET consists of three major steps (Figure 2-14) and uses
four major modules in AUP&ET: (1) Markov Chain Analysis, (2) Criteria Weights Calculation,
(3) Multi-Criteria Evaluation, and (4) CA-Markov simulation. In CA-Markov modeling, the base
year land use image is taken as the model input from which changes are projected. The modeling
further considers transition area objective, as produced by Markov analysis, and a collection of
suitability images that express the suitability of a pixel for each of the land use types from MCE
criteria. Then the modeling begins with an iterative process of reallocating land use until it meets
the area totals predicted by the Markov analysis. The modeling process and underlying principles
are as follows (Eastman, 2009).
¦ The total number of iterations is based on the number of time steps, namely the projection
time frame. For example, if the projection is for 10 years into the future, the time steps
might be chosen to complete the model simulation in 10 steps. The time step is chosen to
strike a balance between model precision and computation time. It also needs to be
appropriate for the rate of urban development in the past and, potentially, in the future.
¦ Every land use type in model iteration typically will lose some of its lands to one or more
of the other classes. It may also gain land area from others. For each modeling iteration,
claimant classes select the land from the host according to the suitability map for that
class.
¦ The CA component arises, in part, from the iterative process of land allocation. It also
results, in part, from a filtering stage with each iteration that reduces the possibility of
unsuitable changes. The net result of this iterative process is that the land use changes
occur in response to the growth in the areas of high suitability spatially proximate to
existing areas.
3.1.2. Calibration and validation of the land use simulation model
Model calibration is important for the projection of urban land use because of its dynamic
evolution with time. Calibration aims at obtaining values of the transition rule parameters that
enable the most accurate reproduction of the past evolution in land uses. There are two
traditional methods to calibrate CA-based models: (1) methods based on trial and error, and (2)
methods based on statistical techniques.
The first category does not require a set of strict mathematical formulas. It assesses the
results obtained from alternative combinations of parameter values (Ward et al., 2000) and the
sequential multistage optimization by an automated exploration of combinations of parameters
(Silva and Clarke, 2002). For the second category, the most frequent statistical method is logistic
regression that provides the weights of the variables involved. However, the statistical equations
35
-------
might not reflect the actual relationships or explain the underlying mechanisms (Sante et al.,
2012). The first method is used in the case study in Cincinnati (see Section 4.2).
The general validation method consists of the visual comparison of model results and
observed data in a historical period/point of time. The method usually is complemented by
quantitative methods that evaluate overall accuracy. For the accuracy measurement, the most
frequent metrics in increasing order of complexity are (i) ratio of simulated to the real number of
cells (or clusters) for given land use, (ii) overall accuracy measured by the percentage of
correctly classified pixels, (iii) regression analysis between simulation results and real data, and
(iv) a coincidence matrix and the Kappa index (Sante et al., 2012, and references therein).
Because the method based on trial and error is applied in the calibration process, overall
accuracy and the Kappa index are popular measurements in comparing simulated land use with
reference land use. Therefore, the overall accuracy and the Kappa index were adopted for land
use calibration and validation in the AUP&ET simulations.
3.1.3. AIR-SUSTAIN system for transportation simulation
The other major AUP&ET component is for urban transportation planning in adaptation
(Figure 2-9). This scenario-based adaptive planning has basic objectives for high transportation
efficiency and reduced air emission, energy usage, and carbon footprints under the current and
future land use scenarios. The land use types and spatial relations are the basis for defining the
population, employment, and urban activity distributions in transportation modeling (see Section
3.1.1 above).
The scenario analysis for transportation planning is hosted within a newly developed
simulation tool, "Air Impact Relating Scenario-Based Urban Setting and Transportation Asset in
Network" or AIR-SUSTAIN (Yao et al., 2014). Figure 2-15 shows the program's architectural
structure. The current version consists of three application modules: (1) scenario development,
(2) regional level analysis, and (3) project-level analysis. The scenario development module is
built upon the base-year land use, demographic and socioeconomic factors, and transportation
infrastructure data. It further considers the assumed changes in the demographic and
socioeconomic factors for a target year. The target year land use is projected by the CA-Markov
land use model. For regional level analysis at county-resolution, the ICLUS model (U.S. EPA,
2010b) also can be used. Land use projections are described in the preceding subsection.
In general sequence, the AIR-SUSTAIN's regional level analysis can be used to assess
the impacts of a growth scenario on transportation system performance at urban scales (Figure 2-
15). Here the traffic projection results are used to assist in identifying the traffic congestion area
of the road links, where the transportation efficiency deteriorates in traffic flow and CO2
emission. It is noted here that the traffic congestion analysis in Figure 2-15 is different from the
EPA's regulatory conformity hot-spot analysis.
Once the traffic-congested areas are identified, a project-level analysis can be conducted
to identify the most appropriate traffic control measures and other engineering solutions (Figure
2-15). The analysis is centered on options to improve transportation performance and reduce on-
road traffic emissions. Together, technical results on transportation performance at both the
regional level and project levels enable users to quantify environmental benefits in an urban
36
-------
planning scenario. Again, the tool and models are not designed for regulatory compliance
analysis. For the latter, EPA's transportation conformity regulation (40CFR Part 93)2 applies.
The AIR-SUSTAIN modeling and analysis start with the computation of target-year
demographic and socioeconomic distributions based on projected land use. The modeling basis
and techniques were described in preceding Sections 3.1.1 and 3.1.2. Using the user-specified
growth rates, a linkage model in AIR-SUSTAIN populates the spatial distribution of the future
land use changes within an urban area of analysis. The linkage model projection includes future
population, employment, university enrollment, and high school enrollment in each traffic
analysis zone (TAZ) in a target year. The principal variables in the regional-level analysis for
travel demand forecasting include employment, student enrollment, etc. Subsequently, traffic
emissions based on the travel demand forecasting (TDF) outputs are estimated using the EPA
program MOVES. Section 3.2 further describes data flows and model simulation using the AIR-
Figure 2-15 AIR-SUSTAIN modeling framework for transportation analysis of efficiency and carbon
dioxide emission in urban infrastructure adaptation. Abbreviation: SE - socioeconomic
2 EPA has guidance for transportation conformity available on its web site: https://www.epa.gov/state-and-local-
transportation/current-law-regulations-and-guidance-state-and-local-transportation.
37
-------
SUSTAIN tool. Section 4.1 presents a real-world case study in the Cincinnati metropolitan
region for illustration.
3.1.4. The linkage to water infrastructure simulations
The adaptive planning framework within SUD also contains water infrastructure modules
for drinking water supply, wastewater and stormwater management (see Figure 2-9). Because
water infrastructure normally follows the population and land use changes, its planning and
design are assumed to be sequential after transportation infrastructure evaluation.
3.2 The AIR-SUSTAIN simulation tool for transportation
Following the SUD overview in Section 3.1, this section details the functions of the
transportation simulation tool AIR-SUSTAIN. AIR-SUSTAIN is a software interface developed
in this research to integrate land use projection, traffic simulations and optimization. The purpose
is to evaluate development scenarios of land use and transportation by modeling and analyzing
CO2 emissions over a transportation network, fuel usage, and transportation performance in
terms of excess travel time, traffic congestion, etc. The tool utilizes a GIS platform to provide the
urban-wide spatial information on model projections of land use, employment, residential
development, travel demand, and automobile-based travel conditions. Other environmental
performance criteria in the modeling include fuel consumption and total carbon emissions.
AIR-SUSTAIN software contains functions in data flows and linkages among the model
components (Figure 2-15). In Appendix A, details of the program structure and model input and
output are described. Major modeling components include those below:
¦ Linkage model. The linkage model combines the land use model output, the target year
population and employment projections, and base year population and employment data.
This prepares the target year population and employment for each TAZ as the model
inputs for traffic simulation.
¦ Travel demand forecasting (TDF) model using VISUM software. VISUM is
comprehensive flexible software widely used worldwide for metropolitan, regional, state,
and national planning applications. The TDF model simulates the link (i.e., roadway
segment) traffic volume and speed. Simulation results then are used as inputs for the
traffic-related emission estimation in AIR-SUSTAIN.
¦ Microscopic traffic simulation model using VISSIM software. The commercial traffic
analysis software enables the analysis of traffic measures designed to improve traffic
capacity. It also is used with AIR-SUSTAIN to evaluate engineering options to reduce
carbon and pollutant emissions.
¦ Automobile vehicular emissions calculation using EPA's regulatory model MOVES3
(U.S. EPA, 2010a; 2015c) for different scenarios of urban-wide transportation or specific
traffic measures in adaptation.
3 https://www.epa.gov/moves/latest-version-motor-vehicle-emission-simulator-moves
38
-------
3.2.1. Basic functions and interfaces of AIR-SUSTAIN
The AIR-SUSTAIN software interface integrates transportation and land use models for
scenario analysis. The use of scenario-based planning analysis helps assess sensitive interactions
among travel demand, the impact of transportation activities on-road emissions, and urban
development policies. The quantitative analysis is executed in the AIR-SUSTAIN software
through interfaces embedded in a GIS environment (Figure 2-16). The main functions and
interfaces of the AIR-SUSTAIN include Scenario Information Specification, Scenario
Development, Regional Level Analysis, Project Level Analysis, and Results Comparison.
¦ Scenario Development
Scenario development for transportation adaptation is set by importing base year data and
developing demographic and socioeconomic attributes of a scenario. The base year demographic
and socioeconomic data are imported with the feature class TAZ. The target year demographic
and socioeconomic data are contained in the feature class TargetYearTAZ of the computer
program (its format is shown in Appendix A's Tables A1 -1 to 1-4). It is computed based on the
assumed demographic and socioeconomic changes, target year land use projection, and base year
demographic and socioeconomic data.
Among those datasets, the demographic and socioeconomic changes are projected, and
often assumed, for a future scenario. They depend on urban development policies or objectives.
This group of data is specified by the user through the functions in the Scenario Development
software module. The target year land use data are projected by the land use model embedded in
the AIR-SUSTAIN model. The data inputs, specification of demographic and socioeconomic
factors, and model execution are implemented in the Scenario Development module.
Scenario | Data
New Scenario
Load Scenario
Scenario-based Urban"Settings and Transportation Assets In Network
Scenario Design Regional Level .Analysis Project Level Analysis Results Comparision
Modelina Year Selection
Figure 2-16. AIR-SUSTAIN graphic interface for scenario modeling and analysis.
Modeling Year
¦ Regional Level Analysis
The Regional Level Analysis module is used to estimate travel demand and on-road
emissions for the base and target year in a project area. It should be noted that this traffic
analysis is different from the regional emissions analysis done for transportation conformity.
Overall, AlR-SUSTAIN's Regional Level Analysis consists of two major elements: (1) Travel
Demand Forecasting, and (2) Emission Estimation. When performing the regional level analysis,
a TDF model first simulates trips on roadway links for the entire area of concern based on
demographic and social-economic data, as well as transportation infrastructure, such as road
network, parks, water bodies, TAZs, etc. Subsequently, the forecasted traffic data are used to
generate inputs for traffic emissions modeling to estimate vehicle emissions for each road link.
39
-------
In the emission analysis, CO2 equivalent (CO2.1), criteria pollutants, and energy consumption are
estimated by using the EPA's MOVES model (U.S. EPA, 2010a; 2015c).
¦ Project Level Analysis
In the Project Level Analysis, the traffic congestion links are identified from the regional
level analysis results. As aforementioned, the traffic congestion analysis is different from
assessing whether a project needs a transportation conformity hot-spot analysis. The microscopic
traffic simulation model VIS SUM is used to estimate the traffic flow operations on the
congestion links under alternative traffic control measures. The assumed traffic control measures
then can be assessed in terms of traffic operation performance and their influence on the on-road
traffic emission rates. Emission rates for each congestion link are calculated using the MOVES
model in the project-level analysis module.
¦ Results Comparison
After performing the scenario design, regional level analysis, and project level analysis,
the AIR-SUSTAIN simulation outputs from the base year and target year can be compared and
visualized in ArcGIS by the Results Comparison tab (Figure 2-17). In the subsequent sections,
the AIR-SUSTAIN tool is described on its data structure; model linkage; travel demand; traffic
r^ii b 11 b~i
Scenario Data
Air Impact Relating Scenario-based Urban Settings and Transportation Assets In Network
Scenario Development Regional Level Analysis Project Level Analysis Results Comparision
Land Use
Land Use
View Res
Demographic and Socioeconomic Results
Demographic and Socioeconomic Data
Travel Demand Forecasting Results
Travel Demand Forecasting Data
Emission Estimation Results
Emission Estimation
View
O Data of modeling years
O Changes between modeling years
O Data of modeling years
O Changes between modeling years
O Data of modeling years
O Changes between modeling years
Modeling Year
yyyy
Process Status
O Data of modeling years
View
O Changes between modeling years
Hotspo
Figure 2-17 The Results Comparison module interface in the AIR-SUSTAIN tool.
40
-------
congestion identification; the microscopic simulation of adaptation options; and, lastly, the
emissions and energy consumption estimation.
3.2.2. Travel demandforecasting - VISUM
Travel demand forecasting (TDF) is essential in the traffic analysis that links
transportation to land use and socioeconomic factors at a regional level. In the AIR-SUSTAIN
tool, the TDF model is used to forecast travel demand for the base year and target year. This
projection is primarily dependent on the settings of land use based on the socioeconomic
datasets. TDF model outputs include the link (i.e., roadway segment) traffic volume and speed,
which further can be used as inputs for the traffic-related emission estimation.
3.2.2.1. Modeling
Travel demand analysis was first developed in the late 1950s for highway planning using
a four-step model. This model, using the conventional trip-based approach, is a primary tool for
modeling future travel demand and performance of a regional transportation system. The AIR-
SUSTAIN tool adopts this traditional four-step model for the travel demand forecasting. The
demand forecasting involves these four basic steps:
¦ Trip generation. This is the process in estimating the number of person-trips that will
begin from or end in each TAZ within the region on a typical day. The traditional trip-
based approach considers each trip as the unit of analysis. When an individual makes a
series of trips, each trip is treated as a separate, independent travel event (McNally, 1996,
2007).
¦ Trip distribution. This modeling process allocates the trips generated in one zone to other
zones.
¦ Mode split. It estimates modal percentages of the travel according to the time and cost
characteristics of various competing modes based on demographic and socioeconomic
characteristics of the urban residents.
¦ Traffic assignment. This last step in travel demand forecasting assigns trips to the
transportation network.
Notably, the four-step models rely on average transportation behavior between and
within traffic zones. More sophisticated activity-based models attempt to represent focus groups
of populations. This approach considers underlying travel behavior (Jones et al., 1990), and thus
explicitly recognizes and addresses the limitations of the conventional trip-based approach. The
model analysis is shifted from rough aggregates to the level of the individual traveler (Zmud et
al., 2014). This recent development makes it possible to incorporate detailed demographic data.
However, considering the data needs in modeling, the current VISUM software in AIR-
SUSTAIN is based on the four-step model approach. The same approach is used in other traffic
modeling packages such as Cube and TransCAD.
3.2.2.2. Model calibration and validation
Model calibration and validation are fundamental to travel demand forecasting. Model
calibration and validation data may include:
41
-------
¦ reliable estimates of base-year TAZ household characteristics and employment
information,
¦ an accurate representation of the base year highway (and transit, if any) network, and
¦ a reliable base-year travel survey or monitored traffic data based on main permanent
stations.
Model calibration and validation can proceed after the model parameters are estimated in
the AIR-SUSTAIN simulation. Model calibration enables model parameters to be adjusted until
the predicted travel matches the surveyed travel (e.g., origin-destination [O-D] survey data)
across the region for the base year. The model calibration assumes that these calibrated
parameters will remain constant overtime (Pedersen and Sandahl, 1982). Furthermore, model
validation tests the model predictability of the future. In many areas, traffic counts commonly are
used for model validation. Validation requires comparing the TDF model predictions on specific
roadways with the traffic counts data (e.g., Annual Average Daily Traffic [AADT]) that occurred
on the same roadways in a validation period.
TDF calibration and validation is based on the Mean Absolute Percentage Error (MAPE)
in Eq. 2.2. In the travel demand model, parameters, such as the utility functions' parameters and
network capacity are adjusted to calibrate the model.
1 T
mape = —Y
rji
t tr
Mabs{t)~Mstn{t)
Mabs(t)
(2.2)
where Mabs(t) and Mstn(t) are the field measured time-series values and the simulated time-series
values during a period of time t respectively. In the AIR-SUSTAIN, MAPE is calculated for the
measured and simulated traffic count and model volume. The MAPE ranges for total error by
functional classification (the type of road) are set by FHWA (1990):
Freeway <7%
Expressway <10%
Arterial <15%
Collector <25%
Frontage/Ramps <25%
3.2.3. Assistance in traffic congestion identification
AIR-SUSTAIN in the SUD tools has the technical capability in analyzing traffic
conditions and identifying traffic congestion areas for given land use and urban developmental
scenario. It uses MOVES2014 (EPA, 2015c, 2010a) to calculate pollutant emission for the
identified traffic congestion areas in need of traffic management in adaptation. This analysis
provides technical data such as link traffic volume, speed, and emission estimate. The traffic
congestion analysis in AIR-SUSTAIN is not equivalent to and not applicable to regulatory
analysis for transportation conformity.
Using AIR-SUSTAIN, the locations of traffic congestion in future urban development
options can be identified from the land use and traffic simulations. Theoretical basis and
application examples of traffic and air emission modeling are given in this section and later in
42
-------
Section 4.2. Details of the modeling program steps are also provided in Appendix A. In
summary, the AIR-SUSTAIN simulation program identifies all corridors for the traffic
congestion area analysis according to the regional-level analysis results. Their link information is
saved in the database and marked in ArcGIS for the microscopic analysis in the subsequent
modeling step.
3.2.4. Microscopic simulation using VISSIM
The AIR-SUSTAIN simulation identifies links of traffic congestion and high emission
rates based on the regional-level analysis results. Identified locations can be further modeled
using VISSIM software for evaluating traffic measures on highway corridors, local arterial roads,
and other road segments. VISSIM was developed at the University of Karlsruhe in Germany and
is distributed by PTV Transworld AG for microscopic simulation at a higher spatial resolution.
The model enables analysis of the traffic flow by modeling each entity (car, train, or person)
within a traffic stream and of the interaction between drivers (Barcelo et al., 2005). This
modeling capability makes it possible to simulate the traffic control and management systems at
all levels, from traffic control platform to individual traffic controllers (Gettman and Head,
2003). This type of analysis facilitates urban adaptation planning and engineering down to a
local project level. The current version of the AIR-SUSTAIN tool has this microscopic
simulation capability added by emission calculations using EPA's MOVES software. Appendix
A provides the details of principles and modeling steps, including input data structure, scenario
determination, model calibration and validation. Both high-resolution traffic results and
evaluation results can be input values in MySQL and Geodatabase available in AIR-SUSTAIN.
Possible simulation results include the following:
¦ high-resolution traffic condition at the link, including second-by-second speed and
acceleration. Such results can be used as the inputs for emission calculation, for example,
using MOVES at the project level.
¦ evaluation results on average speed, delay, and queue length of each link. This modeling
output can be used to compare the differences and effectiveness of possible transportation
control measures.
3.2.5. Emission estimation using MO VES
AIR-SUSTAIN incorporates EPA's MOVES2014 as the energy and emission analysis
tool. In 2010, the MOVES model and software were released by the U.S. EPA (U.S. EPA.,
2010a) for estimating air pollution emissions from on-road mobile sources. U.S. EPA (2015c)
released the updated MOVES2014 program. At the time of development, the AIR-SUSTAIN
based on MOVES2014 uses traffic data from the regional level and project-level simulations to
estimate the emissions factors. This model integration makes it possible for users to evaluate air
quality, carbon emission, and energy consumption for competing for urban adaptation scenarios.
In emission calculation, vehicle activity inputs can be defined at three levels of data
requirement and precision: (1) average speed, (2) drive cycle, and (3) operating mode
distribution. Each is associated with different levels of model accuracy (Figure 2-18). Average
speed is a basic parameter in traffic operation. It is calculated from AIR-SUSTAIN's TDF
model. Using the average speed as the traffic input, MOVES modeling in AIR-SUSTAIN selects
the operating mode distribution based on vehicle characteristics. On the other hand, the second
43
-------
option Drive Cycle is a
second-by-second description
of vehicle activity over time.
Such data are usually
collected by using a GPS-
equipped probe car. It is
assumed that every on-road
vehicle is following the same
trajectory of the probe car.
This option can better
represent the traffic operation
than average speed but
requires extensive data
collection.
•C
s
©
o
e
Oh
Low
Operating Mode Distribution
Driving Cycle
Average Speed
Data Requirement
High
Figure 2-18
General relationship between model precision and
data requirements in traffic modeling.
At the highest level,
the operating mode
distribution method takes a
different approach. It
assumes a fraction of vehicle operation mode bins based on its instantaneous operating mode
distribution that is determined by vehicle specific power (VSP) and speed. This method describes
the entire vehicle population's operation in the study area and provides the highest individual
vehicle level in data resolution. It is worth noting that the MOVES program internally converts
all the average speed and drive cycle inputs into the operating mode distribution for the use of
the MOVES emission rate database.
A general comparison of the relative model accuracy and data requirement for the three
methods are schematically illustrated in Figure 2-18. Considering traffic data availability in
general applications, the AIR-SUSTAIN tool uses the average speed option for regional level
analysis and the operating mode distribution for project-level analysis. This technical approach is
illustrated schematically in Figure 2-19. Traffic inputs for the regional level analysis are
extracted from TDF model outputs. In addition to the average speed, other traffic inputs include
link traffic volume and vehicle composition. Traffic inputs for the project-level analysis are
generated by the microscopic simulation model. They consist of traffic volumes, link average
speed, operating mode distribution, and vehicle composition. Details of model inputs, governing
equations, and model simulations for traffic flow, vehicle compositions, and operating modes are
contained in Appendix A.
Commonly used parameters for model inputs include emission source type, road type,
vehicle age distribution, and operating mode:
Emission Source Type
The source type in MOVES is a combination of vehicle type and how the vehicle is used.
For example, long-haul and short-haul trucks tend to be very similar in size and design, but the
way they are used defines their source use type in the emission category. Table A2-5 in
Appendix A shows the source types, descriptions, and their equivalents as defined by the
Highway Performance Monitoring System (HPMS).
44
-------
Figure 2-19 Modeling framework for emission estimation using both the macroscale VISUM and
microscopic VISSUM traffic simulation models.
Road Type
The list of Road Types is contained in the MOVES database. The default database has
Road Types that represent urban and rural driving on roads with restricted and unrestricted
vehicle access. Restricted access road types usually are used to model freeways and interstates;
ramps are considered part of restricted access road types. In the modeling program, the Ramp
Fraction tab of the County Data Manager only will become available if an unrestricted road type
(i.e. 2 or 4) is selected. Table A2-6 in Appendix A shows the MOVES road type.
Vehicle Age Distribution
The MOVES model uses vehicle age information, groups the vehicle-specific power
(VSP) for light-duty vehicles and the scaled tractive power for heavy-duty vehicles into the age
groups. Table A2-7 in Appendix A shows the age categories used in the MOVES model.
Operating Mode
The operating mode bins are predefined in the MOVES model, as shown in Table A2-8
in Appendix A. Each operating mode, categorized by vehicle source type, road type, and vehicle
age group, is assigned an emission rate that is determined previously in the MOVES database.
4. Adaptive Urban Planning in SUD Case Studies
Urban-scale adaptation, as described in Section 2.0, aims to develop sustainable
infrastructure under current and future hydroclimatic and land use conditions. Adaptation co-
benefits can be achieved simultaneously for increased water infrastructure resilience and
emission reduction. The two benefits do not conflict with each other but are achievable through
45
-------
adaptive planning in urban infrastructure development. The sequence for such planning analysis
is shown in Figure 2-6. To illustrate urban adaptation and considerations, two SUD application
case studies in Cincinnati, Ohio, and Manatee County, Florida, are described in Section 4.0. The
case studies are drawn from the existing publications of this research (Liang and Keener, 2015;
Liang et al., 2013; Yao et al., 2014; Wang et al., 2013; Yang et al., 2013; Chang et al., 2012).
The case study in Cincinnati shows the relationships among land use, population,
transportation under the present urban form, and possible future adaptation scenarios in
development. The scenarios are compared using urban efficiency parameters, including fuel and
energy consumption, emission and air quality, the UHI effects, and commuting times. In the
second case study, this type of scenario analysis for water systems is showcased in master
planning for Manatee County.
4.1 Urban form and urban infrastructure
4.1.1. Urban form and land use patterns
Cincinnati metropolitan area hosts approximately 2.1 million people in 15 counties over
7350 km2 of land on the banks of the Ohio River (Figure 2-20). Traffic patterns in the region are
characteristic of a monocentric urban framework centered on downtown Cincinnati. Rolling hills
with limited topographic relief follow the Ohio River and the NNE-SSW oriented Mill Creek.
The surrounding suburban area is dominated by flat to moderately hilly farm lands and forests to
the east and south. Valleys along the Ohio River and the Mill Creek are about 100 m lower in
elevation.
The north-south trending, narrow high-density urbanized zone with heavy surface
pavement was delineated from a U.S. Geological Survey urban land use map and by
interpretation of a Google® satellite map dated 2013. The high-density zone is shown in Figure 2-
20; O-O' marks its long axis. Inside the zone, the land use and cover are characterized by a large
fraction of surface pavements and roofs (Figure 2-21 a,b). Small patches of green lands and lawns
are interspersed among the man-made structures. Besides the continuous large area of the heavily
urbanized Mill Creek corridor, two small areas of high-density development appear in the
Western Hills area west of 1-75, and in the Blue Ash-Mason area along 1-71. These small and
isolated patches of high-density development are surrounded by residential development of
single houses and forest reserves.
Beyond the high-density urbanized urban center is a mixed zone of dispersed low-density
developments characteristic of detached residential houses and commercial areas separated by
lawns and tree zones. Typical development pattern is shown in Figures 2-21c,d. Figure 2-22
schematically shows the spatial transition from the high-density urban core, to the mixed zone
and, ultimately, to the exurban farm lands. In recent decades, urban development in the
Cincinnati metropolitan area has been concentrated in several areas or transformation districts
leading to the direction of a polycentric urban form.
¦ Significant urban development has occurred in the West Chester area and the Mason area
along north 1-75 and north 1-71, respectively. Substantial development also has been seen
in northern Kentucky along these two interstate highways. These developments sprouted
from the significant establishment of commercial activities, introducing satellite urban
centers through the process of ribbon and leapfrog sprawl in Figure 2-5.
46
-------
Legend
Urban land use boundary
Mason
High-density urban boundary
17-0016
Forest / preservation area
West Hi
61-0010
'61-8001
61-004$/*' 61-7001
25-0022
117-0007
'eater Cincinnati -
®rthem KY Int'l Airport
Aurora
Amelia
Erlanj
Figure 2-20 Major transportation traffic routes and the urban physical footprints of the Cincinnati metropolitan region. Also shown are
the locations of four radio sounding locations (S-1 to S-4) and 15 EPA's NAAQS air quality monitoring stations in filled
blue circles. The I-75 site location refers to the study area by Liang et al. (2013). Urban footprint and high-density
pavement areas are delineated from the 2007 USGS land use maps. Modified from Liang and Keener (2015).
47
-------
Figure 2-21 Different land use patterns in areas among the 12 EPA's NAAQS monitoring stations. (A)
high-density urban core of residual/commercial area at NAAQS station 061-0040; (B)
urban core of industrial/commercial area at station 061-0043; (C) low-density residual area
at station 037-0003; (D) greenness in urban perimeter at station 037-3002. Each photo's
long side is ~2.0 km. Maps obtained from GoogleMap™. From Liang and Keener (2015).
¦ Cincinnati downtown has been redeveloped over the past decades, with the increasing
development of high-density residential communities. The recent development of a street
car system solidifies the current development further into a walkable urban center.
The city's preference on infill development along 1-75 and 1-71, and continued
development in the northern Kentucky region, have led to significant transformation of the
commercial activities in the region. These developments further lead to a formation of the
polycentric form with implications in both transportation and water management.
In the following subsections, the urban form and the land use and cover types are
investigated for their relations to the urban climate, transportation demands, and the unique
atmospheric structure above the urban center. The most important property is the UHI formation
shown in ambient temperature (Ta) and thermal inversion in the urban boundary layer (Figure 2-
22). These unique environmental phenomena affect air quality as well as water consumption and
hydrology. Thus they are the constraints in urban adaptation.
4.1.2. Transportation and traffic distribution
Current transportation in the Cincinnati metropolitan area mostly relies on automobiles.
The limited bus-based mass transit and the recently constructed street car system that started
operation in 2016, provide mass transportation mostly limited in the downtown area. The major
48
-------
Figure 2-22 A schematic diagram of three-dimensional model for the urban form, traffic and
atmospheric structure in the Cincinnati metropolitan region. Ta-ambient temperature;
ABL - Atmospheric boundary layer; UBL - urban boundary layer; subscripts N and St
for neutral and very stable atmosphere, respectively, and Z is height. NAAQS stations
are indicated for their relative locations. From Liang (2014).
road network in the metropolitan region consists of interstate freeways and arterials (1-71,1-75,1-
74, and 1-275), collectors (SR-126, SR-129), and local roads. The road network connects north-
south high-density industrial-commercial zone to low-density residential and commercial
districts in the urban perimeters and exurban areas. The high-density zone is extended along the
Mill Creek valley with the automobile as the primary transportation means. The Ohio-Kentucky-
Indiana Regional Council of Governments (OKI) collected the 2009 traffic data and provided
traffic counts and composition of 20 traffic stations for this research.
Analysis of the year-2009 traffic data indicates strong diurnal and spatial variations of
daily traffic counts and traffic composition at interstate freeways and arterials (1-71,1-75,1-74,
and 1-275), collectors (SR-126, SR-129) and local roads. The traffic is generalized into the five
time periods of different traffic compositions in Table 2-4. Similar traffic diurnal variability
Table 2-4 Four daily periods of traffic compositions on the highway in Cincinnati, OH*
Period
Time
Traffic Composition
Night period
Morning rushing hours
Daytime period
Afternoon rushing hours
Evening period
11 pm - 6 am
6 am - 8 am
8 am - 3 pm
3 pm - 5 pm
5 pm - 11 pm
Diesel truck dominant
Gasoline car dominant
Mixture
Gasoline car dominant
Increasing diesel truck
Note: * - from Liang (2014).
49
-------
and spatial distributions were reproduced by Yao et al. (2014) in a detailed area-wide trip
generation and traffic volume modeling. For the analysis, hourly traffic profiles during
weekdays were constructed for each of the stations.
Figure 2-23 shows the year-2009 averaged traffic volume for passenger cars in
automobile class C1-C3, diesel trucks including single-unit trucks (C4-C7) and multi-unit
trucks (C8-C13) on the highways and local roads. Average weekly traffic compositions for
selected major monitoring stations are listed in Table 2-5. The highest traffic volume and large
variations occurred along 1-71 and 1-75. The average and standard deviation of weekday total
traffic volumes were 69485±26590 vehicles/day (N=13 stations), 34770±14180 vehicles /day
(N=3 stations), and 43452±22661 vehicles /day (N=3 stations), for the interstate freeways,
collectors, and local roads, respectively. The level of service is consistent with the field traffic
measurements Liang et al. (2013) reported for October 2010 during the 1-75 black carbon
dispersion studies.
In the Cincinnati area, most multi-unit truck traffic is concentrated along the interstate
highways. Traffic volume was 8159±4339 vehicles/day or approximately 10 times more than in
the collector and local roads. Truck volume above 13300 vehicles /day was measured at north
1-75 serving the industries and in I-75/I-71 after merger leaving Ohio into northern Kentucky of
mixed land use in perimeter and exurban. Representative land use examples are shown in
Figure 2-21.
4.1.3. Urban form and air quality
The transportation system described above, and current centralized water services
facilitate the formation of the present monocentric urban form in Cincinnati. The environmental
impacts of this urban planning are shown by ambient air quality variations through the
metropolitan region. From Liang (2014), the spatial correlation between air quality and the urban
form is evident:
¦ by the analysis of the decade-long measurements of PM2.5 for 13 U.S. EPA National
Ambient Air Quality Standards (NAAQS) monitoring stations, and
¦ by quantitative modeling of black carbon dispersion experiment for 10 days (October 6 to
October 15, 2010) at the roadside of northern highway 1-75.
The locations of the 10-day experimental study and the 15 NAAQS stations over
different land use types are shown in Figure 2-20. Station 17-061-0040 is in the high-density
urbanized zone at the center of the Cincinnati metropolitan area. It is used as the reference
station for analysis of spatial relationships among UHI effects, ambient air temperature and air
quality variations. The statistics of 10.5 years of ambient temperature and PM2.5 measurements at
the reference station 17-061-0040 is given in Table 2-6. The yearly temperature and PM2.5 means
have a range of standard deviation. Frequency distribution of PM2.5 concentration measurements
are asymmetric, with a bias toward small concentrations (Kurtosis =2.21 and Skewness>l; see
Table 2-6). In general, yearly minimum and maximum occurred in the winter and summer
season, respectively.
Despite the large seasonal variations, the daily temperatures and PM2.5 concentrations are
highly correlated among the 15 NAAQS stations. The correlations using Eq.2.1 were obtained
for four temperature parameters (daily maximum, daily minimum, daily average, and diurnal
50
-------
Daily Average Total Traffic Volume
• 0.000003 - SOS5.OXXX50
S 90S5.000001 -18870.000000
® 1SB70.0D«»1 -32135.00KW0
# 32135.0300)1 -56163.O0OOOO
0 £6163.000001 -120387.00XX»
Truck Traffic \folume
0.000000 - 1699.000000
1609.000001 -
5609.000001 -
Figure 2-23
Truck and passenger car
traffic volume distribution in
the Cincinnati metropolitan
region. Heavy truck traffic
concentrated in i-75, I-74 and
the confluence of 1-75/1-71
leading to Kentucky in the
south. Relatively, 1-71 has
greater car traffic. From
Liang (2014).
51
-------
Table 2-5 Locations and traffic flow in 2009 for selected locations in the Cincinnati road network (From Liang, 2014).
OKI Station
Traffic
Roads
Average Weekday Traffic'*
Location
Target
Cross-by
Auto
SU Truck
M J Truck
T Ottal
HAM3343
I-71 (E)
Kennedy Ave
96907
6371
6791
110039
1
WAR0422
1-71 (E)
SR-48
25164
3376
9989
38528
2
WAR0509
I-71 (E)
SR-123
32617
3833
9048
45498
3
WAR0471
I-75 (N)
Central Ave
62956
11445
13943
88344
4
WAR0548
¦-75 jN}
5R-63
54975
7265
11432
73671
5
BUT0475
1-75 (S>
Kylei Station Rd
61260
5878
11664
73802
6
BUT0670
!-75 (S)
Union Center Blvd
53124
4347
6245
63715
7
BUT0701
l-75{S)
Cincinnati-Dayton Rd
47391
2996
5776
56164
8
KEN0458
l-71/75(S)
Fifth St
91842
12444
16101
120386
9
HAV12246
l-74-275(W)
1-74
62234
4797
5303
72334
10
CLE0180
!-275{M)
5R-32
66355
2385
1964
71204
11
CLE0211
l-275(S)
5R-125
54706
2412
1868
58987
12
HAM 3 383
l-74(W)
New Haven Rd
18106
1576
5948
25631
13
HAM0970
SR-126{W)
1-75
51627
2386
580
54593
14
BUT0437
SR-129(W)
SR-747
32603
1901
876
35380
15
BUT0842
SR-129{W)
SR-4 Bypass
22117
1204
469
23790
16
BUT0479
SR-129{W)
1-75
23310
1237
771
25318
17
HAV12022
Norwood Lateral (E)
1-75
63692
4099
1510
89301
18
HAW381B
Lebanon Rd (US-42) (S)
Cottingham Dr
25565
1110
331
27006
19
HAM3408
Winton Rd (5)
Fleming Rd
32653
1115
279
34048
20
Note; * - Data from OKI.
«» - Vehicle types fol'ovving ODOT: Auto - 4-axial passenger cars; SU Truck - Single unit truck; and
MU truck - rru'ti-unit truck.
52
-------
Table 2-6 Statistics of daily temperature measurements at NQAAS Station 17-061-00040
Statistics
Tavg (°C)
Tmin (°C)
Tmax (°C)
AT (°C)
PM2.5 (mg/m3)
Mean
14.22
9.20
20.09
10.88
13.89
Standard Deviation
10.17
9.66
11.14
3.83
7.27
Kurtosis
-0.91
-0.84
-0.90
-0.51
2.21
Skewness
-0.35
-0.32
-0.35
-0.03
1.25
Minimum
-15.2
-20.4
-10.2
1.1
1.2
Maximum
34.3
27.4
42.1
22.8
52.1
Count
1661
1458
1458
1457
1717
Note: Raw data from the EPA NAAQS monitoring network.
temperature range) and PM2.5 concentrations. Compared to the reference station 17-061-0040
(Cref), temperature and PM2.5 measurement data of other stations (C,) are correlated by:
Ci = aiCref+ei (2.1)
The obtained slope («,) and intercept (e, ) are statistically significant (see Liang, 2014)
with a large correlation coefficient (R2~0.99). Departure from 1:1 relationship indicates
atmospheric differences among stations rather than measurement errors.
4.1.4. Thermal inversion and mixing height
Liang and Keener (2015) analyzed atmospheric sounding data from NOAA/NESDIS4,
and constructed atmospheric temperature profiles using the method by Ma et al. (1999). Two
satellites, GOES-8 and GOES-9, equipped with filter wheel radiometers, collected radiance
measurements from the on-board thermal infrared channels, while allowing retrieval of the
atmospheric temperature and moisture profiles. The data were retrieved at a 10-km spatial grid
and in hourly intervals for sounding data locations S-l to S-4 in Figure 2-20.
Figure 2-24 shows a typical diurnal atmospheric profile in the Cincinnati metropolitan
region. The tropopause layer separates the turbulent troposphere from the temperature-inverted
laminar stratosphere above. A nocturnal temperature inversion is evident in the lapse rate5 in
the near-ground urban boundary layer. At this location (39°14'43", -84°26'46"), thermal
inversion reached its maximum in the early morning, followed by inversion destruction and
then the recovery to normal lapse rate as a slightly stable boundary layer (SBL) in the early
afternoon. The daytime lapse rate returned to a level of neutral stability close to the dry
adiabatic lapse rate (DALR) at 9.8 °C/km (Figure 2-24). The nocturnal temperature inversion
was then re-established by the late evening. This diurnal variation is evident for all four radio-
sounding locations S-l to S-4. From the temperature profiles, the lapse rate (Lh) and mixing
height (Zim) were determined for each day.
4 http://www.star.nesdis.noaa.gov/smcd/opdb/goes/soundings/skewt23L/html/skewhome.html
5 Lapse rate is defined as the gradient of temperature change per unit distance from ground surface.
53
-------
T(°C)
Figure 2-24 Representative temperature profiles showing the boundary inversion and capping
inversion, Temperature data were obtained from NOAA for the northern Cincinnati site.
Altitude 0 is set at surface elevation. DALR is the atmospheric dry adiabatic lapse rate.
From Liang and Keener (2015).
The determined Lh and Zinv values in a 10-day period of October 2010 are shown in
Figure 2-25. Clearly, a sequential occurrence of nocturnal thermal inversion with the strongest
phase in the days of October 7-12. Changes in temperature gradients in altitude became gradual
in the tropopause. Returning of lapse rate in the daytime to a level of neutral stability close to
neutral DALR of 9.8 °C/km is found across the observation period. The near-surface boundary
layer above the urban canopy marks the extent to which thermal and mechanical mixing occurs.
A maximum inversion strength with a lapse rate of -29.2 °C/km occurred at 4 am on October 9
for the near-surface boundary layer thickness of 421-607 m.
4.1.5. Urban and exurban differences
Thermal inversion development in the region had similar overall diurnal Lh and Zm\
variability. However, a small difference exists between the Lunken airport station and others
(Figure 2-25). The measured inverse lapse rates are lower at Lunken airport compared to the
other locations inside of the high-density urbanized area. The smaller profile slope reflects
weaker inversion strength in the peak inversion phase. Liang (2014) further showed the
difference was persistent based on linear correlations of Lh values at different locations.
54
-------
Figure 2-25 Temporal Lr and Hjnv variations showing diurnal thermal inversion in the urban boundary
layer in October 2011. The inserts (a1) and (b1) show the observed difference of hourly
variation among the sounding sites in the period of October 9-10. From Liang (2014).
4.1.6. Urban form effects on urban heat island and air quality
4.1.6.1 Long-term changes in the urban center
The unique structure of urban boundary layer is considered responsible for causing UHI
formation and related air quality deterioration (Rotach et al., 2005; Liang et al., 2013; Trompetter
et al., 2013; Wang et al., 2012). This relationship is found from analysis of the long-term
ambient temperature and PM2.5 concentration data from 1999 to 2013 when all stations of the
Cincinnati metropolitan area are compared to the reference station 17-061-0040 in the urban
core.
55
-------
1/1998
1/2000
1/2002
1/2004
1/2006
1/2008
1/2010
1/2012
1/2014
1/2016
Figure 2-26 Tmin, AT, and PM2.5 variations with time at NAAQS monitoring station 061-0040. After
wavelet denoise, the seasonal variations are shown in heavy lines. From Liang and
Keener (2015).
The 1457 daily measurements at the reference station show no statistically significant
change over time for daily Tmax and Tavg However, long-term changes in Tmin and AT can be
convincingly identified beyond the noise of seasonal variations using the so-called continued
wavelet transformation techniques. At a data noise threshold db=0.80, wavelet-denoising
(Torrece and Compo, 1998; Farge, 1992) of the Tmin and AT data captured nearly 80% of the
variation in Figure 2-26. Wavelet-transformed Tminmd AT maxima occurred in May-June of
each year, and the minima in the winter period. The seasonal cyclic variation is evident.
These temperature highs and lows after denoise show an increase of-1.6 and -2.1 °C
over 10 years, respectively, by linear regression (Figure 2-27). These long-term changes
correspond to night-time maximum and minimum temperature, respectively, in the summer and
winter seasons. Because of the increase in night-time temperature, diurnal temperature range AT
56
-------
25
20
15
10
O
-5
-10
~
Tmin' maX
~
AT, max
~
Tmin' min
~ ~
0.156±0.007 yr1; R2=0.30
-0.122±0.004 yr1; R2=0.41
1/98 1/00 1/02 1/04 1/06 1/08 1/10 1/12 1/14 1/16
Date
Figure 2-27 Temporal change of ambient temperature Tminand AT at station 061-0040 in the
Cincinnati urban core. The regression slopes are statistically significant with
p<0.0001. Adopted from Liang (2014).
decreased by 1.2 °C over a decade (Figure 2-27). These long-term changes are consistent with the
other publications on urban microclimate (Wang et al., 2012; Braganza et al., 2004).
4.1.6.2 Urban-wide co-variations in temperature andPM2.5
• Ambient temperature
The UHI effect across the region is shown by ambient temperature measurements.
Ambient temperatures measurements each year are correlated among stations, and the slope of
the correlation quantitatively is determined according to Eq.2.1. An example is shown in Figure
2-28 between the reference station 17-061-0040 at the urban core and other stations. The average
square coefficient of correlation (R2) for the 91 to 116 station-year correlations is >0.993 (0.941-
0.999). Because the data covers a 10.4-year long period and for all seasons, the strong linear
correlation indicates an effective and time-persistent urban-scale heat flux and air circulation
above the canopy layer.
Based on the correlation, temperature difference T' between a location and the reference
station in the urban core is calculated for the decade-long measurements. The results are
presented in Table 2-7. Apparently, the calculated T'values are spatially correlated with the
delineated urban land use. Quantifiable and statistically significant UHI effects coincide with
high-density urbanized zones. In cross-section A-A' (see location in Figure 2-20), ambient
temperatures above the canopy layer are consistently higher inside the high-density zone than in
the surrounding suburb and exurban areas (Figure 2-29). For three stations outside of the zone,
annual mean T'avg and the largest T'avg in summer are lower by 0.89±0.14 °C and 1.55±0.30 °C,
57
-------
•
~
A
25-0022
37-0003
17-0016
?7.nnm
y=0.961x-0.288
R2=0.97
17-nnifi
y=0.993x-1.783 :
ejp #
R2=0.97
D nl
-p^-nn-p?
"
y=0.961x-0.652
-
>7
R2=0.96
0 10 20
Tmin (°C)> 61-0040
0 10 20
Tmin (°C)> 61-0040
0 10 20
Tmin (°C), 61-0040
.
i i i i I
i i i i I i i i i I i i i i I i i i i
y=0.992x-0.870 ;
•
61-0042
R2=0.98
~
61-8001
A
61-7001
y=0.985x-0.518 :
R2=0.97
y=0.991x-1.036
, , , ,
, , , , i
R2=0.99 -
, , , , i , , , , i , , , , i , , , ,
Figure 2-28
-20 -10 0 10 20 30
Tmi„(°c), 61-0040
Tminat 17-061-0040 station is linearly correlated with those of other stations in the year
2005 measurements. Modified from Liang and Keener (2015).
respectively. The largest AT' also occurred in summer when highest night-time Tmin=21 A °C and
highest day-time Tmax=42.1 °C were measured in the 17-061-0040 station. The average AT'min
and AT'max were -1.69 °C and -1.71 °C.
In contrast, temperatures are relatively uniform inside of the high-density zone. The mean
T'aVgis —0.09(±0.27) °C. The annual mean T'avg increases slightly from its southern tip at
station 037-3002 toward station 061-0040 in the urban core.
• PM2.5 variability
Like ambient temperature, the observed PM2.5 concentrations are linearly correlated
between the reference station 39-061-0040 and all other stations (Liang, 2014). The correlation is
persistent for all years of measurements at the sampling height above the canopy layer. This
correlation covers all PM2.5 concentration range [1.2-52.1 mg/m3 (fn=13.89, N=1717)]. See
Table 2-6.
58
-------
Table 2-7 Temperature differences between the reference station and other stations abstracted from the >10-year daily temperature
measurements (From Liang, 2014).
Annual Mean
Winter
Summer
Station
T'avg (°C)
T' (°C)
1 min V W
T' (°C)
1 max V W
T'avg (°C)
T' (°C)
1 min V w
T' (°C)
1 max V W
T'avg (°C)
T' (°C)
1 min V W
T' (°C)
1 max V W
14.22
9.30
20.09
-15.20
-20.40
-10.20
34.3
27.4
42.1
39-061-0006
-0.79 ± 0.45
-1.82 ± 0.20
-1.52 ± 0.19
-1.17 ± 0.93
-2.50 ± 0.65
-1.19 ± 0.87
-0.53 ± 0.82
-1.40 ± 0.72
-1.76 ± 0.96
39-061-0010*
-1.05 ± 0.22
NA
NA
-0.67 ± 0.74
NA
NA
-1.31 ± 0.78
NA
NA
21-037-3002*
-0.58 ± 0.48
-1.18 ± 0.82
-0.34 ± 0.54
0.06 ± 0.90
-1.15 ± 1.06
0.64 ± 1.11
-1.03 ± 0.91
-1.19 ± 1.07
-1.05 ± 1.09
21-117-0007
-0.25 ± 0.38
-0.33 ± 0.42
-0.58 ± 0.54
0.08 ± 0.47
-0.33 ± 0.63
0.48 ± 0.75
-0.48 ± 0.74
-0.32 ± 0.91
-1.34 ± 0.99
39-025-0022*
-0.82 ± 0.65
-1.40 ± 0.51
-0.37 ± 0.36
0.11 ± 0.85
-0.50 ± 1.05
0.80 ± 1.65
-1.46 ± 0.86
-1.95 ± 0.19
-1.22 ± 0.57
39-017-0016
0.39 ± 0.54
0.91 ± 0.64
-0.61 ± 0.87
-0.16 ± 0.88
0.22 ± 1.27
-0.90 ± 0.86
0.77 ± 0.60
1.33 ± 0.81
-0.40 ± 1.02
39-061-0014
-0.28 ± 0.58
-0.69 ± 0.65
0.02 ± 0.69
0.17 ± 0.87
0.16 ± 1.34
1.13 ± 2.19
-0.58 ± 0.77
-1.21 ± 0.74
-0.78 ± 0.94
39-061-8001
-0.05 ± 0.31
-0.31 ± 0.43
-0.18 ± 0.51
0.18 ± 0.61
-0.12 ± 0.53
0.51 ± 0.46
-0.21 ± 0.64
-0.43 ± 0.69
-0.68 ± 0.75
39-061-7001
-0.22 ± 0.52
-0.39 ± 0.41
-0.36 ± 0.70
0.29 ± 0.71
0.25 ± 0.86
1.54 ± 3.29
-0.57 ± 0.77
-0.78 ± 0.49
-1.73 ± 1.55
39-061-0041*
-0.48 ± 0.38
-1.04 ± 0.41
-0.38 ± 0.41
0.38 ± 0.46
-0.24 ± 0.84
0.41 ± 0.49
-1.07 ± 0.80
-1.52 ± 1.12
-0.95 ± 0.63
39-061-0043
-0.36 ± 0.42
-0.60 ± 0.46
-0.14 ± 0.58
0.05 ± 0.69
-0.32 ± 0.91
0.26 ± 0.62
-0.63 ± 0.65
-0.77 ± 0.65
-0.43 ± 0.82
21-037-0003*
-0.79 ± 0.93
-0.86 ± 1.08
-0.77 ± 0.60
0.81 ± 0.71
0.08 ± 0.82
1.20 ± 0.53
-1.88 ± 1.18
-1.43 ± 1.36
-2.20 ± 0.92
39-061-0042
0.11 ± 0.57
-0.03 ± 0.71
-0.11 ± 0.79
0.84 ± 0.88
0.95 ± 1.49
0.98 ± 0.76
-0.39 ± 0.79
-0.63 ± 0.70
-0.90 ± 0.94
Note: * Stations are outside of the high-density urban area.
#Stations are outside but nearthe high-density urban area.
NA - Data not available.
59
-------
1.0
0.5
0.0
O -0.5
o
> -1.0
h-
-1.5
-2.0
-2.5
annual
summer
0
o
10
Distance (km)
25
O'
Figure 2-29 Spatial variations of temperature difference for mean and maximum Tavg and PM2 5 in
cross section O-O'. The profile starting point a is station 037-3002 at southern tip of
the high-density zone. See Figure 2-20 for the cross-section locations. From Liang
(2014).
The high degree of linear correlation among the stations is significant for the long
duration of monitoring. The correlation coefficient (R2) for the 130 station-year correlations
ranges 0.53-0.99 with an average of 0.92. Nearly 92% of PM2.5 variability can be explained by
the urban-wide correlations. Similar conclusions on area-wide PM2.5 variations were made by
Martuzevicius et al. (2005) using hourly monitoring data, instead of daily, of the 13 NAAQS
network stations in the Cincinnati area
In summary, the intra-station correlations both in temperature and PM2.5 strongly suggest
atmospheric mixing and mass communication at the station's sampling height. The UHI effect is
evident at the urban core along with the air quality variations.
4.1.6.3. Thermal inversion and wind conditions
The frequent and high-strength thermal inversion in the Cincinnati metropolitan area is
linked to the weakened wind field and deteriorating air qualities. Liang et al. (2013) reported
onsite measurements of black carbon, PM2.5. and other air pollutants in a field study at the 1-75
highway. The study was in the high-density urbanized zone of northern Cincinnati (see Figure 2-
20 for location). Their results clearly showed the weak to stagnant wind conditions in early
morning hours, associated high black carbon concentrations near roads, and a high ratio of
organic carbon (OC) over elemental carbon (EC).
The co-variation between field-observed wind speed and the determined temperature
lapse rate (Lh) is shown in Figure 2-30. Here Lh values quantitatively measure the inversion
strength; the SBL, weak SBL, and very weak SBL are defined as Lh = 5 to -15, 5-10 °C/km, and
60
-------
E
O 0 -
10/7
o
-------
induced upslope and downslope airflow is needed to produce nocturnal inversion in complex
terrain. For areas with a gentler topographic slope like the Cincinnati metropolitan area, other
possible mechanisms are likely. One possible mechanism may involve UHI-induced thermal flux
and horizontal air movement. Upward sensible heat flux and air aloof from the warmer urban
interior can induce movement of colder air masses from surrounding rural areas, leading to
evening urban breeze, colder air at the ground surface, and hence the concurrence of UHI and
thermal inversion (Rotach et al., 2005; Rendon et al., 2014; Hidalgo et al., 2010). Temperature
condition for this UHI formation was observed in the higher night-time temperature T'min, in the
high-density urban zone (see Table 2-7).
The UHI effect and temperature variations are not uniform across the Cincinnati
metropolitan area. They depend on land use and land covers, and ultimately, the urban form.
Similar conclusions were made by several recent studies using detailed thermal mapping of the
UHI effects (e.g., Liu et al., 2012; Buyadi et al., 2013). As shown in this case study, the high-
density urbanized zone is associated with an increase of ambient daily temperature by 0.89-1.55
°C. Night-time temperature increase is larger at -1.7 °C. The increase closely follows the high-
density zone boundary in cross-section A-A'. The N-S trending high-density zone of varying
width is found to have a varying degree of UHI effects (Figure 2-29 and Table 2-7). The
temperature increase is the largest in the Cincinnati downtown area, around stations 37-061-0040
and 37-061-0042. The UHI effect reduces in the north, where the high-density zone narrows, and
nearly disappears in the southern tip at station 21-037-3002.
The variation trend and its association with land use types are further corroborated by a
negligible small temperature increase found outside of the high-density zone in residential areas.
Three stations in the perimeters of Cincinnati metropolitan area show a smaller daily average
T'aVg than in the high-density zone. Similarly, the station 37-061-0006 is located in a small and
isolated high-density urban area along 1-71 (Figure 2-20). Calculated T'avg values are close to
that of three exurban stations. Not coincidently, these residential areas are characteristic of
detached single houses with large trees, large yards, and acres of natural area in between (Figure
2-22). This type of suburban region with less UHI effect was common for over 38 U.S. urban
centers that Imhoff et al. (2010) studied using LANDSAT satellite imagery data.
UHI effects measured in this study are much smaller than one derived using the empirical
formulation of Oke (1976 and references therein). Based on the formula, the Cincinnati
metropolitan area of -2.1 million population would yield a 4.91 °C temperature increase. The
majority of the Cincinnati area is typical of the medium to low-density suburban areas in the new
classification Oke (2006) proposed for urban climate zone. The urban form, the use of green
space, and the elongated narrow shape of the high-density zone may be contributing factors for
the observed smaller UHI effect.
4.1.6.5. Adaptation and potential effects
Urban-scale UHI occurrence affects both air pollutant transport and water demand. The
planning of the high-density urbanized area and green spaces affect UHI occurrence and
atmospheric circulation in the boundary layer. The studies in the Cincinnati metropolitan area
indicate the potential of adaptation co-benefits in the following two areas.
62
-------
¦ A ir pollutant transport
The Cincinnati case study shows that the urban form and its physical structure
configurations can affect urban microclimate and, thus, the air pollutant transport. Figure 2-31
illustrates three typical types of canopy layer settings that can affect near-ground pollutant
distribution. These include the open-field setting at the highway 1-75 site (Liang et al., 2013), the
street canyons among the low- and high-rise buildings of the urban interior, and lastly, residential
areas with significant tree canopy effects. Among the three types, highway roads in open fields
are most common in the Cincinnati metropolitan area. Both the UHI and PM2.5 levels are a
function of the canopy, transportation infrastructure, vehicle numbers, and emission rates.
vJ
Residential
o c
n-i
Open highway (1-75 site)
NAAQS
)>
c ^
V
Street canyon (0610040)
Figure 2-31 Schematic diagram showing major types of microclimate conditions in the
surface roughness layer (SRL) equivalent to the urban canopy layer (UCL).
The NAAQS stations above the UCL are affected by urban boundary layer
(UBL) circulations. Other symbols: G - Gaussian dispersion, NG - Non-
Gaussian dispersion. Modified from Liang and Keener (2015).
U.S. EPA has published guidelines on quantitative modeling and assessments for these
urban settings (U.S. EPA, 2017b, 2004b). Numerous literature also elucidated adaptation actions
that have the potential to mitigate the negative consequence in air pollution, for example, the use
of tree barriers along traffic routes (e.g., Baldauf et al., 2008). Urban adaptation can likely be
planned to affect and even modify the microclimate settings, including those in Figure 2-31, and
their locations and spatial distribution.
¦ The UHI and controlling factors
The Cincinnati case study shows interrelationships among urban form, air quality, UHI
formation, and population distributions. The high-density urbanized zone along the 1-75 highway
and the Mill Creek has many properties of UHI effects: night-time temperature increased by -1.7
°C compared to exurban areas, a long-term night temperature increased by 2.0 °C per decade, and
a higher PM2.5 concentration occurred above the urban canopy layer. The UHI formation and
thermal inversions are attributed to high concentrations of air pollutants near ground levels. The
UHI occurrence can increase water consumption, altering water demand variations in space and
seasons. Although the exact impact is not quantified in the case study, it is generally understood
that the higher daily temperature, smaller diurnal temperature AT, weak winds can produce
changes in evapotranspiration rate for lawns and vegetation and lead to greater water
63
-------
consumption per capita. This association has been described in the literature (e.g., Guhathakurta
and Gober, 2007).
Major factors affecting the UHI effects include spatial continuity of high-density areas
with altered ground surface (e.g., concrete pavements), size and locations of green space, forest
and native land coverage, non-continuous multiple urban centers, even tree canopy barrier that
helps modify the interactions between the urban canopy and the overlying urban boundary layer.
On this basis, subsequent Section 4.2 outlines how adaptive urban planning could be made for
future development scenarios.
4.2. Adaptive urban planning modeling and analysis in Cincinnati
The Cincinnati metropolitan area follows development trajectories of many very large
U.S. metropolitan regions. Notable development actions include downtown revitalization, infill
developments, the Ohio River bank development, and a series of land use policies to improve the
urban efficiency including transportation and initiatives like the street car system. These
development initiatives have changed the population and urban activity distribution. The change,
when coordinated in planning, could lead to reduce the UHI occurrence and to positively affect
water and transportation infrastructure planning and operation.
4.2.1 Three development scenarios
The scenario-based adaptive planning was conducted for the Cincinnati metropolitan area
to assess development options. It was focused on land use changes and their downstream effects
on transportation performance and benefits in carbon emissions. Parts of the research have been
published in Wei et al. (2017, 2012) and Yao et al. (2014). Through this example, the step-by-
step process is illustrated for using AUP&ET tools.
Figure 2-32 shows the distribution of base-year population, household, and employment
in the metropolitan area in 2010. It is noted that the classic monocentric urban form is starting to
evolve into multi-centers of employment, with the population and households scattered and
distributed across the region. Transformation districts in Mason north of the city, Norwood, and
Downtown are further reshaping the population distribution along the north-south tending 1-71
and 1-75 corridors along with changes in the employment distribution.
To explore the potential future developmental scenarios, three options were analyzed
using AIR-SUSTAIN for Hamilton County, Ohio. The year 2010 was chosen as the base year,
and the year 2030 was set as the target year. The three developmental scenarios were analyzed
using the AIR-SUSTAIN tool on transportation first; the results can be used later for water
infrastructure planning and adaptation. The three scenarios are:
¦ Scenario 1 (SI) is referred to the single-center development pattern. The single center is
taken to occur in the Downtown and Uptown Cincinnati areas, as shown in Figure 2-33a.
¦ Scenario 2 (S2) is referred to the multiple-center development pattern. Two-center
development is assumed in this scenario; one in the Downtown Cincinnati and the other
in Mason area in the northern Hamilton County and southern Warren County. The
development case is shown in Figure 2-33b.
¦ Scenario 3 (S3) adopts the same development pattern as S2. However, it differs by having
two Rapid Bus Transit lines connecting these two centers, as shown in Figure 2-33c.
64
-------
# People
¦ 0-732
¦ 733-1806
~ 1807 - 3323
¦ 3324 - 6038
¦ 6039-17026
¦ 0-279
m 280 - 687
~ 688 -1235
¦ 1236 - 239
¦ 2399 - 588
Figure 2-32
The base-year distribution maps for
the Cincinnati metropolitan area in
2009: A) population; B) household;
and C) employment.
Generally, this type of planning and engineering analysis requires stakeholder
engagement, economic analysis, and engineering evaluation. The analysis presented here is
simplified and intended to show how the AIR-SUSTAIN tool can be used in scenario-based
adapti ve planning.
4.2.2 Transportation and emission analysis using AIR-SUSTAIN tool
4.2.2.1 The modeling processes
All three competing scenarios assume the same 15% increase of population and
employment to occur from the base year to the target year 2030. The population and employment
increases are allocated and distributed around the activity center(s). The process for a scenario
analysis was developed through 19 steps in the analysis from a new scenario setup (Figure 2-34),
regional and project level traffic analysis, to emissions modeling using MOVES to obtain the
final simulation results. Details of these simulation steps are contained in Appendix A. Important
modeling steps are discussed below.
After setting up the adaptive planning project, a new scenario was created and saved with
an AIR-SUSTAIN database in MySQL and ArcGIS. Figure 2-35 shows a graphic user interface
for the project scenario setup. The subsequent Steps 2-4 specify the existing TAZ, road network,
65
-------
(A) (B) (C)
Figure 2-33 Three development scenarios for the Cincinnati metropolitan area in the target year 2030. A) monocentric development
around the downtown; B) two-center configuration in downtown and Mason area; and C) two-center configuration with mass
transit between the centers.
Mason
Mason
66
-------
New Scenario
Senario Name
Project Directory
Modeling Year (yyyy)
Analyst
Date
IDRISI Director/
MOVES Director/
Example
|CAUser3\Ting\Desktop\Example
Base: 2010
Target: 2030
UC
Saturday . September 20. 2014
C:\Users\Ting\Desktop
~ Q
C:\UsersVPublic\MOVES2012041] | ...
Scenario Description (optional):
and boundaries of the activity centers
in adaptive urban planning. Activity
centers or incentive districts function
as the transformation districts (see
Figure 2-6) important to urban
adaptation, and, by design, they
introduce changes to the urban form
and associated changes in population
and urban activities. These projected
changes define the technical basis for
transportation and water infrastructure
adaptations. In the modeling, the tool
has the capability of specifying the
changes in population and employment
from the base year at the TAZ level.
Population Change and Employment
Change in the program were specified
for regions inside and outside of the
incentive boundaries separately
(Figure 2-35).
Land use for the base year and the target year is generated from modeling in another
AUP&ET module. Urban land use projection is implemented using the CA-Markov method or in
combination with the ICLUS available in AUP&ET (see Figure 2-14 and Section 1.3.1 for
details). In the AIR-SUSTAIN, the land use projection maps are used as a GIS layer for trip
generation modeling and forecasting. See Appendix A and Wei et al. (2017) for more details.
Save
Scenario
Cancel
Figure 2-34 Setup of a new scenario in AIR-
SUSTAIN
Target Year Scenario Design —
1. Assumed Changes in Demographic and Socioeconomic Factors
a. Population O Edit Q Load File c. University Enrollment Load File
b. Employment Q Edit ® Load File d. High School Enrollment Load File
View
Base Year Data
Target Year Scenario Design
Travel Demand Forecasting
(A)
& Population Data Editor | <=> || m || S3 |
Incentive Area Population Percentage Change
Non-incentive Area Population Percentage Change
Save
4 Employment Data Editor I ° |l 0 II £3 |
Incentive Area Household Percentage Change
15 | %
Non-incentive Area Household Percentage Change
|o I %
Save
(B) (C)
Figure 2-35 Program interface for A) importing the Base Year data; B) assigning population
change; and C) assigning employment changes at TAZ levels.
67
-------
One potential option for using the adaptive urban planning is to specify the maximum population
density in the transformation districts. Such planning measures can potentially transform a
rnonocentric urban form into a polycentric configuration (see Figure 2-6). Examples of high-
density development and urban transformations can be found in the literature (U.S. EPA, 2007b,
2013a; Oh et al., 2005; Gim, 2012; and Sukkoo, 2007).
For illustration, Figure 2-36 shows an example of setting up the scenario-based
population and demographic simulations. By defining thq Maximum Population Density in the
incentive area, i.e., 15,000 (person/mile2), the target year demographic and socioeconomic data
are generated by using the linkage model according to the base year data and the specified
demographic and socioeconomic changes. Alternatively, one can input the population density
projected in other population-based land use models such as ICLUS (U.S. EPA, 2010b). After
completing a demographic projection for a developmental scenario, the adaptive planning
process continues in several consecutive steps shown below to help understand the adaptation
attributes:
¦ Projecting the future trip generation or the "travel need" in a future time (Steps 7-10).
This analysis is based on scenarios of growth policies in anticipation of future economic
status and the conditions specified at the beginning of an urban planning cycle (Figure 2-
6).
3. Socioeconomic Data Update Base on Assumed Data
a. Maximum Population Density 115OQ0|
b. Linkage Model | Run
c. Target Year Demographic and Socioeconomic Data 1. Population
Target year demographic
and SE data
! = S II S3
Q
File Edit View Bookmarks Insert Selection Geoprocessing Customize Windows Help
~ BBS 6* + - '**>¦«» 4E1 BBBBD
t o p 2 * ^ w-
Figure 2-36 Simulation module of demographic analysis for a development scenario. High
population density is specified for analysis of urban adaptation options.
68
-------
¦ Analyzing the transportation impacts on air emissions (Steps in 11-13). These simulations
yield technical information on traffic vehicle-to-capacity ratio (V/C), locations of traffic
congestion areas, and also allow users to explore potential planning or engineering
solutions in adaptation. An example of the congestion analysis in the Cincinnati
metropolitan area is shown in Figure 2-37.
¦ Developing traffic management solutions and evaluation of adaptation limits relying on
traffic management and improvement of the existing transportation network. Normally
this adaptation analysis (Steps 15-16) relies on the analysis of infrastaicture optimization
using the microscopic simulation model VISSUM.
¦ Evaluating environmental and economic consequences for urban planning scenarios
(Steps 17-19). In these final steps, the adaptive urban planning is evaluated using
quantitative analysis of carbon emi ssion emissions, transportation performance (traffic
delay, fuel consumption, etc.), and economics of adaptive measures.
In the step-by-step analysis, transportation performance is reviewed and evaluated against
the master planning or adaptation objectives. The selected developmental scenario is then
examined further in the analysis of the infrastructure adaptation for water supply, wastewater,
and stormwater management. See the AUP&ET process diagram in Figure 2-9.
Congestion area
55.421996
I.247707
99.993349
7051.641836
0.097149
Polyline
0.091775
5759.179056
45.670909
23066
30.65859
Polyline
23408
5632.664272
0.094315
"31.087014
399.28471
Polyline
0081338
52.736179
Polyline
23724
7927141442
29.281309
64.974474
31.815801
7129.445341
.084971
60.951904
0.085321
Polyline
23768
7986.814954
0.088107
57.048753
23870
B859.0226S3
Polyline
23876
7104.868745
.534304
63.813635
Polyline
23948
Polyline
(0 out of 23 Selected)
Hotspots
Legend
RoadNetwork
Figure 2-37 Traffic congestion areas identified for typical peak-hour traffic for 2009 in Hamilton
County, showing concentrations along 1-71, I-75, I-275N, and Ronald Reagan Highway.
Annual traffic data from Ohio-Kentucky-Indiana Council of Governments.
69
-------
4.2.2.2 Comparison of future developmental scenarios
The scenario analysis for the Cincinnati metropolitan area yields a quantitative basis to
compare and evaluate the three development scenarios for the target year 2030.
1) Demographic and socioeconomic changes
The changes in population, household, and employment from 2010 base year to 2030
target year are shown in Figures 2-38, 2-39, and 2-40, respectively. The result shows that future
population distribution is significantly dependent upon the developmental scenario. In the SI
scenario, all anticipated population growth is allocated in the downtown area. Such an increase is
most likely to occur throughout downtown, particularly in the Over-the-Rhine area to the north.
In comparison, the changes in downtown are less prominent in S2 and S3 scenarios, because of
significant change in the Mason-West Chester region along the northern 1-75 and 1-71 highways
and their connectors (Figure 2-38).
Figure 2-38 Population changes by 2030 for the three developmental scenarios (S1, S2, and
S3 in the inserts) in comparison to the distribution of base year 2010 (background).
Land boundaries indicate census blocks.
Future household numbers in 2030 would experience similar changes as the population
(Figure 2-39). The household number is projected for the largest increase in the number of small
downtown TAZs, where only a moderate-high rate of population increase is anticipated. This
disparity is found to be related to the smaller households in the downtown area. The intercity re-
development initiatives by the City of Cincinnati may have facilitated these changing trends.
Figure 2-40 shows the employment changes at the TAZs level. The change shows a large
increase in northern Cincinnati (West Chester - Springdale - Mason area), the downtown (Ohio
River bank, and university hospital area), and in northern Kentucky. These projected increases
70
-------
Figure 2-39 Changes in number of household by 2030 for the three developmental scenarios
(S1, S2, and S3 in the inserts) in comparison to the distribution of base year 2010
(background). Land boundaries indicate census blocks.
are considered reasonable based on the current developmental trends; for example, there are
projected increases of employment in the Ohio River bank area that has been redeveloped in the
past two decades and continues to experience relocation of business operations into the area - an
example is the recently relocated GE business operation headquarter. In S2 and S3
Figure 2-40 Employment changes by 2030 for the three developmental scenarios (S1, S2, and S3
in the inserts) in comparison to the distribution of base year 2010 (background). Land
boundaries indicate census blocks.
71
-------
developmental scenarios, a large increase in employment is predicted in the Mason area in the
northeast (Figure 2-40).
Table 2-8 Trip generation results in the number of trips per day by the target year 2030
Daily Trips-
Base Year
S1
S2/S3
HBO
2,237,609
2,795,451
2,802,678
HBSC&HBU
66,635
93,287
93,553
HBW
1,127,146
1,347,873
1,359,091
NHB
1,541,498
1,794,372
1,780,907
Total Trips
4,972,888
6,030,983
6,036,229
Note: HBO - home-based-other trips; HBW - home-based-work trips; NHB - nonhome-based
trips; HBSC - home-based school trips; HBU - home-based university trips
2) Travel demandforecasting results
The travel demand forecasting results are illustrated by trip generation (Table 2-8) and
trip distribution (Table 2-9) for the three scenarios. Based on the AIR-SUSTAIN modeling
results, there would be a -20% increase in trip generation from the base year (2010) to the target
year (2030). This increase is largely due to the targeted population increase of 15% and related
business activities. The daily trip generation would be around 6 million for the SI and S2/S3
scenarios. Furthermore, the model projection shows the nature of the trips would change.
Comparing the two sets of scenarios, the multi-center configuration in S2/S3 favors home-based
work trips (HBW) with trips within the home TAZs. In contrast, the monocentric SI scenario has
a greater number of non-home-based (NHB) trips (Table 2-8).
Table 2-9 Trip distribution results for the number of trips per day originated from and attracted
to centers*
Daily Trips
Base Year
S1
S2/S3
Intracenter
181,232
398,746
422,073
External
432,667
425,816
435,525
Total
613,899
824,562
857,598
*Note: Centers are future incentive areas as shown in Figure 2-33.
Table 2-10 summarizes traffic performance for the three developmental scenarios
simulated for the year 2030. Average queue length and wait time would increase because no
expansion of the transportation network was assumed for the modeling period (2010-2030). This
assumption was used to evaluate the potential capacity reserve of the current roadways. The
analysis results allow one to evaluate the maximum potential of existing network utilization,
namely the threshold, when adapted through traffic management tools. For the same queue
length, the S3 scenario with mass transit between two future centers would decrease the average
72
-------
Table 2-10 Average queue length, average wait time, total delay, and average delay
during Morning Peak Hours (7:00 ~9:00 am)
Scenario
Base Year
S1
S2
S3
Average queue length
(vehicle per link)
9
12
12
12
Average wait time
(minute per link)
2.03
2.87
2.56
2.08
Total Delay
(vehicle-hour)
113,456.4
205,121.3
179,796.6
153,018 4
Average Delay
(minute per vehicle)
10.8
16.1
14,1
12.0
Note: Link refers to model road segment in model space.
time by 27.5% or 2.87-2.08 = 0.79 min/'link. The average total delay would be reduced from 16.1
to 12.0 min/vehicle or by 25.4% (Table 2-10).
The traffic improvement by using mass transit in the two-center configuration is
graphically shown in the peak-hour (7:00-9:00 am) traffic volume distribution over the
metropolitan's road network (Figure 2-41). Compared to the current condition, the 2030 peak-
Figure 2-41 Simulated peak hour (7:00-9:00 am) traffic volume distribution over the Cincinnati
road network for the base year (2010) and under three development scenarios in
the target year (2030).
73
-------
hour traffic pattern is characterized by increased traffic along the 1-75 and 1-71 highways. The
traffic around the Cincinnati downtown area (1-71,1-75,1-562, and connectors) would become
increasingly heavy in SI of the concentrated downtown development. The condition would
improve for the two-center configuration particularly with the help of the mass transit
development in the S3 scenario.
3) Energy and emission redaction as adaptation co-benefits
Computer-simulated city-wide CO2 emissions and energy consumption (per day) from
the transportation sector are shown in Figure 2-42. They are compared among the development
scenarios. From 2010 to 2030, the SI developmental scenario is projected to have the largest
increase in CO2 emissions and energy or gallons of fuel consumed. These two variables
(energy/fuel and emissions) are internally related to each other. The CO2 emissions would
increase by 13.6%. This degree of increase is comparable to and slightly less than a 17.5%
increase in the trip generation (see Table 2-10), mostly due to the higher population density and
employment in the downtown area. The improvement is attributed to the use of the smart
development approach in favorite of high-density residential and "walkable" communities.
The largest and most significant improvement in CO2 emission reduction and the energy
consumption is predicted for the S3 development scenario with two centers and mass transit. The
emissions are 15.6% less compared to that in the SI scenario that continues the monocentric
development (Figure 2-42). Transportation efficiency would also increase in the multi-center
configuration. The average traffic delay per person is calculated to be 25% less than in the SI
scenario, and only slightly higher than the base year in 2010 while at over 17.5% increase in
travel demand (see Table 2-10).
C02 Equivalent (kg)
Energy Consumption (kJ)
2300000
2200000
2100000
2000000
1900000
1800000
1700000
1600000
I Series 1
I
Base year
SI
1
S2
.685 1996739.C
I
1904870.854 2203241.685 1996739.032 1857874.362
S3
31000000
30000000
27000000
26000000
23000000
Figure 2-42 Comparison of three development scenarios (S1, S2, and S3) in peak hour (7:00-9:00)
vehicular CO2 emission and energy (fuel) consumption.
4.2.2.3 Implications on adaptation co-benefits
The case study in the Cincinnati metropolitan area illustrates the utility of an integrated
analysis tool AIR-SUSTAIN in AUP&ET for the analysis of urban-scale development scenarios.
The interactions among travel demand from population and land use change, demographic and
74
-------
socioeconomic distributions, as well as the transportation activities and their on-road emissions
can be quantified in model simulation to assess the effects of urban infrastructure adaptation
policies.
The scenario-based analysis reveals the co-benefit of urban adaptation and highlights
critical elements of adaptive planning through transportation optimization. Even with no large
road network expansion, multi-center urban transformation is projected to reduce emissions and,
at the same time, improve urban efficiency. Transportation measures such as mass transit can
facilitate the urban form transformation from the current monocentric form to polycentric
development.
It should be noted, however, the urban-scale adaptation case study in Cincinnati is a first-
order analysis in urban planning. The model simulation only considered land use and
transportation with assumptions for simplification given the data availability. For example, the
CO2 emission calculation assumes average speed under normal driving conditions in adaptation
scenarios. The model simulation did not differentiate light passenger trucks (source type 31)
from passenger cars (source type 21) in emission calculation. The MOVES engine in AIR-
SUSTAIN has the capability of performing detailed emission analysis. For this extended
capability, please refer to MOVES user's manual and related technical references (U.S. EPA,
2010a; 2015e; 2017b).
Another important element in adaptive urban planning is water infrastructure and service
functions. After transportation planning, subsequent analysis of water infrastructure adaptation
may lead to further refinement of the developmental scenarios. See the adaptive process in
Figure 2-6. In the next Section 4.3, a case study in Florida is used to illustrate the use of
adaptation in master planning to evaluate and optimize planning options for county/urban-scale
water supply expansion.
4.3 Adaptation analysis for water master planning in Manatee County, Florida
The case study on master planning for water infrastructure expansion was conducted in
2009-2011 for Manatee County in Florida (Figure 2-43). The research results have been
published in Chang et al. (2012).
The large growth in population, tourism, and economic development occurred in the past
two decades preceding this study. This trend was expected to continue in the future. The
combination of increasing water demand, climate-related chronic droughts, and depletion of the
Upper Floridian Aquifer as the main source water was the central concern to local water resource
managers who were tasked to provide adequate and sustainable water supplies for the future.
In response to the chronic drought conditions, the Southwest Florida Water Management
District (SWFWMD) designated the entire western portion of Manatee County as the Most
Impacted Area (MIA) and managed it as a part of the Eastern Tampa Bay Water Use Caution
Area. In May 2008, the Manatee County Board of County Commissioners adopted the Water
Supply Facilities Work Plan (Master Plan hereafter) that describes alternative capital
improvement options for water resource development. The long-term strategies are documented
in the county's development master plan.
75
-------
Legend
^ ManateeWTP
N
o wellfields
Lake Manatee Watershed
W-
E
0 4 8 16
16
S
Kilometers
Figure 2-43 Location of the Manatee County water supply system along the upper Manatee River
in Florida. From Chang et a!. (2012).
The adaptation study documented here was designed to analyze the Master Plan options
for water supply expansion and to identify the most feasible and effective adaptation solutions.
Life-cycle analysis (LCA) approach was used to analyze carbon emissions, energy consumption,
and cost/cash flows. The study systematically considered each phase of planning, design,
construction, and operation for the existing and planned new water infrastructure facilities. The
focus was to find alternatives that reduce CO2 emissions and at the same time, achieve
socioeconomic objectives. Details of the analysis can be found in Chang et al. (2012). By
implementation of the Master Plan and other adaptation measures, the county successfully
provided uninterrupted water supply in recent years, even during the severe drought of 20176.
4.3.1 Water supply assessment
4.3.1.1 Water supplies
Water supply for the county is sourced from surface water and groundwater in the area.
Surface water from Lake Manatee, a man-made reservoir on the Manatee River, provided an
average of 132,111 nr d"1 (34.9 million gallons per day). The permit for withdrawal was
governed by permits issued by the Southwest Florida Water Management District according to
Florida water law (Chapter 373 FS).
6 http://www.bradenton.coni/news/local/articlel45043929.html
76
-------
$
Manatee County
A2
A20
A3
A4
A5
A12 All
A7
A6
Lake Manatee
Reservoir
WTP
ASR
A10
ECWF-I
A1
A8
MPWF
•S1
Si
Ol
Si
A9
County Boundary
A13
A14
A15 A16
A18 .
Al^ A19I
Figure 2-44 Locations of WTP, ASR, well fields (ECWF-1, MPWF), and the twenty potential water
supply alternatives A1-A20. Not drawn to scale. From Chang et al. (2012).
Groundwater for water treatment was derived from two local wellfields: East County
Wellfield I (ECWF-1) and the Mosaic Phosphate Wellfield (MPWF). Their relative locations are
schematically shown in Figure 2-44. The ECWF-1 wellfield was permitted for average annual
withdrawals at 60,514 m3d_1 (16.0 million gallons per day), while MPWF was permitted for
7,419 m3d_1 (2.0 million gallons per day). The Lake Manatee Water Treatment Plant (WTP)
located in the southwest of the Lake Manatee was the only WTP in the Manatee County,
providing all potable water supplies from the Manatee County Utility Department (MCUD). The
treatment plant had a maximum operating capacity of 317,975 m3d"' (or 84 million gallons per
day); 204,412 m3d_1 (or 54 million gallons per day) was for surface water treatment and 113,562
m3d"' (or 30 million gallons per day) was for groundwater treatment.
A total of six (6) aquifer storage and recovery (ASR) wells were located next to the Lake
Manatee WTP. The ASR wells were used to store treated potable water in the Floridian Aquifer
and for withdrawal to augment water supply during the drought season. These ASR wells had
been in operation since 1986. This operation was permitted to provide up to 11,356,235 m3 (3
billion gallons) of storage with a combined capacity of 37,854 m3d_1 (10 million gallons per day).
Figure 2-44 schematically shows the locations of water supply system components,
including Lake Manatee WTP, the ASR wells, Lake Manatee surface water system, and the two
groundwater wellfields. Manatee County also connected three (3) regional wastewater treatment
plants to a 32-mile regional distribution system called Manatee Agricultural Reuse Supply
(MARS) for customers to use reclaimed wastewater for agricultural irrigation. The use of
reclaimed water saved groundwater from the Florida Aquifer that would otherwise be used for
irrigation. The saved credits from reduced groundwater use became the net benefits that could be
used as future potable water sources.
77
-------
4.3.1.2 Water demand
MCUD provided water to retail customers, significant users, and wholesale customers.
Retail customers distributed in both incorporated (e.g., administrative) areas and unincorporated
areas of the County. Significant users refer to those with demands over 94.635 m3d"' (or 25,000
gallons per day). In 2006, this category of customers accounted for approximately 8782 m3d"' (or
2.3 million gallons per day). Wholesale customers included the cities of Bradenton, Palmetto,
Longboat Key, and some regions in Sarasota County south of the Manatee County. Detailed
water demand for wholesale customers is listed in Table 2-11. Reserve capacities available to the
wholesale users remained constant over time as defined in the fixed water supply agreements.
Table 2-11 Water demand in 2006 and projections for wholesale customers in annual average (Board
of County Commissioners, 2008)
Wholesale Customers
Water Demand (cubic meters per day)
2006
2010
2015
2020
2025
2030
City of Bradenton
1892.7
1892.7
1892.7
1892.7
1892.7
1892.7
City of Palmetto
7570.8
7570.8
9463.5
10,409.9
11,356.2
12,113.3
Town of Longboat Key
9463.5
9463.5
9463.5
9463.5
9463.5
9463.5
Sarasota County
37,854.1
30,283.3
22,712.5
18,927.1
NA
NA
Note: NA - Not available
Future water demands for retail customers and significant users were generally unknown
because of the uncertainty in socio-economic development. Detailed population projections
using historical population trends (Board of County Commissioners, 2008) were used as the
basis to calculate future water supply needs. A constant water usage rate per capita, according to
the Master Plan, was assumed for the period of analysis. The water usage per capita in the
MCUD service area was set by permitting and planning. The total municipal water demand
estimated for MCUD corresponding to the population growth is listed in Table 2-12.
Table 2-12 Water demand projections for retail and significant users in annual average (Board of
County Commissioners, 2008)
Customers
Water Demand (cubic meters per day)
2006
2010
2015
2020
2025
2030
Retail customers
115,455.1
115,303,6
132,186.6
149,864.5
168,299.4
187,605.0
Significant users
8,782.2
14,346.7
16,466.5
18,662.1
20,933.3
23,356.0
The historical data show an average daily demand of 181,018 m3 (or 47.8 million gallons)
in 2006. The demand total included 115,455 m3d_1 (or 30.5 million gallons per day) for domestic
water usage, 65,563 m3d_1 (or 17.3 million gallons per day) for wholesale customers and
significant users. With the projected population growth, annual average potable water demand
would increase to 234,317 m3d_1 (or 61.9 million gallons per day) by 2030 as specified in the
Master Plan (Board County Commissioners, 2008). MCUD had an annual average of the
permitted water supply of 200,059 m3d_1 or 52.9 million gallons per day in 2009. According to
78
-------
the Master Plan, a total of 34,447 m3cT' (or 9.1 million gallons per day) of additional water
supply would be required by the year 2030.
4.3.1.3 Future water supply alternatives
MCUD identified twenty potential water supply alternatives to meet the increased water
demands in the future. The master planning called for a combination of surface water and
groundwater source expansion options. They are grouped into five categories: groundwater
options, surface water options, water permit transfer options, regional water options, and other
options. Table 2-13 lists the twenty competing water supply alternatives. Groundwater options
included building new wellfields in various locations of Manatee County. By operating the
MARS system with less groundwater for irrigation, MCUD could increase the permitted
groundwater pumping for potable water supply. Overall, the MARS projects consist of four
phases: MARS-I, MARS-II, MARS-III, and MARS-IV; MARS-I and MARS-II projects had
been implemented as of 2012 (Chang et al., 2012).
Table 2-13 Twenty alternatives for water supply expansion in the county master planning
Alternative
No.
Brief Description
Groundwater Options
MARS-I
1
This option is to supply new groundwater by developing a new
wellfield in the central Duette Park area near the existing ECWF-1.
MARS-II
2
This option is to supply new groundwater by developing a new
wellfield in the Erie Road Tank site.
MARS-III
3
These options are to supply new groundwater by developing a new
wellfield. The location of the new wellfield has not yet been decided.
MARS-IV
4
Surface Water Options
Lake Parrish
Reservoir
5
This option is to divert more surface water from the Little Manatee
River into the existing Lake Parrish Reservoir located in the northern
part of Manatee County as a cooling pond for a power plant. The
increased water storage in the Lake Parrish Reservoir is used for
irrigation to obtain well credits. Improvements in the existing systems
include upgrading diversion pumps, distribution pumping, and piping
facilities.
Dredging of Lake
Manatee
6
This option is to increase the storage of the Lake Manatee Reservoir
to increase the surface water annual yield from Lake Manatee. The
capital investment includes the creation and maintenance of a new
reservoir and dam, wetlands mitigation costs, and water transmission
and treatment at the existing water treatment plant. This alternative
may or may not be funded by the Southwest Florida Water
Management District (SWFWMD).
79
-------
Alternative
No.
Brief Description
Gilley Creek
Reservoir
7
This option is to build a new reservoir upstream of Lake Manatee at
the Gilley Creek location to yield more annual surface water. This
alternative may or may not be funded by SWFWMD.
North and East
Fork Reservoir
8
This option is to create an upstream impoundment at the North and
East Fork locations to increase storage and yield available at the
Lake Manatee intake. The capital investment includes the creation
and maintenance of a new reservoir and dam, wetlands mitigation
costs, and water transmission and treatment at the existing water
treatment plant. This alternative may or may not be funded by
SWFWMD.
Tatum Reservoir-
Lake Manatee
WTP
9
This option is to develop a reservoir to store surface water diverted
from the Myakka River located in the southeastern portion of
Manatee County. The stored surface water from the Tatum Reservoir
is used for irrigation purposes so that the water credits that originally
are used for irrigation can be transferred for potable water supply.
The facilities to be built include an impoundment structure and
distribution pumping and piping.
Transferred Water Use Permit Options
Well Credit from
Current Reuse
Customers
10
This option is to renegotiate with the current reclaimed water
customers for increased reclaimed water flows in the new agreement
term. The cost associated with this alternative is for pumping to and
treatment at the existing water treatment plant.
Developer-
Provided Water
Use Permits (WUP)
Transfer
11
This option is to implement a management option that will require
new farmland developers to obtain the previous landowner's water
use permit as a part of a land purchase. In this way, the Manatee
County Utility Department (MCUD) can take off the burden of
increasing the water supply to the new potable water demand of new
developers.
Direct Purchase of
WUP
12
This option is to buy water use permits from permittees who are
discontinuing farming operations instead of making new developers
purchase the water use permit. This alternative conflicts with option
No.11; Manatee County wishes to forego the option if option No.11
can be implemented.
Regional Water Options
Peace River Water
Treatment Facility
Expansion
13
This option is to improve the existing Peace River water treatment
facility in Desoto County by the construction of a new 6.0-billion gal
reservoir and expansion of the water treatment facility's production
capacity from 12- to 24- and, finally, to 48-million gal/day.
Shell Creek
Restoration
14
This option is based on improvements on the existing Shell Creek
water system by restoration and enhancement of natural water
storage areas. This alternative is for potable water supply to the City
of Punta Gorda and the region. An environmental benefit is identified
for this alternative because of the restoration of natural conditions.
Dona Bay/Cow Pen
Slough Restoration
(Option A)
15
This option is to build a new surface water supply system located
within Sarasota County. Dona Bay Option A is a two-phase project.
The first phase is to build a new reservoir and a new water treatment
80
-------
Alternative
No.
Brief Description
plant at the Dona Bay site, and the second phase is to expand the
size and capacity of the reservoir and the water treatment plant.
Dona Bay/Cow Pen
Slough Restoration
(Option B)
16
This option is to build a new surface water supply system located
within Sarasota County. Dona Bay Option B is a single-phase project.
This alternative conflicts with option No. 15.
Flatford Swamp
Restoration
17
This option is to build a new water supply system at the Flatford
Swamp area located in the southeastern portion of Manatee County.
The water source comes from the excess irrigation runoff in Flatford
Swamp that causes widespread tree mortality. This alternative
conflicts with options No.18 and No.19.
Other Options
Flatford Swamp -
Stored and Treated
at Tatum Reservoir
18
This option is to pump the surplus water stored in the Flatford
Swamp, which is located in the southeastern portion of Manatee
County immediately north of Myakka City to the Tatum Reservoir for
storage and to build a new water treatment plant to treat the water to
potable water standards at the Tatum Reservoir site. This alternative
conflicts with options No.9, No.17, and No.19. This alternative may or
may not be funded by SWFWMD.
Flatford Swamp
supplemented with
Diversion from the
Myakka River -
Stored and Treated
at Tatum Reservoir
19
This option is similar to option no. 18. The difference is that this
option will divert seasonal surface water from the Myakka River to
supplement the Flatford Swamp irrigation runoff. Diversion structure,
pumping facilities, and additional capacity of the new water treatment
plant will be needed. This alternative conflicts with option No.9,
No.17, and No.18. This alternative may or may not be funded by
SWFWMD.
Seawater
Desalination
20
This option is to treat seawater to potable water standards. New
seawater desalination facilities at the Port Manatee site need to be
built. High operation and maintenance costs may be experienced.
However, potential price reduction equipment and funding from
SWFWMD may make this alternative a competitive one.
Surface water options refer to those alternatives that involve new or expansion of existing
reservoirs, by which additional surface water could be diverted from rivers into the reservoirs
during the wet season. Some of the surface water could be used for irrigation purposes without
treatment at Manatee WTP. This amount was then counted as groundwater credits for MARS-I
expansion. The expansion timing for MARS-III and IV were unknown. Groundwater credits
could be reserved for the MARS-I expansion when replaced with surface water sources. Water
permit transfer options were possible where a water use permit holder no longer needed the
water or where reclaimed water became available.
Regional water supply is another option. The Peace River Manasota Regional Water
Supply Authority (PR/MRWSA) intended to integrate and improve water resource management
in Charlotte County, DeSoto County, Manatee County, and Sarasota County in order to provide
the region with an adequate, reliable, and sustainable water supply into the future. Starting in
2014, the PR/MRWSA had begun providing water to Manatee County. Other water options
considered in the master planning process included seawater desalination and swamp restoration
at the Flatford Swamp in southeastern Manatee County. The Flatford Swamp received a
81
-------
significant amount of irrigation runoff. Reducing the irrigation runoff flow into the swamp was
predicted to help re-establish hardwood trees in the swamp and reduce environmental impact.
Seawater desalination involved building a seawater treatment plant at the Tampa Bay.
Figure 2-44 schematically illustrates the relative locations of all twenty potential water
supply alternatives. Among them, locations of alternatives #10, #11, and #12 are shown only for
illustration; these three alternatives require no physical facilities. Some of the twenty alternatives
may be eligible for SWFWMD funding, thus decreasing the county's capital investment and,
thereby, the unit cost of potable water. It was noted, however, that the SWFWMD funding was
not guaranteed even when all required criteria were met. In the comparative analysis, the highest
(conservative) unit cost was used for the alternative following the common practice of
engineering feasibility analysis.
Table 2-14 summarizes the maximal water credit and unit cost for each of the 20 water
supply alternatives. The maximum water credit was defined as the maximum permitted water
withdrawal. Unit cost was calculated as the present value for a cubic meter in U.S. dollars based
on the 2007 value. It includes the amortization of the estimated initial capital investments and the
operation and maintenance (O&M) costs.
4.3.2 Expansion scenario analysis
Most decision-making systems nowadays rely on a single attribute; for example,
economic cost or water supply capacity. Traditional decision-making mostly relies on the
outcome of a cost and benefit analysis in the context of single-objective optimization, which was
of particular interest to both water supply (Urbaniak, 1988; Slowinski et al., 1985) and
wastewater treatment (Ong and Adams, 1990).
Table 2-14 Maximum water credit and unit cost of the twenty water supply alternatives*
1
2
3
4
5
6
7
8
9
10
Max Water Credit
8.21
11.36
7.57
18.93
15.52
44.29
34.83
40.13
17.79
17.03
Unit Cost (dollars
per cubic meter)
0.34
0.53
0.31
0.50
0.51
1.09
0.67
0.74
1.08
0.50
11
12
13
14
15
16
17
18
19
20
Max Water Credit
ot
ot
45.42
75.71
75.71
75.71
56.78
30.28
43.15
37.85
Unit Cost (dollars
per cubic meter)
0.53
0.60
0.30
0.51
0.76
0.62
0.72
0.61
0.55
1.07
Notes: *Adopted from the 2008 Manatee County Water Supply Facilities Work Plan (Board of County
Commissioners, 2008)
|The maximum (max) water credits for alternative #10 and #11 are not available and the value of zero
was assigned as the default. The maximum water credit is 1000 m3/day.
However, the cost-saving alone does not reflect all sustainability attributes in evaluating
the adequacy of competing for water supply expansion options. In the case study of Manatee
County's water expansion, a decision-making framework included both carbon footprint and
economic cost, in addition to the management objective of adequate water supplies. Two
82
-------
approaches in a systems evaluation are common in finding the global optimal solution among
competing alternatives. One is a top-down modeling assessment; the other is bottom-up
threshold analysis. Optimization models for the top-down water supply system planning have
been developed to address multiple planning goals (Harrington and Gidley, 1985; Yamout and
El-fadel, 2005). Various analysis techniques are long available including nonlinear programming
models (Mulvihill and Dracup, 1974) and multicriteria decision analyses (Slowinski et al., 1985).
The framework was based on the LCA method for comparative alternative evaluation.
The purpose was to select the best expansion options. LCA is a well-established and
standardized method of analysis for cost comparisons and can be applied to evaluate and reduce
possible environmental impacts as a part of sustainability analysis. For example, some LCA
investigations use greenhouse gas (GHG) emissions in form of carbon footprint, as one of the
evaluation categories to evaluate multiple technical solutions or alternatives for municipal
wastewater treatment systems (Tillman et al., 1998; Dennison et al., 1998; Lundin et al., 2000;
Peters and Lundie, 2001) and in water supply assessment (Voivontas et al., 2003; Lundie et al.,
2004).
Evaluation of the expansion alternatives in Manatee County followed the LCA principles
in an analysis of the cost and life-cycle GHG emissions. Based on these determined parameters,
a multi-objective optimization scheme was developed to identify the global optimal planning
solutions (Chang et al., 2012).
4.3.2.1. Carbon footprint estimates
The carbon footprint is a sum of CO2 equivalents in all phases of each expansion
alternative. The time duration for this analysis was twenty years (from 2011 to 2030) during
which the construction, production, use, and recycle phases were analyzed sequentially as shown
in Figure 2-45. The systems analysis diagram shows material and energy flows, where each
Input; steel
Input: ccmcnt
A &
Input; energy
A
Construction Phase
Raw materials
acquisition (steel
and cement)
Facility
construction
Production Phase
I
I
I
Use Phase
ja
rr^
TV
I
I
I
I
Raw water
transportation
n-LK
Trv'
Potable water
distribution
©
Treatment of raw
water at WTP
©
Sewage and
wastewater
collection
Input; raw water
A
Recycle Phase
Treatment of
wastewater at
wastewater WIPs
jSJ
Reclaimed
water reuse
for irrigation
Discharge to
sea or river
System Boundary1
Output: CO; equivalent
Data Layer
Model Layer
Output analysis
Figure 2-45 The life-cycle system analysis flow diagram for determining carbon footprint in water
infrastructure expansion alternatives. Process is divided into three layers. Adopted
from Chang et al. (2012).
83
-------
block represents material stocks and is connected by arrows to surrounding blocks indicating
critical material flows. Materials, or raw water in this analysis, are extracted at the beginning of a
life cycle, pass through intermediate phases, and finally return to the environment at the end of
the life cycle. In this LCA analysis, the end-of-life phase of water facilities is not included
because water facilities usually have a service life far beyond the analysis period.
Chang et al. (2012) analyzed the carbon footprint for all twenty water supply alternatives.
Their results were used in this LCA analysis. The emissions in construction and operation phases
were calculated in the 20-year period. The total CO2 equivalent emission in a 20-year period was
the sum of CO2 equivalent emissions for construction and operations. As shown in Figure 2-45,
the construction phase includes the processes © and ©. When a potential water supply
alternative was selected and implemented, the CO2 equivalent emissions were determined for
both facility construction and operation. The operation phase included the processes ®, @, ©,
©, ®, ®, and ©.
The LCA analysis results are listed in Table 2-15. The CO2 equivalent emissions were
proportional to the amount of water supplied, but the total energy usage in the water
infrastructure life-cycle computation varied significantly among the expansion options (Table 2-
15).
Table 2-15. Life cycle analysis of carbon footprint for the twenty water supply alternatives (modified
from Chang et al., 2012)
Alternative
CO2 equivalent emissions in
constructional phase Process
® + ®
(kilograms)
CO2 equivalent emissions in
operational phase. Process
®+©+®+©+®+®+®
(kilograms per cubic meter)
Groundwater
1
1.96x1010
2.35
2
2.85x1010
2.68
3
2.08x1010
2.48
4
4.11 x1010
2.87
Surface water
5
3.40x1010
2.71
6
1.88x1010
1.16
7
2.67x1010
1.99
8
8.91 x1010
3.75
9
4.63x1010
3.13
Water use
permit
transfer*
10
Negligible*
1.16
11
Negligible*
1.16
12
Negligible*
1.16
Regional
water
13
1.81xio11
5.89
14
2.72x1011
6.85
15
1.07x1011
3.35
84
-------
Alternative
CO2 equivalent emissions in
constructional phase Process
® + ®
(kilograms)
CO2 equivalent emissions in
operational phase. Process
®+®+©+@+®+®+®
(kilograms per cubic meter)
16
1.07x1011
3.35
17
6.56x1010
2.71
Others
18
4.88x1010
2.71
19
5.75x1010
2.71
20
6.28x1010
3.28
*Water permit transfer is simply an administrative action with almost no obvious carbon footprint
relative to other options.
4.3.2.2. Multi-objective evaluation
A multi-objective mixed integer programming modeling was conducted to assess these
multi-stage expansion strategies based upon the LCA and cost evaluation results. The analysis of
future water supply scenarios covered a 20-year timeframe from 2011 to 2030. The trade-off
analysis for a compromised solution was based on two objectives. One was to minimize the total
system costs required for the water supply expansion. The other was to minimize the total GHG
emissions expressed as CO2 equivalent, reflecting overall energy consumptions. Both objectives
were applied through modeling to screen and order the relevant water supply alternatives.
Simulations using a compromise programming model yielded the best Pareto frontier
solutions for the alternative expansion options. The model computation was a function of the
total number of planning periods in the multi-stage framework of infrastructure planning,
construction, operation, and disposal (or decommission) (see Figure 2-45). The time interval
was, generally, location-specific depending on local decision-making objectives. More stages
and greater implementation details lead to more decision variables and parameters that require
greater computation time. For this illustration, a 5-year time span was assumed for the
construction phase in the case-study decision analysis. The 5-year duration is generally in
agreement with the capital expenditure process.
4.3.3. Quantitative modeling and systems analysis
In quantitative analysis, the multi-stage planning horizon was divided into four time
periods with each having a 5-year time span. Decisions in each period of expansion to meet the
growing water demand were evaluated in a trade-off analysis between the two objectives. The
Multi-objective and Multistage Mixed Integer Programming model used in the Manatee County
evaluation is described below.
Objective Functions
Two governing objective functions were implemented for each 5-year interval. The
carbon emission equivalent (C02,eq) was calculated using the LCA procedures in Figure 2-45.
Monetary values of cost were discounted to the year 2007. According to Chang et al. (2012),
Objective function 1:
85
-------
Minimize Zi = total C02,eq emissions (unit: g) =
20 4 20
^(l000i4ilCO2 eo. X 1825 + Yi±C02^ + ^ ^[l00(MitCO2,ee. X 1825 + (Yit - yi(t_1))C02,ecJ
i=l t=2 £=1
Objective function 2: Minimize Z2 = total cost (unit: $) =
20 4 20
^(lOOO^C; X 1825 + Y^Fi) + ^^[1000^ X 1825 + % - Y^.^F,]
i=1 fc=2 £=1
where I# is 1 if the alternative z is implemented in and after time stage /; otherwise Yu = 0, z = 1,
2, ..., 20; ^ = 1, 2, 3, 4 for the period of 5-year implementation interval. C02,ec; is the amount of
C02,eq emission in the construction phase of alternative z in unit of g, and CCh.eo/is the amount of
C02,eq in the operational phase of alternative z in unit of g m"3, z = 1,2, ..., 20. An is actual water
withdrawn (103m3d_1) from alternative z'(= 1,2, ..., 20), and t= 1,2, 3, 4. G is unit water cost of
the alternative solution z in $ m"3 (z = 1, 2, ..., 20). Ft is the fixed capital investment for the
alternative solution z (= 1,2, ..., 20).
Model Constraint Setting
Constraint setting in the compromise programming model included definitional
constraints, water demand constraints, capacity limitation constraints, availability constraints,
sequencing constraints, mutually exclusive constraints, irreversible constraints, screening
constraints, and non-negative and binary constraints. These constraints provided different
functionalities in an intertwined solution space that narrowed down the dynamic selection and
ranking according to streamlined logic described by the coupled objective functions and
constraints over the planning time horizon.
Seven model constraints below defined both the current maximum water supply and the
projected water demand in the unit of 103 rnVday in each time period:
S = 200.04 1 03m3d_1
(2.3)
Di = 192.19 1 03m3d_1
(2.4)
Di = 209.14 1 03m3d_1
(2.5)
£>3 = 211.83 103m3d_1
(2.6)
Da = 234.43 103m3d_1
(2.7)
Fi = 0.001 $
(2.8)
G = a large dumb number (e.g., 9999999)
(2.9)
where S is the limit in current water supply upper; Dt is water demand in time period t (= 1, 2, 3,
4); Ft is the virtual fixed-cost, an artificially assigned small number relative to all cost parameters
to support screening logic in the cost-effectiveness objective and associated constraints. The use
of dumb number G in programming is to assure computing stability for the If-Then logic
screening in constraints by Eqs.(2.10)-(2.12) below. The settings of Ft and G also help avoid the
selection of an alternative with no additional water supply over the planning horizon:
86
-------
The constraints between demand (D) and supply (S) were applied to the entire 20-year
period in the modeling space:
20
~ l)< ~S f°r all t
(2.10)
;=1
¦ The water amount supplied by each future water source could not exceed its
predetermined supply limit due to water rights:
Ait < for all t and all i
max¦
(2.11)
in which a "iX is the maximum water credit (103m3d_1) for At, i = 1,2,..., 20.
¦ Only MARS-I and MARS-II were available in the time period 1 and the rest of future
water supply alternatives were available only after time period 1:
Construction sequencing constraints
This set of constraints assured that the MARS-II project was not be implemented until the
completion of the MARS-I project according to the County's infrastructure expansion work plan.
Similarly, the MARS-II project implementation could only occur before the MARS-III project.
This forward-looking sequence applies to MARS-III project that might be implemented ahead of
the MRAS-IV project. Mathematically,
Mutually exclusive constraints:
Some future water supply alternatives were mutually exclusive according to the county's
original work plan. This set of constraints assured that only one of the exclusive future water
supply alternatives could be implemented in any time period. For example, Alternatives 11 and
12 are mutually exclusive because the water use permit allocation is either transferred from
developers to the county or otherwise acquired by Manatee County through other means. An
example was to exchange the county's reclaimed water for groundwater currently used for
agricultural irrigation. The MARS-III project conflicts with the other regional water supply
alternatives because any one of the regional water supply sources or implementation completion
of the MARS projects could provide adequate water supply (Board of County Commissioners,
2008). Alternatives 15 and 16 are mutually exclusive because both alternatives intended to use
the same water supply sources but in different implementation schedules. Similarly, Alternatives
17, 18, and 19 are mutually exclusive because all three alternatives rely on Flatford Swamp as
the water source. The three alternatives differ because of assumptions on the construction of a
new WTP as a part of the regional water supply option. Alternatives 9, 18, and 19 are mutually
exclusive because all three were related to a new reservoir site at Tatum. The difference is
/ = 1,2
/ = 3,4,...,20
(2.12)
for all t
(2.13)
87
-------
whether the new reservoir site would be used to store water pumped from the Myakka River or
the Flatford Swamp.
The construction sequences and mutually exclusive constraints were mathematically
expressed in modeling. Eqs.2.19-2.22 define the need for MARS-I implementation before
considering relevant Alternatives #5, #9, #10, and #11, because of the constraints from
sequential water credit transfer. These constraints are as follows:
for all t
(2.14
Yz+Yn+Yn+Yn+Yn+Y^l
for all t
(2.15
for all t
(2.16
Ym+Ym+Ym -1
for all t
(2.17
Y +Y +Y <1
29t TJ18f TJ19f —1
for all t
(2.18
Y
-------
YJYnAr-(D2-S) 0 (2.27)
Yi,t = 0,1 7=1,2, ...,20, f= 1,2, 3, 4 (2.28)
Yi, 72, 73 = 0, 1 (2.29)
4.3.4. Adaptation analysis results on cost and carbon/energy footprint
4.3.4.1. Carbon/energy footprint and cost optimization
Optimal solutions were identified using the multi-objective model simulation by solving
each of the individual objective equations sequentially. The solution (shown in Table 2-16) was
considered optimal when each objective is optimized individually and achieved as a whole. One
caveat is that the ideal solution may not be feasible or practical because the objectives may be
competing, even conflicting in the decision space. In this type of application, the "Pareto
Optima" solution set is commonly used. The solution optimization can be found in the Pareto
Optima frontier in the solution space of the compromise programming model. Alternatively, the
compromised solution can also be obtained by applying the distance-based metrics defined in a
compromise programming model (Zeleny, 1973).
The solution space for the Manatee County case study is two-dimensional as defined by
the two objective functions. The x-axis was selected for CO2 equivalent emissions (Zi) and the
axis for total system cost {Zi). In all cases, a Pareto Optimal solution in global optimization space
represents the best alternative that may perform better for both objectives. For the exclusive
optimization decision (n = 1), five sets of solutions are found in Table 2-17. Solution #1 is the
89
-------
GHG-effective solution; Solution #5 is the most cost-effective; and the other three are
compromised solutions. For alternative management decisions at n =2 and n = 3, the Pareto
Optimal solutions were found and described in Chang et al. (2012).
Table 2-16 Optimal solutions of the multi-objective model
Minimize Zi
Minimize Zi
(kg)
($million)
n = 1
1.15xio11
223
3
II
ro
7.55x1010
172
3
ii
CO
7.54x1010
172
Note: n is the number of alternatives allowed in one set of ideal solutions.
A sensitivity testing for the optimal solution was conducted at assumed 10% uncertainty
in the estimates of future water demand change at all four stages. For water managers, future
prediction is the basis for decision making and often contains uncertainty. The sensitivity
analysis results are shown in Table 2-18. The best case is that future water demand is 10% less
than the prediction or 0.9A; conversely, the future water demand of more than 10% than
predicted or 1.1 A is the worst-case scenario in planning, requiring attention in further analysis.
Table 2-17 The Pareto optimal expansion strategies (n = 1)
Solution
Number
Zi
Z2
Expansion Strategies
(kg)
($million)
Period 1
Period 2
Period 3
Period 4
1
1.15X1011
313
1,2
17
-
6
2
1.42x1011
295
1,2
17
-
7
3
1.56X1011
260
1,2
16
-
10
4
2.14x1011
258
1,2
16
-
19
5
3.22x1011
223
1,2
14
-
10
Note: Best compromised solution #3 is in bold and italic.
For n= 1, the Pareto Optimal solution sets were examined for the best case 0.9!), (Di =
172.97, Di = 188.23, Z)3 = 190.65, and Da = 210.99) and the worst case of 1.1A (Di = 211.41, Di
= 230.05, Di = 233.01, and £>4 = 257.87). Solutions marked by or "+" represents 10% lower
or higher water demand than the predicted level in the Master Plan, respectively. The results
indicate robust analysis conclusion insensitive to 10% uncertainty in water demand projection, as
the Pareto Optimal frontier remains unchanged in shape.
4.3.4.2. Optimal expansion solutions and construction sequence
Water supply system expansion normally takes place in phases considering water service
needs and economic factors such as capital flow and construction cost. The preceding analysis
for the Manatee County water infrastructure expansion showed multiple compromised solutions
in the trade-off between the overall system's cost and life-cycle carbon footprints. In engineering
practice, master planning of water infrastructure improvement often considers other factors such
90
-------
as land availability, engineering feasibility, capital expenditure, and cash flow, among the others.
For these reasons, one can further assess the best options in the Manatee County case study,
which offer the optimal management options. Such an assessment is discussed below.
Table 2-18 The Pareto optimal expansion strategies for the best and worst cases (n = 1)
Solution
Number
Zi
(kilograms)
Z2
($million)
Optimal Expansion Strategies
Period 1
Period 2
Period 3
Period 4
1-
4.87x1010
2.77
1,2
6
-
10
2-
7.54x1010
193
1,2
7
10
6
3-
9.63x1010
185
1,2
7
10
3
4-
1.06X1011
172
1,2
19
-
10
5-
2.31 *1011
111
1,2
13
-
10
1 +
1.57x1011
347
1,2
16
10
6
2+
1.84x1011
339
1,2
16
10
7
3+
1.91 x1011
335
1,2
16
10
5
4+
2.14x1011
333
1,2
16
10
19
5+
3.23x1011
303
1,2
14
10
6
6+
3.50x1011
296
1,2
14
10
7
7+
3.57x1011
293
1,2
14
10
5
Note: Best compromised solutions 4" and 4+ are in bold italics. From Chang et al. (2012).
Best Compromised Solution
The optimal solutions represent the best combination of systems' cost and carbon
footprints for the projected future water demand at the 10% uncertainty bounds. To find the best
compromised solutions for all three sets of future water demands, the two objective functions are
normalized for Zi and Zi in the same scale between 0 and 1. The normalized objective functions
(NZ\ and NZi) are given by:
y ymin
NZ, =—1 . (2.30)
i y max y mm v y
1 ~~ L\
ry ^min
NZ,= 2~ 2 . (2.31)
^ y max 7mm v y
2 ~~ 2
The normalized solution space for optimal water infrastructure solutions is shown in
Figure 2-46. By the normalization, the best solution can be found by the distance to an
imaginable solution of zero cost and zero carbon emission or the origin (0,0) in Figure 2-46. A
91
-------
widely accepted definition of such distance is based on Minkowski's La metrics (Zeleny, 1973),
where 1 < a < qo .
L. =
(2.32)
Normalized Zy
Figure 2-46 Pareto solution fronts for the best compromised solutions to meet the projected future
water demand (base case) and the demand with 10% uncertainties (best case and
worst case). From Chang et al. (2012).
For water managers, a = 1 means equal weighing for both objectives; a = 2 implies a
weighted geometric distance between the solution (NZi, NZi) to the ideal solution (0, 0); and
a=cc implies minimization of the maximum NZi when La is to be minimized. The parameters in
the Manatee County case study were set at a = 2 and wi = wi = 1, for an illustration. Using these
assumptions, the best compromised solutions are:
¦ For the projected water demand, the best compromised solution #3 would cost $2 million
or 0.8% more than the next less expensive option. The CCh.eq emissions would decrease
by 27.1%.
¦ For the best case with lower future water demand, the best solution #4+ would cost more
by 55%) or $61 million than the next less expensive option. It would result in a 54%
reduction in C02,eq emissions.
¦ For the worst case of 10% higher future water demand, the best solution #4- would
increase the cost by $30 million or 9.9% more than the next less expensive option. The
carbon emissions reduction would be 33.9%.
92
-------
Decision support in master planning
It is noteworthy that the Pareto front is not continuous because practical water
engineering solutions are discrete. The trade-offs of the best discrete alternatives, as described
above and in Tables 2-20 and 2-21, present a quantitative basis for managers to use for decision-
making. The co-benefits and compromise among emission reduction, cost saving, and
engineering feasibility are obvious when adaptive planning is considered for water infrastructure
expansion.
Based on the optimization results, Figure 2-47 shows the optimal facility expansion
strategies for each of the five-year implementation periods. Both the cost and carbon emissions
are considered with the following conclusions:
¦ In the base case and the best case, the current water supply would be self-sufficient in the
first five-year period. The modeling results indicate that if water demand is higher than
the forecast, extra water resources would be needed. Then the MARS-I and MARS-II
projects could provide sufficient water supplies to meet the demand until 2025.
¦ According to Chang et al. (2012), the need for and the nature of optimal expansion
strategies in this time-period are sensitive to the forecasted water demand. The regional
water option offers larger water supply capacity and, at relatively lower unit costs, than
other alternatives. It could be needed in the worst case. In the best case, however,
regional water supply options could be avoided due to their relatively larger carbon
footprints due to the long-distance water transfer. Other alternatives available within the
Manatee County could provide better performance to satisfy both objectives, and
therefore could be used as a contingency.
¦ Water demand is anticipated to increase further starting from 2026. The modeling results
indicate a variety of expansion strategies available for selection. In all cases, the water
use permit alternative (e.g. alternative #10) would be always preferred due to its zero-
carbon footprint, or energy neutral, and low unit cost. Not coincidently it was considered
as a priority in the county's master planning.
¦ For the worst case in future water demand, MARS-I and MARS-II would be still the most
desirable alternatives by 2016 (Figure 2-47). Regional supply alternatives would be cost-
effective compared to other alternatives except for the MARS projects. They may not
represent the most favorable solution in carbon emissions, because of necessary facility
expansion/construction and long-distance water transfer. Instead, Dona Bay/Cow Pen
Slough Restoration Option B (alternative #16) could be selected as a compromised
solution. It has the lowest carbon emission impacts among all the regional alternatives.
Worth noting, there are several limitations in this analysis. These include uncertainties
surrounding water pricing, a discount of the potential to receive SWFWMD funding, and thus
arbitrarily higher unit costs, among the others. Their impacts on the determined optimal solutions
were not evaluated.
5. System-Scale Adaptation for Existing Urban Water Infrastructure
As shown in Figure 2-6, urban planning is one major element in infrastructure adaptation.
Urban-scale adaptive planning of water and transportation infrastructure has the potential to
generate adaptation co-benefits and improve the urban performance and resilience against the
93
-------
2011-
2016-
2021-
2026-
Figure 2-47 Suggested optimal facility expansion strategies in each of the five-year periods based on the optimization modeling of water
infrastructure expansion options for Manatee County, Florida. From Chang et al. (2012).
(a) Worst Case Base Case Best Case
A13
94
-------
impact of hydroclimatic and land use changes. Examples of co-benefit potential were examined
in preceding Section 4.0. The next step is adaptation at local system-scales. This critical
adaptation step is focused on specific water systems and infrastructure components (Figure 2-2)
in the engineering steps (construction and operation) after master planning (Figure 2-6).
One key element of system-scale adaptation is to define the limit of adaptation actions.
Felgenhauer and Webster (2013) defined the adaptation limit as the point beyond which
adaptation's economic return is diminished and a paradigm shift is necessary. To determine this
threshold, engineering assessments with an accurate technical basis are necessary to evaluate
adaptation potentials and their feasibility. When necessary, additional rounds of planning-
engineering-adaptation may be conducted in periodic planning revisions.
Here in Section 5.0, these topics and basic considerations for adaptation are detailed in
terms of the system-scale engineering approach. The concept of the capacity reserve (CR) (Yang,
2016; Levine et al., 2016) is introduced in the analysis of the threshold to define adaptation
engineering needs and limitations. Other contents are withdrawn from publications of this
research (Yang and Goodrich, 2014; Yang, 2010, 2016). In the subsequent Sections 6.0 and 7.0,
case studies will be used to illustrate the engineering approach for CR improvement in drinking
water treatment and distribution. Associated adaptation tools will be described.
5.1. Basic considerations in adaptation engineering
5.1.1. Adaptation engineering for water infrastructure
Engineering design and implementation for adaptation of existing infrastructure can be
broken down into stages of (1) system assessment of adaptation feasibility, (2) adaptation design
and implementation, and (3) effectiveness monitoring and adaptation update. These adaptation
elements are schematically shown in Figure 2-48 in the context of existing water planning and
engineering processes.
The first step is to understand the capacity of existing water infrastructure and its
resilience against the impacts of projected hydroclimatic and land use changes. The process,
marked as Stage "1" in Figure 2-48, follows the traditional water infrastructure planning. Before
the step is the quantification of projected global changes to assess the needed improvement or
renovation of existing infrastructure (see Stage "0" in Figure 2-48). The capacity assessment of
the existing structure is commonly known as the "bottom-up" approach. The purpose is to
evaluate the design capacity and remaining capacity reserve of water infrastructure and then
determine the threshold beyond which the water infrastructure service would be compromised.
An example of this "bottom-up" approach is taken by the EPA's Climate Resilience Evaluation
and Assessment Tool (CREAT)7. In Section 5.3.2, the CREAT tool and its applications will be
discussed.
Adaptation engineering takes place in Stage "2" (Figure 2-48). This second stage is
focused on the improvement of the system's CR by adapting and improving the existing water
infrastructure. New infrastructure or system revitalization often requires a substantial initial
capital investment. Thus, common management practice is to first improve the resilience of
existing infrastructure through capacity improvement before new capital projects for new
7 http://water.epa.gov/infrastructure/watersecurity/climate/creat.cfm
95
-------
infrastructure or systems. This management consideration is also shared in urban transportation
infrastructure management. Nevertheless, as described in Sections 3.0 and 4.0, scenario analysis
for transportation and water infrastructure offers a practical means to identify the most cost-
effective options; mostly by capacity improvement rather than new construction.
By considering climate as a variable, the adaptive planning and engineering approach
deviates from the traditional water infrastructure practice (Figure 2-48). Hydroclimatic variables
important to water infrastructure services include the rate of precipitation change, changes in
watershed hydrologic variables such as runoff properties, and water quality changes. Precipitation
IDF, and ambient and water temperature are the most fundamental hydrologic parameters. All of
them are currently assumed to be constant in hydrological design (Figure 2-48); for example, in
the Atlas-14 IDF design charts (Bonnin et al., 2006, 2011). Small rates of change and
hydrological uncertainties can be managed by using engineering safety factors, a traditional way
to manage engineering uncertainty.
Monitoring, repair,
rehabilitation for
capacity reserve
Hydroclimatic
projections, hydrologic
modeling
Periodic revision
to planning &
operation
o
Vulnerability assessment and adaptation
Revision and adaptation update
Urban development
planning
Hydroclimatic
projections, hydrologic
modeling
Figure 2-48 Assessment-adaptation process for water infrastructure planning and engineering. The
box in dashed line contains the elements of climate and land use projections in
infrastructure master planning. Arrows indicate process direction. Numbered labels
indicate the stages of engineering analysis (See text for explanation).
96
-------
Planning time horizons, normally as long as 30-50 years, is comparable to the time in
which change in hydroclimatic conditions may materialize. Potential implications to water
infrastructure and water programs were discussed in U.S.EPA (2015a). A large rate in
hydroclimatic changes may invalidate the current design basis for existing infrastructure,
prompting the development of a proper design basis for adaptation (Yang and Goodrich, 2014;
Yang et al., 2017). However, the current model projections of future climate and land use often
have substantial degrees of uncertainty (Miller and Yates, 2006; IPCC, 2014, 2007) due to error
promulgation from multiple sources (Hosseinzadehtalaei et al., 2017). The risk of projection
uncertainty is also compounded by other decision factors such as capital investment. Because the
uncertainty is likely to decrease time, flexibility in infrastructure design and program
implementation is important to properly manage the climate uncertainties (Fletcher et al., 2017).
One approach is the iterative adaptive approach outlined earlier in Figure 2-1. For water
infrastructure, this iterative process and associated adaptation strategy are detailed in Figure 2-48
for water systems engineering. This approach can lead to better management of climate risk and
adaptation economics. Case studies are provided for illustration in subsequent Section 6.0-7.0.
5.1.2. Adaptation attributes of three types of water infrastructure
Existing water infrastructure has a large physical footprint that is difficult to change,
without large capital investment. Over the past century, water infrastructure was designed and
constructed mostly underground for anticipated population growth and land use changes to meet
the water needs, while climate and precipitation regimes were assumed to be stationary (see
Figure 2-48). The properties of this infrastructure and its urban service functions, developed
under the stationarity assumption, can severely limit adaptation approaches and engineering
options.
Three principal types of water infrastructure are prevalent in the U.S.: wastewater
collection and treatment, drinking water treatment and distribution, stormwater collection and
management (Figure 2-49). While the service function varies geographically and differs among
types of water infrastructure, general engineering and management principles follow a triple
bottom line of management objectives: protection of public health, safety and welfare; system
reliability; and engineering economics. In this specific context, the drinking water treatment and
distribution in the U.S. are designed to meet regulatory compliance with drinking water quality
standards and to provide an uninterrupted water supply. Centralized wastewater systems serve to
collect wastewater from individual users, transfer it to a location for treatment and subsequent
discharge into a water body under a regulatory permit while minimizing public health risks and
exposure. Onsite small wastewater systems and decentralized wastewater management are the
alternative systems serving small communities and individual households (U.S. EPA, 2002); they
are not discussed here because, by their very nature, they are not used generally in dense urban
communities. Additionally, stormwater infrastructure has been constructed on a massive scale to
provide drainage, sanitation, and flood control in an urban catchment area, primarily to protect
public infrastructure and private property. In the U.S. Northeast and the Great Lakes region,
stormwater and wastewater networks often share the same pipe networks in a combined sewer
system (CSS). Combined sewer overflows (CSOs) occur during high-intensity precipitation,
causing untreated or partially treated wastewater to be by-passed of the treatment plants for
discharge at CSO outfalls. The result is pollution in the receiving water bodies (U.S. EPA, 2001,
2008, 2009a; Weinstein, 2009; Capodaglio, 2004).
97
-------
A) Drinking water
Pre-oxidation
Flocculation
Sedimentation
\ r \///////r
w
Drinking water treatment
Drop manhole
Wastewater collection
transfer
Secondary clarifier
Wastewater treatment
Filtration,
Nitrification /
denstrification
Source water
B) Wastewater
CL, Cft
"Q
C) Stormwater
Stormwater collection / conveyance
q„, c„
BMPs, LIDs in Catchment
Discharge
Figure 2-49 Process schematic diagrams for typical centralized drinking water, wastewater, and stormwater infrastructure in an urban
watershed. Combined sewer system (CSS), stormwater, and wastewater treatment effluent discharges (Qc?, Cd) are regulated
for stream flow (Q0) and pollutant concentration before and after the discharge point (C0 and Cm). Solid arrow indicates water
flow directions. I/O is water inflow and outflow in the buried pipes through infiltration and exfi It ration. Solid triangle indicates a
process unit potentially vulnerable to future precipitation and hydroclimatic changes. From Yang (2016).
98
-------
In the urban water cycle of water-wastewater-storm water, extreme precipitation impacts
on surface urban watersheds can potentially make several infrastructure components vulnerable to
failure, i.e., unable to provide desired service functions. These vulnerable locations are markedin
Figure 2-49. The nature of the vulnerability is explained in Tables 2-22 and 2-23. Further
technical discussions for each are presented subsequently in the aspect of infrastructure CR.
5.1.3. The capacity reserve concept and climate resilience
The sustainability of water infrastructure is shown in its resilience and adaptability to a
changing environment. Resilience is defined here as the ability for a system to recover its
physical state and service functions after an external impact (Milman and Short, 2008; McDaniels
et al., 2008). Capacity reserve (CR) is an important physical attribute that quantifies the resilience
as discussed extensively in sustainability science literature (Tillman et al., 2005, 1998;
Dominguez and Gujer, 2006; Yang, 2016). Some (e.g., Oh et al., 2005; Chen et al., 2008) have
discussed CR in the context of urban carrying capacity.
J
k
*>¦ y*
^ Design capacity
Capacity 1
Reserve
V | Capacity loss
\
„ y Desired service
W
\ in
V
^ ">»» No service
| Aging ,
Damaging | Recovery
to
tj t2
Time
Figure 2-50 Four types of infrastructure vulnerability under the threat of external impact event
(e.g., storm surge). In all cases, capacity reserve is the capacity difference between
the minimum service required and the design capacity. See text for more
explanations. From Yang (2016) and Levine et al. (2016).
The CR concept has long been used in civil, structural and process engineering, referring
to extra capacity or flexibility for assurance of desired structural integrity or performance (e.g.,
Tillman et al., 2005; Matos et al., 2013). A commonly used term is the margin of safety or safety
factor in design. The system resilience in CR is measured by the ability of a water system to
provide a service level above the desired level of service. In engineering, one ultimate goal is to
balance CR, the risk of failure, and adaptation cost. Note that the risk of failure of water
99
-------
infrastructure is the overarching concern. This is why significant redundancy or CR is often built
into these systems. Similar social, environmental, and economic objectives, known as the "triple-
bottom-line," apply to hydroclimatic adaptation in general (Cromwell III et al., 2007).
Figure 2-50 schematically shows the CR concept and its relationship with infrastructure
resilience. The installed CR is a parameter to quantitatively measure the vulnerability threshold
from which the ecological system resilience concept of Marshall and Toffel (2005) is modified
for infrastructure adaptation analysis. For capital-intensive water infrastructure, at the core of
adaptation is the ability to increase its resilience for unanticipated changes and to build-in
adequate flexibility for control of the uncertainty-related risk. This capacity is shown for the
scenario I and II upon an external impact from hydroclimatic or other environmental changes
(Figure 3-51). Scenario II represents the temporary vulnerability of the infrastructure "out-of-
service" below the desired capacity. This condition in urban water supply and sanitation
happened in an increasing frequency in the past decade; examples include the impacts of the 2012
Hurricane Sandy in New York City and the adjacent coastal states, the droughts in Florida during
the 2000s, flooding in Houston in 2017 from Hurricane Harvey, and the ongoing droughts and
wildfires in California. After such hydroclimatic disruptions, some urban water infrastructure
may not recover to the original design capacity. The difference is capacity loss (Figure 2-50),
requiring management attention.
In contrast, the Types III and IV changes in infrastructure service are not fully recoverable
(Figure 2-50). The impact at ti results in permanent impairment of the infrastructure service
functions, while Type IV leads to the failure of water services, a condition that water managers
strive to avoid. In both cases, the infrastructure service functions are significantly impaired in
their ability to provide the desired service. A service recovery requires capital investment for
rebuilding at a significant cost or a paradigm shift to avoid future recurrence of such service
disruptions. Examples of these potential scenarios include the damage by coastal hurricanes,
storm surge, sea level rise and periodic inundation (Comfort, 2006; Turnipseed et al., 2007;
Gesch, 2005; Wing et al., 2002), impacts from water pollution resulting from a climate event
(e.g., Wing et al., 2002; Cann et a., 2013), as well as preventative measures taken for adaptation
and mitigation (e.g., Rosenzweig et al., 2007). In the 2012 Hurricane Sandy, boil water advisories
were issued to a large number of customers and local health agencies, during and after the
disruptive event8.
Significant functional damage to water infrastructure in Type III and IV situations needs
attention in "no-regret" adaptation. It requires adaptation planning because of the long-lasting
effects. While conventional rebuilding and reconstruction are often effective measures, long-term
sustainability has been discussed to avoid repeated system failures. Examples of adaptation
include water supply and sanitation paradigm changes (Gleick, 2000; Pahl-Wostl, 2007), urban
system re-planning, and avoidance of disaster areas (Bull-Kamanga, et al., 2003; Godschalk,
2003; Comfort, 2006), and coordination with urban-scale or region-scale water management. In
urban-scale adaptive planning, the urban resiliency is analyzed through a systematic analysis of
land use, population distribution, in which water infrastructure improvement can be made in
conjunction with transportation infrastructure. Examples of these adaptive planning were
discussed in preceding sections 2.0 and 3.0.
8 https://www.health.ny.gov/environmental/emergency/weather/hurricane/
100
-------
One further complication in the systems analysis is the evolving nature of urban
management objectives. The required service capacity f(t) often varies with time. The f(t)
increases with urban population and economic activities while decreasing as water conservation
takes place. This places a challenge on adaptation engineering. When the CR limit is exceeded,
the water structure functions are compromised (Figure 2-50) with partial or complete loss of
service capacity (Type III and IV). Then the central question is how to take necessary and
proactive adaptation measures and to minimize or avoid the hydroclimatic impacts that lead to
Type II, III, and IV changes. This subject is discussed below.
5.1.4. CR and engineering practice
Engineering practices use different approaches to define and use CR for various water
infrastructure. More details will be provided in Section 5.2. In current engineering practice,
accurate determination of the design variables is emphasized to minimize the uncertainty and
ensures adequate system capacity at a reasonable cost for a margin of safety or safety factor. By
doing so water managers can minimize the excess capacity that could later become stranded
(unused) capacity for economic considerations, or on the opposite side avoid the lack of capacity
for intended services. For many water managers, engineering such systems commonly assumes
stationarity whereby the climate and hydrological design parameters can be specified with
appreciable degrees of "perceived" certainty, for which CR is then often defined as a constant.
Progressive refinement of design basis and engineering objective is widely used to minimize
uncertainty and thus the system costs.
This traditional engineering practice is challenged for non-stationary climate and
hydrological variables. Because failure is so judiciously avoided, excess capacity is common in
the water industry, in the form of redundant systems. The large uncertainties in consideration of a
non-stationary climate would ensure excessive, if not prohibitive, capital and operational cost.
The alternative approach is to use adaptive engineering, by which the modification of water
infrastructure CR is planned, but not installed until the uncertainty is adequately reduced. One
pre-requisite for this adaptive approach is the general modeling-monitoring framework shown in
Figure 2-1 and, specifically, for water system engineering in Figure 2-48.
Several widely used engineering practices have a potential for the adaptive engineering,
such as modular design and phased construction (Girard and Mortimer, 2006; Chung et al., 2009),
decentralized water supply, wastewater and stormwater management (Weinstein, 2009; Gikas and
Tchobanoglous, 2009), as well as model-driven water reservoir operations for river flow
management under changing hydroclimatic conditions. For existing water infrastructure,
adaptation potential can be pre-installed during retrofitting, realignment, and expansion of
existing infrastructure assets, process optimization, as well as operational changes. All of these
adaptation techniques may require substantial physical asset alteration which may be reasonably
managed as a part of the renovation and replacement of aging infrastructure. The consideration of
the adaptation can be necessary under the following three conditions:
¦ The infrastructure planning horizon is long, for which future precipitation, land use, and
population changes are not precisely determined. Only by using this timeline can one
evaluate whether the rate of hydroclimatic change is too small to be "tangible" for
adaptation, or too excessive for the infrastructure to adapt at a reasonable cost. In this
report, the adaptation need analysis is set for the next 30-50 years;
101
-------
¦ The rate of precipitation change is larger than assumed in the original engineering design,
or the rate is comparable to those of the other two non-stationary variables - population
and land use changes;
¦ Large uncertainty in precipitation projection is translated and further propagated into
infrastructure engineering parameters, affecting the CR determination. The uncertainty is
also time-dependent, decreasing over time as the climate (precipitation) projection
improves.
Similarity and differences among the three engineering approaches are summarized in
Table 2-19. The engineering methods and techniques are different in improving the infrastructure
CR and performance of the existing water infrastructure. They are further described below for
each of the three major water infrastructure types.
5.2. Water infrastructure capacity reserve and resilience
5.2.1. Stormwater infrastructure functions and design tolerances
Storm water, drinking water, and wastewater infrastructure in an urban catchment are
schematically shown in Figure 2-49. Stormwater infrastructure manages overland runoff and
channel flows. Its major components, service functions, and likely vulnerability to precipitation
change are listed in Table 2-20. In a nonstationary climate, future runoff time-flow (1-0)
variations can significantly differ from that of the original engineering basis. This difference
affects the designed hydraulic conveyance capacity of a built stormwater network. The difference
can also adversely affect the hydraulic and water quality design functions of low-impact-
development (LID) and stormwater control measures (SCMs)9 in urban stormwater management.
5.2.1.1. Realized hydraulic capacity reserve
Carrying capacity and hydraulic profiles of a stormwater network are designed to limit the
nominal pipe flow to a range of 0.6-4.6 m/sec. This design criterion is intended to prevent
excessive sedimentation in the conveyance pipes or erosive damage to the pipe and receiving
water bodies. For a fixed topography, the runoff t-Q profile in a stormwater pipe depends on the
precipitation intensity, pre-storm soil moisture content, vegetation cover, and land-use patterns.
Among these factors, precipitation intensity and soil moisture are climate-dependent. Design
precipitation intensity at a given return interval (e.g., 10-year design storm) is commonly
determined from categorized IDF charts such as NOAA precipitation Atlas 14 (Bonin et al., 2006,
2011), National Weather Bureau Technical Paper 40 (Hershfield, 1961), and the SCS 24-hour
rainfall curves (Guo and Hargadin, 2009). These current methods are all based on assumed
precipitation stationarity.
Hydraulic CR of a stormwater pipe is realized from two primary sources. Because of the
stochastic hydrologic process and the uncertainties in hydrologic parameters, a large empirical
safety factor around 1.5-2.0 is often used in hydraulic design. For example, Schaad et al. (2009)
described an approach of using large safety factors in hydraulic engineering of a holistically
managed stormwater system. The other primary source of hydraulic CR comes from the fact that
stormwater pipes are available only at fixed nominal diameters, and that a minimum diameter
9 https://www.nap.edu/catalog/12465/urban-stormwater-management-in-the-united-states
102
-------
Table 2-19. Water infrastructure design and engineering domains, and their attributes.
Deterministic Engineering Domain (1)
Adaptation Engineering Domain (2)
Re-design & Re-construction Domain (3)
Attribute
Potential action*
Attribute
Potential action*
Attribute
Potential action*
New Infrastructure
Hydraulic capacity
Specific value
Process adjustment and
retrofitting;
No large-scale asset
modification;
Go to Domain (2) or (3)
in severe CR limitation
Range; capacity adaptively
installed
Assessment-adaptation-
monitoring for optimal
cost-benefit balance;
Go to Option (3) for
severe CR limitation
Specific value
Optimization,
Management
objective re
retrofitting;
and
-evaluation
Engineering flexibility
Limited in quantity.
Realized at construction
Flexible timing for extra
capacity installation
Large CR expansion
after re-construction
Water quality capacity**
Specified value
Range; capacity adaptively
installed
Specified value
Engineering flexibility
Limited in quantity.
Realized at construction
Flexible timing for extra
capacity installation
Large CR expansion
after re-construction
Techniaues and examples
Stormwater
infrastructure
Hydraulic design using
runoff rational methods for
facilities (e.g., retention
ponds and storm sewer)
Satellite retention
facilities;
Slice gate automation;
Go to Domain (2) or (3)
in severe CR limitation
Structure, stormwater control
design for non-stationary
precipitations;
Module design, phased
installation;
System monitoring and
forecasting
Adaptive capacity
installation;
Go to Domain (3) for
severe CR limitation
New infrastructure
network with or without
use of existing assets
Optimization,
Management
objective re
retrofitting;
and
-evaluation.
Wastewater
infrastructure
Drinking water
infrastructure
Ten-State design
standards, other
design protocols
Unit process and system
modeling and
specifications (e.g.,
disinfection chamber)
Process automation;
Flow detention facility;
Go to Domain (2) or (3)
in severe CR limitation.
Disinfectant, dosage
change;
Go to Domain (2) or (3)
in severe CR limitation
Module design, phased
installation;
Decentralized wastewater
system;
Onsite wastewater reuse;
System monitoring and
forecasting.
System optimization,
retrofitting;
Module design, phased
installation;
System monitoring and
forecasting.
Adaptive capacity
installation;
Go to Domain (3) for
severe CR limitation
Network expansion;
Adaptive capacity
installation;
Go to Domain (3) for
severe CR limitation
New designs and use of
revolutionary
technologies
and concepts
New designs and use of
revolutionary
technologies
References
ASCE (2004), Lin (2001),
USEPA (1994; 2002a;
2008), Salvato et al.
(2008), engineering
codes and guidelines
Carter and Jackson (2007);
Chung et al. (2009);
Semadeni-Davies et al.
(2008); Gikas and
Tchobanoglous, (2009);
Oron et al., (2007), Gupta
and Shrivastava (2006),
and USEPA (2009b)
Chang et al., 2006;
Neuman (2009);
Neuman and Smith
(2010).
103
-------
Table 2-19 continued.
Deterministic Engineering Domain (1)
Adaptation Engineering Domain (2)
Re-design & Re-construction Domain (3)
Attribute
Potential action*
Attribute
Potential action*
Attribute
Potential action*
Existina Infrastructure
Fixed
Infrastructure optimization,
retrofitting;
Go to Domain (2) or (3)
for severe CR limitation
Range of values
Iterative assessment-
adaptation-monitoring
for optimal cost-benefit
ratio;
Go to Domain (3) for
severe CR limitation
Specific value
Optimization, retrofitting;
Management and
objective re-evaluation
Hydraulic capacity
Engineering flexibility
Water quality capacity**
Limited, and deteriorated
after construction
Fixed
Large, adaptively installed
Range of values
Large CR expansion
after re-construction
Specified value
Engineering flexibility
Limited and deteriorated
after construction
Large, adaptively installed
Large CR expansion
after re-construction
Techniaues and examples
Stormwater
infrastructure
Wastewater
infrastructure
Drinking water
infrastructure
Operation and
maintenance
CSO division adjustment;
Go to Domain (2) or (3)
for CR expansion
Operational adjustment
for CR increase;
Process optimization
without large asset
change;
Go to Domain (2) or (3)
for severe CR limitation
Operational adjustment
for CR increase;
Process optimization
without large asset
change;
Go to Domain (2) or (3)
for severe CR limitation
Urban SCMs including Gl
designed for non-stationary
precipitation;
Structure retrofitting;
Recursive monitoring-
adaptation-assessment
Model-based system design
and upgrading;
Adaptive system retrofitting
and improvement;
Recursive monitoring-
adaptation-assessment
System optimization;
Process retrofitting without
large asset alteration;
Recursive monitoring-
adaptation-assessment
Adaptive CR installation
(new infrastructure);
Go to Doman (3) for
severe CR limitation
Adaptive CR installation
(new infrastructure);
Go to Doman (3) for
severe CR limitation
Adaptive CR installation;
Network expansion;
Go to Doman (3) for
severe CR limitation.
New infrastructure
network with or
without use of
existing assets
Application of new and
revolutionary
technologies
Application of new and
revolutionary
technology;
New infrastructure
expansion for CR
Optimization, retrofitting;
Management and
objective re-evaluation.
References
ASCE/AWWA (2004),
USEPA (2004),
engineering codes
and guidelines
Chung et al. (2009), Gikas
and Tchobanoglous
(2009), Montalto et al.
(2007); and Donofrio
et al. (2009).
Chang et al., 2006;
Neuman (2009);
Neuman and Smith
(2010).
Note: * - Potential actions at the upper limits of infrastructure CR and flexibility.
** - Refers to the capacity of a water infrastructure in maintaining performance on specific water quality criteria.
104
-------
Table 2-20. Important engineering attributes for stormwater infrastructure adaptation
Unit Operation
Function
Major Design Criteria*
Vulnerability **
Adaptation
Physical
Chemical, biological
Physical Hydraulic Water Quality
damage Function Function
Function
Example
Stormwater collection
Stormwater collection
Stormwater gravity drain
and conveyance
Stormwater runoff collection
in urban area for reliable
drainage and sanitation
Stormwater transfer by pipe
network to discharge
locations or retention
facilities
Drain inlet spacing <183 m;
Manhole spacing: 122-183 m
(varied with pipe diameter);
25-year design storm (varied)
I/O design limit in per day-
-km-cm;
Flow velocity: 0.6-4.6 m/s
for gravity sewer
Prevent methane and
sewer gas generation;
Remove oil and grease,
debris and large
objects.
Prevent methane and
sewer gas generation
Likely Likely Likely
medium high low
Likely Likely Likely
medium high low
Stormwater ponding, urban
flooding, and drainage
management.
Infiltration / exfiltration (I/O)
management;
Pipe flow velocity control
Stormwater inlet design for non-
-stationary precipitation.
Pipe repair, I/O management;
Drop manhole alignment for new
Q-t profiles;
In-line degritter for debris
Stormwater control measures n
Hydraulic retention
Increased water retention in
urban catchment basin for
reduced peak flows
Varies, based
precipitatio
on assumed
i stationary
Varied Likely Varied
high
Increased retention function
for non-stationary
precipitation
Detention pond, stormwater swirl,
and permeable pavements
Stormwater treatment
ponds and bioretention
facilities
Groundwater recharge
or evaportranspi ration
Enhanced water quality
improvement within
an urban catchment
Diverting water from the
urban catchment and
channel flows
Performance design for
target pollutant
removal.
Varied Varied Likely
high
Varied Likely Likely
low low
Enhanced water quality
improvement within
an urban catchment
Reduced stormwater channel
flow and discharge
Distributed stormwater retention
and treatment ponds
Permeable pavement, green roof,
recharge sewer.
Stormwater reclamation
Reclamation and reuse of
stormwater diverted from
channel flows
Contaminant prevention
for source water in
reclamation
Varied Likely Likely
high high
Collection and treatment of
stormwater for beneficial
reuse
Cisterns, rain barrels, rain
gardens
CSS and CSO control
Stormwater diversion
Prevent hydraulic overloading
of wastewater treatment
plant in high-intensity
precipitations
Flow rate and water level for
diversion valves in CSS;
Water level control in CSO
retention facilities.
Likely Likely Likely
low high high
Reduce CSO impacts to both
wastewater treatment
plants and discharge
receiving water
System engineering of retention
and CSO treatment facilities;
Extreme precipitation forecasting
and emergency responses.
Ddischarge at
stormwater outfalls
Stormwater discharge into
a water body under a
NPDES permit
Flow rate and discharge
velocity
Varied in water quality
parameters
Likely Likely Likely
low medium medium
Reduce discharge impacts
on receiving water in
erosion, temperature,
turbidity, nutrients and
other pollutants.
Discharge swirl and detention;
Sensor-based monitoring-
controlled discharge
Note: * Summarized from civil engineering manuals and U.S. engineering codes and guidelines. These design criteria are for general guidance.
** - Qualitative rating for anticipated major changes in precipitation and hydrology, excluding the extreme meteorological events.
# - Stormwater control measures are organized in the four groups by primary functions.
105
-------
(typically 15 inches) is often required. This means that the hydraulic carrying capacity (O2) of
installed pipe with a diameter (c/2) can be greater than the design peak flows (Oi) for a pipe
diameter (dmm). The maximum increase for the installed carrying capacity
Q2 -Q1
Q1
is:
Q2 ~Qi
Q1
\2I
J2
-1
(2.33)
D (in)
in
1
>
O
<1
For storm pipe mains
larger than 0.60 m (24 in) in
diameter, this engineering
practice potentially offers a
hydraulic CR of 31% on
average while satisfying the
design criteria on pipe flow
velocity. This design
consideration is shown in
Figure 2-51. In this simple
envelope calculation, AO% is
calculated using Eq.2.33 for
pipes at a hydraulic slope (S) of
0.2% and 0.5%. Pipe flow
velocity (V) in a range of 0.6-
4.6 m/sec by engineering
standards, is calculated using
Manning's equation. The
maximum capacity increase is
approximately 60% for lateral
pipes of diameter <0.61 m (24-
in). Therefore, combining with a safety factor of 1.5-2.0, the pipe engineering practice could have
installed a maximum hydraulic capacity up to 230% of the design value.
Of course, one reason for doing this is that the storm water pipes often contain sand and
other material building up on the bottom since the flows are often not continuous. The pipe
sedimentation reduces the capacity of the pipe and this the available CR.
Figure 2-51
D (m)
Maximum percentage increase (AQ%) in hydraulic
capacity of stormwater conveyance using
commercial concrete pipes of discrete nominal
diameters (D). Eq. 2.33 is used for the calculations.
5.2.1.2. Water quality limitations
Climate-driven water quality changes can significantly limit the infrastructure CR in
stormwater adaptation. Studies (e.g., Horowitz, 2009; Whitehead et al., 2009; Yang et al., 2002)
have linked the intensity of peak runoff to the increased turbidity, and higher metals, chemical,
and dissolved organic carbon loading in urban streams. Peak pipe flow and high discharge
velocities are also found to be responsible for soil erosion, water quality change, and ecological
deterioration at stormwater outfalls and their immediate downstream segments (see
McCorquodale, 2007; Novotny and Witte, 1997). These hydrologic and water quality changes
106
-------
can be attenuated or amplified within an urban catchment of higher paved ground surfaces (Table
2-20).
Combined sewer overflow (CSO) events during intense precipitation in many U.S.
communities are a major factor limiting infrastructure CR otherwise available for adaptation.
Storm water runoff and untreated, but diluted, sewage are diverted for discharge when storm water
flows exceed the hydraulic capacity of the wastewater treatment plants and available retention
facilities. The peak flow, on the other hand, is a function of the precipitation duration and
intensity, catchment basin hydrograph, and the groundwater infiltration rate into the pipes (Black
and Endreny, 2006; Lai, 2008; and Diaz-Fierros et al., 2002). More intense precipitation events
projected as future climate conditions will likely yield greater peak flows and more frequent CSO
events unless efforts are undertaken to separate flows (U.S. EPA, 2009a, 1994; Capodaglio, 2004;
Alp and Melching, 2009). The EPA 20-watershed study (U.S. EPA, 2013b) showed significant
hydrological flow modifications across the U.S. that can worsen the CSO occurrence (Johnson et
al., 2015).
Land use and the degree of impervious surface in the urban watersheds can amplify
hydrological responses to future climate-related precipitation changes, often in the form of
increasing peak flows and runoff. On the other side, stormwater control measures including the
LIDs are used for enhanced stormwater retention and reduced peak runoff. They are often
engineered assuming precipitation stationarity (e.g., Lai, 2008; Montalto et al., 2007; U.S. EPA,
2004a; Marsalek and Chocat, 2002; Dietz, 2007; Carter and Jackson, 2007; and Gilroy and
McCuen, 2009). Thus, the stormwater control measures are vulnerable under a non-stationary
climate (Table 2-20). For example, Semadeni-Davies et al. (2008) and Sun et al. (2016) suggested
the need to consider climate and precipitation changes in stormwater structure designs.
The U.S. EPA's National Stormwater Calculator estimates and evaluates SCMs
applicability in reducing stormwater runoff (U.S. EPA, 2014). While the tool includes
precipitation projections under future climate scenarios, a comprehensive nationwide evaluation
has not been completed. With this data not yet available, the maximum CR of 230% of the design
value was taken as the upper limit for stormwater infrastructure.
5.2.2. Drinking water infrastructure functions and design tolerances
5.2.2.1. Engineering resilience in a distribution network
Community water systems in the U.S. provided water supplies to over 292 million people
in 2008. Engineering attributes of major community water system components and their potential
vulnerability to precipitation changes are shown in Table 2-21. Drinking water distribution
following the treatment (Figure 2-49) is engineered to meet water demand for both domestic
consumption and firefighting throughout a service area. Long-term water demand variations, a
prime engineering factor in water distribution design and operation, are linked to demographic
and land use changes, urban microclimate, and the transformation of water-intensive industries
(Levin et al., 2002; Pires, 2003; Hummel and Lux, 2006). Such water demand changes are
commonly captured in urban development master plans and regional economic development
projections (see Figure 2-48) that may have intrinsically included hydraulic capacity reserve
adequate for adaptation.
Water quality changes within a distribution system have been extensively studied.
However, little is known about the water quality change in pipes under future climate and
107
-------
hydrological conditions. In a study of climate adaptation for a large U.S. Midwest utility, Li et al.
(2009, 2014) and Clark et al. (2009) reported that an increased total organic carbon (TOC) level
in (surface) source waters under future climate scenarios could lead to higher TOC concentrations
in produced water and subsequently greater disinfection by-product (DBP) formation even at
levels in violation of the U.S. drinking water standards. This type of potential water quality effect
can significantly reduce the available infrastructure CR, making adaptation necessary. In Table 2-
20, a variety of adaptation options are listed for changes in unit process, such as enhanced TOC
removal using GAC or chemical flocculation (e.g., Jarvinen et al., 1991; Crozes et al., 1995; Li et
al., 2009; and Clark et al., 2009), water age reduction and chlorine addition optimization for DBP
control (Carrico and Singer, 2009; Prasad et al., 2004; Boccelli et al., 2003). In addition, higher
surface water and associated drinking water temperatures likely in future climate will change the
disinfection kinetics, DBP formation rates, and biological stability in a distribution system. These
areas of indirect hydroclimatic impacts are worthy of further investigations.
5.2.2.2. Realized capacity reserve in drinking water treatment
Water intake and water treatment are vulnerable to the direct impacts of precipitation
changes (Table 2-21). Detailed modeling-monitoring studies have shown the degree of these
impacts in surface water bodies of different sizes across the U.S. (e.g., U.S. EPA, 2013b; Chang
et al., 2006, 2014a,b; Neil et al., 2019; Imen et al., 2016). The impacts vary among watersheds,
different types of land use and land cover, as well as the nature of precipitation and temperature
changes. A resilient water treatment process is required to accommodate these source water
variations and to provide finished drinking water in compliance with the SDWA regulations.
As shown in Figure 2-49, a typical surface water treatment process in the U.S. consists of
pre-oxidation, rapid mixing, flocculation and sedimentation, granular filtration, advanced
treatment if necessary [e.g., GAC filtration, membrane separation], and finally disinfection in
clear wells before distribution to consumers. In the design of these treatment process units, a
simple empirical safety factor of 1.2-1.5 is often used; some larger values are possible. For
example, Kim and Bae (2007) proposed a safety factor of 2.0 in the hydraulic design of a baffled
GAC contactor for odor control. More advanced probability-based methods are now developed
for systematic reliability- cost tradeoff evaluation. Boccelli et al. (2007) described process
optimization guided by a cost-performance ratio to determine safety factors in the flow rate
design of an infiltration-based surface water treatment plant. Gupta and Shrivastava (2006)
introduced a water treatment design method using Monte-Carlo simulation to quantify
performance uncertainties in suspended solids removal. Li et al. (2009) developed a Monte Carlo
methodology to simulate the cost-probability relationship in GAC contactor process modification.
While these advanced design methods better quantify the capacity and cost cumulative
density function (CDF) curves, they often require extensive input data and computation. Instead,
the traditional safety factor method is widely used in field engineering of the deterministic
domain. This practice alone yields a maximum treatment capacity at 150% of the design value to
permit redundancy when units are out of service. For impacts exceeding the CR limits, adaptation
is needed to increase infrastructure CR, mostly through treatment plant retrofitting, process
modification, change of unit operations, or installation of a new process (Table 2-21).
An engineering adaptation example is given by Li et al. (2014; 2009) and Clark et al.
(2009). These investigations led to the development of the adaptation engineering model "Water
108
-------
Table 2-21. Important engineering attributes and likely vulnerability in drinking water treatment and distribution systems for community water supplies.
Major Design Criteria*
Vulnerability **
Adaptation
Unit Operation
Function
Physical
Chemical, biological
Physical
Hydraulic
Function
Water Quality
Function
Function
Example
Source water protection
Source water intake
Protect source water quality
at water intake
Assure water availability for
drinking water production
Water level at water
intake
Minimize daily and seasonal
water quality variations;
Minimize biological growth
at intake (e.g., mussel).
Likely
Likely
Likely
Adaptive change of intake
elevation and location;
Multi-elevation intake
aprons
Intake security against
physical damage
High
Medium
Medium
Physical damage
protections
Enhanced structure
supports
Drinkina water treatment
Rapid mixing
Rapid dispersion of
coagulants in water
<1 min retention time
Likely
Low
Likely
Low
Likely
Low
Coagulation &
Flocculation
TOC and particulate removal
15-20 min and 18-25
min residence time for
high-energy and low-
energy flocculation.
Varied dosage among
coagulants: alum,
chlorine, polymer, and
potassium permanganate
Likely
Low
Likely
Low
Likely
High
Inflow TOC variations
monitoring and chemical
dosing control.
Sensor-based TOC
monitoring and process
adaptive control
Clarification
Remove settleable solids
after flocculation.
Alternative unit processing
by membrane and
particulate filtrations
32.6-48.9 rrf/d/m2 for
turbidity removal
20.4-32.6 rrf/d/m2 for
algae removal
Likely
Low
Likely
Medium
Likely
Medium
Reduce high-turbidity effect
on downstream units;
Remove excessive algae
present in raw water.
Process monitoring and
control;
Unit process optimization
Dissolved air floatation
(DFA)
Remove solids and odor with
ascending fine bubbles
10-12 m/h air flow;
5-10% recycle flow.
Follow coagulation /
flocculation unit process
Likely
Low
Likely
Low
Likely
Low
Adjust particle surface
charge for enhanced DFA
Unit process optimization
High rate filtration
Remove various constituents,
including turbidity, coliform,
color, taste, metals, and
toxic chemicals
hydraulic loading: 83
L/nfmin (rapid sand)
Backwash monitoring
and operation.
Likely
Low
Likely
Medium
Likely
Medium
Reduce shock loading of
high turbidity;
Optimize backwash
scheduling, operation.
Process monitoring and
control
Oxidation and
disinfection
Biological inactivation and
oxidation of organic
matters
Disinfectant concentration
limit:1.0 mg/L CI";
Contact time
Likely
Low
Likely
Low
Likely
High
Reduce TCC concentration
and variations;
Unit process optimization.
Retrofitting for higher
contact efficiency;
Change of oxidants.
Ion exchange
Membrane filtration
Cation or anion exchange to
remove nitrate, Fe, Mn,
and hardness
Remove organic and
inorganic contaminants
by using membranes
Service flow rate: <668
L/m3 for N+2 removal;
Backwash rate: 81-122
Urn2 for N+2 removal
Hydraulic loading rate;
Temperature;
Suspended solids.
Inflow pH range;
Membrane anti-degradation;
Biological growth.
Likely
Low
Likely
Low
Likely
Low
Likely
Low
Likely
Medium
Likely
Low
Remove excess turbidity
in pretreatment;
Process unit arrangement,
optimization, retrofitting.
Pretreatment to remove
excessive turbidity;
Backwash operations.
Process adjustment;
Enhanced water
pretreatment;
Process monitoring
Pretreatment with coarse
membrane filter;
Back-wash automation
GAC adsorption
Absorb chemicals onto
absorbent media
10-12 m/h loading;
Bed depth and volume.
Regeneration time;
DOC, odor, and other
contaminant removal.
Likely
Low
Likely
Low
Likely
High
Increase GAC adsorption
efficiency and prevent
break-through
Adjust GAC regeneration
cycle;
Operation optimization.
109
-------
Table 2-21 continued.
Design Criteria*
Vulnerability **
Adaptive Engineering and Management
Major Operation Unit
Function
Physical
Chemical, biological
Physical
Hydraulic
Function
Water Quality
Function
Functions
Example
Treatment process
Overall specification of
each process unit for
treatment objectives
Process flow rate;
Flow variations.
Drinking water treatment
guidelines
Drinking water quality
standards.
Likely
Low
Likely
Medium
Likely
High
Increase treatment capacity
reserve for new source
water variations and water
demand changes
Process optimization,
retrofitting, or change
and expansion
Drinkina water distribution
Water demand
Spatial and temporal demand
variation affect network
operation and water age
Not applicable
Not applicable
Not
Applicable
Likely
High
Likely
High
Water demand management
under high temperature
and heat stress of future
climate
Water pricing, lawn
irrigation timing and
management
Pipe network
A network of pipes in different
diameters and materials to
deliver water from treatment
plant to consumer's tap
Pressure management:
413 kpa (241-689 kpa)
Flow velocity: 1.2-1.8
m/s in mains.
Corrosion protection;
Water age management;
Water quality standard
compliance at user's tap.
Likely
High
Likely
Low
Likely
High
Prevent pipe corrosion and
leaks under future climate;
Water quality management;
In-network water
treatment such as
chlorine addition
and THM stripping.
Note: * Summarized from ASCE/AWWA (2004), Lin (2001), Salvato et al. (2008).
** - Qualitative rating given for major changes in precipitation and hydrology, excluding the extreme meteorological events.
110
-------
Treatment Plant - Climate Adaptation Model" or WTP-cam. The model and its application at the
Greater Cincinnati Water Work's Miller WTP will be described in Section 6.0.
5.2.3. Wastewater infrastructure functions and design tolerances
5.2.3.1. Realized capacity reserve in hydraulic loading
Important engineering functions and physical attributes for wastewater infrastructure are
shown in Table 2-22. A general wastewater treatment process in the U.S. includes
physiochemical pretreatment, biological oxidation of macronutrients (primarily biological
oxygen demand [BOD], N and P), possible filtration to reduce suspended solids, optional tertiary
treatment (N and P removal), and finally effluent disinfection before discharge (see Figure 2-49).
Hydraulic loading capacity is often specified for future wastewater generation within a service
area and to account for groundwater infiltration into wastewater collection pipes (Lai, 2008; Lin,
2001). These variables are lumped into a single parameter - wastewater generation rate per
capita in engineering designs [e.g., 1900-4550 lpd/person (500-1200 gpd/person)]. In addition,
an empirical safety factor of 1.2-4 is used to accommodate unexpected hydraulic variations (peak
flows). Values up to 4.0 are justified for special engineering conditions, such as complex
hydrogeological regions, aged water collection networks with extensive infiltration and
exfiltration, very small service areas, or service areas of the large variation in wastewater
generation rates.
5.2.3.2. Realized capacity reserve in biological systems
Space-demanding aerobic and anaerobic biological treatment is often a limiting unit
process that determines available CR at a wastewater treatment plant. Most wastewater plants
have significant CR to permit unit operations to be taken out of service for maintenance due to
the corrosive environment in which they operate. Since the early study of Kincannon and Gaudy
(1966), biological wastewater treatment is known for its sensitivity to both hydraulic and
contaminant shock loading (Jing et al., 2009; Chen et al., 2008), leading to treatment process
upset (Ray and Peters, 2008; Capodaglio, 2004) and performance deterioration (O'Reilly et al.,
2009). Other causes for reduced treatment capacity include aging treatment equipment and
wastewater infrastructure, poor process control, and operational inefficiencies.
Here the limitation and vulnerability are illustrated in the design or retrofitting of an
aeration tank, a principal unit in the activated sludge wastewater treatment process. BOD
removal rate (n ) in an aeration basin/clarifier combination, or the ratio of tank influent
v ' BOD 7 5
(C0) and effluent (C) is a function of flow rate (Q), tank volume (V), BOD oxidation rate (Kd),
biomass cell age (0C), microorganism concentration in the tank (X), waste rate, and maximum
yield coefficient (Y). Following Lin (2001), the removal rate can be written as:
„ =
-------
Table 2-22. Important engineering attributes and potential vulnerability of wastewater infrastructure
Major Design Criteria*
Vulnerability**
Adaptation
Major Operation Unit
Function
Physical
Chemical, biological
Physical
Hydraulic
Function
Water Quality
Function
Function
Example
Wastewater collection
Wastewater collection
Wastewater collection from
all users in a sei\ice area
WW yield: 0.38 m3/person-day;
Flow velocity: 0.6-4.6 m/s;
Flow rate: 1.5 m3 /person-day
(laterals and branches)
Sulfur and methane gases
generation
Likely
High
Likely
High
Likely
Low
Pipe I/O flow management;
Wastewater reuse and
separation.
Pipe leak detection;
Dual pipe system;
Onsite wastewater
treatment
Wastewater pumping
and conveyance
Wastewater transfer to a
central location(s) for
treatment
I/O rate: < 0.45 m3/day-km-cm;
Flow: 0.95 m3/ca-day (main);
Flow velocity: 0.6-4.6 m/s
Sulfur and methane sewer gas
management;
Fire hazard prevention.
Likely
High
Likely
High
Likely
Low
I/O management;
Flow velocity & abrasive
damage control.
Pipe leak detection
Drop manholes;
In-line degritter
Wastewater treatment
Preliminary treatment
(screening, degritting)
Solids and debris removal in
headworks
Screen debris removal: >5.1-cm
Flow (grit chamber): -0.328 m/s;
Aerated grit chamber: 2-5 min
residence time
Not applicable
Likely
Low
Likely
Low
Likely
Low
Primary treatment -
Sedimentation tank
Removal ofsettleable solids
and 25-35% BOD
Peak flow <0.71 lps/m2;
Maximum weir load: 2.16 lps/m;
Water depth: >2.1m.
Target removal rates:
BOD: 20-40%, TSS: 35-65%;
Settleable biosolids: 50-75%.
Likely
Low
Likely
Medium
Likely
Medium
Flow equalization facilities
to smooth flow variations;
Process monitoring
Monitoring and increased
maintenance
Secondary treatment -
Trickling filters
Biological treatment to
remove BOD and
macronutrients
Filter depth: 1.5 - 3.0 m;
Hydraulic loading:
0.012-0.047 lps/m2, or
0.047- 0.47 lps/m2 (high rate).
Normal: 0.08 - 0.40 kg
BOD/m3-day;
High-rate: 0.48 - 1.44 kg
BOD/m3-day.
Likely
Low
Likely
High
Likely
High
Process control for
resilience in shock loading
Process flow stabilization
Trickling filter retrofitting;
Change recirculation ratios;
Process monitoring and
control for weir loading.
Secondary treatment -
Activated sludge process
High efficiency of BOD and
nutrient removal
Weir loading: 1.44 lps/m;
Hydraulic loading:
0.47-0.57 lps/m2
0.38 lps/m2 with nitrification
Maximum BOD loading:
0.24-0.64 kg/day/m3;
Aeration rate:
93.5-125 m3 oxygen / kg BOD
Likely
Low
Likely
High
Likely
High
Process control for
resilience in shock loading
Increase treatment
capacity reserve.
Modify cell age and sludge
return rate;
Improve aeration efficiency;
Increase aeration capacity.
Secondary and final
clarifier
Settleable biosolid removal
Surface settling rate:
50-62 lps/m2
Not applicable
Likely
Low
Likely
Low
Likely
Low
Enhance biomass setting
Operational adjustment
Nitrogen removal
Successive nitrification
and denitrification
~Varies"l^^^
VahesT^See'uX^PAT2009b)T"
Likely
Low
Likely
Low
Likely
High
Chlorination
Treatment effluent
disinfection
>15 min contact time in
chlorination contact basin
<200 fecal coliform / 100 ml
Likely
Low
Likely
Low
Likely
Low
Treatment process
Overall specifications of
each process unit for
treatment objectives
Process flow rate;
Flow rate variance.
Surface water quality standards
for discharge control
Likely
Low
Likely
Medium
Likely
High
Increase treatment capacity
reserve to against source
water variations and water
demand changes
Process optimization,
retrofitting, or change
and expansion
Wastewater effluent discharae
Treatment effluent
discharge
Treatment effluent discharge
under a permit
Varies depending on discharge
regulations
Varies depending on discharge
regulations
Likely
Low
Likely
Medium
Likely
High
Discharge limits sensitive to
the impacts on receiving
streams;
Compliance to discharge
limits.
Adjust treatment process
for likely to-be-revised
discharge limits.
Note: * Summarized from "10-state" wastewater treatment standards and Lin (2001). These design criteria are for general guidance.
** - Qualitative rating given for major changes in precipitation and hydrology, excluding the extreme meteorological events.
I/O - wastewater inflow and outflow by infiltration and exfiltration; WW - wastewater.
112
-------
precipitation. An increase in the flow is a likely sign of such overloads. These events are
responsible for the treatment process upsets and discharge violations (e.g., Tafuri and
Selvakumar, 2002). Efforts to split combined systems and seal the piping at the surface are
appropriate adaptation measures to address this problem. Other potential adaptation actions are
listed in Table 2-22.
Capacity reserve in biological treatment is recognized by using an empirical design safety
factor of commonly 1.2-1.3, and by modifying unit operations without large physical asset
alteration (Table 2-22). In addition, the treatment CR is also made available through the
optimization of the biological process. One operational adjustment, for example, is to increase
the capacity by changing biomass cell age, aeration rate and efficiency. When cell residence age
and aeration rates are adjusted for higher aeration capacity |^—-j from 2.27 to 2.83, the BOD
removal rate is increased theoretically by 56.9%, or Ai -> A2 and Bi -> B2 (Figure 2-52).
For a treatment plant of 100 m3/day design capacity, a 166% increase in flow rate can
potentially decrease the BOD removal rate from 75.5% to 45.5%. This decrease from A0 to Ai is
illustrated in Figure 2-52. Similarly, treatment efficiency decreases from A0 to Bi as a result of
increased BOD concentration and mass loading into a plant. An increase in flow and BOD mass
loading reduces the BOD removal rate from starting position A0 to Ai and Bi, respectively. In
process adaptation, aeration capacity adjustment from 2.27 to 2.83 can partially recover the lost
performance, or Bi to B2 and from Ai to A2 in Figure 2-52. Labels 30, 50, and 70 mg/L are plant
inflow BOD concentrations. This is a partial recovery of the capacity loss due to the future
increase in flow rate and BOD mass loading. These are many engineering measures currently
available in the market to make these aeration changes.
CO
>
o
E
0)
oa
Q
o
m
100
80
60
40
20
\
BOD Removal Rate
in Aeration Tank
\AoX
f v ^
\ \
\ 1 \ \
= wL=2,27
\f\x
\*b2\
\ \
(—) =2.83
¥.9
V 0 'night
\
XA\
30 mg/L
v \V\
^7
\ \\ x
\ \\
\ / 50 mg/L
70 mg/L
100
1000
Q (m3/day)
Figure 2-52 BOD removal efficiency of a wastewater activated aeration tank as a function of flow,
BOD mass loading, and aeration capacity.
113
-------
By a combination of using design safety factors and operational adjustments, the total
realized CR could reach 30-80% of the design value in an activated sludge process. In CR
evaluation, however, one should also consider performance deterioration over time for aging
wastewater treatment facilities. This portion of the treatment CR is recoverable by process
monitoring, control and adjustment, or by using advanced engineering techniques such as fuzzy
logic control (e.g., Miiller et al., 1997; Peng et al., 2007). The analysis here assumes that the
performance reduction is minimized through process adjustment and optimization. Thus, the
realized CR of 30-80% design value is considered as a reasonable estimate.
5.2.3.3. The CR efficacy in current system design
Based on the analysis in Sections 5.2.1.-5.2.3., Figure 2-53 schematically shows a general
range of percent CR installed in the current infrastructure engineering. The adaptation need can
be seen by comparing the installed CR against the rates of precipitation change in the contiguous
U.S. The change in precipitation can result in changes in watershed hydrology, including both
stream flow and water quality. It is assumed that the rate of precipitation change in the next 50
years is proportionally translated into hydraulic design parameters (e.g., runoff).
In the integrated watershed simulation of future climate and land use change, the
investigations over three watersheds in Ohio and Nevada (Tong et al., 2012; Sun et al., 2013; Fu
et al., 2018) showed similar degrees of change in stream flow and water quality. For example, in
the Little Miami River watershed, 20% precipitation increase or decrease in 2050 would result in
a 43.83% increase and 53.08% decrease in stream flow, respectively. The total phosphorus
Figure 2-53 Relative magnitude of infrastructure CR installed in current engineering practice
(left) in comparison with the relative precipitation change (solid bar) and its
uncertainty (pattern and solid line with whisker) by 2060. PCT - percentile. From
Yang (2016).
114
-------
increases in both cases by 21.35% and 6.73%. Total nitrogen concentrations change by a smaller
amount <11.55% and 2.91% respectively. Thus, the hydroclimatic impacts on watershed
hydrology would likely be in the same order of magnitude as precipitation changes. This
generalization is also reported by numerous climate model simulations.
On the national average, the precipitation change is notably smaller almost by an order of
magnitude in some places (Rajagopalan and Lall, 1998; and IPCC, 2014). In a nationwide
analysis, this research analyzed the likely precipitation changes recorded in long-term historical
measurements. The average and range for approximately 1100 weather stations in the U.S. are
shown in Figure 2-53. Along with the average (U.S. Mean) are 90% and 10% percentiles, the
maximum and minimum, and their associated uncertainties. In comparison, the infrastructure CR
installed by current engineering practice is a magnitude of order larger than the national average
rate of 5% for precipitation changes in the next 50 years (Figure 2-53).
However, the change of extremes with the projection uncertainties far exceeds the CR%
in current practice. Therefore, the adaptation is very likely needed in places with extreme
precipitations. This relatively simple evaluation has two noteworthy implications:
¦ As a national average, the future precipitation changes of -5% by 2060 can likely be
managed by the installed CR in existing infrastructure. This generalized conclusion
supports the current engineering practice that has been applied worldwide for decades.
However, the national average cannot represent all local conditions because of uneven
changes across geographic areas (IPCC, 2007; Rajagopalan and Lall, 1989; and this
study).
¦ The second implication is important to adaptation at the local watershed levels. Climate
stations with precipitation increase in the 90% percentile are spatially clustered in many
areas such as the eastern Texas-Oklahoma region. For areas in Arizona and New Mexico,
precipitation decrease in the <10% percentile is compounded by the high rate of
population growth. The combined effect makes water availability the dominant
adaptation factor for these regions. Thus, the degree of these vulnerabilities is a focus of
infrastructure assessments in the location-specific analysis. There are "bottom-up"
analysis tools available, one of which will be described subsequently in Section 5.3.2.
5.3. Water infrastructure vulnerability analysis for adaptation
5.3.1. The resilience assessment and two approaches
In a bottom-up approach, the climate vulnerability of water infrastructure is assessed to
determine the CR threshold. Below the threshold, the water infrastructure service function is
impaired either in the short- or long-term. The result is a basis to determine infrastructure
resilience against specific impacts of hydroclimatic and land use changes.
The U.S. water utilities have taken both top-down and bottom-up approaches in the
threshold analyses (e.g., Miller and Yates, 2006; Freas et al., 2008; Stratus Consulting and MWH
Global, 2009). Most utility water managers who are engaged in climate vulnerability analysis
have a strong technical understanding of their water systems, including local hydrology,
historical operating conditions, and standard operational practices, but have little access to
climate model projections tailored to their specific regions. Interestingly, the U.S. EPA
conducted a review of 50 water utilities nation-wide in 2010 on their analysis methodology.
115
-------
Among them, eight utilities had conducted their climate vulnerability analysis, and only two
followed the bottom-up approach (Table 2-23).
Table 2-23 Types and approaches of eight water utilities in climate vulnerability assessment*.
Utility
Service Provided
Vulnerability Assessment
Type
Population
State
East Bay Municipal
Utility District (EBMUD)
Water,
Wastewater
1.3 million
CA
Bottom-up
City of Boulder Utilities
Division
Water,
Wastewater
113,000
CO
Top-down
Denver Water
Water
1.3 million
CO
Top-down
Massachusetts Water
Resources Authority
Water,
Wastewater
2.2 million
MA
Top-down
New York City
Department of
Environmental
Protection (NYCDEP)
Water,
Wastewater
9.2 million
NY
Top-down
Portland Water Bureau
Water
860,000
OR
Top-down
San Antonio Water
System (SAWS)
Water,
Wastewater
1 million
TX
Bottom-up
Seattle Public Utilities
Water
1.35 million
WA
Top-down
Note: * - according to a U.S. EPA 2010 study (see U.S. EPA, 2015a)
The bottom-up approach generally includes a component to quantify the likely
vulnerability and identify the most vulnerable critical assets in the water systems. For example,
the East Bay Municipal Utility District (EBMUD), a water and wastewater utility in the Greater
Oakland, CA area, used an approach adopted from the AwwaRF (now the Water Research
Foundation) publication "Climate Change and Water Resources: A Primer for Municipal Water
Providers" (Miller and Yates, 2006). The EBMUD analysis consists of several steps:
¦ Identify the vulnerability of potential portfolio components (e.g., new reservoirs,
expanded reservoir storage, increased conservation, conjunctive use, water reclamation,
desalination, inter-basin transfers) and screen those components for technical,
environmental, and economic feasibility in adaptation;
¦ Develop alternate portfolios of multiple components that could meet projected demands
(e.g., increased conservation and conjunctive use, or water reclamation and inter-basin
transfers).
¦ Conduct a preliminary portfolio analysis using a combination of the Water Evaluation
and Planning (WEAP) system model and the district's EBMUDsim model - known
116
-------
collectively as the "W-E model." Portfolios that performed poorly under current
hydrological conditions were eliminated. The remaining portfolios were subjected to
detailed analyses under anticipated climate conditions using the W-E model.
¦ Identify portfolios with adaptation potentials and use sensitivity analysis to evaluate
critical vulnerabilities and ways to address the vulnerabilities.
The San Antonio Water System (SAWS) in western Texas, serving a population similar
in size to EBMUD, also used a bottom-up approach. This threshold approach identifies system
components that are dependent on the status of climate variables (precipitation, temperature, etc.)
and the overall system risk under the future hydroclimatic conditions. The preliminary risk
assessment is also based on the professional judgment of experts who know the system and the
planning area (see U.S. EPA, 2015a, and reference therein). The qualitative or semi-quantitative
analysis consisted of:
¦ identifying the climate variables of importance and exploring the sensitivity of SAWS to
these variables;
¦ determining water system responses to a range of potential future climate conditions;
¦ assessing the vulnerability of SAWS to hydroclimatic impacts;
¦ assessing system performance according to the uncertainty associated with hydroclimatic
factors driving SAWS vulnerability;
¦ evaluating overall system risk and identifying areas in need of further analysis.
5.3.2. Water resilience evaluation and resilience tool — CREAT
A systematic examination is considered a necessary process in evaluating the threshold
for adaptation. While many utilities take various approaches according to their own needs, a
systematic process for vulnerability analysis has emerged from the U.S. EPA's Climate Ready
Water Utilities (CRWU) program10. The program conducted case studies and vulnerability
analysis at participating facilities. These actions led to the establishment of an adaptive response
framework (U.S. EPA, 2012b), and the publication of Climate Resilience Evaluation and
Awareness Tool (CREAT) Version 2.0.
CREAT is a software tool that guides users through a series of investigative steps (Figure
2-54). CREAT Version 3.0 is the most recent software available. As a stand-alone risk
assessment product, CREAT allows users to assess potential impacts of future climate on their
utility and to evaluate adaptation options to address those impacts. It follows a structured
approach with the threat analysis leading to the adaptation actions. Major features are:
¦ A library of drinking water and wastewater utility assets (e.g., water resources, treatment
plants, reservoirs, distribution system components, pump stations) for one-by-one
evaluation of hydroclimatic impacts
¦ A list of hydroclimatic impacts (e.g., sea-level rise, precipitation changes, reduced snow
pack) covering a broad range of future conditions that potentially affect water utilities
10 http://water.epa.gov/infrastructure/watersecurity/climate/index.cfm
117
-------
Setup
Regional and
Historical &
Climate Infor
if
Local
'rojected
mation
I *
Threats
Baseline
Adaptive
Analysis
1
Measures
Resilience
|
Adaptation
Analysis
¦- ¦
1
Planning
•u
Results and Reports
Figure 2-54 The process of climate change vulnerability analysis using the EPA tool GREAT,
Adopted from EPA CRWU website.
¦ Adaptation suggestions that can be implemented to adapt to the hydroclimatic impacts
that can be customized by the user
¦ A series of risk-reduction cost reports that will allow the user to evaluate various
adaptation options
Water sector utility owners and operators can use information about their utility in
CREAT to identify climate and hydrological threats, assess potential consequences, and evaluate
adaptation options. This approach allows utilities to assess impacts and identify the thresholds
where asset or mission failure could occur. Users can also consider existing climate science data
to evaluate the plausibility of climate-related impacts and how soon these impacts may affect the
utility. CREAT has been applied to many case studies in the contiguous U.S. A few examples of
the case studies conducted by the EPA CRWU program include:
New York City Department of Environmental Protection (NYCDEP)
In 2010-2011, a CREAT pilot study was conducted for the NYCDEP water and
wastewater systems. The pilot site is located in Corona, New York. Through the study, CREAT
was used to assist utilities in making risk management and planning decisions, and to identify
areas of potential refinement for the tool before it is finalized for broader use in the water sector.
Climate information embedded in CREAT was used to assess the risk and likelihood of climate
threats. Through a desk-top exercise and technical data analysis using CREAT, the vulnerability
to future hydroclimatic conditions, particularly sea level rise and storm surge, was identified for
the district's water and wastewater assets. In fact, the water and wastewater transfer facilities in
low-lying areas identified by CREAT later experienced operational difficulties during Hurricane
Sandy in October 2012.
New York/New Jersey Harbor
One of the first CREAT pilot tests was conducted in conjunction with The New
York/New Jersey (NY/NJ) Harbor Estuary Program and the North Hudson Sewerage Authority
118
-------
(NHSA). The receiving waters for the NHSA system are part of the NY/NJ Harbor Estuary
ecosystem providing critical habitat, recreation and transportation services. The NHSA system
includes 107 miles of combined sewers, 17 combined sewer overflow (CSO) regulators, 11 CSO
outfalls, and 6 pump stations. This system serves many communities and is fed by a system of
rivers draining five states and flowing through several metropolitan areas. It also serves as an
important habitat for over 300 species of migratory birds, spawning ground for several species of
fish, and provides recreation and transportation services. In the pilot study, the CREAT tool was
used to identify potential future impacts for water utilities and helps the utilities catalog potential
actions in adaptation planning. The exercise fostered dialogue among stakeholders that share a
common interest in climate resilience.
Manteo and Columbia, North Carolina
Another CREAT pilot testing was conducted with a workgroup comprised of town
officials and water managers from Manteo and Columbia, North Carolina, as well as
representatives from the Albemarle-Pamlico National Estuary Partnership. The towns of Manteo
and Columbia are located in the Albemarle-Pamlico watershed with Manteo along the coast and
Columbia on the banks of the Scuppernong River, 40 miles inland to the west. Both towns have
suffered damage to natural resources and water-sector infrastructure from heavy precipitation
events along with coastal and inland storm surge. A major goal of this exercise was to determine
how CREAT can best provide a framework and tool for small communities.
Morro Bay, California
The CREAT pilot study was located on the west coast with representatives from the
Morro Bay National Estuary Program, Los Osos Water Purveyors, and contractors. The aim was
to identify strategies for the Los Osos Groundwater Basin Management Plan. The Los Osos
aquifer system only has one freshwater input and no inter-basin transfers, and thus is very
sensitive to nitrate pollution from septic systems, overdraft, and hydroclimatic impacts on
precipitation. Morro Bay is located along the central coast of California. Two communities in the
area only provide limited wastewater and stormwater infrastructure serving approximately
25,000 residents. Through the desk-top studies, the U.S. Geological Survey's SEAWAT model
was added to assess potential changes in groundwater quality due to salt water intrusion and
changes in recharge dynamics.
The Ohio River basin case study
The Great Miami River watershed is located in southwestern Ohio and drains an area of
5,300 square miles including portions of fifteen Ohio counties. Principal tributaries to the Great
Miami River (170.3 miles in length) include the Stillwater River, the Mad River, and Loramie
Creek. The watershed has a population of 1.5 million people and more than 75% of the
population resides in the urban areas surrounding Dayton, Cincinnati, Hamilton, and Troy.
Approximately 83% of the land within the watershed is used for agriculture, primarily row-crop
production of corn, soybeans, and wheat. Typical livestock includes swine, cattle, and poultry.
Residential, commercial, and industrial lands account for approximately 12% of land use in the
watershed, with the remaining area consisting of forests (4%) and water bodies or wetlands (1%).
119
-------
Major industries located in the watershed produce automobile parts, chemicals, household goods,
paper products, and processed foods and beverages. CREAT was tested for a medium-sized
utility (serving approximately 20,000 people) located in the Great Miami River watershed. The
results indicate the vulnerability over turbidity and water quality deterioration in flood and
related events.
5.3.3. From vulnerability analysis to adaptation engineering
The vulnerability analysis on a water system, combined with climate information derived
from the watershed-scale modeling-monitoring framework (Figure 2-1), can provide actionable
data to support adaptation engineering. This process is represented by Steps 1, 2, or 4 in Figure
2-1 leading to adaptation design and implementation. In the CREAT process (Figure 2-54), the
vulnerability assessment is followed by adaptation planning.
An effective adaptation action may take place in watershed-scale or urban-scale for
fundamental changes of an urban system. It can also occur in system scale, aiming to improve
the infrastructure's resilience and service functions under future climate and land use conditions.
These three levels of adaptation were shown in Figure 2-2 and described in Section 1.2. In the
subsequent sections 6.0-8.0, specific adaptation engineering tools and engineering methods for
water systems are described with illustrations of case studies:
¦ Water treatment plant - climate adaptation model (WTP-cam) developed for adaptation
analysis of water treatment plants;
¦ Water distribution adaptation methods to quantify and analyze the risk of elevated
disinfection by-products (DBPs) in the distribution system;
¦ Surface water management to tackle the climate-induced increase of surface runoff in
storm water systems;
¦ Managed aquifer recharge and water reuse in adaptation to climate-induced water
availability problems.
6. SUD Methods and Tools for Drinking Water Treatment
For existing drinking water treatment plants, adaptation engineering involves the
probability-based projection of the future source water changes as well as the adaptive
engineering of unit processes to accommodate the change. This section describes one major SUD
component - the EPA WTP-cam (Water Treatment Plant-Climate Adaptation Model) computer
program, and its application for water supply system adaptation at the GCWW Richard Miller
Treatment Plant. The description is focused on new features including Monte Carlo analysis,
customization of the granular activated carbon (GAC) unit process, and GAC adaptation cost
analysis. Much of the content is drawn from publications from this research (Clark et al., 2002,
2009; Li et al., 2009, 2012, 2014; and Levine et al. 2016).
The WTP-cam version 1.0 is based on the climate adaptation models published by Li et
al. (2009; 2014) and Clark et al. (2009). The computer program was developed from the Water
Treatment Plant (WTP) model that was originally proposed by the U.S. EPA for support of
drinking water disinfection rule promulgation (U.S. EPA, 2005). The original WTP model was
designed for single case runs with deterministic solutions. In other words, when model inputs are
120
-------
defined, the model outputs will be a single value for modeled water parameters. More details on
the WTP program can be found in Appendix B.
Given the uncertainties in defining hydroclimatic impacts on source water, the WTP
model was upgraded to accept stochastic inputs from source water quality through a Monte Carlo
simulation. The latest model was renamed to be WTP-cam. WTP-cam also includes features to
examine the performance and associated cost of incremental adaptation to the treatment process
in response to changing source water quality. At present, the model is only applicable to surface
water sources subject to the climate and land use changes. Major features include the ability to:
¦ Predict natural organic matter (NOM), disinfectant residuals, and DBP concentrations.
¦ Predict the impact of the water treatment process on water quality parameters affecting
disinfectant residual decay and DBP formation.
¦ Assist utilities in evaluating the possible effects of source water variation and treatment
process operations on DBP formation.
¦ Simulate the impact of uncertainties in raw water qualities through Monte Carlo analysis.
¦ Design treatment process adaptation and estimate adaptation cost.
¦ Assist regulatory programs in evaluating adaptation design or new requirements.
6.1. Principle, models and algorithms in WTP-cam
6.1.1. Conventional treatment unit processes
For conventional treatment processes, WTP-cam version 1.0 uses the statistical
regression equations in the existing WTP2.0 program. The latter utilizes empirical correlations
intended to predict central tendencies in NOM removal, disinfection, and DBP formation for
conventional water treatment and GAC adsorption. The user manual in Appendix B (U.S. EPA,
2005) and references therein provide details of these linear regression equations and data used
for the unit process analysis. The empirical correlations were established from regressions of
water plant treatment data from usually consist of independent variables and empirical constants.
Such results are informational to national and regulatory analysis (see Figure 2-10). However,
the modeling results may not be accurate predictions for a specific water plant.
Figure 2-55 schematically shows unit processes of a conventional water treatment plant,
modeling data needs, and outputs in WTP2.0 as adopted in WTP-cam. The original WTP model
uses empirical algorithms to predict TOC and UVA removal, disinfectant chlorine decay, and
DBP formation. Underlying modeling algorithms were established from the regression analysis
of observed water plant data in the EPA's Information Collection Rule (ICR) database (see
Appendix B). For the coagulation-flocculation-filtration unit process, the TOC removal rate
(Atoc) and UVA removal rate (Auva) in alum ad ferric-based coagulation were derived from ICR
data of 39 and 21 plants, respectively. The removal rate is a linear combination of raw water
quality (SUVAraw, TOCraw) and operating parameters (pH, and coagulant dose):
&toc= f(pH,SUVAraw, TOCraw, dose)
A similar relationship is also found for the softening process. Differently, the GAC
performance in TOC removal is based on a semi-empirical model on TOC breakthrough in GAC
121
-------
Raw Water
PH
TOC
UVA
alkalinity
temperature
Br
Ca hardness
Mg hardness
ammonia
turbidity
Giardia
Crypto.
flow rate
Input
Type
Baffling characteristics
Detention times*
Chemical doses
Output from previous
process
Unit Process
Conventional
Softeninq
GAC
Membranes
Ozone
Biofiltration
Output
PH
TOC
UVA
alkalinity
temperature
Br"
Ca hardness
Mg hardness
ammonia
Disinfectant residual
DBPs
inactivation ratio
solids
Figure 2-55 Unit process, inputs and outputs in model simulation of WTP2.0 adopted in WTP-cam
program.
columns. The TOC breakthrough curve for a single GAC contactor is given by the classic
logistic function (U.S. EPA, 2005),
/(') =
TOC
roc.
1 + be
-d-t
(2.35)
where, f(t) is TOC fraction remaining; TOCin and TOCe^ are TOC influent and effluent
concentrations at the GAC unit, respectively; t is GAC service time; a, b and d are model
parameters estimated by statistical regression. The model constants a, b and d are mostly a
function of influent TOC, pH, and empty bed contact time (EBCT). Based on statistical
regression, these parameters can be estimated by (U.S. EPA, 2005),
a = 0.682 (2.36)
b = 0.167pH2 - 0.808pH + 19.086 (2.37)
d = TOCin[pH(-0.0000058 ¦ EBCT2 + 0.000111EBCT + 0.00125) +
+0.0001444 ¦ EBCT2 + 0.005486EBCT + 0.06005]
(2.38)
6.1.2. Customization of GAC unit process
For TOC removal, the use and customization of the GAC unit process is a viable
technical approach for many U.S. water plants (Levine et al., 2016; Clark et al., 2009; Li et al.,
2014, 2012). In WTP-cam, the GAC unit process is modeled with a new feature to estimate
parameters a, b and din Eqs.2.36-2.38. The new model parameterization relies on a non-linear
regression method for a given plant instead of the statistical values. This improvement allows
122
-------
one to customize site-specific conditions in estimating the model parameters. Either TOC
treatment monitoring data or long-duration bench-scale studies can be used (Li et al., 2014). The
model when calibrated aims to examine the treatment effects of different raw water sources,
GAC size, pretreatment configuration, and bed depth/empty bed contact time (EBCT).
Total organic carbon removal in GAC is often characterized using TOC breakthrough
experiments using GAC column experiments. This testing is normally conducted when changing
GAC suppliers or in pretreatment configurations. Roberts and Summers (1982) also found that
complete removal of TOC by GAC cannot be achieved under common water treatment
conditions. An immediate, partial breakthrough of TOC, can be observed even with virgin GAC,
indicating that a portion of the influent TOC is not amenable to removal by GAC treatment.
Roberts and Summers (1982) observed that the GAC effluent TOC is always lower than
the influent level, even if the GAC reactor is saturated with organics. This degree of removal
under the steady-state is attributed to biodegradation (U.S. EPA, 1996) in a way similar to the
biofiltration process (Levine et al., 2016). The ratio of TOC concentration between effluent and
influent, called "fraction remaining," generally ranges from 0.1 to 0.5 during the early stages of
operation. The ratio depends on the compositions of organic constituents in water and the
EBCT/bed depth of a GAC contactor. For steady-state removal, the fraction remaining varies
from 0.6 to 0.9 with service times from 3,000 to 14,000 bed volumes.
To obtain site-specific model parameters for design a pilot-plant or full-scale study of
GAC adsorption processes is often required, which is both time-consuming and expensive.
Instead, rapid small-scale column tests (RSSCT) (Crittenden et al. 1991; Zachman and Summers
2010) are widely used as a substitute. The RSSCT method is based on mass transfer models to
scale down a full-size GAC contactor. The hydraulic and kinetic similarity is assured by properly
selecting the GAC particle size, hydraulic loading, and EBCT of the small contactor. For this
purpose, U.S. EPA (1996, 2000) described the standardized guidelines for GAC treatment
studies to obtain high-quality TOC breakthrough data in RSSCT.
WTP-cam provides a new feature to custom-parameterize a, b and din Eq.2.35 for
specific water plant operations. For accurate modeling, the model uses a non-linear regression
method of site-specific TOC treatment data when collected. When training data are not available,
one can opt to choose default statistical values. In WTP-cam, plant-specific parameterization is
based on the modified Gauss-Newton method to estimate the model parameters a, b and d. The
procedure (Li et al., 2009 and Clark et al., 2009) relies on the non-linear regression function
through the least square analysis. This modeling technique was developed and validated using
the ICR treatment database for 63 treatment studies nationwide (U.S. EPA, 2000), including 44
RSSCT studies, 18 pilot studies, and 1 full-scale study. For a given plant, the training historical
dataset is analyzed in WTP-cam using the fitting objective function below:
n 2
Min Q(a,b>d) = TJ(yk -f(tk^,h,d)) (2.39)
k=\
where, f{t',a,b,d) = ^ + in Eq.2.35; a, b and d are the model parameters to be estimated; A =
tkand yk are the known field values, representing GAC service time and TOC fraction remaining
respectively; n is a known number of field samples.
123
-------
6.1.3. A daptation cost and economics
The economic analysis of adaptation can be estimated for engineering options. For
treatment adaptation, WTP-cam allows estimating the costs to the changes in GAC process
design or operation. In general, the GAC process cost includes the cost for GAC contactor cost,
initial GAC cost in setup, annual GAC make-up cost, and GAC reactivation cost. The initial
GAC cost is a one-time charge for GAC required to fill the contactor, equal to the product of the
total volume of contactors, the density and unit cost of virgin GAC. The annual GAC make-up
cost is the yearly cost of GAC lost due to reactivation. It is the product of the GAC loss rate in
reactivation, GAC reactivation rate, and virgin GAC cost. The GAC contactor cost can be
estimated from the cost model by Adams and Clark (1988):
y = a + b{USRT)cdz (2.40)
where y is the capital, operational or maintenance cost; USRT is the process design or operating
variable that is the total surface area of the GAC filter for contactors (total hearth area for GAC
reactivation) or the total effective volume of the GAC unit for capital cost; a, b, c and d are
empirical parameters determined from nonlinear regression analysis, and z is either 0 or 1 for
adjusting the cost functions for a given range of USRT values.
The costs in Adam and Clark (1988) are based on the year-1983 dollar value. In WTP-
cam model computation, all costs are converted to the year 2009 dollar using the Producers Price
Index (US BLS, 2008). The same method can be used for other years of interest; revision to
other benchmark years will be made later. The contactor cost can be further categorized by the
costs of capital, process energy, building energy, maintenance material, and operational and
maintenance (O&M) labor. The computational parameters are listed in Table 2-24. GAC
reactivation cost is the other variable estimated using a similar algorithm (Eq.2.40). The model
parameters are different and are listed in Table 2-25.
Table 2-24 GAC contactor cost estimate parameters
Type of
Cost
Capital
Process
energy
Building
energy
Maintenance
Material
O&M Labor
USRT
volume
area
area
area
area
a
93700
0
15150
540
1160
b
1999.1
12
350
23.6
0.3
c
0.712
1
0.916
0.753
1.068
d
0.958
1
1
1
1.152
z
1
1
1
1
1
Unit cost
Construction Cost
0.08 $/kwh
0.08 $/kwh
--
9 $/hr
1.3y
(in 2009)
(in 2009)
(in 1983)
Ratio of
2009to1983
cost
2009ENR/1983ENR=
R=2.16
--
--
2009PPI/1983
PPI
= 2.56
2009
PPI/1983
PPI
= 2.56
124
-------
Table 2-25 GAC reactivation cost
Type of
Cost
Capital
Process
energy
Building
energy
Maintenance
Material
O&M Labor
Natural Gas
USRT
area
area
area
area
area
area
a
144000
354600
12250
0
2920
648400
b
198300.4
6387
312.1
4456.6
282
287714.9
c
0.434
0.755
0.649
0.401
0.7
0.899
d
1
1
1
1
1
1
z
1
1
1
1
1
1
Unit cost
Constructio
0.08
$/kwh
0.08 $/kwh
9 $/hr
$0.0035 /scf
n Cost 1,3y
(in 2009)
(in 2009)
(in 1983)
(in 1983)
Ratio of
2009ENR/
2009PPI/1983
2009PPI/1983
2009PPI/1983
2009/1983
1983ENR
PPI
PPI
PPI
cost
= R = 2.16
= 2.56
= 2.56
= 2.56
WTP-cam also introduces annualized cost in economic analysis. In a capital recovery
analysis, for example, a 20-year return period and a 5% annual interest rate can be assigned to
construct a cost curve that illustrates the total annual cost of the GAC system in adaptation
options in different GAC service time or reactivation period. For illustration, Figure 2-56 shows
an example cost curve developed for the GCWW's Miller WTP. The Miller WTP has 12 down-
flow gravity GAC contactors and two multi-hearth furnaces for onsite reactivation. Each of the
Miller WTP contactors
has a volume of 595
m3 and a surface area
of 181 m2 The overall
GAC loss rate through
the system is about
8%. The carbon
loading rate is 482
kg/day of GAC per
square meter of the
hearth area. This cost
curve is the basis for
GAC adaptation
analysis at the
treatment plant.
Details of this
adaptation case study
will be described later
in Section 6.3.
&
o
Uf
"c3
3
100 200
Reactivationperiod, days
300
400
Figure 2-56
Cost curve for annual cost of GAC unit, indicating the
cost associated with a given GAC reactivation period
for the Miller water treatment plant in Cincinnati.
125
-------
6.1.4. Other unit processes for adaptation
Other unit processes in advanced water treatment are applicable for adaptation to the
changes in source water quality. Examples include membrane and advanced oxidation for
removal of emerging and trace contaminants such as endocrine-disrupting compounds, algae
toxins, herbicide, and pesticides, etc. Such advanced tertiary processes will be added to WTP-
cam in the future.
6.2. Adaptation Analysis using WTP-cam
A WTP-cam simulation can proceed in the following sequential steps: (1) source water
quality definition, (2) product water quality projection, (3) model validation, and (4) result
analysis and visualization. Appendix B provides program illustrations and instructions for the
WTP-cam simulation. The program is Windows-based running in a Windows 8 or newer
operating system.
6.2.1. Treatment process and compliance targets
6.2.1.1. Treatment processes in simulation
In a WTP-cam simulation, the first step is to develop a physical model of a treatment
plant for examination. Figure 2-57 shows an example for the GCWW Miller WTP. Raw water
from the Ohio River is pumped into two large equalization basins at the plant. Then the water
(A)
(B)
Raw Water
Alum
Rapid Mixing|
1
Flocculation
Pre Settling
Reservoir
Lime
Settling
Filtration
GAC
Chlorine (Gas)
Clearwell
31
WTP Effluent
JL
Average Tap
End of System
Figure 2-57 Schematic diagram for (A) treatment unit process at the GCWW Miller water treatment
plant, and (B) WTP-cam program flow in the example simulation.
126
-------
enters treatment units in coagulation, sedimentation, rapid sand filtration, followed by GAC
processing. The spent GAC is reactivated in two large on-site furnaces. After chlorine
disinfection, the treated water is stored in a clearwell before being pumped into the distribution
system. More details will be provided in Section 6.3. In WTP-cam simulation, these unit
processes are arranged into a process train as shown in the sequential block diagram (Figure 2-
57b).
Data input into the simulation program includes physical and process information for the
treatment plant. The data for conventional treatment processes - sedimentation, coagulation, and
flocculation, are shown in Figure 2-58a. A set of physical parameters need to be measured for
specific treatment systems; for example, volume and geometry of the flocculation basins, rapid
mix basin geometry, and settling basin geometry. For GAC absorption, parameter input in the
simulation setup is shown in Figure 2-58b.
6.2.1.2. TOC simulation and compliance targets
WTP-cam program is focused on TOC removal and the assessment for DBP stage-II
regulation compliance under future conditions. For this purpose, the simulation of Quarterly
Running Average is designed to assess TOC and hence DBP compliance in water treatment. The
influent
pH ,
influent to pn? tu e 18,6 (Celsius)
us)
2.6 (itg/L)
0.096(l/cm)
0.069(mg/L)
72 (mg/L as caco3)
77 (mg/L as caco3)
141 (mg/L as cacoB)
0,21 (mg/L as N)
Alui
Rapiei
Minimum Tenrceratiire
Total oraamc tar Don
uv Absor[
Bromide ,,,,,
Alkalinity
calcium Hardness ....
Total Hardness
AMiioma
Turbidity 151,0 (ntu)
Peak Flow 220.000(MGD)
Plant Flow 120.600(MGD)
Sirface rfater by satr true (true "false^
sou! ce aster crypto, concentration 0,000 foocysts Liter)
ltz Rule watershed control Prog, credit" . false sjwik false)
:f gw system, is virus cisinfecticn Peq'd0 false Ttrue falser
virus Disinfection for gw. if Req d 4,0 (loos)
1.1 (mg/L as a12(so4)3*14h2o)
Filtration
Liquid volume
Ratio of T50;Detention Tine
Ratio of TlO1 Detentior "'"ine
chlorinated Backwash water"
Filter Media -'Anthracite ''sand or gac) ,,
Giardia Removal credit - conv, Filters ,
virus Removal credit - conv. Filters ...
crypto. Removal credit - conv. Filters ,
Giardia Removal credit - Direct Filters
virus Removal credit - Direct Filters ,,
Crypto. Rerro\al credit - Direct Filters
cfe Turb. Meets LT2 ""oolbox criteria" ,,
:fe Turb. Meets LT2 "oolbox criteria" ,,
crypto, credit as 2nd Stage Filt
\i life of Basin 0.0084(mg)
Rntio of T50'Detention Time 1.00 (ratio)
Patiu of TlO/Detention Time 1,00 (ratio)
Fl0CC
lire of Basin 1.9400(MG)
Ratm of T50/'Detention Tine 1,00 (ratio)
Ratii nf T10/Detention Time 0,50 (ratio)
Presed,
\olu..e of Basin 2.2300(KG)
Ratio of 150 cetertion Tire 1 0u ^ "tio
Retio nf no Detention ^ire u 44 r tio"1
Eligible for LT2 Toolbox cr\pto. credit" . falce ,trie fal
L~2 -onlbox crypto. Rero al credit ,, u 4 Jnq
Reservoir
\oh re of Basin 373.0000(MG)
- * T50/Detention Time 1.00 (ratio)
Ratio of TlO/Detention Time 0,32 (ratio)
GAC
Empty Bed cortact Ti*e k'at "Flant Flo*'
gac React! ati on inter .al ,
gac contacting sister (Single Blenaed;
"OC Breakthrough for Single unit ,Va\ A.g)
crypto, Reno.al edit as 2rd stage
chlorine iGasi
U-lorihe Dose
contact ~= -
volume of Basin
Ratio of T50/Deterttion Time
Ratio of TlO/Detention Time
wtp Effluent
Average Tap
Average Residence Time (For Average Flow)
End of system
Maximum Residence Time (For Average Flow)
A/S
iTRUE/FALS
is or G)
5 (logs)
2.0 (logs)
3 u ibq ^
2.0 (logs)
1.0 (logs)
; 0QSJ
FALSE ("RUE F^L
FALSE TRUE FhL
0.5 (logs)
nl (minutes)
160 (days)
Blended(s or B)
A\g_TOC(M or A)
0.5 (logs)
3.0 (mg/L as C
28.3000(MG)
1.00 (ratio)
0.20 (ratio)
1.0 (Days)
3.0 (Days)
Lime
5,0 (ig/L as ca(OH)2)
PH_ADJ.(P or s)
Lime Dose
For pH adjustment (P) or softening (s)
settling Basin
volume of Basin 2-
Ratio of T50/'Detention Tine 1 "0 f?tm
Ratio of HO/Detention Time
(A) (B)
Figure 2-58 Original input data for the example process train at the Miller WTP. (A) inputs for plant
and conventional treatment process; (B) parameters for filtration, GAC and chlorine
disinfection in advanced treatment.
127
-------
compliance criterion for TOC is <2.0 mg/L calculated quarterly as a running annual average.
Other regulatory targets or treatment objectives are not considered in WTP-cam version 1.
In a WTP-cam simulation, four computed running averages are calculated; one for each
quarter of a year. The running annual average is defined as the arithmetic average of TOC
concentrations at a current season and previous three seasons based on the U.S. EPA
disinfectant/disinfection by-product (D/DBP) regulations. For example, the running annual
average for TOC is calculated for the GAC treatment effluent of the GCWW's Richard Miller
Treatment Plant (Table 2-26).
The means, variances, and cross-correlations of raw water parameters vary with seasonal
changes in most cases. Thus, four sets of input parameters for raw water qualities need to be
prepared as modeling inputs, for which four simulations are conducted each year responding to
the four seasons (Table 2-26). The TOC concentration is recomputed for each season as the
running average.
Table 2-26 Illustration of calculating the running annual average for finished water TOC
Year
Season
TOC concentration
Running annual average
Spring
1.3
--
2009
Summer
1.7
--
Autumn
2.2
--
Winter
1.7
1.7
Spring
1.2
1.7
2010
Summer
1.4
1.6
Autumn
2.4
1.7
Winter
1.5
1.6
6.2.2. Monte Carlo methods in modeling source water quality
6.2.2.1. Incorporation of hydroclimate uncertainties
Several source water quality changes, such as increasing levels of turbidity, nutrients,
blue algae, and biological contaminants, can affect the removal of TOC and other contaminants
in water treatment. A newly developed modeling-monitoring platform in Figure 2-1 utilizes near
real-time (daily) high-resolution satellite monitoring to provide real-time water quality
information in assisting water plant operations (Imen et al., 2016). Now many water quality
parameters relevant to drinking water treatment can be quantitatively monitored, including major
nutrients, turbidity, TOC, chlorophyll-a, and microcystin (Chang et al., 2014a,b). Furthermore,
the modeling-monitoring platform integrates climate and land use models to project future
changes of major water quality parameters such as TOC, total phosphorous, total nitrogen, and
turbidity. If the change is significant, these water quality projections are needed for design basis
development in water infrastructure adaptation. However, future water quality projections often
contain large uncertainties. Even with the use of the integrated modeling-monitoring techniques,
128
-------
Start
the full range of water quality parameters of necessary accuracy for engineering planning and
design may not be readily available. Water managers in charge of water adaptation will likely
face the uncertainty challenge in the foreseeable future.
In this context, Li et
al. (2014, 2012) proposed
the use of Monte Carlo
analysis as a practical tool to
characterize the range of
future water quality changes.
In this approach, Monte
Carlo analysis is used to
obtain sample solutions by
repeating a simulation
process for problems
involving random variables
of known probability
distributions. The
correlations among water
quality parameters are
assumed to remain for the
future period of interest in
order to establish the water
quality parameters such as
TOC and turbidity for WTP-
cam simulation. This Monte
Carlo simulation to account
for climate-related
hydrological uncertainty is a
new feature not previously
available in WTP v2.0.
Stage 1: Parameter preparation
If "Quarterly Running Average" is checked, prepare four different
sets of parameters such as raw water statistics for spring, summer,
autumn and winter seasons. Otherwise, prepare one set of
parameters.
If "Preserve Correlation" is checked, read corresponding data file(s)
to compute four/one set(s) of parameters for multivariate modeling.
If raw water quality statistics are provided by data file(s), read
corresponding data file(s) to compute four/one set(s) of raw water
statistics.
Initialize the random number generator by Seed for Random Number.
Obtain Raw Water Probability Dis tribution.
Stage 2: Monte Carlo loop from 1 to Number of Runs.
Simulation of raw water quality.
If "Quarterly Running Average" is checked, compute raw water
qualities using raw water statistics and correlation matrixes in
turn from spring, summer, autumn and winter.
If "Preserve Correlation" is checked, compute raw water quality
based on multivariate modeling. Otherwise, simply based on
raw water probability distribution.
Performing a WTP run for this realization. If "Quarterly Running
Average" is checked, compute the quarterly running average using
the simulated water quality from this realization and previous three
realizations.
'Contaminant Control" is checked and for a non-compliance
realization:
First to estimate the maximum permitted concentration of
"Controlled Contaminant" for this realization using
"Regulation Standard" and "Margin of Safety".
Second to seek a proper control variable for the "Controlled
Processing Unit" that make the "Controlled Contaminant"
to be the maximum permitted concentration.
Compute the adaptation cost with the current control variable.
Save outputs to files.
If
Figure 2-59 shows
the key steps of the Monte
Carlo analysis in WTP-cam
simulation; new inputs are
marked in bold. Three key
considerations in the Monte
Carlo analysis are: (1) the
Quarterly Running Average,
(2) Preservation of the
correlation among
parameters, and (3) Pollutant
removal targets. The
Quarterly Running Average parameter is specially designed for the regulation consideration over
TOC concentrations. TOC <2.0 mg/L in finished water, calculated quarterly as a running annual
average, is an important compliance criterion according to the EPA disinfectant/disinfection
Figure 2-59
End
Program logic sequences in Monte Carlo simulation of
future source water quality variations using the
correlation matrix method (from Li et al., 2014).
129
-------
byproduct (D/DBP) rule. Furthermore, WTP-cam applies four quarterly statistics to represent
annual seasonal variations. Seethe preceding Section 6.1.1.
The modeling option for "Preserve Correlation" is designed to preserve the joint
correlation among raw water quality parameters when simulating stochastic raw water quality
variables. In the presence of cross-correlation, concentrations of correlated reactants vary
simultaneously in the source water. This assumed cross-correlation among raw water quality
parameters can affect the calculation of DBP formation during water treatment and distribution.
A first-order multivariate seasonal autoregressive model (Bras and Rodriguez-Iturbe, 1984) was
used in the WTP-cam. This seasonal model preserves all seasonal means and variance of water
quality parameters, all cross-correlations among all water quality parameters, and lag-one
correlations between adjacent seasons and between all water quality parameters. Section 6.2
describes the theoretical basis for the applied multivariate analysis.
The Pollutant Option in the modeling is designed to modify the design and operation of
the current processing train when a non-compliance realization is detected in simulation. For
example, when a TOC
non-compliance in
finished water occurs,
the WTP-cam program
can be used to design
operation modification
by increasing the
frequency of GAC
regeneration. Such an
adaptation measure
aims to bring the TOC
excursion within
acceptable limits. The
inputs for this option
are made for a given
contaminant,
regulation standard,
margin of safety, and
unit processes in a
treatment plant. For the Figure 2-60 Graphic user interface for inputs in Monte Carlo simulations of
WTP-cam version 1, future water qualities,
the engineering option
has been developed for TOC control in the GAC treatment process.
6.2.2.2. Inputs for Monte Carlo Setting
The input parameters for Monte Carlo analysis may be divided into three groups: analysis
options, control parameters, and the source of influent water quality statistics/correlation. Figure
2-60 shows a graphic user interface (GUI) for these inputs in the example process train of Figure
2-57.
Monte Carlo Setting
-Options—
p" Preserve Correlation
p" Quarterly Running Average
p" Contamination Control
Controlled Contaminant
|TEE 3
Controlled Processing Unit
| GAC V]
Raw WQ Probability Distn
|l_ogNormal
"3
Control Parameters
Number of Runs, >1
Seed for Random Number, 1-50000
Regulation Standard, mg/L
Margin of Safety, mg/L
Source of Influent WQ Statistics
11000
|"l68
[5^
Or Input manually, Please Click Here
Correlation Matrix
Please Provide Data File(s) Here if Preserve Correlation is Checked
Computed by Available Data File(s), Please Click Here
Default Example
130
-------
Analysis options: the options are designed to govern the flow of Monte Carlo simulation.
Table 2-27 lists the name of the option, range of available values, and description of the option.
Control parameters: there are four control parameters used in the Monte Carlo simulation:
¦ Number of Runs - a user-defined integer to specify the number of runs required.
¦ Seed for Random Number - a positive number to initialize the random number generator
in the program. The Monte Carlo simulation can be repeated using the same random
number seed.
¦ Regulation standard - a value representing the compliance standard for the controlled
contaminant selected in Options.
¦ Margin of Safety - refers to the difference between the compliance standard and the real
controlled concentration that provides extra reliability for compliance. The margin of
safety is usually set within 1-10% of the regulation standard.
Table 2-27 Options for Monte Carlo analysis
Control
Range of value
Description
Preserve Correlation
TRUE/FALSE
Multivariate analysis will be used to
simulate stochastic raw water quality if
TRUE (checked).
Quarterly Running Average
TRUE/FALSE
Simulation will be based on four seasons
of variation if TRUE.
System Adaptation
TRUE/FALSE
Loading adaptation program for the non-
compliance realizations if TRUE.
Controlled Contaminant
TOC/None
Determining the contaminant to be
controlled by adaptation.
Controlled Unit Process
GAC/None
Determining the unit process that can be
adapted for controlled contaminant.
Raw Water Probability
Distribution
Norma l/Lognormal
Determining the probability distribution for
all raw water quality parameters
6.2.2.3. Source of influent water quality statistics/correlation
Influent water quality statistics are essential to generate raw water quality parameters for
the input of each simulation. Two methods provided by WTP-cam are available to obtain these
parameters. The first is to simulate source water quality using the correlation matrix (see Section
6.2 for details). The second method is to input these parameters manually through the manual
input function. Four dialogue windows appear one at a time for the four seasons if Quarterly
Running Average is checked. Figure 2-61 shows an example of a manual input window for the
Spring in the example process train at the Miller WTP in the Cincinnati case study. These
datasets are saved in separate files for retrieval and simulation. See Appendix B for program
details.
131
-------
To project future
water quality, it is
assumed that the
covariations among
water quality parameters
in the Ohio River source
water will remain. This
assumption allows one to
compute all other
important parameters
from a target TOC level,
which are modeling
parameters in WTP-cam
simulations. The joint
correlations among raw
water quality parameters
are preserved when
computing the stochastic
raw water quality in the
future. This statistics-
based seasonable
multivariate analysis was
conducted through Figure 2-61 Manual input window for influent water quality statistics.
Monte Carlo simulations.
Detailed principles and
mathematical relations are contained in Section B4.1.1 of Appendix B.
In summary, the first-order multivariate seasonal autoregressive model AR(1) (Bras and
Rodriguez-Iturbe, 1984; Salas et al., 1980) was adopted in the WTP-cam algorithm for the model
simulation. For each season of a year, water quality measurements are Log-transformed in WTP-
cam simulation into variables x[ and y[. These two variables become normally distributed with
means mx.and my., standard deviations Sx.and Sy and the correlation coefficient among them.
The sample means, standard deviations and correlation coefficients of the transformed variables
x'i and y[ are calculated. These parameters are then used to build the necessary auto-covariance
and cross-covariance matrices (see Appendix B, Section B4.1.1.). The final results are projected
water quality value and its associated range in the probability distribution. This analysis follows
the following steps:
¦ Define a domain of possible inputs.
¦ Generate inputs randomly from the domain using a specified probability distribution.
¦ Perform a deterministic computation using the inputs.
¦ Aggregate the results of the individual computations into the final result.
6.2.3. Advanced unit process and adaptation cost
Adaptation analysis using WTP-cam is conducted after a noncompliance event is
identified. Adaptation refers to necessary changes in the design and/or operation of the current
Raw Water Quality Statistics Input Window
Time Horizon: Spring
OK
|Parameter
Average
Standard Deviation
pH, -
7.7
0.17
Alkalinity, mg/L
55.5
18.2
Turbidity, NTU
43.4
38.0
Calcium Hardness, mg/L
63.5
23.3
Total Hardness, mg/L
110.4
18.4
TOC, mg/L
2.3
0.6
UVA, 1/cm
0.12
0.06
Bromide, mg/L
0.03
0.01
Ammonia, mg/L
0.29
0.41
Temperature, Celsius
12.4
0
Flow Rate, MGD
108.4|
0
Cancel
132
-------
water treatment train or unit process. At this time, the adaptation module is fully developed for
TOC treatment in GAC adsorption. Adaptation for other treatment unit processes has not yet
been programmed into the WTP-cam software.
The margin of safety is an option in adaptation analysis using WTP-cam. The margin of
safety refers to the difference between the compliance cut-off point and the calculated
concentration. For example, if the margin of safety is O.lmg/L, the controlled TOC concentration
will be 1.9 mg/L. The simulated running annual average of TOC concentration should be <1.9
mg/L in engineering analysis.
When a TOC noncompliance event is identified, one effective adaptation technique is to
reduce the GAC service time through treatment process adjustment (Li et al., 2014). In the WTP-
cam simulation, the appropriate GAC service time is calculated and the process control is
adjusted to ensure TOC <1.9 mg/L. The computation procedure is as follows:
¦ Reducing current GAC service time by one day.
¦ Using the new service time to re-compute the TOC concentration for each of four seasons
without change to the other operating conditions in each season.
¦ Calculating the new running annual average of TOC.
¦ Comparing the new calculated TOC to the controlled concentration of 1.9 mg/L. The new
service time is adopted if new TOC is less than 1.9 mg/L; otherwise, repeat computation
from the first step until the solution is found.
Treatment adaptation by modifying the GAC process can reduce the likely future risk of
TOC noncompliance. Such potential options are further evaluated in WTP-cam on adaptation
cost and treatment effectiveness. Apparently, a reduced GAC reactivation period and operational
adjustment may increase energy consumption a major item in the primary adaptation cost. The
cost for GAC replacement and reactor optimization can be estimated using the equations and
procedures described in the preceding section 6.1.3. An example of such engineering and
economic analysis is described next.
6.3. GCWW Richard Miller Treatment Plant case study using WTP-cam
GCWW provides drinking water at -5.26 m3/s or 120 million gallons per day (MGD) to
~ 235,000 customer accounts through 5,100 km of water mains. Built in 1907, the GCWW's
Miller WTP treats surface water from the Ohio River and provides 88% of the drinking water
supply to the customers at a maximum summer capacity of 9.65 m3/s (220 MGD). In this
adaptation case study, the WTP-cam tool was used in the simulation to assess the likely
hydroclimatic changes in the future on drinking water treatment at the Miller WTP. The
investigation results have been published by Li et al. (2014, 2012) and others. The technical
questions for adaptation study include:
¦ How the climate-related risk to drinking water standard violations can be assessed?
¦ What adaptation limit or climate impact threshold can be established?
¦ What is the probable cost associated with the adaptation scenarios?
133
-------
6.3.1. Miller water treatment plant operation and performance
Table 2-28 Miller WTP unit process design parameters
Unit Process
Volume, m3
T10, min
Rapid mixing
32
2
Flocculation basin
7343
14
Pre-settling
8441
14
Reservoir settling
1411805
1,728
Coagulation basin
98410
144
Filtration
9352
4
GAC Contactors
9311
--
Clear-well
107116
77
6.3.1.1. Treatment process and modeling
Figure 2-57 in Section 6.2.1 shows the treatment process at the Miller WTP. Raw water is
taken from the Ohio River. At the time of the investigation, the treatment process consisted of
coagulation, sedimentation,
biologically active rapid sand
filtration, GAC adsorption, and
water disinfection. A new UV
disinfection facility started
operation shortly after this
adaptation analysis. It replaced
the conventional chlorination.
All data acquired before the
UV unit operation are used in
this analysis based on chlorine
disinfection (see Figure 2-57).
The intent was to examine the
conventional use of
chlorination as the basis for the
adaptation analysis.
TOC and turbidity are
the subjects of the adaptation
study. These DBP-formation
precursors or potential
indicators are removed by the
conventional coagulation, sedimentation, biologically active rapid sand filtration, and GAC
adsorption at the plant. The Miller WTP design parameters are listed in Table 2-28. In the table,
Tio value is the hydraulic retention time required for the effluent tracer concentration to reach
10% of the inflow tracer concentration. It is normally determined in a step-dose conservative
tracer test of the treatment unit.
The plant performance and treatment efficiency of each unit process were evaluated using
the U.S. EPA's ICR database. The database was designed to obtain water quality, water
treatment, and occurrence information needed for the development of Safe Drinking Water Act
regulations. ICR data include detailed information on plant design, treatment processes, and
operations for all large public water utilities in the U.S., each serving a population >100,000. The
data collection covered an 18-month monitoring period from July 1997 through December 1998.
The ICR database also provides water quality measurements at various sampling locations along
the water treatment train and in a water distribution system.
For the Miller WTP, specific data utilized for the treatment simulation cover three
sampling periods: sample period 10 (April 1998), sample period 13 (July 1998), and sample
period 16 (October 1998). Raw water inflow rate and chemical feed doses during each period are
listed in Table 2-29. Based on the information provided by GCWW, the GAC reactivation during
these periods was set at 8 months for the winter-spring season and as 4 months for the summer-
early autumn season.
Note: - data not available.
Data source: U.S. EPA ICR database.
134
-------
6.3.1.2. GAC absorption and TOC removal
Table 2-30 shows statistics of the performance parameters for the GAC unit based on
weekly samples for the 76-month period from January 2004 to April 2010. The data includes the
influent and blended effluent TOC concentrations, the number of active GAC contactors, the
plant inflow rates, and the
GAC reactor empty bed
contact time (EBCT).
Influent TOC concentrations
follow an annual cycle with
seasonal extremes ranging
from 1.01 mg/L (March 24,
2004) to 2.76 mg/L
(September 22, 2004).
Blended effluent TOC
varied from 0.26 mg/L (July
13, 2005) to 1.44 mg/L
(November 1, 2006); all
concentrations were below
the compliance standard of
2.00 mg/L.
The GAC plant flow rates averaged 5.26 m3/s (120 MGD), ranging from 3.26 m3/s (74
MGD) on December 27, 2006 to 7.61 m3/s (174 MGD) on September 5, 2007. Among the 12
available GAC contactors, 6 to 11 contactors were in operation at any given time. GCWW's
operational strategy for the GAC process demands that GAC contactors be brought on-line,
reactivated, and taken off-line in a staggered sequence. This operation aimed to balance a variety
of operational goals including total trihalomethane (TTHM) reduction, water production, furnace
operation schedules, GAC storage, and the effective removal of Spring pesticide runoff in the
source water. To meet these operational goals, the monthly average EBCT was set consistently
around 17 minutes
Table 2-30 Statistics of full-scale field measurements
Field Measurements (units)
Average
Standard
Deviation
Coefficient
of Variation
Minimum
Maximum
Sample
Size
Influent TOC, (mg/L)
1.72
0.36
0.21
1.01
2.76
289a
Blended effluent TOC, (mg/L)
0.85
0.26
0.31
0.26
1.44
289a
Number of active GAC
contactors
9
1.22
0.14
6
11
289a
EBCT, (minute)
17.1
0.8
0.05
12.1
24.4
279b
Plant water inflows, (m3/s)
5.26
0.81
0.15
3.26
7.61
279b
Plant TOC mass inflow, (g/s)
9.17
2.57
0.28
4.72
16.6
279b
Note: a 2296-day sample period from Jan 7, 2004 to April 21, 2010 (one sample every 7.94 days)
b2184-day sample period from Jan 7, 2004 to Dec 30, 2009 (one sample every 7.83 days)
Table 2-29 Inflow and chemical feed levels for the Miller WTP
Parameter
Sampling period
10
13
16
Inflow rate, m3/s
4.41
5.27
5.76
Alum at RM, mg/L
0.87
1.82
0.87
Lime at COAG, mg/L
6.73
7.92
4.62
Chlorine at CLR, mg/L
1.26
1.56
1.46
Note: RM-rapid mixing; COAG-coagulation basin; CLR-clearwell.
Data source: U.S. EPA ICR database.
135
-------
Figure 2-62 shows the temporal pattern of influent TOC and blended effluent TOC
concentrations at the GAC unit process during the 76-month sampling period. Seasonal changes
of influent TOC to the GAC reactor are evident. Higher concentration always occurred in the
second half of the year compared to the first half, and this seasonal variability is consistent with
TOC levels in the Ohio River.
Influ
ent
O)
£
cf
o
c
0
o
c
o
o
O
O
Blended efflue
nt
Compliance standard
Figure 2-62
3
2.5
2
1.5
1
0.5
0
1/1/04 1/1/05 1/1/06 1/1/07 1/1/08 1/1/09 1/1/10 1/1/11
Date
Temporal variations of influent and blended effluent TOC in the GAC unit.
Influent TOC concentrations and blended effluent TOC concentrations are not
significantly correlated with each other (Figure 2-62). Pearson product-moment correlation
coefficient R is only 0.08. However, as shown in Figure 2-63, the number of active GAC
contactors is highly correlated with plant inflow (R=0.75) and mass inflow (R=0.65). Due to
increased water demands during warm weather, summer months had higher plant inflow rates
than winter months, and, hence, more GAC contactors were active in the summer time. The
number of active GAC contactors is negatively correlated to the blended effluent TOC
concentration (R=-0.69). See Figure 2-64. This indicates that TOC in the finished water is
controlled mainly by the number of GAC contactors in service, not TOC concentration entering
the treatment unit. This result strongly shows the effectiveness of GAC operation in TOC
removal.
CO
o
¦5
c
o
o
a)
>
-I-'
o
ro
a)
-Q
E
3
12
9
6
3
0
ive
Water In'
Mass Infl.
ntactors
low
ow
20
15
U
0J
in
nr
oj
m
in
£,
10
_o
_o
4—
_c
Ll_
in
5
0J
4—'
ru
in
ro
0
5
1/1/04 1/1/05 1/1/06 1/1/07 1/1/08 1/1/09 1/1/10 1/1/11
Date
Figure 2-63 Temporal variations of inflow, mass inflow and active number of GAC contactors.
Figure 2-65 compares temporal variations in EBCT and plant inflow. While the plant
inflow rate displayed pronounced seasonality, the overall average EBCT across the bank of
136
-------
12
¦C 6
ro
b conta
lended effl
ompliance
ctors
uent
standard
O
O
0 L] Ld 0
1/1/04 1/1/05 1/1/06 1/1/07 1/1/08 1/1/09 1/1/10 1/1/11
Date
Figure 2-64 Temporal variations of active contactors and blended effluent TOC concentration.
active GAC contactors was relatively stable. This quasi-steady EBCT is achieved in operation
successfully by adjusting the number of active GAC contactors to meet GCWW's operational
goals including the offsetting of the seasonal inflow variations.
40
30
O 20
m
LU
10
0
EBCT
Water inflow
4 °
-------
Logp) LosdOQ
Figure 2-66 Normal probability plots for source water pH (107 samples) and TOC (93 samples) for
Ohio River from the ICR database (July 1997-December 1998).
running annual average of TOC in the finished water. The mean and standard deviation become a
stable constant after 500 and 2,000 runs, respectively; the skewness becomes constant after 5,000
runs. Therefore, 5,000 runs were chosen for all Monte Carlo simulations.
Another assumption is that the confounding effects of population growth can be
neglected for plant flow rates. This simplification allows identification of the climate-induced
water quality changes impacting water treatment performance. The design and operation
conditions for the Miller WTP under future scenarios were initially kept unchanged from the
baseline period. In addition, it was assumed that the coefficients of variation for all water quality
parameters in 2050 would remain the same as those for the baseline data. This similarity is
guided by the ratios of
-------
6.3.2.2 Source water characterization
Surface water quality at the plant intake from the Ohio River varies significantly in
response to upstream hydrological changes and watershed management. Factors affecting the
water quality variation include upstream watershed management, river spills from ships, and the
hydroclimatic factors such as seasonal and long-term precipitation changes.
Baseline Condition in 1998
Water quality variability in the river is characterized for a period July 1997 to December
1998 using water quality data from the ICR database. Averages and standard deviations for the
baseline period are shown in Table 2-31. Because the SDWA TOC regulation requires reporting
of quarterly running annual average, the raw water quality at the plant is divided into spring (March
to May), summer (June to August), autumn (September to November) and winter (December to
February). The source water quality exhibited a statistically significant difference among the
seasons.
Joint Correlation of Source Water Quality Parameters
For the lognormal distribution (see Figure 2-65), a Monte Carlo simulation was used in
modeling source water quality parameters. The basis for the modeling, such as the joint
correlation, was described in preceding section 6.2.1. Monte Carlo simulations after 5000 runs
established correlations among the 9 water quality parameters for the Ohio River water. The
numbers in italics (Table 2-32) illustrate that more than half of the pairs of source water quality
parameters are statistically correlated; their correlation coefficient is >0.2 and the p-value is
<0.1.
6.3.2.3 Projecting raw water quality in 2050
To assess future source water quality, the 1998 baseline data was modified to project
possible water quality scenarios in the Ohio River in 2050. The 2050 water quality projection
considered the following aspects of anticipated changes.
¦ TOC. Alkalinity, and Total Hardness
Skjelkvale et al. (2005) studied the regional trend of surface water chemistry and
acidification for 12 geographic regions in Europe and North America from 1990 to 2001. As one
of the 12 regions, the Appalachian Plateau includes the upstream reaches of the Ohio River.
Therefore, the regional trends in their study for alkalinity, total hardness, and dissolved organic
carbon (DOC) were adopted to estimate these parameters for the period 1998 to 2050. The trends
for alkalinity, total hardness, and DOC are equivalent to a change by +0.036, -0.22, and +0.03
mg/L per year, respectively. Because DOC is usually the main component of TOC, the trend for
TOC is assumed the same as that for DOC.
¦ Ammonia
The most important sources of ammonia are from decomposed plant and animal matter,
fertilizer, sewage, and industrial effluents. Whitehead et al. (2006) investigated the hydroclimatic
impacts on ammonia in the River Kennet of the U.K. for the period from 1961 to 2100. A 25%
increase in ammonia from 1998 to 2050 can be assumed based on their study. It is believed that
139
-------
Table 2-32 Correlation matrix for source water quality parameters (for Ohio River from July 1997 to December 1998)
Parameter
Statistics
Alkalinity
Turbidity
Ca
hardness
Total
hardness
TOC
UVA
Bromide
nh3_n
PH
P
p-value
0.63
0.00
-0.11
0.27
0.04
0.67
0.43
0.00
0.36
0.00
0.11
0.32
-0.02
0.87
-0.02
0.83
Alkalinity
P
p-value
1
-0.15
0.13
0.06
0.51
0.80
0.00
0.63
0.00
0.25
0.02
0.26
0.02
-0.08
0.45
Turbidity
P
p-value
1
-0.15
0.11
-0.27
0.01
0.32
0.00
0.54
0.00
-0.38
0.00
0.32
0.00
Ca
hardness
P
p-value
1
0.25
0.01
0.02
0.85
-0.29
0.01
0.20
0.06
-0.17
0.14
Total
hardness
P
p-value
1
0.36
0.00
0.10
0.35
0.59
0.00
-0.26
0.02
TOC
P
p-value
1
0.65
0.00
-0.04
0.71
0.18
0.13
UVA
P
p-value
1
-0.42
0.00
0.29
0.02
Bromide
P
p-value
1
-0.12
0.31
140
-------
the 25% increase in ammonia is reasonable for the source water quality in this study, but the
effect of ammonia on TOC in finished water is negligible (see later discussion).
¦ Bromides, UVA, pH, Turbidity, and Calcium Hardness
Bromides occur naturally in both surface and groundwater but are particularly high in
areas of saline intrusion. Cromwell III et al. (2007) pointed out that sea level rise in future
climate would increase bromide levels in coastal regions. However, there is presently no
evidence to indicate changes of bromide levels in inland regions because of future climate
conditions. This finding was assumed to apply to the Miller WTP in 2050.
Similarly, there is no evidence found yet to quantify changes on the levels of water
quality parameters UVA, pH, turbidity, and calcium hardness under future climate conditions,
these parameters in 2050 were assumed the same as the 1998 baseline values.
¦ Temperature
Cromwell III et al. (2007) predicted increases in surface water temperatures ranging from
1.1 to 6.6°C from 1990 to 2100. The average water temperature in 2050 is estimated to be 2°C
higher than the baseline values for all seasons during the 52-year period.
The analysis above generates an estimate of the key parameters for the 2050 raw water
quality. Furthermore, the changes are translated to the other water quality parameters using the
correlation matrix in Table 2-32. This empirical statistical analysis leads to a proposed source
water quality in the Ohio River intake in 2050. The result is shown in Table 2-33.
Table 2-33 Projected raw water quality parameters for the Miller WTP in 2050
Unit
Spring
Summer
Autumn
Winter
Parameter
Mi
<7l
Mi
(Tl
Mi
(Tl
Mi
(Tl
PH
—
7.7
0.17
7.7
0.20
7.8
0.22
7.8
0.18
Alkalinity
mg/L
57.3
18.9
79.1
22.1
83.3
21.7
64.1
23.7
Turbidity
NTU
43.4
38.0
26.9
36.9
8.5
7.6
41.5
64.7
Ca
Hardness
mg/L
63.5
23.3
76.2
31.6
87.1
35.6
74.2
33.7
Total
Hardness
mg/L
98.8
16.8
128.9
24.5
149.9
28.5
121.5
32.8
TOC
mg/L
3.8
1.0
4.4
0.9
4.1
0.5
4.1
0.9
UVA
cnr1
0.12
0.06
0.11
0.06
0.08
0.02
0.09
0.05
Bromide
mg/L
0.03
0.01
0.05
0.02
0.10
0.04
0.07
0.04
NH3 n
mg/L
0.36
0.50
0.25
0.13
0.23
0.13
0.23
0.13
Temperature
°C
14.4
27.7
22.8
11.8
Flow
m3/s
4.75
5.01
5.75
5.30
Note: is average and a1 is standard deviation in 2050; - Not applicable.
6.3.2.4 WTP-cam model calibration and validation
Before applications to treatment process modeling were performed, the WTP-cam for the
Miller WTP was calibrated and validated using input data extracted from the ICR database.
141
-------
Model calibration and validation were based on sample period 10 (April 1998), sample period 13
(July 1998), and sample period 16 (October 1998). Plant operation data for the three periods
were described in detail in preceding Section 6.3.1.
Results from the WTP-cam simulation for the field data of validation periods are shown
in Table 2-34. Reasonable agreements for most water quality parameters are achieved including
pH, alkalinity, total hardness, TOC, free chlorine residual, and TTHMs. The measured and
simulated TOC concentrations were in good agreement in the finished water:
¦ The relative projection error is <10% for bulk water parameters pH, alkalinity, and total
hardness. The error was 8.4±8.3% in the coagulation basin and filters, 7.5±7.4% for the
GAC units and finished water, and 1.5±5.0% in the distribution system. These
uncertainty assessments include all the data analyzed without excluding the period-13
data that are statistically different from the others of the calibration periods. Excluding
period 13 data, the projection errors are only a half.
¦ For the sampling periods 10 and 16, model-simulated TOC concentrations were projected
higher than measured concentrations by 26.2±2.7% in the coagulation basin and filters,
but very close at the GAC units and finished water. The model estimates are >50% than
the measured for the sampling period 13 when TOC concentrations were low (Table 2-
34).
¦ The ICR data showed that UVA was removed by coagulation and GAC in the Miller
WTP, while the WTP-cam predicted most UVA removal by GAC. The simulated UVA
agreed well in the finished water for sampling period 10.
¦ Excellent agreement was achieved between the simulated and the sampled chlorine
residuals in the finished water.
Table 2-34 Comparison of sampled and modeled water quality results
Water quality
Sampling
Data
Influent
Coag.
Filtration
GAC
Finished
AVG1*
AVG3**
parameter
period
type
Basin
water
10
Sampled
7.7
8.7
8.6
8.0
8.5
8.6
8.6
Modeled
7.7
9.4
9.4
9.4
9.1
9.1
9.1
pH
13
Sampled
7.6
7.9
7.8
7.8
8.2
8.4
8.5
[-]
Modeled
7.6
9.2
9.2
9.2
8.9
9.0
9.0
16
Sampled
7.7
8.3
8.1
8.0
8.4
8.3
8.6
Modeled
7.7
8.8
8.8
8.8
8.2
8.2
8.2
10
Sampled
56
59
58
58
58
60
60
Modeled
56
64
64
64
62
62
62
Alkalinity
13
Sampled
63
56
59
58
57
64
68
[mg/L]
Modeled
63
72
72
72
69
69
70
16
Sampled
75
77
80
77
81
81
82
Modeled
75
81
81
81
78
78
78
10
Sampled
113
128
120
119
120
121
115
Modeled
113
122
122
122
122
122
122
Total
13
Sampled
98
106
107
108
106
115
120
Hardness
Modeled
98
109
109
109
109
109
109
[mg/L]
16
Sampled
164
162
166
164
169
164
165
Modeled
164
170
170
170
170
170
170
142
-------
Table 2-34 cont'd
Water quality
parameter
Sampling
period
Data
type
Influent
Coag.
Basin
Filtration
GAC
Finished
water
AVG1*
AVG3**
10
Sampled
Modeled
1.8
1.8
1.5
1.8
1.4
1.8
1.0
0.8
1.0
0.8
0.8
0.8
TOC
[mg/L]
13
Sampled
Modeled
3.6
3.6
2.55
3.6
2.2
3.6
1.3
0.51
1.3
1.3
1.3
16
Sampled
Modeled
2.3
2.3
1.9
2.3
1.7
2.3
0.6
0.54
0.6
0.6
0.6
10
Sampled
Modeled
0.069
0.069
0.028
0.061
0.024
0.061
0.012
0.009
0.010
0.006
0.006
0.006
UVA
[cm-1]
13
Sampled
Modeled
0.178
0.178
0.068
0.151
0.054
0.151
0.018
0.013
0.013
0.013
16
Sampled
Modeled
0.075
0.075
0.067
0.067
0.004
0.003
0.003
0.003
10
Sampled
Modeled
0.0
0.0
0.0
0.0
1.0
1.0
0.9
0.7
0.7
0.5
Free chlorine
residual
13
Sampled
Modeled
0.0
0.0
0.0
0.0
1.2
1.2
0.8
0.7
0.7
0.5
[mg/L]
16
Sampled
Modeled
0.0
0.0
0.0
0.0
1.3
1.2
1.0
1.0
0.7
0.9
10
Sampled
Modeled
0.0
0.0
0.0
0.0
11.9
9
23.7
16
28.8
22
TTHMs
[M9/L]
13
Sampled
Modeled
0.0
0.0
0.0
0.0
8.1
20
30.9
37
47.8
53
16
Sampled
Modeled
0.0
0.0
0.0
0.0
8.5
8
29.6
16
46.5
23
Note: *AVG1 refers to average retention time 1 day. **AVG3 refers to the maximum retention time, 3 days. —:
data not available.
6.3.3. Engineering analysis for water treatment adaptation
6.3.3.1. Adaptation feasibility evaluation
Tables 2-31 and 2-33 list the source water quality at the plant water intake in the 1998
base year and the 2050 target year, respectively. For these projected source water changes, the
adaptation feasibility of the treatment plant was evaluated using the calibrated and validated
WTP-cam model. In this evaluation, the plant treatment processes were assumed to remain
unchanged. However, adaptation took place by modifying the GAC treatment operations because
GAC was projected to be the most effective process for TOC removal (see Section 6.3.1.2).
Figure 2-67 compares the TOC and TTHM results between the baseline and the future
scenarios. The cumulative density function (CDF) was defined by adding the probability of
simulated TOC or TTHM concentrations in the finished water. The CDF curves displayed the
vulnerability of potential exceedance of the drinking water standards when the treatment
processes and GAC operation remain unchanged. Under the baseline conditions, the Miller WTP
meets the TOC compliance criteria of 2.0 mg/L (see Figure 2-67a). Under future climate
conditions, however, the source water would likely have higher TOC concentrations and
different water chemistry. 2765 of the 5000 Monte Carlo runs in the WTP-cam simulations show
TOC concentration >2 mg/L in finished water. This result indicates a 55% probability of
violating the TOC compliance criterion under the same TOC and TTHM regulation limits. If the
143
-------
TTHM maximum contaminant level (MCL) became more stringent in future, greater risk of a
violation would be higher: if the TTHM MCL decreased to 60 [j,g/L or 40 (J,g/L, the risk of an
MCL violation would increase to 4% or 36% under the future scenario, respectively.
Figure 2-67 Modeled treatment performance of the Miller WTP in baseline (1998) and future (2050)
scenarios. (A) TOC running average at finished water; (b) TTHM after 3 days residence
in distribution system. Results based on 5000 Monte Carlo simulations.
Li et al. (2014) analyzed these model projections and discussed potential engineering
options in the system-scale adaptation. They noted two potential engineering options for current
plant configuration and operation:
¦ One option is to replace the chlorination system with the newly deployed UV disinfection
treatment unit in the plant. After UV disinfection, chlorination takes place in the clear
well before the product water is distributed at the Miller WTP. Compared to sorely
relying on chlorination, the modified process using UV disinfection can significantly
reduce DBP generation in the treatment and the subsequent distribution. When re-
chlorination is required to satisfy the contact time (CT) rule for biological control, the
DBP formation may still become a technical challenge when TOC is not adequately
removed during water treatment. Boccelli et al. (2003) proposed a mathematical model to
analyze the re-chlorination effects.
¦ The engineering feasibility also depends on the water age in the distribution system. The
GCWW water distribution system is a single network that supplies water to the
population in a monocentric urban form. As described in Section 4.2, the Cincinnati
metropolitan area started to evolve toward a multi-center configuration with increased
expansion in the northern portion of the city. How to manage the water age and system
efficiency is a central subject in a detailed adaptation feasibility analysis involving both
the GCWW treatment and the distribution system. Currently, for assurance of contact
time rule compliance in the distribution system, the post-UV product water is disinfected
using chlorination at clear well at a reduced level.
¦ Another option is to optimize the Miller WTP operation by the configuration of the GAC
absorption process for the current and future climate conditions. This option requires no
significant capital investment, representing a practical and attractive adaptation solution.
Two variables are important to the adaptation feasibility analysis. One is the adaptation
threshold of the current system beyond which new GAC contactors or other treatment
144
-------
units are required. The second is energy consumption in the GAC regeneration and its
CO2 emission in the life-cycle analysis.
6.3.3.2. Adaptation economics in TOC treatment
Currently, the Miller
WTP has a total of 12
downflow gravity contactors
and two multi-hearth furnaces
for onsite reactivation. Each
of the contactors has a
volume of 595 m3 and a
surface area of 181 m2 The
overall GAC loss rate through
the system is 7-8%.
According to the past
operational data, the carbon
loading rate is 482 kg/day of
GAC per square meter of the
hearth area in GAC
reactivation.
— Safety margin: 0.02mg/L
— Safety margin: 0.20mg'L
4 6
Net Annual Cost, $miUion
Figure 2-68 Accumulative probability of net annual adaptation cost
for the source water change scenario in year 2050.
The energy cost can be readily converted into CO2
emissions.
A capital recovery
analysis assumes a return period of 20 years with an interest rate of 5%. The resulting cost curve
between the reactivation period and the annualized cost is shown previously in Figure 2-56. The
annual cost of the GAC system is expected to decrease with increasing reactivation period. For a
reactivation period shorter than 90 days, the annual cost increases rapidly at a shorter
reactivation period. The implication of this reactivation period threshold is obvious in adaptation
economics.
The net annual adaptation cost is defined as the difference between annual cost calculated
using the cost curve in Figure 2-56 and the base annual cost at the current operation. The annual
cost for the current operation was $13.6 million for an average GAC reactivation period of 180
days at the plant. Based on the cost calculation outlined in Section 6.1.3, the cost at a given level
of TOC removal - namely, a probability of meeting the compliance level of 2 mg/L can be
calculated. The results are shown in Figure 2-68. The net annual cost to control TOC <2.0 mg/L
in the 2050 climate scenario would decrease to $7.0 million for a 0.02 mg/L safety margin or
$7.8 million for a 0.20 mg/L safety margin. If the plant performance criterion allows a 10% risk
for TOC above the 2.0 mg/L limit or a 0.9 compliance probability, the net annual cost would
further reduce to $3.4 million for a 0.02 mg/L safety margin and $4.4 million for 0.20mg/L
safety margin (Figure 2-68).
6.3.3.3. Implication for engineering practice
The adaptation case study at the Miller WTP shows an example of the quantitative
analysis that examines engineering approaches in system-scale adaptation. The results show the
system's adaptability to offset source water changes projected for the year 2050. The adaptation
reliability is quantified by evaluating and comparing the ability to achieve regulatory
compliance, adaptation economics, and climate co-benefits in energy reductions.
145
-------
Quantitative engineering analysis using WTP-cam allows one to project the likely future
changes in source water quality and engineering options in a successful adaptation. It is
important to note, however, that several assumptions were built into the analysis. These
assumptions include: (1) the correlation matrix among water quality parameters used in WTP-
cam modeling is assumed to remain unchanged through time; (2) compliance criteria in SDWA
regulations such as DBP standards in drinking water remain unchanged in the future; and (3) the
current treatment technologies in the removal of TOC are effective and remain deployed. One
can observe from these assumptions that such adaptation analysis depends on location-specific
conditions, anticipated future technological and regulatory environments.
7. Adaptation Engineering for Drinking Water Distribution
Water quality management in drinking water distribution networks is another important
area in system-scale infrastructure adaptation to address the changes with climate and land use
conditions. Efficient and adaptive water distribution (this Section 7.0) and treatment (Section
6.0) are the two essential components of the SmartWater system (Figure 2-7).
Drinking water quality at consumers' tap depends on both the water demand management
over a service area and the control of water quality variations in the treatment and the
distribution itself. This relationship is schematically shown in Figure 2-11. The changes in
climate and land use have produced and likely will continue to produce impacts to surface water
quality. Some examples include long-term hydrological changes, short-term disruption of
meteorological extremes, seasonal variations of water quality parameters (e.g., TOC, turbidity),
as well as the occurrence of eutrophication conditions (e.g., high temperature and nutrients) with
the prevalent presence of chlorophyll-a and cyanobacteria. These types of source water changes
are pertinent challenges to the planning and operation of a water treatment plant and distribution
system. Despite the multi-barrier approach utilized to protect public health in the U.S.,
perturbations or changes in source water may affect the performance of the conventional
drinking water treatment plant.
Water demand variation, both in space and time, is the primary factor affecting drinking
water quality in the municipal water supply (Figure 2-11). Water demand is a function of urban
development, urban adaptation, asset management, and socioeconomic factors. These factors are
often dependent upon a collection of urban management and policies, rather than a simple
technical issue on the distribution network itself. For example, the urban adaptation facilitated by
land use planning and transportation infrastructure can significantly affect the spatial distribution
of population and business activities, and thus result in substantial changes in the location of
water demands.
Municipal development goals play an important role. Many water utilities are now
focusing on two water conservation measures (see U.S. EPA, 2015a). One is water conservation
through the reduction of water loss or non-revenue water in the water distribution pipelines to
customers' taps. This measure is a part of the water utility asset management described
extensively in the literature and U.S. EPA reports (e.g., Barlett et al., 2017; EPA, 2007c).
Because of the aging water infrastructure, a nation-wide average rate of water loss is around 18-
20%. Some utilities with old water pipes and complex pressure zones to manage may experience
water loss as high as 50% in some distribution network segments. The other measure relies on
water use reduction on the per capita basis through market and management actions. Some U.S.
146
-------
water utilities in water-poor regions are actively developing management and economic
incentives to encourage water-saving practices, such as artificial lawns. It is noted that this
option may bring other implications. For example, strict per capita goals may impact water utility
revenue and thus have a negative response in limiting the ability of commercial development and
job growth.
For water distribution, the SmartWater system for adaptation leverages technological
advances in sensor-based model-driven system controls. Drinking water distribution modeling
and system control based on the EPANET hydraulic model of Rossman (2002) and extensions
(Uber et al., 2004; Shang et al., 2008) are widely used. Water quality control in distribution
systems has been also investigated extensively since the 1990s. A wide range of technical data,
models, and management methodologies are now available (e.g., Rossman, 2002; Uber et al.,
2004; Shang et al., 2008; Boulos et al., 2006; Mays, 1999; and references therein). However,
these existing advances are mostly without consideration of changes in future climate-dependent
demand and urban conditions.
When considering the impacts of climate and land use changes, a confluence of factors
can all affect the water quality management in a distribution system. To reduce the water quality
deterioration in distribution is the focus of adaptation. Important factors include high TOC in
source water and potentially in the finished water of a treatment plant, rising water temperatures,
higher ground temperatures surrounding buried water pipes, as well as changes in the reactivity
of organic matter under future climate. There are several U.S. EPA guidelines on water
distribution systems; those practices will be not repeated in this report. Instead, the report here is
focused on adaptation methods and tools in three areas below:
¦ modeling DBP concentrations in water supply for vulnerability assessment;
¦ in-network water treatment to manage DBPs formed in the distribution system; and
¦ modeling of the water demand changes for developmental scenario analysis.
7.1. Water age and water quality changes in distribution: The need for adaptation
Disinfection by-product (DBP) formation during drinking water distribution is an
inevitable outcome due to reactions between residual disinfectants and DBP-formation
precursors in the organic matter. The well-known water quality impacts are regulated under the
SDWA contact time rule/DBP Stage-II rules (see U.S. EPA, 2015a). The need for adaptation in
water distribution can be assessed through compliance monitoring that is often guided by
EPANET-based simulation of residual chlorine and DBP concentrations. This section is based on
recent publications by Zhao et al. (2018a, 2018b, 2017) and others cited therein.
7.1.1. EPANET-based risk assessment on DBPformation
Model simulation has been widely practiced for the water distribution system since its
inception in the late 1980s. Chlorine as an oxidant in drinking water reacts with TOC to form
DBPs, including the regulated trihalomethanes (THMs) and haloacetic acids (HAAs). At the
same time, chlorine is also simultaneously transferred to pipe walls and consumed in reactions
with pipe wall materials and biofilms. In the simplest terms, total free chlorine [CI] - the sum of
hypochlorous acid [HCIO] and hypochlorite ion [ClO~] in a flowing pipe, reacts with natural
organic matters [NOMDBP](e.g., humic and fulvic acids) in the bulk water demand [BCD] to
147
-------
form DBPs. A fraction of [CI] also reacts with other bulk demand [BCD'] without DBP
formation. Bulk water demand consists of organic materials and other chlorine reactants and
those forming from a detachment of biofilms and pipe scales in water distribution. Chlorine also
reacts with pipe wall materials and attached biofilm, with both terms lumped together as the wall
demand. Generally, the multi-component chlorine reactions can be written as:
Cl2 + H20 -> HCIO + H+ + Cl~
pKa ,
HCIO -^H+ + ocr
k-DBP
HCIO + C10~ + BCDdbp > DBP + cr
k!
HCIO +C10~ + BCDother -» others
kw
HCIO+C10 + [wall] —> scale + others
DBPs, represented by trihalomethane (THM) compounds - trichloromethane,
bromodichloromethane, dibromochloromethane, tribromomethane, collectively as total THMs
(TTHMs), form during chlorination through stepwise NOM-oxidation and hydrolysis as
illustrated below. Using the model compound propanone CH3COCH3 and m as the Br molar
fraction:
CH3COCH3 + (3 — m)HOCl + mBr~ -> CH3COCBrmCl^3_m) + (3 — m)H20 + mH+
CH3COCBrmCl^3_m^ + H20 CH3COOH + CH BvmCl3_m
The simultaneously occurring processes in a flowing water pipe are schematically shown
in Figure 2-69. Analytical solutions for the chlorine transport and DBP formation have been
published (e.g., Biswas et al., 1993; Rossman, 2002; and Clark, 1998). Clark and Haught (2005)
developed a mass transfer limited chlorine model and compared it with others in Rossman et al.
(1994) and Biswas et al. (1993). Clark et al. (2010) further analyzed the competing chemical
reactions in the modeling of chlorine decay and DBP formation. In the center of discussion are
the water quality models of Clark (1998) and other subsequent publications (e.g., Clark and
Sivaganesan, 2002; Boccelli et al., 2003). These models stipulate that in second-order kinetics,
the DBP concentration increase from an initial condition (CB — CB 0) is proportional to the decay
of chlorine concentration (CA 0 — CA) by the proportion constant T' in the distribution pipe:
Cb = Cb,o + T'(Cao — CA
rjl f kD
k.w+k'i7
The proportion constant T' is simply a ratio of DBP formation kinetic rate (kD ) over the
total chlorine decay (kE = kw + k'b); kwand k'b are the reaction rates for wall demand and bulk
demand, respectively. In this research, the analytic solution of Clark (1998) was further refined
based on kinetic theory and comparative experiments conducted at the U.S. EPA Test and
Evaluation Facility. The study led to the proposal of new analytical models in Eqs.2.43-2.44,
respectively, for chlorine decay and DBP formation (Zhao et al., 2018a; Yang et al., 2008). The
(2.41)
(2.42)
148
-------
Figure 2-69 Schematic diagram showing the simultaneously occurring chlorine reactions in
bulk and wall demands, and mass exchange between the bulk water and pipe
wall. Implications on reaction kinetics for bulk decay (k'b), DBP formation (kD), and
wall decay (kw) are shown on the right. From Zhao et al., (2018a).
DBP analytical equation can be simplified to Eq.2.45 under common conditions in water
distribution when DBP-forming fractions in the bulk demand are very small or 0—0:
"TIRP 1
ACg _ DBP[^ / Ca,o J „ kDBP9 ¦ CAfi ( Cj/ p /|3 \
7- (kb+K) ^ 1 ( }
^,0
^ <244)
AC's7'C""Iife]'Cj <245)
The DBP formation in distribution systems depends on not only initial chlorine
concentration but more importantly, the reactivity of DBP-forming precursors and their
reactivity. This is expressed as the kinetic ratio k° in Eq.2.45, and 6 in Eq.2.43. The models
will be further reviewed and incorporated into the EPANET for the SUD's SmartWater module.
First, the simultaneous occurrence of chlorine decay and THM formation depends on the kinetic
ratio k° that, in turn, is a function of pipe flow hydrodynamics. Several common water
kw+k-'fo
quality parameters, including total organic carbon (TOC), residual chlorine, UV256, water pH,
and temperature, have been used to estimate the THM formation potential (Clark and
Sivaganesan, 2002). As shown in Eqs. 2.43-2.45, the non-DBP forming bulk demand and the
149
-------
wall demand compete for the finite chlorine residual in water and thus affect the THM formation
potential in competitive reaction.
Second, NOM properties and specifically the chlorine-reaction fraction are the
controlling variables in competitive reactions. Removal of reactive NOM fractions in treatment
is the effective approach to decrease 9, T and Y in order to simultaneously maintain the required
chlorine residual level and reduce the THM formation potential. For aged NOM with small 6,
kD/
such as GAC-treated tap water, the hydrodynamic effects on kE and + k ) cannot
neglected. The reaction competition from flow-dependent wall demand becomes comparable
with THM-forming and other bulk demand.
Finally, the kinetic constants and the time to reach a pseudo-steady state are all related to
pipe flow hydrodynamics. Many parts of a distribution network may have water ages exceeding
24 hours, with re-chlorination possibly necessary to compensate for the excessive loss of
chlorine residuals for biological control. Re-chlorination, however, will further increase the DBP
levels (Boccelli et al., 2003) even to the extent of violating drinking water standards. Therefore,
it is of fundamental importance to reduce water age through adaptive engineering measures, such
as through tank operations and better monitoring water demand in real-time throughout a
distribution network, and perhaps structurally by changing the routing of water, pipe sizes, and
other network configurations. The need for adaptation to water demand and water age
management is described next.
7.1.2. Water age variations, modeling and adaptive control
Water age varies significantly in a distribution network. An extended water age (t) in
Eqs.2.44-2.45 can result in low residual disinfectant levels and elevated DBP concentrations, a
phenomenon that has been widely documented. One central adaptation objective is to assess the
vulnerability and to plan and design corresponding adaptation measures.
For vulnerability analysis, this research recently completed real-time water demand
measurements over 25% of the network nodes for 2 months for an independent distribution
segment in Cincinnati, Ohio. Subsequently, a hydraulic and water age simulation was conducted
using the EPANET model (Zhao et al., 2017). The ~ 38.6 km2 network serves 8,485 buildings,
consisting of 4,843 pipes, two elevated water tanks, four booster pumps, three control valves,
and one water reservoir for water supply. In the network, the north and south supply areas
supplied by the two elevated tanks contain numerous local pipe loops many in "Ft"
configurations and dead-end branches of < 8-in diameter. The study results clearly showed large
water age variations that can be monitored and analyzed using all-pipe and all demand (APAD)
techniques (Zhao et al., 2018b). In comparison, the hourly demand variation curve (HDVC)
modeling widely used currently is incapable of assessing the water age variability. Two
conclusions are particularly noteworthy with implications to the water age assessment and
management:
¦ The pulse nature of water demand is prevalent among individual water users throughout
the network. In the one-week 68-hr period, measured pulse demand in most network
nodes is zero for approximately 70% of the time (Figure 2-70). In the analysis, the time-
discontinuity in water demand starts to disappear at the level of 31-home demand
150
-------
Single home
One simulate demand in APAD
5/21/15 5/22/15 5/23/15
Time
5/25/15
2.5
2.0
3 15
c 1.0
CO
E
® 5
Q 5
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170
Time (hours)
Time (hours)
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170
Time (hours) Time (hours)
Figure 2-70 Water demand and computed Re variations in a two-week period for a single home, 31 and 114 homes of a pipe dead-end
section, showing significant differences between the APAD model and the generalized water demand pattern. From Zhao et al.
(2018b).
31 homes
—<— APAD model
—¦— Generalized model
151
-------
aggregation. It is replaced by time-continuous variation patterns for a block of 114 homes
(Figure 2-70).
¦ There is a large range of water ages among all network nodes (Figure 2-71). Simulated
water ages during the two-week validation period average at 35.8 and 34.6 hours given
by the APAD and HDVC models, respectively. Both demand models yield a large spread
of simulated water ages from <15 hours near the pump stations to over 180 hours in dead-
end branches.
¦ In all cases, the large spreads in water age and their spatial association with the network
configurations (Zhao et al., 2018b) point to the need for network optimization and
adaptations. Some adaptation measures that were discussed in the literature include
reconfiguration of local pipe loops, synchronized tank operations, re-chlorination, and in-
network water treatment. The latter is one major technical approach as described in the
next Section 7.2.
Figure 2-71 Probability distribution and corresponding CDF of simulated waterage for the
network. From Zhao et al. (2018b, 2017).
7.2. In-network water treatment as adaptation measure
Except for TOC removal in water treatment plants, other adaptation approaches for
effective THM management rely on water quality management in finished water distribution as
the last barrier to protect human health. Water quality management in distribution is not new.
Decades of research and practical engineering have produced a suite of distribution system
models (e.g., EPANET, EPANET-MSX, etc.11) and technological innovations in the in-network
water treatment. For the latter, examples can be found in re-chlorination, in-network GAC
absorption, and aeration.
11 http://www.epa.gov/nrnirl/wswrd/dw/epanet.html
152
-------
Unique for source water changes is the focus of adaptation measures on extreme conditions
giving arise from climate and land use changes. These changes are not considered in traditional
water supply engineering. For example, some water supply systems have experienced a rapid
decrease in water demand due to socioeconomic changes or the loss of major employment
centers, resulting in an oversized distribution system. The hydroclimatic changes also can
generate conditions resulting in high TOC concentrations in source water and finished water, as
well as leading to warm water temperatures in pipelines. Higher water temperature, high-
concentration of reactive NOMs in water will likely make DBP control a necessary but difficult
task. It will concurrently increase biofilm formation. For these impacts, the in-network aeration
and GAC treatment during distribution have been investigated to remove THM and other volatile
contaminants.
In-network aeration
The use of aeration to remove volatile organic compounds, including THMs, relies on the
principles of air-liquid two-film mass transport. In water distribution networks, the air stripping
process is commonly used by retrofitting the existing water storage tanks and in-ground
reservoirs. Such an application was being tested and investigated at the Las Vegas Valley Water
District (LVVWD). Figure 2-72 shows a schematic illustration of the aeration system
constructed at the Alphas twin tank storage basins.
In the EPA-LVWWD joint research, one retrofitted water aeration system in the
LVVWD alpha tank was investigated. The system consisted of low-profile fine bubblers, air
manifold, and tank mixer (Figure 2-72). The mixer and other media (e.g., plastic cubicle) were
used to improve the stripping efficiency of fine bubblers. However, because of the limited water
depth above the bubbler, these types of aeration systems tend to have low stripping efficiencies
or use high air-to-water ratio for greater removal rates; for the latter, the improvement is at the
expense of energy consumption, a major consideration in adaptation design.
McDonnell (2012) investigated the mechanisms and modeling of in-network aeration for
THM removal. The investigation included experimental testing of the THM stripping in an
Air manifold
—
j^vxyy
Concrete Cover
¦vvyyv
Aeration tank
0
0
IW 0
0
0 0 Q
0
0 0 0
QlP 0 0
S
Concrete column
3 ° 0 (
0
0 0 0 0
0 0
0 0 Fine bubbler
Profile View
Figure 2-72 Schematic views of in-network aeration in the LVVWD water distribution network to
remove volatile THM from drinking water in the alpha tank reservoir. Illustration after
actual tank-retrofitted aeration system: left - plain view of air sparging pipes lines and
tank mixers (M1-M6, and N1-N6); right - profile view showing air bubble plume
geometry and the two-film transport mechanisms, h - effective water depth; r - diameter
of air bubble plume.
153
-------
experimental water column, and mass transport modeling of field testing in the LVVWD's tank
Alpha. The results led to the development of a THM stripping model as a program extension for
the EPANET Multi Species Extension (MSX) v. 1.2.0 (Shang et al., 2008). The effect of in-
network aeration on water quality was further modeled using EPANET and the extension.
The aeration in the Alpha tank has significant effects on THM concentrations in the
network. As shown in Figure 2-73a, pressure zone 2 receives 75% of its water from the Alpha
Tank during an average day. In response, pressure zone two received the greatest reduction in
average total THM concentration. The TTHM removed in nearly half of the nodes is 90 mg/L
(Figure 2-73b). This is expected since pressure zone two is heavily influenced by the Alpha tank.
Water quality network modeling using EPANET 2.0 for the study is provided in McDonnell
(2012).
Figure 2-73 EPANET simulation of flow and THM distribution in the Western Hill portion of the
LVVWD water distribution system. (A), fractions of water from Alpha Tank; (B).
distribution of THM reduction after aeration at normal airflow rate of 2 standard cubic
feet per second in the Alpha Tank. Adopted from McDonnell (2012).
154
-------
GAC treatment and other post-formation treatment processes
GAC and other absorbents (e.g., zeolite) have been used in the removal of DBP
compounds from drinking water; this technology has also been applied at the point-of-use and
point-of-entry (Stubbart, 2004) as a part of the small systems12. The adsorption logistic models
for TOC simulation were presented in Sections 6.2-6.3. Similarly, GAC has been used to remove
THMs, whereas it is less effective for mono and dihaloacetic acids (Tung et al., 2006). In
addition, membrane filtration such as reverse osmosis and nanofiltration (Kimura et al., 2003;
Uyak et al., 2008) have been tested and studied for removal of THM species from drinking
water.
7.3.Water conservation, storage and reuse through adaptive planning
Water availability and water shortage are other characteristic impacts of the
hydroclimatic and land use change. Water conservation, storage and reuse of reclaimed water are
valuable practices in adaptation for many water-stressed regions. Examples can be found in Las
Vegas and other cities in the U.S. southwest and southern California. Even in water-rich regions,
water conservation is often a technique for the reduction in water and energy usage. These
adaptation techniques to relieve hydroclimatic impacts are described in Ranatunga et al. (2014),
Wang et al. (2013), Neil et al. (2012, 2014), Yang and Restivo (2010). Details of these
adaptation techniques are contained in U.S. EPA (2017a, 2015b).
8. SUD Applications in Coastal Regions: Water Infrastructure and Emergency Planning
Existing data and research results show that changes in precipitation patterns and
overland runoff hydrographs will almost certainly impact the drainage capacity, stormwater
control measures, LID, and green infrastructure, as well as stormwater discharges, including the
long-standing CSO challenges facing many U.S. cities (U.S. EPA, 2013b; Johnson et al., 2015).
Coupled with land use changes, the vulnerability of these infrastructure assets cannot be
underestimated. Some specific analysis is shown in Tables 2-23 and 2-25. Impacts on unit
processes are illustrated in Figure 2-49 of Section 5.2.2.2. Specific adaptive engineering
solutions are location-specific, mostly related to changes in precipitation, runoff, disruptive
storms such as hurricanes, storm surge, and sea level rise.
In this section, vulnerability and adaptation analysis for the stormwater and wastewater
infrastructure in coastal regions are briefly discussed. A full analysis of the coastal water
resources and water infrastructure will be published separately, where case studies along the
Atlantic coast and the Gulf of Mexico will be examined.
8.1. Water infrastructure vulnerability in coastal regions
The U.S. coastal zone hosts over 80% of the population, vast built infrastructure, over
90% economic outputs, and invaluable ecological resources. Nearly 39% U.S. population in
2010 lives within 50 miles (-90 km) of coastal lines13. In the low-lying Atlantic coast and the
Gulf coast, the built and future infrastructure and sensitive environmental assets are vulnerable to
extreme meteorological events such as Hurricane Sandy. Tropical cyclones, hurricanes, and
12 http://www.mae.gov.nl.ca/waterres/reports/cwws/BMPs_for_Control_of_DBPs_Apr_13_2009.pdf
13 http://oceanservice.noaa.gov/facts/population.html
155
-------
storm surges have demonstrated the potential to compromise structural integrity and service
functions of critical aboveground assets by wind damage, flooding, and the change of surface
water and groundwater hydrology. The impacts and destruction of "soft" environmental assets
such as coastal marshes and wetlands cannot be neglected either.
Figure 2-74 schematically shows three principal types of short-term disruptive and long-
term hydroclimatic threats to the coastal infrastructure. Near the coast, sea level rise changes the
hydraulic gradient for communities in low lying areas. For example, according to the City's
Department of Environmental Protection, the City of New York has experienced the sea level
rise effects on drainage systems and wastewater pump stations in the low-lying Queens district.
Storm surge, particularly those associated with hurricanes, are shown to repeatedly result in
severe inundation of the coastal areas. The combined effect of sea level rise and storm surge is
even more disruptive. To above-ground infrastructure, wind damage associated with hurricanes
Global/regional climate systems (AMO, etc.)
Orographic and local
precipitations
Winds
Storm surge
Mountains
Infrastructure
Estuary
wetlands
Ocean
Figure 2-74 Schematic illustration of long-term climate and short-term meteorological and disruptive
storm surge events in a typical coastal zone.
can be disruptive. The electric grid damage and supply disruption are particularly important to
the water infrastructure services, during and after the events.
In the inland region away from the coastline, the coastal mountains and other geophysical
features induce strong atmospheric interactions with global/regional climate systems such as
Atlantic Meridian Oscillation. This atmospheric interaction can introduce localized moisture
circulations such as orographic precipitations and rain shadows (e.g., McKenny et al., 2006;
Konrad II, 1997; Wallis et al., 2007; Changnon, 2006). These localized rainfall anomalies and
changes in the future climate may not be captured in conventional hydrological design guidelines
such as NOAA's Atlas-14 precipitation design tables.
Table 2-35 lists major hydrological impacts and adaptation design variables for water
infrastructure and other environmental assets. These engineering considerations of intense
precipitation, wind speed, and storm surge include a revision to engineering parameters for wind
(average speed, gust speed, and direction), precipitation (duration, depth, and intensity), and
inundation level (depth and duration). Specifically:
¦ Design precipitation (duration, depth, and intensity). Urban infrastructure systems or
components are planned and designed to assure adequate hydraulic capacities providing
156
-------
Table 2-35 Selected hydrological impacts and adaptation variables in coastal area.
Design Precipitation
Design Wind*
Design Inundation*
Infrastructure assets
Roadways
Movement of vehicles
Runoff management
Pavement damage
Damage to light fixtures and signs
Pavement damage
Base damage
Damage to structures/bridges
Water supply
- Source water quality,
- Damage to power supply
- Flooding and inundation
Stormwater and
wastewater facilities
- Equipment flooding
- Runoff management;
- Hydraulic capacity of structures
(e.g., pipe, culvert, sluice gate)
- Stormwater quality and discharge
- Flooding
- Physical damage to structures
- Damage to power supply
- Physical damage to structures
- Salt water intrusion
- Flooding and inundation
Solid and hazardous
- Flooding
- Cover design
- Flooding and inundation
waste facilities
- Cover design
- Groundwater level and control
- Dust dispersion and control
- Disruptive wind damage
Environmental assets
Estuary wetlands
- Nutrient flux
- Changing hydraulics of flows
- Disruptive wind damage
- Flooding and inundation
Riverine
- Base flow and drought
- Nutrientflux and flora ecohealth
- Peak flow and erosion
- Disruptive wind damage
- Flooding and inundation
Note: * Refer to wind and inundation in coastal shores are related to cyclones and storm surge.
157
-------
desired water services. The design basis is often in the form of a design storm such as 10-
year 24-hour precipitation. Application examples include culvert sizing in road
construction, pipe sizing and grading for drainage of urban runoff, and retention pond
design for stormwater management (Table 2-35). Note roadways are not exempt from
this type of damage to the pavement and electrical fixtures.
¦ Design wind (average speed, gust speed, and direction). Design values for average wind
and gust wind vary among engineering conventions and often are specific to the design
objective and its risk category. For example, the 25-year, 50-year, and 100-year wind
speeds are used in determining the minimum strength load requirements for Occupancy
Category I, II, and III infrastructure, respectively (ASCE, 2014; Simiu, 2011; Cook et al.
2011). The U.S. EPA specification requires a temporary landfill cover to be designed
against a gust wind speed of 2 m/s (Table 2-35).
¦ Design inundation (depth and duration). In coastal areas, the inundation level is the sum
of flooding and wave action or surge. The surge actions increase the flooding level and
spatial extent, and the surge-related inundation is temporary. Acute hydraulic impacts
recede after the disruptive storm surge event. The schematic in Figure 2-75 shows the
concept of water action and storm surge height during a hurricane event. Thus, both
inundation depth and duration are two primary design parameters (Table 2-35). Road
bases and pavement may be permanently damaged by inundation.
How to plan and manage these valuable assets under the current and future conditions are
essential to coastal risk assessment and management. For this purpose, the SUD methods and
tools are designed to analyze the hydrological and transportation impacts in coastal
hydroclimatic events. Detailed analysis and technical basis for these tools will be presented in
subsequent publications. Flere in this report, the case study at the town of Mattapoisett in the
Figure 2-75 Schematic diagram showing wave action and storm surge height as a function of storm
surge, tidal cycle, and sea level rise. Numbers are for illustrative purpose. From NOAA
website (https://www.nhc.noaa.gov/surge/).
158
-------
Massachusetts southeast coast is described to show how SUD is used to develop the technical
basis for wastewater adaptation and emergency evacuation planning
8.2. Wastewater vulnerability and adaptation in storm surge
Mattapoisett is a small fishing town at the shore of Mattapoisett Harbor. Wastewater from
the residence and commercial entities is collected by a network of gravity sewer pipes and then
pumped to a regional wastewater plant using a transfer pump station at the side of the harbor
(Figure 2-76). One objective for the risk assessment was to quantify inundation and its impacts
on wastewater system operations.
In this analysis, the NOAA's Sea, Lake, and Overland Surges from Hurricanes (SLOSH)
model was used to simulate the storm surge height under specific local conditions including
topography, Mattapoisett Harbor bathymetry, likely storm tracks in the Mattapoisett area, and
atmospheric profiles at the origin of storm surge in the open sea. Table 2-36 shows the ranges of
major variables, yielding a total of 432 model runs. The modeling results yield estimates of water
depth, wind speed, and direction at a point of 50x200 m spatial grids. For each geographic
location, the projected water depth for all model runs form an envelope of inundation depth
estimates. The Maximum Envelope of Water (MEOW) provides the worst-case basin snapshot
for a particular storm category, forward speed, trajectory, and initial tide level, incorporating
uncertainty in the forecasted landfall locations. The Maximum of MEOW (MOM), on the other
hand, provides the worst-case snapshot for a particular storm category under "perfect" storm
conditions described by a combination of forwarding speed, trajectory, and initial tide level. In
practice, MEOW can be used for planning while MOM would be for emergency planning and
evacuation.
Table 2-36 SLOSH modeling parameters for storm surge modeling at Mattapoisett, MA
Parameters
Values
No. of Variations
Landfall location
1 (Hurricane Bob)
1
Pressure (mb)
40, 60, 80
3
Radius of maximum wind (mile)
25, 40, 55
3
Forward wind speed (mph)
30, 45, 60
3
Track direction (degree)
NNW, N, NNE, NE
4
Sea level rise (ft)
0, 1, 2, 4
4
Total Number of Model Runs
432
The location-specific SLOSH modeling was calibrated against the inundation extent of
historical Hurricane Bob in August 1991. During Hurricane Bob, storm surge pushed salt water
over the salt-lock dam into the Mattapoisett River northwest of the town. The overtopping
resulted in salt water intrusion into the river, and consequently into the unconfined aquifer and
impacted groundwater at the Fairhaven Tubular well field immediately north of the dam. The
aquifer is source water for the regional drinking water treatment plant, approximately 3.5 miles
southwest of the town. Using the calibrated model, the inundation map based on MEOW results
for a Category-4 hurricane is constructed (Figure 2-76). Major findings are:
159
-------
Legend
Active Sewer M ains
Low: 4.76837e-007
Low: 4.76837e-007
JVIattapofsktt
I . _ !\A. >,,¦
station impacted
Logger
x-;sectioFt-6'
Portion of the drinking water and
Kilometers
0 0.3 0.6
x-section A'
jtermep, increment P Corp.. GEBCO, USGS, FAO, MPS.
WlL. Ordnance Survey. Esri Japan. METI. Esri China !Hong
OpenStreetMsp contributors. and the GIS User Community
Drinking water piant
operation impacted
SLRO Cat4
High 7.06136
Unaffected area with sLR2Cat4
groundwater wells »H,9h :7 54M4
Grou
nd water
oilfields
undated
The only one
wastewater pump
ocwastewater system inundated
x-section B
Mattapoisett Harbor
x-section A
Figure 2-76 Location of the water infrastructure at the Town of Mattapoisett aside of the
Mattapoisett Harbor.
¦ The vulnerable areas of Mattapoisett (the southeastern region) that are inundated by
storm surge remain approximately the same under the different hurricane and sea level
rise scenarios. This is due to the topographic slope towards the harbor. Storm surge and
rate of inundation from a Category 4 hurricane at the current sea level could result in
inundation depths over 13.4 feet in some locations in Town. The maximum inundation
depth can be reached within 5 hours of the time of landfall.
¦ The wastewater pump station at Eel Pond in the southwestern corner of the town is also at
significant risk. The pump station could be submerged under a Category 2 hurricane or
above. In a Categoiy 3 hurricane and at the current sea level, the SLOSH simulation
shows 5.8 feet water depth at the Eel Pond (Figure 2-77). The water depth would increase
to over 13 feet under a Category 4 hurricane.
¦ Such inundation and physical damage can make the critical wastewater transfer station at
Eel Pond inoperable. Currently, the wastewater transfer station has no backup. Loss of
service could affect the town's residents after the hurricane during the recovery phase.
8.3. Emergency evacuation and water supplies
The AIR-SUSTAIN module of the SUD system was further applied to assess the traffic
conditions during hurricane evacuations, likely evacuee migration paths and bottled water supply
at shelters. In this simulation, SLOSH-model generated inundation maps were used as inputs to
the AIR-SUSTAIN module to estimate the inundated area, affected population, and potential
evacuation routes under four categories of hurricanes (Table 2-37). The population and
160
-------
Figure 2-77 A cartoon illustration of SLOSH modeling results on likely inundation risk for the
wastewater transfer station at Mattapoisett, MA.
households are based on the 2010 population census data. Overall, there would be over 2200
people and over 1500 households likely in the estimated evacuation area when a Category 4
hurricane landed directly in Mattapoisett Harbor.
Traffic simulation using AIR-SUSTAIN considers evacuees and the traffic flow from the
two large cities to the west: New Bedford and Fairhaven. It is further assumed that 80% of the
evacuates elect to travel to family and friends outside of the inundation zone, while the other
Table 2-37 Population affected for evacuation under four categories of hurricane
Sea Level Rise
Hurricane Category 1
Hurricane Category 2
(Ft)
Traffic analysis
Affected
Affected
Traffic analysis
Affected
Affected
zones
Population
Household
zones
Population
Household
SLR 0
121
15,995
8,429
136
28,117
14,589
SLR 1
122
16,086
8,487
136
28,235
14,675
SLR 2
122
16,196
8,561
136
28,335
14,735
SLR 4
122
16,380
8,677
136
28,514
14,845
Sea Level Rise
(Ft)
Hurricane Category 3
Hurricane Category 4
Traffic analysis
Affected
Affected
Traffic analysis
Affected
Affected
zones
Population
Household
zones
Population
Household
SLR 0
144
39,325
20,150
183
78,030
39,323
SLR 1
144
39,488
20,269
183
78,159
38,395
SLR 2
144
39,563
20,313
183
78,276
38,461
SLR 4
144
39,749
20,422
184
78,754
38,731
161
-------
20% of the population travels to public shelters. In Mattapoisett, the interstate highway 1-195 and
Main Street were determined to be the main evacuation routes according to the Mattapoisett
transportation and police departments.
Under these assumptions, the total clearance time from the affected area is estimated for
three cases of the emergency evacuation activation: slow (8 hours), moderate (6 hours), and fast
(4 hours). Adding about approximately 1 hour and 20 minutes of transportation, the total time for
complete clearance of the inundation area (emergency activation plus transportation) ranges from
5 hr 14 min to 9 hr 18 min (Figure 2-78). Main traffic delays were projected to occur on 1-195
north of Mattapoisett due to emergency traffic from the west. Most of the traffic congestion
would occur on 1-195 as the main regional evacuation route. For the fast 4-hr evacuation
activation starting at noon, the hourly traffic maps are shown in Figure 2-78. Route traffic
management is necessary to ensure a fast and smooth evacuation ending at 8 pm (Figure 2-78).
Emergency water supply would be required for evacuees in public shelters or
friends/families. Because the Mattapoisett drinking water treatment plant is likely to be adversely
affected under a Category 4 hurricane, water sources for emergency supply need to be arranged
in emergency preparedness planning.
Legend
Links
Volume-TSys [veh] (C.Curlnt)
<= 1000
<= 1500
i <= 2000
>2500
Zones
~
0^^^^ mi
New
Becifor
Mattapoisett
Mattapoisett
airheaven
New
Bedford
New
Bedfon
Mattapoisett
Mattapoisett
ittapoisett
Fairheaven
71
5:00prrr6i0Qpm
Figure 2-78 Hourly traffic map in the Mattapoisett region after evacuation order activated at noon
time.
9. Summary and Recommendations
This Part II infrastructure adaptation report describes methods, techniques, and case
studies for adapting water infrastructure to and improving its resilience against the projected
162
-------
impacts of hydroclimatic and land use changes. The focus is to establish actionable science for
adaptation planning and engineering at local scales.
As a part of this investigation, the relationship between climate, land use, transportation,
energy, pollution, and water management were shown to intersect. Urban development can lead
to a substantial UHI formation, which increases energy use. Urban sprawl leads to less heat
island effects, but higher use for energy in transportation, both urban residents and material flows
including water, thus adding to pollution. Development can alter rainfall and runoff
characteristics, which have subsequent impacts on water supplies and water quality. The latter
changes can challenge water plant operations, requiring process engineering adaptation. This can
result in increasing energy needs. Urban development patterns also impact water distribution and
sewer collection. Ultimately to address the sustainability of our communities, this EPA Office of
Research and Development research has analyzed the inter-connectedness and developed
systematic adaptation strategy and tools to better inform decision-makers for informed decisions.
Such guided decision-making can help to find optimal investments to protect their economy,
property, social systems, and infrastructure in anticipation of future conditions.
Water infrastructure adaptation can be planned and evaluated in three different levels:
adaptation at the watershed scale, urban scale, and water system scale. This spatial boundary
helps define the adaptation objectives that often require inputs with stakeholder involvement and
identifies the adaptation parameters to investigate in planning and engineering design.
Adaptation at the three-scale levels share the same iterative adaptation process (see Figure 2-2).
Following defining the adaptation objective and adaptation physical boundary, the adaptation
process begins with analyzing the water infrastructure vulnerability to the concurrent and future
changes of hydroclimatic and land use conditions. This analysis is conducted in the context of
urban developments. This first step is followed by technical and engineering analysis to define
specific adaptation planning and engineering options. Upon design and implementation with
consideration of the future change uncertainties, the last phase is centered on adaptation
effectiveness monitoring and evaluation. For the iterative process, the monitoring-evaluation
results lend a basis to revise the adaptation planning and, if necessary, urban development
policies and management objectives. This proposed adaptation framework is readily adopted into
the current urban planning practice. Figures 2-4 and 2-6 show the current practice and adaptive
urban planning, respectively.
To support the water infrastructure adaptation in three spatial scales, the developed Smart
Urban Design (SUD) methods and tools are described in this report with case studies for
illustration. At the watershed scale, the adaptation aims to protect source water quality. The
developed integrated watershed modeling tools and methods inside of SUD can be used to
project water quality in response to future climate and land use management options. On the
urban scale, the SUD provides an integrated analysis of land use, transportation, and water
infrastructure in scenario-based simulations that quantify basic urban functions and efficiency in
transportation and water services. Specific evaluation metrics are defined to evaluate the urban
development options, including air quality, water resources and utilization, and transportation
access for a given urban development scenario. In the local system scale, SmartWater models
and simulation tools were developed to provide specific engineering analysis of water system
vulnerability and engineering options to adapt. Through case studies covering different climate
regimes in the U.S. Midwest, Southwest, and coastal areas, the following major findings were
made:
163
-------
¦ It is clear from the studies across the country that water infrastructure and, to a larger extent,
urban adaptation can be effective in improving infrastructure resilience, and importantly, can
offer potential climate-economic-compliance co-benefits. To achieve these outcomes, the
adaptation must be planned and designed in a systems approach considering interactions
among urban systems.
¦ Scenario-based adaptive monitoring and planning are essential to urban and water
infrastructure adaptation. As described earlier, the large physical footprints and inflexible
infrastructure assets have created a "locked-in" condition for which alteration and changes to
the infrastructure can often be cost-prohibitive and difficult to overcome social and political
barriers. Model-aided computer simulation, when conducted appropriately, can provide
managers the tool to examine urban-scale adaptive planning, water infrastructure master
planning, water treatment adaptation, as well as engineering options to improve water system
services. The results can provide an effective venue for water professionals to communicate
to stakeholders involved.
¦ Urban environments present one of the important potential areas for adaptation and
mitigation of hydroclimatic change impacts (IPCC, 2014; Yang and Goodrich, 2014). This
potential is evident in the case studies described in this report. For example, a multi-center
transformation of the Cincinnati metropolitan area calls for a mix of automobile and mass
transit framework that can reduce fuel consumption and air emission by 15.6% and average
traffic delay by 25% in 2030. A large degree of carbon/energy reduction can be also achieved
by selecting optimal water infrastructure expansion actions in Manatee County, Florida when
adaptation is incorporated in the master planning.
¦ Water supply adaptation is effective when water treatment and distribution are considered
together through systems investigation. Only by this approach, future changes can be
grouped into those affecting source water quality and water demand. Then, the systems
analysis using the SmartWater tools can identify the most effective and economic
engineering solutions to adapt the water system for better compliance at a reasonable cost. A
practical example in the GCWW's Richard Miller Treatment Plant shows the feasibility by
changing and optimizing GAC reactor operations under future source water conditions.
¦ Coastal areas host multiple dimensions of long-term hydroclimatic impacts and short-term
meteorological disruptions in the narrow coastal zone. As a result, integrated modeling and
qualitative analysis are often necessary to develop options for long-term adaptation options
and emergency preparation plans against disruptive events like storm surges.
The central question for many urban managers and decision-makers is as what
Timmerman and White (1997) described, namely, how the urban growth can be planned
adaptively to reduce the negative impacts of urban metabolism and ensure sustainable growth.
This question is especially important at this stage because of the nation's impending strategic
investment in infrastructure. For this purpose, SUD methods and tools need to be developed
beyond the initial stage with applications in different climate and socioeconomic settings.
164
-------
10. References
Adams, J.Q., and R.M. Clark (1988). Development of cost equations for GAC treatment systems.
J. Environ Engineering, 114(3): 672-688.
Adegoke, J.O., R. Pielke Sr., and A.M. Carleton (2007). Observational and modeling studies of
the impacts of agriculture-related land use change on planetary boundary layer processes
in the central U.S. Agricultural and Forest Meteorology, 142:203-215.
Alp, E., and C.S. Melching (2009). Evaluation of the duration of storm effects on in-stream
water quality. J. Water Resour. Ping. & Mgmt., 135(2): 107-116.
ASCE/AWWA (2004). Water Treatment Plant Design, 4th edition. E.E. Baruth (ed.), McGraw-
Hill, 896p.
ASCE (2014). Minimum design loads for buildings and other structures, Standard ASCE/SEI 7-
10. ASCE, Reston, VA.
Ashley, R., J. Blanksby, A. Cashman, L. Jack, G. Wright, J. Packman, L. Fewtrell, T. Poole, and
C. Maksimovic (2007). Adaptable urban drainage: Addressing change in intensity,
occurrence and uncertainty of stormwater (AUDACIOUS). Built Environment, 33(1):70-
84.
Asian Developmental Bank (ADB) (2012). Green Cities. M. Lindfield, F. Steinberg (eds.),
Manila, Philippines, adbpub@adb.org. 412p.
Auch, R., J. Taylor, and W. Acevedo (2004). Urban growth in American Cities. USGS Circular
1252, 52p.
Azadi P, O.R. Inderwildi, R. Farnood, and D.A. King (2013). Liquid fuels, hydrogen and
chemicals from lignin: A critical review. Renewable and Sustainable Energy Review,
21:506-523.
Baldauf, R., E. Thoma, A. Khlystov, V. Isakov, G. Bowker, T. Long, and R. Snowe (2008).
Impacts of noise barriers on near-road air quality. Atmospheric Environment, 42: 7502-
7507.
Barcelo, J., E. Codina, and J. Casas (2005). Microscopic traffic simulation: A tool for the design,
analysis and evaluation of intelligent transport systems. J. Intelligent and Robotic
Systems, 41(2-3): 173-203.
Bartlett, S., H. Cisneros, G. Heartwell, K. McAndrew, A. Warnock, M. Nellenbach, M., S. Kline,
A. Winkler, and J. Varn (2017). Understanding America's water and wastewater
challenges. Bipartisan Policy Center, Washington, DC., 28p.
Baynes, T.M. (2009). Complexity in urban development and management: Historical overview
and opportunities. J. Industrial Ecology, 13(2): 214-227.
Beck, M.B. (2005). Vulnerability of water quality in intensively developing urban watersheds.
Environmental Modelling and Software, 20:381-400.
Biswas, P., C. Lu, and R.M. Clark (1993). A model for chlorine concentration decay in drinking
water distribution pipes. Water Research, 27(12): 1715-1724.
165
-------
Black, J., and T. Endreny (2006). Increasing stormwater outfall duration, magnitude, and volume
through combined sewer separation. J. Hydrol. Engrg., 11(5):472-481.
Board of County Commissioners (2008). Manatee County Water Supply Facilities Work Plan,
Manatee County Administrative Center Commission Office Commission, Bradenton, FL.
Boccelli, D.L., M.E. Tryby, J.G. Uber, and R.S. Summers (2003). A reactive species model for
chlorine decay and THM formation under rechlorination conditions. Water Research, 37:
2654-2666.
Boccelli, D.L., M.J. Small, and U.M. Diwekar (2007). Drinking water treatment plant design
incorporating variability and uncertainty. J. Envir. Engrg. 133(3):303-312.
Bonnin, G., D. Martin, T. Parzybok, B. Lin, D. Riley, and M. Yekta (2006). Precipitation
Frequency Atlas of the United States. NOAA Atlas 14, Volume 2, Version 3.0, National
Weather Service, Silver Spring, MD.
Bonnin, G.M., D. Martin, B. Lin, T. Parzybok, M. Yekta, and D. Riley (2011). Precipitation-
frequency Atlas of the United States, NOAA Atlas 14, Volume 1-9. National Weather
Service, Silver Spring, D.
Boulos, P.F., K.E. Landsey, andB.W. Karney (2006). Comprehensive Water Distribution
Systems Analysis Handbook for Engineers and Planners. MWH SOFT, Pasadena, CA.
Braganza, K., D.J. Karoly, and J.M. Arblaster (2004). Diurnal temperature range as an index of
global climate change during the twentieth century. Geophysical Research Letters,
31(13): L13217,
Bras, R.L., and I. Rodriguez-Iturbe (1984). Random Functions and Hydrology. Addison-Wesley
Publishing Company.
Brown, C., Y. Ghile, M. Laverty, and K. Li (2012). Decision scaling: Linking bottom-up
vulnerability analysis with climate projections in the water sector. Water Resources
Research, 48: W09537.
Bull-Kamanga, L., K. Diagne, A. Lavell, E. Leon, F. Lerise, H. MacGregor, A. Maskrey, M.
Meshack, M. Pelling, H. Reid, D. Satterthwaite, J. Songsore, K. Westgate, and A.
Yitambe (2003). From everyday hazards to disasters: The accumulation of risk in urban
areas. Environment & Urbanization, 15(1): 193-203.
Buyadi, S.N.A., W.M.N.W. Mohd, and A. Misni (2013). Green spaces growth impact on the
urban microclimate. Procedia-Social and Behavioral Sciences, 105: 547-557.
Cann, K.F., D.R. Thomas, R.L. Salmon, A.P. Wyn-Jones, and D. Kay (2013). Extreme water-
related weather events and waterborne diseases. Epidemiol. Infext., 141: 671-686.
Capodaglio, A.G. (2004). Improving sewage treatment plant performance in wet weather. NATO
Science Series: IV: Earth and Environmental Sciences, 43: 175-185.
Carrico, B., and P.C. Singer (2009). Impact of booster chlorination on chlorine decay and THM
production: Simulated analysis. J. Envir. Engrg. 135(10): 928-935.
Carter, T., and C.R. Jackson (2007). Vegetated roofs for stormwater management at multiple
spatial scales. Landscape and Urban Planning, 80: 84-94.
166
-------
Chang, H.-J., R. Hargrove, Y.-X. Long, and D.J. Osborne (2006). Reconstruction after the 2004
tsunami: ecological and cultural considerations from case studies. Landscape EcolEng,
2: 41-51.
Chang, N.-B., C. Qi, and Y.J. Yang (2012). Optimal expansion of a drinking water infrastructure
system with respect to carbon footprint, cost-effectiveness and water demand. J Environ
Management, 110: 194-206.
Chang, N-B, B.W. Vannah, Y.J. Yang, and M. Elovitz (2014a). Integrated data fusion and
mining techniques for monitoring total organic carbon concentrations in a lake.
International Journal of Remote Sensing, 35:3: 1064-1093.
Chang, N.-B., B. Vannah, and Y.J. Yang (2014b). Comparative sensor fusion between
hyperspectral and multispectral satellite sensors for monitoring microcystin distribution
in Lake Erie. IEEE J. Selected Topics in Appl. Earth Observation & Remote Sensing,
7(6): 2426-2442.
Changnon, D. (2006). Regional and temporal variations in heavy precipitation in South Carolina.
Int'lJ. Climatology, 14(2): 165-177.
Chen, Y., J.J. Cheng, and K.S. Creamer (2008). Inhibition of anaerobic digestion process: A
review. Bioresource Technology, 99: 4044-4064.
Chung, G., K. Lansey, and G. Bayraksan (2009). Reliable water supply system design under
uncertainty. Environmental Modelling & Software, 24: 449-462.
Clark, R. M. (1998). Chlorine demand and TTHM formation kinetics: a second order model,
Journal of Environmental Engineering, 124(1): 16-24.
Clark, R.M, and M. Sivaganesan (2002). Predicting chlorine residuals in drinking water: second
order model, J. Water Resour. Ping. andMgmt., 128(2): 152-160.
Clark, R.M., and R.C. Haught (2005). Characterizing pipe wall demand: Implications for water
quality modeling. J. Water Resour. Ping. andMgmt., 131(3): 208-217.
Clark, R.M., Z. Li, S.G. Buchberger, and Y.J. Yang (2009). Evaluating the effects of climate
change on the operation, design and cost of water treatment. WQTC 09, Seattle, WA.
Clark, R.M., Y.J. Yang, C. Impellitteri, R.C. Haught, D. Schupp, S. Panguluri, and E.R.
Krishnan (2010). Controlling disinfectant residual losses in drinking water distribution
systems: Results from experimental and modeling studies. J. AWWA. 102(4): 1-21.
Clarke, J. F. (1969). Nocturnal urban boundary layer over Cincinnati, Ohio. Mon. Weather Rev.,
97(8): 582-589.
Comfort, L.K. (2006). Cities at risk: Hurricane Katrina and the drowning of New Orleans. Urban
Affairs Review, 41(4): 501-516.
Cook, R., L. Griffins, P. Vickery, and E. Stafford (2011). ASCE 7-10 wind loads. Proc. 2011
Structure Congress, Las Vegas, NV.
Cooley, H., J. Fulton, and P.H. Gleick (2011). Water for Energy: Future Water Needs for
Electricity in the Intermountain West. Pacific Institute, ISBN 1-893790-36-3.
167
-------
Cooper, D., M. Leclerc, J. Archuleta, R. Coulter, W. Eichinger, C. Kao, and C. Nappo (2006).
Mass exchange in the stable boundary layer by coherent structures. Agricultural and
Forest Meteorology, 136(3): 114-131.
Crittenden, J.C., P.S. Reddy, H. Arora, J. Trynoski, D.W. Hand, D.L. Perram, and R.S. Summers
(1991). Predicting GAC performance with rapid small-scale column tests. J. AWWA,
83(1): 77-87.
Cromwell, J. E. Ill, J.B. Smith, and R.S. Raucher (2007). No doubt about climate change and its
implications for water suppliers. J. AWWA, 99(9): 112-117.
Crozes, G., P. White, and M. Marshall (1995). Enhanced coagulation: Its effect on NOM
removal and chemical costs. J.AWWA, 87(1): 78-89.
Dennison, F.J., A. Azapagic, R. Clifit, and J.S. Colbourne (1998). Assessing management options
for wastewater treatment works in the context of life cycle assessment. Water Science
and Technology, 38(11), 23-30.
Diaz-Fierros, T.F., J. Puerta, J. Suarez, and F.D. Fierros (2002). Contaminant loads of CSOs at
the wastewater treatment plant of a city in NW Spain. Urban Water, 4(3): 291-299.
Dietz, M.E. (2007). Low impact development practices: A review of current research and
recommendations for future directions. Water, Air, & Soil Pollution, 186: 351-363.
Dodder, R., T. Felgenhauer, W. Yelverton, and C. King (2011). Water and greenhouse gas
tradeoffs associated with a transition to a low carbon transportation system. In ASME
2011 International Mechanical Engineering Congress and Exposition, 2011, pp. 531-
547.
Dodder, R.S. (2014). A review of water use in the U.S. electric power sector: Insights from
systems-level perspectives. Current Opinion in Chemical Engineering, 5: 7-14.
Dodman, D. (2009). Blaming cities for climate change? An analysis of urban greenhouse gas
emissions inventories. Environ. Urbanization, 21: 185-201.
Dominguez, D., and W. Gujer (2006). Evolution of a wastewater treatment plant challenges
traditional design concepts. Water Research, 40: 1389-1396.
Eastman, J.R. (2009). IDRISIHelp System. IDRISI Taiga. Worcester, MA, USA: Clark
University.
Ewing, R.,H. (2008). Characteristics, causes, and effects of sprawl: A literature review. In
Urban Ecology, John M. Marzluff, Eric Shulenberger, Wilfried Endlicher, Marina
Alberti, Gordon Bradley, Clare Ryan, Ute Simon, Craig ZumBrunnen (eds.), 519-535.
Springer US.
Farge, M. (1992). Wavelet transformations and their applications to turbulence. Annual
Reviews, Inc.
Federal Highway Administration (FHWA) (1990). Calibration and Adjustment of System
Planning Models. Washington DC, FHWA.
FHWA (2002). An Introduction to Urban Travel Demand Forecasting - A Self Instructional
Text. Retrieved December 23, 2013, from U.S. Department of Transportation:
http://ntl.bts.gov/DOCS/UT.html.
168
-------
FHWA (2012). Project-level Conformity and Hot-spot Analysis. Retrieved 10 1, 2014,
http://www.fhwa.dot.gov/environment/air qualitv/conformitv/guide/guide09.cfm
FHWA (2017). Transportation conformity: A basic guide for state & local officials. FHWA-
HEP-17-034. 19p.
Felgenhauer, T., and K.C. Bruin (2009). The optimal paths of climate change mitigation and
adaptation under certainty and uncertainty. Int 7 J. Global Warming, 1: 66-88.
Felgenhauer, T., and M. Webster (2013). Modeling adaptation as a flow and stock decision with
mitigation. Climatic Change, 122(4): 665-679.
Feng, Y., M. Liu, Y. Liu, X. Tong, and S. Deng (2011). Modeling dynamic urban growth using
cellular automata and particle swarm optimization rules. Landscape and Urban Planning,
102: 188-196.
Fernando, FLJ.S. (2010). Fluid dynamics of urban atmospheres in complex terrain. Annual
Review of Fluid Mechanics, 42: 365-389.
Flanders, F., J. Yang, R. Dodder, G. Furie, R. Baldauf, L. Bachle, A. Bostrom, L. Berry, C.
Walters, J. Bare, T. Barzyk, R. Bruins, E. Cooter, F. DiCosmo, T. Eason, T. Fontaine, L.
Jackson, N. Schumaker, and J. Weaver (2014). Synthesis paper on sustainable
transportation. EPA Program Briefing Paper. 126p. RTP, NC.
Fletcher, S.M., M. Miotti, J. Swaminathan, M.M. Klemun, K. Strzepek, and A. Siddiqi (2017).
Water supply infrastructure planning: Decision-making framework to classify multiple
uncertainties and evaluate flexible design. J. Water Re sour. Ping. andMgmt., 143(10):
p.04017061.
Freas, K., B. Bailey, A. Munevar, and S. Butler (2008). Incorporating climate change in water
planning. J. AWWA, 100: 6.
Fu, X., X. Wang, and Y.J. Yang (2018). Deriving suitability factors for CA-Markov land use
simulation model based on local historical data. J. Environ Management, 206: 10-19.
Garcia-Lopez, M.-A. (2012). Urban spatial structure, suburbanization and transportation in
Barcelona. J. Urban Economics, 72 (2-3): 176-190.
Gesch, D. (2005). Topography-based Analysis of Hurrican Katrina Inundation of New Orleans.
U.S. Geological Survey, Sioux Falls, SD.
Gettman, D., and L. Head (2003). Surrogate Safety Measures From Traffic Simulation Models,
Final Report. Federal Highway Administration.
Gikas, P., and G. Tchobanoglous (2009). The role of satellite and decentralized strategies in
water resources management. J. Environ Management, 90: 144-152.
Gilroy, K.L., and R.H. McCuen (2009). Spatio-temporal effects of low impact development
practices. J. Hydrology, 367: 228-236.
Gim, T.T. (2012). A meta-analysis of the relationship between density and travel behavior.
Transportation, 39(3): 491-519.
Girard, M., and M. Mortimer (2006). The Role of Standards in Adapting Canada's Infrastructure
to The Impacts of Climate Change, http://www.csa.ca/climatechange, Canadian Standards
Association. 71p.
169
-------
Gleick, P.H. (2000). The changing water paradigm, a look at twenty-first century water resoruces
management. International Water Resoruces Association, 25(1): 127-138.
Godschalk, D. (2003). Urban hazard mitigation: Creating resilient cities. Nat. Hazards Rev.,
4(3): 136-143.
Gordon, P., H.W. Richardson, and H.L. Wong (1986). The distribution of population and
employment in a polycentric city: The case of Los Angeles. Environment & Planning A
18 (2): 161-173.
Guhathakurta, S., and P. Gober (2007). The impact of the Phoenix urban heat island on
residential water use. J. Am. Plann. Assoc., 73(3): 317-329.
Guo, J.C.Y., and K. Hargadin (2009). Conservative design rainfall distribution. J. Hydrologic
Engrg, 14(5): 528-530.
Gupta, A.K., and R.K. Shrivastava (2006). Uncertainty analysis of conventional water treatment
plant design for suspended solids removal. J. Envir. Engrg., 132(11): 1413-1421.
Guzman-Torres, D., A. Eiguren-Fernandez, P. Cicero-Fernandez, M. Maubert-Franco, A.
Retama-Hernandez, R. Ramos Villegas, and A.H. Miguel (2009). Effects of meteorology
on diurnal and nocturnal levels of priority polycyclic aromatic hydrocarbons and
elemental and organic carbon in PMio at a source and a receptor area in Mexico City.
Atmospheric Environment, 43(17): 2693-2699.
Harrington, J.J., and J.S. Gidley (1985). Variability of alternative decisions in a water resources
planning problem. Water Resources Research, 21(12): 1831-1840.
Heikkila, E., P. Gordon, J.I. Kim, R.B. Peiser, H.W. Richardson, and D. Dale-Johnson (1989).
What happened to the CBD-distance gradient?: Land values in a policentric city.
Environment and Planning ,4,21 (2): 221 -232.
Hering, J.G., T.D. Waite, R.G. Luthy, J.E. Dewes, and D.L. Sedlak (2013). A changing
framework for urban water systems. Environ. Sci. Technol.,Al\ 10721-10726.
Hershfield, D.M. (1961). Rainfall frequency atlas of the United States for durations from 30
minutes to 24 hours and return periods from 1 to 100 years. Weather Bureau, Technical
Paper No. 40., Washington, DC.
Hidalgo, J., V. Masson, and L. Gimeno (2010). Scaling the daytime urban heat island and urban-
breeze circulation. J. Applied Meteorology and Climatology, 49(5): 889-901.
Horowitz, A.J. (2009). Monitoring suspended sediments and associated chemical constituents in
urban environments: lessons from the city of Atlanta, Georgia, USA Water Quality
Monitoring Program. J. Soils & Sediments, 9(4): 342-363.
Hosseinzadehtalaei, P., H. Tabari, and P. Willems (2017). Uncertainty assessment for climate
change impact on intense precipitation: how many model runs do we need? International
J. Climatology, 37: 1105-1117.
Hummel, D., A. and Lux (2006). Population decline and infrastructure: The case of the German
water supply system. Vienna Yearbook of Population Research, 2007, pp. 167-191.
Ibrahim, H, A. Ilinca, and J. Perron (2008). Energy storage systems - Characteristics and
comparisons. Renew. Sustain. Energy Rev., 12: 1221-1250.
170
-------
Imen, S., N.-B., Chang, Y.J. Yang, and A. Golchubian (2016). Developing a Model-based
Drinking Water Decision Support System Featuring Remote Sensing and Fast Learning
Techniques. IEEE Systems Journal, 99: 1-11.
Imhoff, M.L., P. Zhang, R.E. Wolfe, and L. Bounoua (2010). Remote sensing of the urban heat
island effect across biomes in the continental USA. Remote Sensing of Environment,
114(3): 504-513.
Intergovernmental Panel on Climate Change (IPCC) (2007). Climate change 2007: The Physical
Science Basis. Cambridge University Press.
IPCC (2014). Climate Change 2014: Mitigation of Climate Change. Draft. 2014,
http://www.ipcc.ch/.
International Energy Agency (1EA) (2013a). World Energy Outlook 2013. www.iea.org. 9 rue de
la Federation 75739 Paris Cedex 15, France. 2013a, 232p.
IE A (2013b). Redrawing the energy-climate map. www.iea.org. 9 rue de la Federation 75739
Paris Cedex 15, France. 2013b, 126p.
Iziomon, M.G., H. Mayer, and A. Matzarakis (2003). Downward atmospheric longwave
irradiance under clear and cloudy skies: Measurement and parameterization. J.
Atmospheric and Solar-Terrestrial Physics, 65(10): 1107-1116.
Jarvinen, A.V.O., M.T. Pelkonen, and T. Vartiainen (1991). Upgrading the Removal ofHumic
Substances and Mutagen Precursors in Water Treatment. In: Humic Substances in the
Aquatic and Terrestrial Environment, B. Allard, H. Boren, A. Grimvall (eds.), Springer
Berlin/Heidelberg, 225-231.
Jiang, G., F. Zhang, and X. Kong (2009). Determining conversion direction of the rural
residential land consolidation in Beijing mountainous areas. Transactions of the CSAE,
25(2): 214-221.
Jing, C., Z. Ping, and Q. Mahmood (2009). Simultaneous sulfide and nitrate removal in
anaerobic reactor under shock loading. Bioresource Technology, 100: 3010-3014.
Johnson, T., J. Butcher, D. Deb, M. Faizullabhoy, P. Hummel, J. Kittle, S. McGinnis, L.O.
Mearns, D. Nover, A. Parker, S. Sarkar, R. Srinivasan, P. Tuppad, M. Warren, C.
Weaver, and J. Witt (2015). Modeling streamflow and water quality sensitivity to climate
change and urban development in 20 U.S. watersheds. J. AWRA, 50(5): 1321-1341.
Jones, P., F. Koppelman, and J. Orfueil (1990). Activity Analysis: State-of-the-art and Future
Directions. New Developments in Dynamic and Acticity-based Approaches to Travel
Analysis (pp. 34-55). Gower Publishing, Aldershot, England.
Kenworthy, J.R. (2006). The eco-city: ten key transport and planning dimensions for sustainable
city development. Environment & Urbanization, 18(1): 67-85.
Kim, Y-I, and B.-U. Bae (2007). Design and evaluation of hydraulic baffled-channel PAC
contactor for taste and odor removal from drinking water supplies. Water Research, 41:
2256-2264.
Kimura, K., G.L. Amy, J.E. Drewes, and T. Herberer (2003). Rejection of organic
micropollutants (disinfection by-products, endocrine disrupting compounds, and
171
-------
pharmaceutically active compounds) by NF/RO membranes. J. Membrane Science, 227:
113-121.
Kincannon, D.F., and A.F. Gaudy Jr. (1966). Some effects of high salt concentrations on
activated sludge. J. Water Pollution Control Federation, 38(7): 1 148-1 159.
Konrad II, C.E. (1997). Synoptic-scale features associated with warm season heavy rainfall over
the interior southeastern United States. J. Climate, 12: 557-571.
Kumar, M.S., V.K. Anandan, T.N. Rao, and P.N. Reddy (2012). A climatological study of the
nocturnal boundary layer over a complex-terrain station. J. Applied Meteorology and
Climatology, 51(4): 813-825.
Lai, F.-S. (2008). Review of sewer design criteria and RDII prediction methods. EPA/600/R-
08/010, USEPA, Office of Research and Development, Cincinnati, OH.
Larson, W., F. Liu and A. Yezer (2012). Energy footprint of the city: Effects of urban land use
and transportation policies. J. Urban Economics, 72 (2-3): 147-159.
Lee, E.J., C.S. Criddle, P. Bobel, and D.L. Freyberg (2013). Assessing the scale of resource
recovery for centralized and satellite wastewater treatment. Environ. Sci. Technol, 47:
10762-10770.
Levin, R.B., P.R. Epstein, T.E. Ford, W. Harrington, E. Olson, and E.G. Reichard (2002). U.S.
drinking water challenges in the twenty-first century. Environmental Health Perspectives,
110 (S.l): 43-52.
Levine, A.D., Y.J. Yang, and J.A. Goodrich (2016). Enhancing climate adaptation capacity for
drinking water treatment facilities. J. Water and Climate Change. 7(3): 485-497.
Li, Z., R. Clark, S. Buchberger, and Y.J. Yang (2009). Assessing the impact of climate change
on drinking water treatment plant design and operation. EWRI2009, Kansas City.
Li, Z., R. Clark, S. Buchberger, and Y.J. Yang (2012). Evaluation of Logistic Model for GAC
Performance in Water Treatment, J. AWWA, 104(9): E489-E500.
Li, Z., R. Clark, S. Buchberger, and Y.J. Yang (2014). Evaluation of Climate Change Impact on
Drinking Water Treatment Plant Operation. J Environ Engrg., 140(9), Special Issue:
Drinking Water Safety, Security, and Sustainability, A4014005.
Liang, M. S., T.C. Keener, M.E. Birch, R. Baldauf, J. Neal, and Y.J. Yang (2013). Low-wind
and other microclimatic factors in near-road black carbon variability: A case study and
assessment implications. Atmospheric Environment, 80: 204-215.
Liang, M. S. (2014). Quantitative Analysis of Major Factors Affecting Black Carbon Transport
and Concentrations in the Unique Atmospheric Structure of Urban Environment. Ph.D.
Dissertation, University of Cincinnati.
Liang, M.S., and T.C. Keener, (2015). Atmospheric feedback of urban boundary layer with
implications for climate adaptation. Environ. Sci. Technol., 49(17): 10598-10606.
Liang, M.S., and S. Julius, (2017). On the coastal topography and storm surge for infrastructure
risk assessment and adaptation. EWRI 2017, Sacramento, CA.
Lin, S.D. (2001). Water and Wastewater Calculations Manual. McGraw-Hill, New York. 854p.
172
-------
Liu, Y., G. Shintaro, D. Zhuang, and W. Kuang (2012). Urban surface heat fluxes infrared
remote sensing inversion and their relationship with land use types. J. Geographical
Sciences, 22(4): 699-715.
Lundie, S., G.M. Peters, and P.C. Beavis (2004). Life cycle assessment for sustainable
metropolitan water systems planning, Environ. Sci. Technol., 38: 3465-3473.
Lundin, M., M. Bengtsson, and S. Molander (2000). Life cycle assessment of wastewater
systems: influence of system boundaries and scale on calculated environmental loads.
Environ. Sci. Technol, 34: 180-186.
Luthy, R.G. (2013). Design options for a more sustainable urban water environment. Environ.
Sci. Technol, 47: 10719-10720.
Ma, X.L., T.J. Schmit, and W.L. Smith (1999). A nonlinear physical retrieval algorithm - its
application to the GEOS-8/9 sounder. J. Appl. Meteorol., 38(5): 501-513.
Mailhot, A., S. Duchesne, D. Caya, and G. Talbot (2007). Assessment of future change in
intensity-duration-frequency (IDF) curves for southern Quebec using the Canadian
Regional Climate Model (CRCM). J. Hydrology, 347: 197-210.
Marsalek, J., andB. Chocat (2002). International report: Stormwater management. Water
Science and Technology, 46(6-7): 1-17.
Marshall, J.D., and M.W. Toffel (2005). Framing the elusive concept of sustainability: A
sustainability hierarchy. Environ. Sci. Technol, 39(3): 673-682.
Martilli, A. (2002). Numerical study of urban impact on boundary layer structure: Sensitivity to
wind speed, urban morphology, and rural soil moisture. J. Applied Meteorology, 41(12):
1247-1266.
Martuzevicius, D., J. Luo, T. Reponen, R. Shukla, A.L. Kelley, H.S. Clair, and S.A. Grinshpun
(2005). Evaluation and optimization of an urban PM2.5 monitoring network. J. Environ.
Monit.,l(l)\ 67-77.
Matos, J.C., I.B. Valente, P.J.S. Cruz, and L.C. Neves (2013). An advanced probabilistic
updating algorithm for life-cycle analysis of civil structures. In: Life-Cycle and
Sustainability of Civil Infrastructure Systems, Strauss, Frangopol, Bergmeister (eds.),
Taylor & Francis Group, London, ISBN 978-0-415-62126-7.
Mays, L.W. (1999). Water Distribution Systems Handbook. McGraw-Hill, New York.
McCorquodale, JA. (2007). Storm-water jets and plumes in rivers and estuaries. Canadian J.
Civil Engineering, 34(6): 691 -702.
McDaniels, T., S. Chang, D. Cole, J. Mikawo, and H. Longstaff (2008). Fostering resilience to
extreme events within infrastructure systems: Characterizing decision contexts for
mitigation and adaptation. Global Environmental Change, 18: 310-318.
McDonnell, B.E. (2012). Controlling Disinfection By-Products Within a Distribution System by
Implementing Bubble Aeration within Storage Tanks. Master Thesis, University of
Cincinnati, 206p.
173
-------
McKenney, D.W., J.H. Pedla, P. Papadopol, and M.F. Hutchinson (2006). The development of
1901-2000 historical monthly climate models for Canada and the United States.
Agricultural and Forest Meteorology, 138: 69-81.
McNally, M. (1996). An Activity-Based Microsimulation Model for Travel Demand Forecasting.
UC Irvine: Center for Acitivity Systems Analysis.
McNally, M. (2007). The Four Step Model. D.A. Hensher (ed.) Elsevier Science, Oxford, UK.
Miller, K. and D. Yates (2006). Climate Change and Water Resources: A Primer for Municipal
Water Providers. Jointly sponsored by AWWA Research Foundation, Denver, CO, and
University Corporation for Atmospheric Research, Boulder, CO. AWWA Research
Foundation, American Water Works Association, and IWA Publishing.
Miller, G.R., D.D. Baldocchi, B.E. La, and T. Meyers, (2007). An analysis of soil moisture
dynamics using multi-year data from a network of micrometeorological observation sites.
Advances in Water Resources, 30: 1065-1081.
Milman, A., and A. Short (2008). Incorporating resilience into sustainability indicators: An
example for the urban water sector. Global Environmental Change, 18: 758-767.
Montalto, F., C. Behr, K. Alfredo, M. Wolf, M. Arye, and M. Walsh (2007). Rapid assessment of
the cost-effectiveness of low impact development for CSO control. Landscape and Urban
Planning, 82: 117-131.
Miiller, A., S. Marsili-Libelli, A. Aivasidis, T. Lloyd, S. Kroner, and C. Wandrey (1997). Fuzzy
control of disturbances in a wastewater treatment process. Water Research, 31(12): 3157-
3167.
Mulvihill, M.E., and J. A. Dracup (1974). Optimal timing and sizing of a conjunctive urban water
supply and waste water system with nonlinear programming. Water Resources Research,
10(2): 170-175.
Neil, C.W., Y.J. Yang, and Y-S. Jun (2012). A critical review of the potential impact of managed
aquifer recharge on mineral-water interactions responsible for arsenic mobilization in
groundwater. J. Environ. Monit., 4: 1772-1788.
Neil, C.W., Y.J. Yang, D. Schupp, and Y-S. Jun (2014). Water chemistry impacts on arsenic
mobilization from arsenopyrite dissolution and secondary mineral precipitation:
Implication for managed aquifer recharge. Environ. Sci. Technol., 48(8): 4395-4405.
Neil, C.W., Y. Zhao, A. Zhao, J. Neal, M. Meyer, and Y.J. Yang (2019). Trihalomethane
precursor reactivity changes in drinking water treatment unit processes during a storm
event. Water Supply, doi.org/10.2166/ws.2019.089
Novotny V. (2013). Water-energy nexus: retrofitting urban areas to achieve zero pollution.
Building. Research & Information, 41(5): 589-604.
Novotny, V., and J.W. Witte (1997). Ascertaining aquatic ecological risks of urban stormwater
discharges. Water Research, 31(10): 2573-2585.
Oh, K, Y. Jeong, D. Lee, W. Lee, and J. Choi (2005). Determining development density using
the urban carrying capacity assessment system. Landscape and Urban Planning, 73(1):
1-5.
174
-------
Oke, T.R. (1976). The distinction between canopy and boundary-layer urban heat islands.
Atmosphere, 14(4): 268-277.
Oke, T.R. (2006). Initial guidance to obtain representative meteorological observations at urban
sites. World Meteorological Organization, TD-No. 1250, 47p.
O'Neal, M.R., M.A. Nearing, R.C. Vining, J. Southworth, and R.A. Pfeifer (2005). Climate
change impacts on soil erosion in Midwest United States with changes in crop
management. Catena, 61: 165-184.
Ong, S.L., and B.J. Adams (1990). Capacity expansion for regional wastewater systems. Journal
of Environmental Engineering, 116(3): 542-560.
O'Reilly, J., C. Lee, G. Collins, F. Chinali, T. Mahony, and V. O'Flaherty (2009). Quantitative
and qualitative analysis of methanogenic communities in mesophilically and
psychrophilically cultivated anaerobic granular biofilms. Water Research, 43: 3365-3374.
Oron, G., L. Gillerman, A. Bick, Y. Mnaor, N. Buriakovsky, and J. Hagin (2007). Advanced low
quality waters treatment for unrestricted use purposes: imminent challenges.
Desalination, 213: 189-198.
Ostrom, E. (2010). Beyond markets and states: Poly centric governance of complex economic
systems. American Economic Review, 100(3): 641-672.
Pacific Northwest National Laboratory (PNNL) (2012). Climate and energy-water-land system
interactions - Technical report to the U.S. Department of Energy in support of the
national climate assessment. PNNL-21185. http://www.ntis.gov.
Pahl-Wostl, C. (2007). Transitions towards adaptive management of water facing climate and
global change. Water Re sour Manage, 21: 49-62.
Parshall, L., K. Gurney, S.A. Hammer, D. Mendoza, Y. Zhou, and S. Geethakumar (2010).
Modeling energy consumption and CO2 emissions at the urban scale: Methodological
challenges and insight from the United States. Energy Policy, 38: 4765-4782.
Pedersen, N., and D. Sandahl (1982). Highway traffic data for urbanized area project planning
and design. NCHRP Report 255. Transportation Research Board, Washington, DC.
Peng, Y., W. Zeng, and S. Wang (2007). DO concentration as a fuzzy control parameter for
organic substrate removal in SBR Processes. Environmental Engineering Science, 21(5):
606-616.
Perrone, D., J. Murphy, and G.M. Hornberger (2011). Gaining perspective on the water-energy
nexus at the community scale. Environ. Sci. Technol., 45: 4228-4234.
Peters, G.M., and S. Lundie (2001). Life-cycle assessment of biosolids processing options. J.
Industrial Ecology, 5(2): 1088-1980.
Pielke Sr., R.A., J.O. Adegoke, T.N. Chase, C.H. Marshall, T. Matsui, and D. Niyogi (2007). A
new paradigm for assessing the role of agriculture in the climate system and in climate
change. Agricultural and Forest Meteorology, 142: 234-254.
Pires, M. (2003). Watershed protection for a world city: the case of New York. Land Use Policy,
21: 161-175.
175
-------
Prasad, T.D., G.A. Walters, and D.A. Savic (2004). Booster disinfection of water supply
networks: Multi objective approach. J. Water Re sour. Ping. andMgmt., 130(5): 367-376.
Rajagopalan, B., and U. Lall (1998). Interannual variability in western US precipitation. J.
Hydrology, 210: 51-69.
Ranatunga, T., S.T.Y. Tong, Y. Sun, J.Y. Yang (2014). A comprehensive approach to total water
management: A case study of the Las Vegas wash watershed, Nevada. Physical
Geography, 35(3): 220-244.
Rao, K.D. (2005). Multi-criteria spatial decision analysis for forecasting urban water
requirements: a case study of Dehradun city, India. Landscape and Urban Planning, 71:
163-174.
Ray, S., and C.A. Peters (2008). Changes in microbiological metabolism under chemical stress.
Chemosphere, 71: 474-483.
Rendon, A.M., J.F. Salazar, C.A. Palacio, V. Wirth, and B. Brotz (2014). Effects of Urbanization
on the Temperature Inversion Breakup in a Mountain Valley with Implications for Air
Quality. J. Applied Meteorology and Climatology, 53(4): 840-858.
Roberts, P.V., and R.S. Summers (1982). Granular Activated Carbon Performance for Organic
Carbon Removal. J. AWWA, 74: 113-118.
Rosenzweig, C., D.C. Major, K. Demong, C. Stantion, R. Horton, and M. Stults (2007).
Managing climate change risks in New York City's water system: Assessment and
adaptation planning. Mitigation and adaptation Strategies for Global Change, 12(8):
1391-1409.
Rossman, L.A. (2002). EPANET Version 2 Users Manual, USEPA, Cincinnati, OH.
Rossman, L.A., R.M. Clark, and W.M. Grayman (1994). Modeling chlorine residuals in
drinking-water distribution systems. JEnv. Engng., 120(4): 803-820.
Rotach, M., W. Vogt, R. Bernhofer, C. Batchvarova, E. Christen, A. Clappier, B. Feddersen, S.-
E. Gryning, G. Martucci, H. Mayer, V. Mitev, T.R. Oke, E. Parlow, H. Richner, and H.
Roth (2005). BUBBLE-an urban boundary layer meteorology project. Theoretical and
Applied Climatology, 81(3-4): 231-261.
Rothausen, S.G.S.A., and D. Conway (2011). Greenhouse-gas emissions from energy use in the
water sector. Nature Climate Change, 1: 210-218.
Salas, J.D., J.W. Delleur, V. Yevjevich, and W.L. Lane (1980). Applied Modeling of Hydrologic
Time Series. Water Resources Publications, Littleton, CO, pp. 484.
Salvato, J.A., N.L. Nemerow, and F.J. Agardy (2008). Environmental Engineering, 8th edition.
Wiley & Sons, New York. ppl544.
Sang, L., C. Zhang, J. Yang, D. Zhu, and W. Yun (2011). Simulation of land use spatial pattern
of towns and villages based on CA-Markov model. Mathematical and Computer
Modelling, 54: 938-943.
Sante, I.S., A.M. Garcia, D. Miranda, and R. Crecente (2012). Cellular automata models for the
simulation of real-world urban processes: A review and analysis. Landscape and Urban
Planning, 96: 108-122.
176
-------
Schaad, D.E., J. Farley, and C. Haynes (2009). Design and routing of storm flows in an
urbanized watershed without surface streams. J. Hydrology, 375(3-4): 334-344.
Schrank, D., and T. Lomax (2009). 2009 Urban Mobility Report. Reports by Texas
Transportation Institute, The Texas A&M University.
Semadeni-Davies, A., C. Hernebring, G. Svensson, and L.-G. Gustafsson (2008). The impacts of
climate change and urbanization on drainage in Helsingborg, Sweden: Suburban
stormwater. J. Hydrology, 350: 114-125.
Shang, F., J. Uber, and L. Rossman (2008). EPANET Multi-species Extension Users Manual.
EPA/600/S-07/021. Cincinnati, OH.
Silva, E.A., and K.C. Clarke (2002). Calibration of the SLEUTH urban growth model for Lisbon
and Porto, Portugal. Comput. Environ. Urban Syst, 26: 525-552.
Simiu, E. (2011). Design of Buildings for Wind: A Guide for ASCE 7-10 Standard Users and
Designers of Special Structures. Second edition. John Wiley & Sons, Inc., Hoboken, NJ.
Skjelkvale, B.L., J.L. Stoddard, D.S. Jeffries, K. Torseth, T. Hogasen, J. Bowman, J. Mannio,
D.T. Monteith (2005). Regional scale evidence for improvements in surface water
chemistry 1990-2001. Environ. Pollut. 137: 165-176.
Slowinski, R., A. Urbaniak, and J. Weglarz (1985). Multicriteria capacity expansion planning of
a water supply and wastewater treatment system. International Journal of Modelling and
Simulation, 5(4): 124-127.
Small, K.A., and S. Song (1994). Population and employment densities: Structure and change. J.
Urban Economics, 36: 292-313.
Stratus Consulting and MWH Global (2009). Implications of Climate Change for Adaptation by
Wastewater and Stormwater Agencies. Water Environment Research Foundation Report
No. CC2R08.
Stubbart, J. (2004). How do we treat to prevent DBPs. Opflow, 30(9): 8-9.
Sudhira, H.S., T.V. Ramachandra, A. Wytzisk, and C. Jeganathan (2005). Framework for
Integration of Agent-based and Cellular Automata Models for Dynamic Geospatial
Simulations. Bangalore: Indian Institute of Science, Centre for Ecological Sciences.
Sukkoo, K. (2007). Changes in the nature of urban spatial structure in the United States, 1890-
2000. J. Regional Science, 47(2): 273-287.
Sun, Y., S.T.Y. Tong, M. Fang, and Y.J. Yang (2013). Exploring the effects of population
growth on future land use change in the Las Vegas Wash Watershed: An integrated
approach of geospatial modeling and analytics. Environment, Development and
Sustainability, 15: 1495-1515.
Sun, Y., S. Tong, and Y.J. Yang (2016). Modeling the cost-effectiveness of stormwater best
management practices in an urban watershed in Las Vegas Valley. Applied Geography,
76: 49-61.
Tafuri, A.T., and A. Selvakumar (2002). Wastewater collection system infrastructure research
needs in the USA. Urban Water, 4(1): 21-29.
177
-------
Theobald, D.M. (2005). Landscape patterns of exurban growth in the USA from 1980 to 2020.
Ecology and Society, 10(1): 32. https://www.jstor.org/stable/26267722
Tillman, A.M., H. Lundertroem, and M. Svingby (1998). Life cycle assessment of municipal
waste water systems. International Journal ofLCA, 3: 145-157.
Tillman, D.E., T.A. Larsen, C. Pahl-Wostlm, and W. Gujer (2005). Simulating development
strategies for water supply systems. J. Hydroinformatics, 7: 41-51.
Timmerman, P., and R. White (1997). Megahydropolis: Coastal cities in the context of global
environmental change. Global Environmental Change, 7(3): 205-234.
Tong, S.T.Y, Sun, Y., and Y.J. Yang (2012). Generating a future land use change scenario with a
modified population-coupled Markov Cellular Automata Model. J. Environmental
Informatics, 19(2): 108-119.
Torrence, C., and G.P. Compo (1998). A Practical Guide to Wavelet Analysis. Boulder, CO:
American Meteorological Society.
Trompetter, W.J., S.K. Grange, P.K. Davy, and T. Ancelet (2013). Vertical and temporal
variations of black carbon in New Zealand urban areas during winter. Atmospheric
Environment, 75: 179-187.
Tung, H., R.F. Unz, and Y.F. Xie (2006). HAA removal by GAC adsorption. J. AWWA, 98(6):
107-112.
Turnipseed, D.P., K. Van Wilson, J. Stoker, Jr., and D. Tyler (2007). Mapping hurricane Katrina
peak storm surge in Alabama, Mississippi, and Louisiana. 37 th Annual Mississippi Water
Resources Conference, Jackson, MS.
Uber, J., F. Shang, and L. Rossman (2004). Extensions to EPANET for fate and transport of
multiple interacting chemical or biological components. Critical Transitions in Water
and Environmental Resources Management, doi: 10.1061/40737(2004)486.
Uno, I., S. Wakamatsu, H. Ueda, and A. Nakamura (1988). An observational study of the
structure of the nocturnal urban boundary layer. Boundary-Layer Meteorology, 45(1-2):
59-82.
Uno, I., H. Ueda, and S. Wakamatsu (1989). Numerical modeling of the nocturnal urban
boundary layer. Boundary-Layer Meteorology, 49(1-2): 77-98.
Urbaniak, A. (1988). Multicriteria capacity expansion planning for an urban water system with
random data. Civil Engineering Systems, 5: 129-136.
U.S. BLS (Bureau of Labor Statistics) (2008). BLS Handbook of Methods, Chapter 14 Producer
Prices-Background, http://www.bls.gov/opub/hom/homchl4_a.htm.
U.S. EPA (1994). Combined sewer overflow control policy. Federal Register [FRL-4732-7],
Part VII, April 19, 1994. 16p.
U.S. EPA (1996). ICR manual for bench and pilot-scale treatment studies. EPA/814/B-96/003,
Office of Water, Washington, DC.
U.S. EPA (2000). ICR treatment study database, version 1.0. EPA/815/C-00/003, Office of
Water, Washington, DC.
178
-------
U.S. EPA (2001). Report to Congress: Implementation and enforcement of the CSO control
policy. EPA/833/R-01/003, Washington, DC.
U.S. EPA (2002). Onsite wastewater treatment systems manual. EPA/625/R-00/008. Cincinnati,
OH.
U.S. EPA (2004a). Stormwater best management practice design guide: Volume 3 basin best
management practices. EPA/600/R-04/12IB.
U.S. EPA (2004b). AERMOD description of model formulation. EPA/454/R-03/004, Research
Triangle Park, NC. https://www3.epa.gov/scram001/7thconf/aermod/aermod_mfd.pdf.
U.S. EPA (2005). Water Treatment Plant Model version 2.2 User's manual. Office of Ground
Water and Drinking Water, U.S. EPA, Cincinnati, OH.
U.S. EPA (2006). Growing toward more efficient water use: Linking development,
infrastructure, and drinking water policies. EPA/230/R-06/001. Washington, DC.
U.S. EPA (2007a). An approach for using load duration curves in the development of TMDLs.
EPA/84l/B-07/006, Washington DC., 68p.
U.S. EPA (2007b). Moving toward sustainability. Retrieved July 12, 2011, from
http://water.epa.gov/infrastructure/sustain/index.cfm.
U.S. EPA (2007c). Aging water infrastructure research program: Addressing the challenge
through innovation. EPA/600/F-07/015, Washington, DC.
U.S. EPA (2008). A screening assessment of the potential impacts of climate change on
combined sewer overflow (CSO) mitigation in the Great Lakes and New England
regions. EPA/600/R-07/033F, Washington, DC, p.39.
U.S. EPA (2009a). A Screening assessment of the potential impacts of climate change on
combined sewer overflow (CSO) mitigation in the Great Lakes and New England regions
(External Review Draft). EPA/600/R-07/033A, U.S. EPA, Washington, DC.
U.S. EPA (2009b). Land-use scenarios: National-scale housing-density scenarios consistent with
climate change storylines. EPA/600/R-08/076F, Washington, DC.
U.S. EPA (2009c). Proceedings of the first national expert and stakeholder workshop on water
infrastructure sustainability and adaptation to climate change. EPA 600/R-09/010.
Cincinnati, OH
U.S. EPA (2010a). Motor Vehicle Emission Simulator (MOVES); user guide for MOVES 2010a.
EPA/420/B-10/036.
U.S. EPA (2010b). Integrated climate and land-use scenarios (ICLUS) vl.3 User's manual:
ArcGIS tools and datasets for modeling US housing density growth. EPA/600/R-09/143F.
19p.
U.S. EPA (2011). Transportation control measures: An information document for developing and
implementing emissions reduction programs. EPA/430/R-09/040. U.S. EPA.
U.S. EPA (2012a). A framework for sustainability indicators at EPA. EPA/600/R-12/687.
Cincinnati, OH.
179
-------
U.S. EPA (2012b). Adaptive response framework for drinking water and wastewater utilities.
EPA/817/F-l2/009, Office of Water. Washington, DC,
U.S. EPA (2013a). Our built and natural environments. EPA/231/K-l3/001. Washington, DC,
139p.
U.S. EPA (2013b). Watershed modeling to assess the sensitivity of streamflow, nutrient, and
sediment loads to potential climate change and urban development in 20 U.S. watersheds.
EPA/600/R-12/058F, Washington, DC.
U.S. EPA (2014). National stormwater calculator - Version 1.1 (Model). EPA/600/C-13/106b,
Washington, DC.
U.S. EPA (2015a). National water infrastructure adaptation assessment. Part I: Climate change
adaptation readiness analysis. EPA/600/R-15/141. Cincinnati, OH
U.S. EPA (2015b). The impact of traditional and alternative energy production on water
resources: Assessment and adaptation studies. EPA 600/R-14/272. Cincinnati, OH
U.S. EPA (2015c). MOVES2014a User Guide. EPA Office of Transportation and Air Quality.
EPA/420/B-15/095. 264p.
U.S. EPA (2015d). Transportation conformity guidance for quantitative hot-spot analyses in
PM2.5 and PM10 nonattainment and maintenance areas. EPA/420/B-15/084, Washington,
DC, 152p.
U.S. EPA (2015e). Using MOVES2014 in project-level carbon monoxide analyses. EPA/420/B-
15/028, 51p.
U.S. EPA (2017a). Decision support system for aquifer recharge (AR) and aquifer Recovery
(ASR) planning, design, and evaluation: Principles and technical basis. EPA/600/R-
16/222.
U.S. EPA, (2017b). Appendix A to Appendix W of Part 51—Summaries of preferred air quality
models. In Revisions to the Guideline on Air Quality Models: Enhancements to the
AERMOD Dispersion Modeling System and Incorporation of Approaches To Address
Ozone and Fine Particulate Matter. 40 CFR Part 51. Federal Register, 82, 10.
U.S. EPA (2019). Better Assessment Science Integrating point & Non-point Sources - BASINS
4.5 User's Manual. RTP, NC. https://www.epa.gov/sites/production/files/2019-
03/documents/basins4.5coremanual.2019.03.pdf
Uyak, V., K. Ozdemir, and I. Toro (2008). Seasonal variations of disinfection by-product
precursors profile and their removal through surface water treatment plants. Science of
the Total Environment, 390: 417-424.
Voivontas, D., G. Arampatzis, E. Manoli, C. Karavitis, and D. Assimacopoulos (2003). Water
supply modeling towards sustainable environmental management in small islands: the
case of Paros, Greece. Desalination, 156: 127-135.
Wallis, J.R., M.G. Schaefer, B.L. Barker, and G.H. Taylor (2007). Regional precipitation-
frequency analysis and spatial mapping for 24-hour and 2-hour durations for Washington
State. Hydrol. Earth Syst. Sci., 11(a): 415-442.
180
-------
Wang, K., H. Ye, F. Chen, Y. Xiong, and C. Wang (2012). Urbanization effect on the diurnal
temperature range: different roles under solar dimming and brightening. J. Climate,
25(3): 1022-1027.
Wang X., A. Burguess, and Y.J. Yang (2013). A scenario-based water conservation planning
support system (SB-WCPSS). Stochastic Environmental Research and Risk Assessment,
27(3): 629-641.
Ward, D.P., A.T. Murray, and S.R. Phinn (2000). A stochastically constrained cellular model of
urban growth. Comput. Environ. Urban Syst, 24, 539-558.
Webster, M., P. Donohoo, and B. Palmintier (2013). Water-CC>2 trade-offs in electricity
generation planning. Nature Climate Change, 3: 1029-1032.
Wei, H., X. Wang, Z. Yao, H. Liu, S. Liang, and Y.J. Yang (2012). Framework for Integrating
Traffic-Source Emission Estimates into Sustainability Analysis. CICTP 2012\ 3612-3619.
doi: 10.1061/9780784412442.364.
Wei, H., T. Zuo, H. Liu, and Y.J. Yang (2017). Integrating land use and socioeconomic factors
into scenario-based travel demand and carbon emission impact study. Urban Rail Transit,
3(1): 3-14. doi: 10.1007/s40864-017-0056-2.
Weinstein, N. (2009). Decentralized stormwater controls for urban retrofit and combined sewer
overflow reduction, Phase 2. WERFReport 03-SW-3A. ISBN: 9781843393535.
Weng, Q., D. Lu, and J. Schubring (2004). Estimation of land surface temperature-vegetation
abundance relationship for urban heat island studies. Remote sensing of Environment,
89(4): 467-483.
Whitehead, P.G., R.L. Wilby, D. Butterfield, and A.J. Wade (2006). Impacts of climate change
on in-stream nitrogen in a lowland chalk stream: An appraisal of adaptation strategies.
Science of the Total Environment, 365: 260-273.
Whitehead, P.G., R.L. Wilby, R.W. Battarbee, M. Kernan, and A.J. Wade (2009). A review of
the potential impacts of climate change on surface water quality. Hydrological Sciences
Journal/Journal des Sciences Hydrologiques, 54(1): 101-123.
Wilby, R.L. (2007). A review of climate change impacts on the built environment. Built
Environment, 33(1): 31-45.
Wing, S., S. Freedma, and L. Band (2002). The potential impact of flooding on confined animal
feeding operations in Eastern North Carolina. Environ. Health Persp., 110(4): 387-391.
Yamout, G., and M. El-fadel (2005). An optimization approach for multi-sectoral water supply
management in the Grater Beirut Area. Water Resources Management, 19: 791-812.
Yang, Y.J., A. Hale, D. Chapman, and D. Keith (2002). Water quality and biological response to
water aeration of glycol-contaminated streams in cold temperatures. Proc. of the Water
Environment Federation, Industrial Wastes 2002, 962-978(17).
Yang, Y.J., J. A. Goodrich, S.Y. Li, and R.C. Haught (2007). AST/R-based water reuse as a part
of the total water solution for water-stressed regions: An overview of engineering
practice and regulatory prospective. EWRI2007, Tampa, FL.
181
-------
Yang, Y.J., C.A. Impellitteri, R.M. Clark, R. C. Haught, D.A. Schupp, S. Panguluri, and E.R.
Krishnan (2008). Chlorine decay and DBP formation under different flow regions in PVC
and ductile iron pipes: Preliminary results on the role of flow velocity and radial mass
transfer. EWRI2008, Atlanta, GA.
Yang, Y.J. (2010). Re-define adaptation of water resource infrastructure to a non-stationary
climate (Editorial). J. Water Re sour. Ping. &Mgmt. 136(3): 297-298.
Yang, Y.J., and A. Restivo (2010). ASR application in climate change adaptation: The need,
issues and research focus. Water & Energy in Changing Climate, Ground Water
Protection Council, Pittsburgh, PA. September 2010.
Yang, Y.J., N.B. Chang, J. Neal, H. Wei, M.S. Liang, and T.C. Keener (2013). Water and carbon
footprints for sustainability analysis of urban infrastructure. EWRI 2013, Cincinnati, OH.
Yang Y.J., and J.A. Goodrich (2014). Toward quantitative analysis of water-energy-urban-
climate nexus for urban adaptation planning. Current Opinion in Chem Engrng, 5: 22-28.
Yang, Y.J. (2016). Adapting Water Infrastructure to Non-stationary Climate Changes. In:
Sustainable Water Management and Technologies, D.H. Chen (ed.), Taylor &
Francis/CRC Press.
Yang, Y.J., M.S. Liang, Y. Zhao, and J. Neal (2017). Establishing design storm values from
climate models in coastal regions: Challenges and opportunities. EWRI2017,
Sacramento, CA.
Yao, Z., H. Wei, X. Wang, H. Liu, and J. Yang (2014). Scenario-based carbon footprint
inventory tool for urban sustainable development decision support: The Cincinnati case
study. Transportation Research Board Annual Meeting 2014 Paper #14-3242.
Zachman, B.A., and R.S. Summers (2010). Modeling TOC breakthrough in granular activated
carbon adsorbers. J. Environmental Engineering, 136(2): 204-210.
Zeleny, M. (1973). Compromise programming. In: Multiple Criteria Decision Making, J.L.
Cochrane, M. Zeleny (eds.), University of South Carolina Press, Columbia, SC, pp. 262-
301.
Zhao, Y., Y.J. Yang, Y. Lee, S. Panguluri, and Y. Shao (2017). Better water demand and Pipe
description improve the distribution network modeling results. EWRI 2017, Sacramento,
CA.
Zhao, Y., Y.J. Yang, Y. Shao, J. Neal, and T. Zhang (2018a). The dependence of chlorine decay
and DBP formation kinetics on pipe flow properties in drinking water distribution. Water
Research, 141: 32-45.
Zhao, Y., Y.J. Yang, Y. Shao, Y. Lee, and T. Zhang (2018b). Demand-driven spatiotemporal
variations of flow hydraulics and water age by comparative modeling analysis of
distribution network. J. Wat. Resour. Plann. &Manag., 144(12): 04018074.
Zhou, J., J. Lin, S. Cui, Q. Qiu, and Q. Zhao (2013). Exploring the relationship between urban
transportation energy consumption and transition of settlement morphology: A case study
on Xiamen Island, China. Habitat International, 37: 70-79.
182
-------
Zmud, J.P., V.P. Barabba, M. Bradley, J.R. Kuzmyak, M. Zmud, and D. Orrell (2014). Strategic
Issues Facing Transportation, Volume 6: The Effects of Socio-Demographic on Future
Travel Demand. TRB's National Cooperative Highway Research Program (NCHRP)
Report 750. Transportation Research Board, Washington, DC.
183
-------
Appendix A
AIR-SUSTAIN Program Input
and Output Structures
184
-------
Table of Contents
List of Tables 186
List of Figures 187
List of Acronyms 188
A 1.1 AIR-SUSTAIN Program and Operations 189
A 1.1.1 Program Interfaces 189
1. Scenario Information Specification 189
2. Regional Level Analysis 195
3. Project Level Analysis 197
4. Result Comparison 201
A 1.1.2 Inputs and outputs 202
A 1.2 Database in AIR-SUSTAIN 213
A 1.2.1 Database Structure 213
A 1.2.2 MOVES Emission Lookup Tables 221
A 1.3 Transportation Analysis Examples Using the AIR-SUSTAIN Tool 226
185
-------
List of Tables
Table Al- 1 Sample of Population Change 194
Table Al- 2 Sample of Employment Change 194
Table Al- 3 Sample of University Enrollment Change 194
Table Al- 4 Sample of High School Enrollment Change 194
Table Al- 5 Inputs for the AIR-SUSTAIN 202
Table Al- 6 Outputs from the AIR-SUSTAIN 203
Table Al - 7 Inputs for the Scenario Development 204
Table Al- 8 Land Use Inputs 205
Table Al- 9 List of Outputs from the Scenario Development 206
Table Al- 10 Inputs for the Regional Level Analysis 207
Table Al- 11 Description of a VISUM File 209
Table Al- 12 Outputs from the Regional Level Analysis 210
Table Al- 13 Inputs for the Project Level Analysis 211
Table Al- 14 Outputs from the Project Level Analysis 212
Table A2- 1 Tables and fields of AIR-SUSTAIN Scenario Database 213
Table A2- 2 Tables and Items of the MOVES Input Database 215
Table A2- 3 Tables and Items of the MOVES Output Database 217
Table A2- 4 Geodatabase 218
Table A2- 5 The MOVES Source Types 221
Table A2- 6 The MOVES Road Type 221
Table A2- 7 MOVES Age Distribution Categories 222
Table A2- 8 MOVES Operating Modes 223
Table A3- 1 Sample of Population Change 228
Table A3- 2 Sample of Household Fraction and Trip Rate 232
Table A3- 3 Sample of Employment Fraction 232
Table A3- 4 Sample of Vehicle Composition 233
Table A3- 5 Sample of Age Distribution 234
Table A3- 6 Sample of Fuel Formulation 234
Table A3- 7 Sample of Fuel Supply 234
Table A3- 8 Sample of Meteorology 234
Table A3- 9 Sample of State and County 235
Table A3- 10 Sample of Selected Traffic Congestion Links for Corridor Level Impact Analysis
237
Table A3- 11 Vehicle Volume 239
Table A3- 12 Vehicle Type Class and Category 240
Table A3- 13 Car Following Behavior Parameters 240
Table A3- 14 Lane Change Parameters 240
Table A3- 15 Ramp metering design criteria of the Federal Highway Administration 241
Table A3- 16 Signal Control Parameters 241
Table A3- 17 VISSIM Calibration Final Parameter Values 241
Table A3- 18 VISSIM Validation Results 242
Table A3- 19 Microscopic Simulation Link ID (Sheetl) 243
186
-------
Table A3- 20 Microscopic Simulation Link ID (Sheet2) 243
Table A3- 21 Microscopic Simulation Results (Sheetl) 244
Table A3- 22 Microscopic Simulation Results (Sheet2 244
Table A3- 23 An Example of Traffic Congestion link Emissions by Different Scenarios 245
Table A3- 24 Variable and parameter coding type in AIR-SUSTAIN program 246
List of Figures
Figure Al- 1 Scenario information 189
Figure Al - 2 Interface for (a) new scenario; (b) and load scenario 190
Figure Al- 3 "Save Scenario As" interface 191
Figure Al- 5 Load base year data 192
Figure Al - 4 Scenario development interface 192
Figure Al - 6 Assumed changes in demographic and socioeconomic factors 193
Figure Al - 7 Population and employment changes 193
Figure Al- 8 Land use projection 195
Figure Al - 9 Socioeconomic data update based on assumed changes 195
Figure Al- 10 Regional level analysis interface 196
Figure Al- 11 Travel demand forecasting panel 197
Figure Al- 12 Emission estimation panel 198
Figure Al- 13 Traffic congestion identification 198
Figure Al- 14 Project level analysis interface 199
Figure Al- 15 Microscopic simulation results import 200
Figure Al- 16 Traffic congestion area emission estimation 200
Figure Al- 17 Updating traffic congestion identification results to regional analysis 200
Figure Al- 18 Results comparison interface 201
Figure A3- 1 Import base year data in example 227
Figure A3- 2 Program interface for (A) importing the Base Year data; (B) assigning population
change; and (C) assigning employment changes at TAZ levels 227
Figure A3- 3 Target year land use in example 228
Figure A3- 4 Target year land use in example 229
Figure A3- 5 VISUM demand set up 230
Figure A3- 6 Example of road network from the input function 230
Figure A3- 7 Example of VISUM matrix 231
Figure A3- 8 VISUM procedure set up 231
Figure A3- 9 Example of trip distribution result from VISUM (Trips between two centers).... 233
Figure A3- 10 Emission results displayed in ArcGIS 235
Figure A3- 11 Traffic congestion identification criteria 235
Figure A3- 12 Identified traffic congestion links in simulation 236
Figure A3- 13 VISSIM links over the base map 237
Figure A3- 14 Schematic for traffic congestion microscopic analysis 238
Figure A3- 15 An example of desired speed distribution for cars and trucks 239
187
-------
List of Acronyms
ft feet
GIS geographic information system
SUD Smart Urban Design
TDF Travel Demand Forecasting
188
-------
AIR-SUSTAIN is a major component of the Smart Urban Design (SUD) program. The
principles and functionalities are described in the main report Sections 3.0-4.0. This Appendix
describes program inputs, outputs, and major program interfaces for program use and project
simulation.
A 1.1 AIR-SIJSTAIN Program and Operations
A 1.1.1 Program Interfaces
The execution of each analysis function within the AIR-SUSTAIN is achieved through
interfaces embedded in a geographic information system (GIS) environment. Main functions and
interfaces of the AIR-SUSTAIN are:
¦ Scenario Information Specification
¦ Scenario Development
¦ Regional Level Analysis
¦ Project Level Analysis
¦ Results Comparison
1. Scenario Information Specification
The AIR-SUSTAIN provides a Scenario Information interface (Figure Al-1). Before
performing a scenario analysis, the scenario information must be set up first either by creating a
scenario (via the New Scenario button on the menu bar) or loading an existing scenario (via the
Load Scenario button on the menu).
¦
_ n
Scenario | Data
New Scenario
Load Scenario
1 Scenario-based Urban'Settjnigs^andfT^sportation Assets In Network
Save Scenario As
*±2 jmBM
Scenario Design Regional Level Analysis Project Level Analysis Results Comparision
¦ i i i. v I Modeling Year
Modelina Year Selection J
Figure Al-1 Scenario information.
Scenario information in the New Scenario and Load Scenario windows (as shown in
Figure A1-2) includes:
1) Scenario Name (required): the name of a scenario analysis specified by user
2) Project Directory (required): the route where user place the scenario folder
3) Modeling Year (required): Base Year and Target Year
189
-------
(A)
(B)
«ii New Project | o || a || a I
Senario Name | |
Project Directory 1 I ••• I
Modeling Year (yyyy) Base: Target:
Analyst
Date Thursday . September 18.2014 QK]
IDRISI Directory | | ... |
MOVES Directory
Scenario Description (optional):
Save Project
Cancel
¦i1 Scenario Information <=> a ;| £3 |
Scenario 2
C :\Users\Tmg\Desktop\Scenario
Base: 2010 Target: 2030
Ting
Thursday .September 4.2014
j C :\Users\Tmg\Desktop
C:\Usere\Public\MOVES201204l1
Scenario Description (optional):
OK
Senario Name
Project Directory
Modeling Year (yyyy)
Analyst
Date
IDRISI Director/
MOVES Directory
Figure Al- 2 Interface for (a) new scenario; (b) and load scenario.
4) Analyst (required)
5) Date (required)
6) IDRISI Directory (required): where IDRISI is installed;
7) MOVES Directory (required): where program MOVES is installed
8) Scenario Description (optional)
After setting up all required data, in the New Scenario tab, by clicking the Save Project button,
a scenario folder and five MySQL databases are created. A scenario folder contains:
¦ GIS.gdb: a geodatabase store feature classes such as TAZ, road network, incentive
boundary, and scenario analysis results
¦ IDRISI: a subfolder to store inputs and outputs for the land use projection
¦ MOVES: a subfolder to store inputs and outputs for the emission estimation
¦ VISSIM: a subfolder to store microscopic traffic simulation input and output files
¦ VISUM: a subfolder to store the TDF model inputs and outputs
¦ ScenarioName map: an ArcGIS map file that contains input maps and analysis result
maps
The MySQL database (see details later) includes:
¦ AIR-SUSTAIN ScenarioName database, including: projectlnfo, IDRISIInfor,
employmentGrowth, em pi ov m en tTri p Rate, populationGrowth, householdTripRate,
190
-------
increasePercentage, unversityEnrollment, HighSchoolEnrollment, baseYearResults,
targetY earResults
ScenarioName_In database: used as the MOVES input database in the regional level
analysis
ScenarioName_Out database: used as the MOVES output database in the regional level
analysis
ScenarioName_Project_In database: used as the MOVES input database in the project
level analysis
ScenarioName_Project_Out database: used as the MOVES output database in the project
level analysis
If a scenario is created, the user can load the scenario information by clicking the Load
Scenario on the menu bar. Afterwards, the scenario information is displayed in a pop-up window
shown as Figure Al-2b.
The Save Scenario As (as shown as in Figure Al-3) window provides a function to save
a new scenario based on current scenario data by specifying a new scenario name and a new
scenario directory in the SaveAsForm window.
Scenario Data
New Scenario
^ Load Scenario
a)
L
Save Scenario As
Scenario Desgn Re^ooal Levd toalyss Project Level Anaiyw Re
Modetng Year Selection
O Ease Year O Tarjet
Base Year Data
Please select the data type
imported Ba
SaveAsForm
s S3
Specify New Scenario Name:
Specify New Project Director/.
OK Cancel
Figure A I- 3 "Save Scenario As" interface.
The Scenario Development tab (Figure Al-4) has three main panels: Modeling Year
Selection, Base Year Data, and Target Year Scenario Development.
¦ The Modeling Year Selection panel is applied to select scenario analysis year by
checking either the Base Year or Target Year.
¦ The Base Year Data panel provides the function (shown as Figure Al-5a) to import base
year feature classes (i.e., TAZ. RoadNetwork, and Incentive boundary). As illustrated by
Figure Al-5, users can select the data type (i.e., TAZ shown as Figure Al-5b) from the
dropdown list and import data by clicking the Import button (Figure Al-5c). The
191
-------
S3
Scenario Data
Air Impact Relating Scenario-based Urban Settings and Transportation Assets In Network
l
Scenano Devetopmert Reponal Level Acefyta Project Level taatyss Resuls Compamon
Modelng Year Selection
O Base Year C TargetYear
Base Year Data
Please select It* data type
Imported Base Data
Modeling Year
my
Process Status
Target Year Scenano Oe*gn
1 Assumed Changes in Demographic and Socioeconomic Factors
a Population Change O Edit
b Employment Change O Edit
O load File
O Load File
c. University Enrollment Change Load File
d High School Enrollment Change Load File
2 Land Use Projection
a Initial Year (yyyy)
b Land Use Inputs Directory
c IDRIS1
d Target Year Land Use
Run
View Results
3 Demographic and Socioeconomic Data Update Based on Assumed Changes
a Allowable Population Density
b Linkage Model Run
e Target Year Demographic and Socioeconomic Data
Figure Al- 4 Scenario development interface.
imported data will be listed in the right box (Figure Al-5d). Users can also remove the
imported data by selecting the data name in the box and then clicking the Remove button
in Figure Al-5c.
The Target Year Scenario Design panel provides functions to set up assumed changes
in demographic and socioeconomic factors, including population change, employment change,
Base Year Data
Please select the data type:
1.TAZ
2. Road Network
3. Incentive Boundary
Base Year Data
Please select the data type:
1 Jn7
2. Road Network
3. Incentive Boundary
Imported Base Data:
TAZ
Road Network
Incentiveboundary
(a)
Base Year Data
Please select the data type:
|0BE v]
Import DC
Incentiveboundary
RnarfNrtwn*
TAZ 1
largei rear i
Figure Al- 5 Load base year data.
192
-------
Targe* Year Scenario Des>gn
1 Assumed Changes in Demographic and Socioeconomic Factors
a Population Change • Edit J O Load File c University Enrollment Change Load File
b Employment Change O Load File^^ d High School Enrollment Change Load File
View
Edit
t
Q Population Data Editor o S £3
Incentive Afea Population Pefcerrtage Change
X
Non-incentive Area Population Percentage Change
%
Save
'« c Univer
-------
Table Al-1 Sample of Population Change
TAZ
Population
330
156
338
191
318
268
Table Al- 2 Sample of Employment Change
TAZ
Employment
330
156
338
191
318
268
Table Al- 3 Sample of University Enrollment Change
TAZ
Enrollment
Name
330
156
Hebrew Union College
338
191
Institute of Technical Careers
318
268
God's Bible College
Table Al- 4 Sample of High School Enrollment Change
TAZ
Enrollment
Name
210
2556
Walnut Hills High School
244
613
Merry Middle School
251
584
Creative & Performing Arts High School
194
-------
In the Target
Year Scenario Design
panel, target year land use
is projected in the Land
Use Projection (see
Figure Al-8). Before
performing land use
projection, the Initial
Year needs to be
specified by the user.
Other land use inputs are Figure Al- 8 Land use projection.
loaded by specifying the
route of a folder that
contains files listed in Table Al-4. Then by executing IDRISI, target year land use is projected,
and land use projection results can be displayed in the ArcMap main window by clicking the
View Results button.
Figure A1 -9 shows the Socioeconomic Data Update Based on Assumed Data panel. In
the Target Year Scenario Design, target year demographic and socioeconomic data are
generated by base year demographic and socioeconomic data, assumed changes in demographic
and socioeconomic factors, base year land use, and target year land use by linkage model. Before
running the linkage model, the allowable population density should be set up. The maximum
population density is used as the maximum unit area population capacity in a TAZ. In this panel,
3. Demographic and Socioeconomic Data Update Based or Assumed Changes
a Allowable Population Density I
b Linkage Mode) Run
c. Target Year Demographic and Socioeconomic Data v View
Figure Al- 9 Socioeconomic data update based on assumed changes.
target year socioeconomic data can be viewed by specifying the data type in the dropdown list
and clicking the View Results button.
2. Regional Level Analysis
The Regional Level Analysis module is used to estimate the base and target year travel
demand and on-road emissions for the study area. The Travel Demand Forecasting and
Emission Estimation panels are highlighted by red boxes in Figure A I-10. When performing
the regional level analysis, a TDF model first simulates trips on roadway links for the entire
study area based on demographic and social economic data, as well as transportation
2 Land Use Projection
a Irabal Year (yyyy)
2000
b Land Use Inputs Directory
[C:Md*N | 1 - |
c IDRISI
Run
d Target Year land Use
View Results
195
-------
0 ! S3
Scenario Data
Air Impact Relating Scenario-based Urban Settings and Transportation Assets In Network
Scenario De*gn Reponal Level toafy» Project Level Analy*s ReedU Companion
Travel Demand Focecaafcng
Please select TOP Model
VISUM Cube [ TransCAD
1. VISUM File Load File
2 Household Fraction and Trip Rate Load ^
3 Employment Fraction Load File
4 VISUM R"*1
( Note Please import a version file condoning
TAZs road nef*ork. and transit network)
View Results
Enwston Estmabon
Please select Emission Estimation Model
MOVES IeMBC]
1 MOVES Imputs
C Import by individual files
Please select the data t> pe
Import
2 MOVES
3 Results
Remove
C Import by a folder
Please specify files directory
Run
Modeling Year
yyyy
Process Status
Project Information
Modeling Year Selector
Base Year Data
Target Year Scenario Design
Travel Demand Forecasting
Emission Estimation
Hots pots Idetificaticci
Microscopic Simulation Result:
Hots pots Emission Estimation
Regional Emissions Update
Figure Al-10 Regional level analysis interface.
infrastructure, i.e., road network, TAZs. Afterwards, the forecasted traffic data are utilized to
generate inputs for a traffic emission model, which is adopted to estimate road link based vehicle
emissions. Particularly in the emission analysis, CO2 equivalent and energy consumption for
individual road links in the study area are estimated by the user selected emission model.
The Travel Demand Forecasting (TDF) panel has five components (Figure Al-11). In
Component 1, the user needs to specify the TDF model. Among the popular TDF tools such as
VISUM, Cube, and TransCAD, the current version of the AIR-SUSTAIN supports VISUM 13.0.
Other models will be included in the software in the future. When the VISUM label is selected
by the user, the VISUM panel is activated. In the VISUM model, a VISUM file, a Household
Fraction and Trip Rate, and an Employment Fraction need to be loaded by Component 2, 3, 4
respectively (Figure Al-11). Component 5 provides functions to execute VISUM and view TDF
results in VISUM.
196
-------
Travel Demand Forecasting
Please select TDF Model: VISUM •»
VISUM Cube | TransCAD
r — —— |
Figure Al-11 Travel demand forecasting panel.
Figure Al-12 shows the Emission Estimation panel, which contains five components.
Similar to travel demand forecasting, the user is allowed to specify the emission model in
Component 1. For the current version of AIR-SUSTAIN, the regional level transportation
emission estimation will be conducted by using the EPA's MOVES model. The functions for
supporting EMFAC model will be developed in the future. Two methods of loading MOVES
inputs are provided in Component 2. When the Import by individual files is checked, MOVES
input files can be imported individually. Alternatively, if the Import by a folder is checked, all
files in the specified folder are imported as MOVES inputs. The steps in the two methods of
loading MOVES inputs are:
• When Import by individual files is checked, the user should specify the input data type,
(i.e., meteorology, age distribution, fuel formulation, fuel supply, and state and county),
and import the corresponding file by clicking Import button. The imported file can be
deleted by clicking the Remove button.
• When the Import by a folder is checked, the user only needs to specify the directory
where all required files (listed in Table Al-1 and Table Al-4) are placed. The user needs
to prepare each file in the folder according to specifications listed in Table Al-1.
Component 3 in Figure A2-12 is for displaying imported data. The user can select
individual data files in Component 3 and click the Remove button to delete it. When all data files
are imported, the user can run the MOVES by using Component 4. When MOVES simulation is
finished, the user can visualize the results in Component 5.
3. Project Level Analysis
In the Traffic Congestion Identification panel (shown in Figure Al-13), the user can
identify traffic congestion links by clicking Run button. This panel and function reside in the
Project Level Analysis shown in Figure A1-14. Then the Traffic Congestion Identification
197
-------
Emission Estimation
Please select Emission Estimation Model;
MOVES EMFAC
MOVES ImpL/ts
Import by individual files
Please select the data type:
Componen 2
© Import by a folder
Please specif)' files directory:
Q
Import
Remove
Imported Data:
Componen 3
2 MOVES
Run
3. Results
~
View
Figure Al-12 Emission estimation panel.
Traffic Analysis
Congestion Analysis
Run
4
View
Identificati'
Please specify KJentofccafcon cntan-i
Oefiutt
V Daily- link volume >•
S3
window will be displayed. The window contains default and optional criteria for identifying
congestion areas. Default criteria include the Daily Link Volume (equal to or larger than
125,000 passenger cars) and
Truck Fraction (equal or
larger than 8%). Optional
criteria include the Average
speed, Delay, Queue
length, D/C ratio, CO2
equivalent, and Energy
consumption. In current
version of the AIR-
SUSTAIN system, only the
default criteria are used.
When criteria are set up, the
traffic congestion link
identification function is
performed by clicking the
OK button.
~ "ruck fraction >-
OjXKtfsst
[H Average speed <•
I I Oefay >•
PI Queue lengih >¦
i ] DC (SilO >-
~ C02 equivalent >•
O Energy consumption >"
125000
COS
OK
Cancel
Figure Al-13 Traffic Congestion identification.
198
-------
In Component 1 of the Microscopic Simulation Results Import panel (as shown in
Figure Al-15), the user needs to first import the Microscopic Simulation Link ID profile,
which records a map between links in microscopic simulation network and links in VISUM.
Then the micro-simulation results can be imported through Component 2. To this end, the user
needs to set up names of traffic control measures in item a and load simulation results in item b.
This process in Component 2 can be repeated if there are multiple files of simulations results to
import. Imported files are listed on Component 3. A function of removing imported data is also
provided in Component 4. Those simulation results by different traffic control measures can be
compared and displayed by clicking the View Results button in Component 5.
With those imported results, emission estimate for traffic congestion area is recalculated
by the emission estimation model, which is similar to the functions of emission estimation in the
regional level analysis. The emission model can be directly executed without requiring users to
import extra data. In fact, the traffic inputs of the emission model are automatically prepared
based on the micro-simulation results and the non-traffic inputs are retrieved from regional
analysis database. Figure Al-16 shows the panel of Congested Area Emission Estimation.
Airjmpact Relating Scenario-based Urban Settings and Transportation Assets In Network
- ~ ~
<
Scenario De»gn Re^onal Level Analyse Prq^ct Level fcntfa Rnii Companion
Conq Area Identification
Conq Area Identification Run
Mookock: Smitten Resets knport
1 High Resdutron U* 10
2 TrafSc Control Management
a Management Strategy
b Load Files
3 Compansion W
Project Emission Estimation
Please select Emission Es»masen Model
moves emf/c
1 MOVES
2 Results
Regcnal Em**ons Update
1 Traffic Control Management Strategy Name
2 Strategy Results
Run
Imported Data
Modeling Yoor
my
Process Status
Project Informa&on
Modeling Year Selectron
Base Yew Data
Target Year Scenario Oes»gn
Travel Demand Forecasting
Emission EsbmatKm
Cong. Area Identification
Microscopic Suniabon Result
Project Emission Estimation
Regwnal Emissions Update
Figure Al-14 Project level analysis interface.
199
-------
Load File
Microscopic Simulation Results Import OIT1 pOllCnt 1
1. Microscopic Simulation Link ID
Traffic taftrfrl Metres
a. Measure N&«|11 pone lit 2
b, Load Files Load File
3. G
"TWflx.nent
View Reults
Imported Data:
Component 3
Remove Data
Component 4
Figure AI -15 Microscopic simulation results import.
The Project Level Analysis panel provides a function to feedback the effects of different
traffic control measures to the regional level results database. This is achieved by updating traffic
and emission results of the congestion links to corresponding regional links. Figure Al-17 shows
the panel of this function. To perform this function, the emission results by which traffic control
measures should be chosen first in the drop list of the Traffic Control Measure Name. Then the
emission results are exported to the regional database by clicking on the Update button. The
results can be displayed by clicking the View button after selecting the emission type in the drop
list of the Results.
Project Emission Estimation
Please select Emission Esbmation Mode):
MOVES EMFAC
j 1 MOVES
: 2. Results
Run
Wew
I
Figure Al-16 Traffic congestion area emission estimation.
Regional Emissons Update
1 - Traffic Control Measue Name
2 Results
Update
View
Figure Al-167 Updating traffic congestion identification results for regional
analysis.
200
-------
4. Result Comparison
After performing the scenario design, regional level analysis, and project level analysis,
results from the base year and target year can be compared and visualized in ArcGIS by the
Results Comparison tab (Figure Al-18). Those visualized results include:
m
| a |[ B II S3 I
Scenario Data
mr
WWnjpaglRelating^cenariq^based'Urban Settings and Transportation Assets In Network
-
Scenario Development Regional Level Analysis Project Level Analysis Results Comparision
Land Use
Land Use
Modeling Year
O Data of modeling years
yyyy
v View Res
O Changes between modeling years
Process Status
Demographic and Socioeconomic Data
O Data of modeling years
v Vie//
Changes between modeling years
Base Year Date
Travel Demand Forecasting Results
Travel Demand Forecasting Data
Target Year Scenario Design
O Data of modeling years
Travel Demand Forecasting
v View
Changes between modeling years
Emission Estimation
Emission Estimation Results
Hots pots Idetification
Emission Estimation
O Data of modeling years
MicroscoDic Simulation Result*
v View
O Changes between modeling years
Hotspots Emission Estimation
Regional Emissions Update
Figure Al-17 Results comparison interface.
¦ Land Use, contains land use in the base year and target year
¦ Demographic and Socioeconomic Data, includes population, household, employment,
university enrollment, high school enrollment
¦ Travel Demand Forecasting Results, views link traffic information including volume,
average speed, D/C ratio, delay, and queue length
¦ Emission Estimation, includes CO2 Equivalent and energy consumption
By clicking the View Results button in the Land Use panel, base year and target year
land uses are displayed in ArcGIS. In the Demographic and Socioeconomic Results panel,
demographic and socioeconomic data are displayed in ArcGIS by clicking the View button when
either the Data of modeling year or Changes between modeling years is checked. The Travel
201
-------
Demand Forecasting Results panel provides options to view the results of base year and target
year or the difference between base year and target year by checking the Data of modeling year
or Changes between modeling years. Similar functions are also provided in the Emission
Estimation Results panel.
A 1.1.2 Inputs and outputs
The summarized data items of the inputs and outputs for the AIR-SUSTAIN are listed in
Tables Al-5 (inputs) and Al-6 (outputs).
Table Al- 5 Inputs for the AIR-SUSTAIN
Module
Function
Data Item
Format
Scenario
Development
Base Year Data
TAZ
Feature Class
RoadNetwork
Feature Class
Incentiveboundary
Feature Class
Assumed Changes in
Demographic and
Socioeconomic Factors
Population Change
Numbers or Excel
Employment Change
Numbers or Excel
University Enrollment Change
Excel
High School Enrollment
Change
Excel
Land Use Projection
Land Use Inputs Directory
Folder
Regional
Level
Analysis
Travel Demand
Forecasting
VISUM File
Version File
Household Fraction and Trip
Rate
Excel
Employment Fraction
Excel
Emission Estimation
Age Distribution
Excel
Fuel Supply
Excel
Fuel Formulation
Excel
Meteorology
Excel
County and State
Excel
Project Level
Analysis
Microscopic Simulation
Results Import
Microscopic Simulation Link ID
Excel
Microscopic Simulation Results
Excel
202
-------
Table Al- 6 Outputs from the AIR-SUSTAIN
Data Item
Description
Data Source
Format
Landuse_Target
Target year land use
Land use
projection
Feature
class
TargetYearTAZ
Target year population, household,
employment, university enrollment, high
school enrollment, residential area,
employment area
Linkage model
Feature
class
BaseTDF/
TargertTDF
Base/Target year travel demand
forecasting results from trip generation,
trip distribution, mode split, and traffic
assignment
TDF
Version file
RoadNetwork
Base year traffic information including
volume, average speed, D/C ratio, delay,
and queue length
TDF
Feature
class
Target year traffic information including
volume, average speed, D/C ratio delay,
and queue length
TDF
Feature
class
Base year emission results including C02
equivalent and energy consumption
Emission
estimation
Feature
class
Target year emission results including
CO2 equivalent and energy consumption
Emission
estimation
Feature
class
Target year emission results including
CO2 equivalent and energy consumption
updated with project level emission
results
Regional
emission Update
Feature
class
Congestion
Links meet Traffic congestion
identification criteria
T raffic
congession
identification
Feature
class
Inputs and Outputs for the Scenario Development
Inputs for the scenario development are summarized in Table Al-7. Details of land use
inputs are listed in Table Al-8. Outputs for the development are listed in Table Al-9.
203
-------
Table Al- 7 Inputs for the Scenario Development
Date Item
Field
Description
Type
TAZ
TAZ
TAZ name, the format is TAZ_TAZ
Number, i.e. TAZ_151
String
TAZ_N
TAZ number
Integer
TAZ_Order
The field to link TAZs in ArcGIS and
TAZs in VISUM
Integer
POP
Target year population
Integer
HH
Target year household
Integer
EMP
Target year employment
Integer
HI
Target year high school enrollment
Integer
UN
Target year university enrollment
Integer
AREA_TYPE
1=CBD&Urban ; 2=suburban; 3= rural
Integer
GEOCODE_Base_1
Base year residential area (ft2)
Double
GEOCODE_Base_2
Base year employment area (ft2)
Double
GEOCODE_Base_3
Base year institutional area (ft2)
Double
GEOCODE_Base_4
Base year undeveloped area (ft2)
Double
GEOCODE_Base_5
Base year other area (ft2)
Double
GEOCODE_1
Target year residential area (ft2)
Double
GEOCODE_2
Target year employment area (ft2)
Double
GEOCODE_3
Target year institutional area (ft2)
Double
GEOCODE_4
Target year undeveloped area (ft2)
Double
GEOCODE_5
Target year other area (ft2)
Double
RoadNetwork
NO
Link number
Integer
Length
Link length (mile)
Double
to
Free flow travel time (s)
Double
Incentive boundary
Name
Incentive area name
String
Shape_Area
Area (mile2)
Double
Population
Change
TAZ
TAZ number
Integer
Population Change
Population change
Integer
Employment
Change
TAZ
TAZ number
Integer
Employment Change
Employment change
Integer
University
Enrollment
Change
TAZ
TAZ number
Integer
University Enrollment
Change
University enrollment change
Integer
Name
University name(s)
String
204
-------
Date Item
Field
Description
Type
High School
Enrollment
Change
TAZ
TAZ
Integer
High School Change
High school enrollment change
Integer
Name
High school name(s)
String
Land Use Inputs
See details in Table 3.4
See details in Table 3.4
N/A
ft, feet
Table Al- 8 Land Use Inputs
Name
Description
Format
21_natural_restrictions
Base year natural restricted areas
Raster
22_administrative_restrictions
Base year administrative restricted areas
Raster
231_residential_zoning
Residential zoning
Raster
234_employment_zoning
Employment zoning
Raster
30_incentive
User specified incentive layer
Raster
111_population_change
Suitability of Average Annual Percentage Change (AAPC)
of population from initial year to base year
Table
112_employment_change
Suitability of Average Annual Percentage Change (AAPC)
of employment from initial year to initial year
Table
113_median_income
Base year median income suitability
Table
114_distance_to_freeway
Base year Distance to Freeway exits suitability
Table
115_distance_to_transit
Base year walkable distance to transit stops suitability
Raster
116_VCratio
Base year D/C ratio suitability
Table
117_CarbonEmission
Base year carbon emission suitability
Table
118_slope
Slope suitability
Table
119_distance_to_employment
Base year distance to employment land suitability
Raster
120_distance_to_residential
Base year distance to residential land suitability
Raster
121_distance_to_vacant
Base year distance to vacant land suitability
Raster
01_ilu
Initial year land use
Raster
02_blu
Base year land use
Raster
Residential suitability image
Base year residential land use suitability score image
Raster
Employment suitability image
Base year employment land use suitability score image
Raster
Institution suitability image
Base year institution land use suitability image
Raster
Undeveloped suitability image
Base year undeveloped land use suitability score image
Raster
Others suitability image
Base year other land use suitability score image
Raster
205
-------
Table Al- 9 List of Outputs from the Scenario Development
Name
Field
Description
Type
TargetYearTAZ
TAZ
TAZ name, the format is TAZ TAZ Number,
i.e. TAZ_151
String
TAZ_N
TAZ number
Integer
TAZ_Order
The field to link TAZs in ArcGIS and TAZs in
VISUM
Integer
POP
Target year population
Integer
HH
Target year household
Integer
EMP
Target year employment (employee)
Integer
HI
Target year high school enrollment
Integer
UN
Target year university enrollment
Integer
AREA_TYPE
CBD&Urban or suburban or rural
Integer
GEOCODE_Base_1
Base year residential area (ft2)
Double
GEOCODE_Base_2
Base year employment area (ft2)
Double
GEOCODE_Base_3
Base year institutional area (ft2)
Double
GEOCODE_Base_4
Base year undeveloped area (ft2)
Double
GEOCODE_Base_5
Base year other area (ft2)
Double
GEOCODE_1
Target year residential area (ft2)
Double
GEOCODE_2
Target year employment area (ft2)
Double
GEOCODE_3
Target year institutional area (ft2)
Double
GEOCODE_4
Target year undeveloped area (ft2)
Double
GEOCODE_5
Target year other area (ft2)
Double
Landuse_Target
GEOCODE
Land use type, Presidential,
2=employment, 3=institutional,
4=undeveloped, 5=other
Integer
Shape_Area
Area (ft2)
Double
ft - feet
1) Inputs and Outputs for the Regional Level Analysis
Inputs for the regional level analysis include the Household Fraction and Trip Rate,
Employment Fraction, VISUM File for travel demand forecasting, Age Distribution, Fuel Supply,
Fuel Formulation, Meteorology, County and State for emission estimation. Fields and data type
206
-------
of those input files are listed in Table Al-10, and details about a VISUM file are provided in
Table Al-11. Output files for the regional level analysis are listed in Table Al-12.
Table Al- 10 Inputs for the Regional Level Analysis
Name
Field
Description
Type
HH_ID
Household ID
Integer
Fraction
The portion of households whose
ID=HH_ID to the total number of
households
Double
Household Fraction and
HBO_Rate
The trip generation rate by trip
purpose HBO
Double
Trip Rate
HBW_Rate
The trip generation rate by trip
purpose HBW
Double
HBSC_Rate
The trip generation rate by trip
purpose HBSC
Double
HBU_Rate
The trip generation rate by trip
purpose HBU
Double
TAZ_N
TAZ number
Integer
Employment Fraction
LEMP
The rate of low trip rate
employment in TAZ_N
Double
MEMP
The rate of medium trip rate
employment in TAZ_N
Double
HEMP
The rate of high trip rate
employment in TAZ_N
Double
VISUM File
See details in Table 4-7
See details in Table 4-7
See
details
in Table
4-7
Age Distribution
SourceType
11=Motorcycle; 21= Passenger
Car; 31= Passenger Truck;
32=Light Commercial Truck;
41=lntercity Bus; 42=Transit Bus;
43=School Bus; 51=Refuse
Truck; 52=Single Unit Short-haul
Truck; 53=Single Unit Long-haul
Truck; 54 Motor Home;
61=Combination Short-haul
Truck; 62=Combination Long-
haul Truck
Integer
YearlD
Calendar year
Integer
AgelD
Age
Integer
AgeFraction
Distribution of AgelDs
Double
countylD
County
Integer
207
-------
Name
Field
Description
Type
fuelYearlD
Fuel year
Integer
monthGroupID
Fuel month
Integer
Fuel Supply
fuelFormulationID
Fuel formulation identification
number. Must be greater than
100 and less than 25000
Integer
markets ha re
Market share
Double
marketShareCV
Null
Double
fuelFormulationID
Fuel formulation identification
number. Must be greater than
100 and less than 25000
Integer
fuelSubtypelD
Fuel Sub-type coding
Integer
RVP
Reid vapor pressure in psi
Integer
sulfurLevel
Fuel sulfur level in ppm Sulfur
Integer
ETOHVolume
Ethanol Volume (% vol)
Double
MTBEVolume
MTBE Volume (% vol)
Double
ETBEVolume
ETBE Volume (% vol
Double
TAMEVolume
TAME Volume (% vol)
Double
aromaticContent
Aromatic content (% wt)
Double
Fuel Formulation
olefinContent
Olefin content (% wt)
Double
benzeneContent
Benzene content (% wt)
Double
e200
Lower volatility percentage (%)
Integer
e300
Upper volatility percentage (%)
Integer
volToWtPercentOxy
Constant based on oxygenate
type
Double
BioDieselEsterVolume
BioDiesel Ester Volume (%)
Double
Cetanelndex
NULL
NULL
PAHContent
NULL
NULL
T50
Temperature (F) where 50% of
the fuel is vapor
Integer
T90
Temperature (F) where 90% of
the fuel is vapor
Integer
mo nth ID
Calendar month
Integer
zonelD
Zone
Integer
Meteorology
hourlD
Hour
Integer
temperature
Temperature
Double
relHumidity
Humidity
Double
208
-------
Name
Field
Description
Type
County and State
State Name
State Name
Integer
County Name
County Name
Integer
vol, volume; wt, weight
Table Al- 11 Description of a VISUM File
Data
Description
Format
TAZ
A shapefile of traffic analysis zones. Attributes which need to
be created by users include TAZ Num, HH ID 1, HH ID 2,
HH ID 3, HH ID 4, HH, EMP, LEMP, MEMP, HEMP,
HIEN, and UNEN.
Polygon
Road Network
A shapefile of street center lines. Attributes which need to be
specified by users are Cap (capacity), V0 (free flow speed).
Line
Transit Network
A shapefile of transit lines and transit stops.
Matrices
Matrices store number of trips, and travel times between
TAZs
Table
a. Trip Generation
The trip generation in VISUM is implemented by connecting
the socioeconomic data to its corresponding trip production
and attraction rates. Trip production rates and attraction
rates including NHBPRATEato),mx, LEMPR4TEI,
MEMPRATEj, HEMPRATEj, HHRATEj, URATEj, and
HRAIE](in Eq. 3.17 and Eq. 3.18 in Section 3.3) are
required.
b. Trip Distribution
Four Step Model Parameters
The trip distribution in VISUM is implemented by assigning
an appropriate distribution model in VISUM. An iterative
procedure is employed to refine trip interchange estimates
until convergence is met, i.e., the estimated zonal trip ends
attracted to each zone closely match the desired zonal trip
attractions calculated in the trip generation phase. To
implement this, the parameter a in Eq. 3.19 needs to be set
up, and k\\ is automatically estimated by balancing the trip
generations and attractions.
c. Mode Split
The trip mode choice in VISUM is implemented by assigning
an appropriate mode split model. The mode split model
adapts the Logit utility function with parameters of >8 in Eq.
3.20.
N/A
209
-------
Data
Description
Format
d. Trip Assignment
The equilibrium traffic assignment utilizes the Wardrop's first
principle and breaks the OD demand matrix into the
proportions per iteration step. The traffic assignment
procedure is an iterative step where a proportion of traffic
will be assigned in each iteration until convergence criteria
meets.
Table Al- 12 Outputs from the Regional Level Analysis
Name
Field
Description
Type
TargetYearTAZ
TAZ
TAZ name, the format is TAZ_TAZ
Number, i.e. TAZ_151
String
TAZ_N
TAZ number
Integer
TAZ_Order
The key field to link TAZs in ArcGIS and
TAZs in VISUM
Integer
POP
Target year population
Integer
HH
Target year household
Integer
EMP
Target year employment
Integer
HI
Target year high school enrollment
Integer
UN
Target year university enrollment
Integer
AREA_TYPE
1=CBD&Urban; 2= suburban; 3= rural
Integer
GEOCODE_Base_1
Base year residential area (ft2)
Double
GEOCODE_Base_2
Base year employment area (ft2)
Double
GEOCODE_Base_3
Base year institutional area (ft2)
Double
GEOCODE_Base_4
Base year undeveloped area (ft2)
Double
GEOCODE_Base_5
Base year other area (ft2)
Double
GEOCODE_1
Target year residential area (ft2)
Double
GEOCODE_2
Target year employment area (ft2)
Double
GEOCODE_3
Target year institutional area (ft2)
Double
GEOCODE_4
Target year undeveloped area (ft2)
Double
GEOCODE_5
Target year other area (ft2)
Double
GEOCODE
Land use type, Presidential,
2=employment, 3=institutional,
4=undeveloped, 5=other
Double
Shape_Area
Area (ft2)
Double
210
-------
Name
Field
Description
Type
Volume_Base/
Volume_Target
Base/ Target year volume (veh)
Double
Speed_Base/ Speed_Target
Base/ Target year speed (mile/h)
Double
DCRatio_Base/ DCRatio
_Target
Base/ Target year demand/capacity ratio
(%)
Double
RoadNetwork
Delay_Base/ Delay_Target
Base/ Target year delay (min/veh/mile)
Double
QueueLength_Base/
QueueLength _Target
Base/Target year queue length (veh/mile)
Double
C02_Equivalent_Base/
C02_Equivalent_Target
Base/ Target year C02 equivalent (kg)
Double
Energy_Consumption_Base/
Energy_Consumption
_Target
Base/ Target year energy consumption (kJ)
Double
Vehicle
LINKID
Link number
Integer
Composition
Car fraction
The fraction of cars to the total number of
vehicles on a link
Double
ft, feet; veh, vehicle
2) Inputs and Outputs for the Project Level Analysis
Details about inputs and outputs for the project level analysis are listed in Table Al-13
(inputs) and Table Al-14 (outputs).
Table Al- 13 Inputs for the Project Level Analysis
Name
Sheet
Field
Description
Type
Microscopic
Simulation Links
Sheetl
GISLinkID
Link ID in RoadNetwork
Integer
Sheetl
MicroscopicLinkID
Link ID in microscopic simulation
Integer
Sheet2
GISLinkID
Link ID in RoadNetwork
Integer
Sheet2
RoadType
(1 =Off-Network;
2=Rural Restricted Access;
3=Rural Unrestricted Access;
4=Urban Restricted Access;
5=Urban Unrestricted Access)
Integer
Sheet2
LinkLength
Link length (mile)
Double
Sheet2
LinkGrade
Link grade
Double
Sheetl
MicroscopicLinkID
Link ID in microscopic simulation
Integer
Sheetl
time (sim sec)
Time stamp
Integer
211
-------
Microscopic
Simulation
Results
VISUM File
Sheetl
Car#
Car number
Integer
Sheetl
Car v
Car speed (mile/sec)
Double
Sheetl
Car a
Car acceleration (mile/sec2)
Double
Sheetl
Truck #
Truck number
Integer
Sheetl
Truck v
Truck speed (mile/sec)
Double
Sheetl
Truck a
Truck acceleration (mile/sec2)
Double
Sheet2
MicroscopicLinkID
Link ID in microscopic simulation
Double
Sheet2
AverageSpeed
Average speed (mile/h)
Double
Sheet2
Delay
Delay (min/vehicle)
Double
Sheet2
QueueLength
Queue length (vehicle)
Double
Veh, vehicle
Table Al- 14 Outputs from the Project Level Analysis
Name
Field
Description
Type
TAZ
TAZ name, the format is TAZ_TAZ
Number, i.e. TAZ_151
String
TAZ_N
TAZ number
Integer
TAZ_Order
The field to link TAZs in ArcGIS and
TAZs in VISUM
Integer
POP
Target year population
Integer
HH
Target year household
Integer
EMP
Target year employment
Integer
HI
Target year high school enrollment
Integer
UN
Target year university enrollment
Integer
AREA_TYPE
1=CBD&Urban; 2= suburban; 3= rural
Integer
GEOCODE_Base_1
Base year residential area (ft2)
Double
GEOCODE_Base_2
Base year employment area (ft2)
Double
GEOCODE_Base_3
Base year institutional area (ft2)
Double
GEOCODE_Base_4
Base year undeveloped area (ft2)
Double
GEOCODE_Base_5
Base year other area (ft2)
Double
GEOCODE_1
Target year residential area (ft2)
Double
GEOCODE_2
Target year employment area (ft2)
Double
GEOCODE_3
Target year institutional area (ft2)
Double
GEOCODE_4
Target year undeveloped area (ft2)
Double
GEOCODE_5
Target year other area (ft2)
Double
212
-------
Name
Field
Description
Type
Landuse_Target
GEOCODE
Land use type, Presidential,
2=employment, 3=institutional,
4=undeveloped, 5=other
Integer
Shape_Area
Area (f2)
Double
RoadNetwork
C02_Equivalent_Updated
CO2 equivalent (kg)
Double
Energy_Consumption_Updated
Energy consumption (kJ)
Double
ft, feet
A 1.2 Database in AIR-SUSTAIN
A 1.2.1 Database Structure
All model outputs and intermediate data are stored in five MySQL databases and an
ArcGIS Geodatabase. The five MySQL databases are automatically generated for each AIR-
SUSTAIN scenario. They are AIR-SUSTAIN scenario database, regional MOVES input
database, regional MOVES output database, project-level MOVES input database, and project-
level MOVES output database. The fields and data sources of data tables in the MySQL
databases are listed in Table A2-1.
Table A2- 1 Tables and fields of AIR-SUSTAIN Scenario Database
Table Name
Fields
Data Source
scenariolD
Specified by the user in the AIR-SUSTAIN GUI
baseYear
Specified by the user in the AIR
targetYear
Specified by the user in the AIR
analyst
Specified by the user in the AIR
projectlnfo
date
Specified by the user in the AIR
projectDir
Specified by the user in the AIR
idrisiDir
Specified by the user in the AIR
movesDir
Specified by the user in the AIR
projectDescription
Specified by the user in the AIR
Increase Percentage
Poplncrease
Specified by the user in the AIR
Emplncrease
Specified by the user in the AIR
PopulationGrowth
TAZ
Imported from Population Change
Population
Imported from Population Change
EmploymentGrowth
TAZ
Imported from Employment Change
Employment
Imported from Employment Change
213
-------
Table Name
Fields
Data Source
UniversityEnrollment
TAZ
Imported from University Enrollment Change
Enrollment
Imported from University Enrollment Change
Name
Imported from University Enrollment Change
HighSchoolEnrollment
TAZ
Imported from High School Enrollment Change
Enrollment
Imported from High School Enrollment Change
Name
Imported from High School Enrollment Change
HouseholdTripRate
TAZ
Imported from Household Fraction and Trip Rate
Fraction
Imported from Household Fraction and Trip Rate
HBO
Imported from Household Fraction and Trip Rate
HBW
Imported from Household Fraction and Trip Rate
HBSC
Imported from Household Fraction and Trip Rate
HBU
Imported from Household Fraction and Trip Rate
Employment! ripRate
TAZ
Imported from Employment Fraction
LowRate
Imported from Employment Fraction
MediumRate
Imported from Employment Fraction
High Rate
Imported from Employment Fraction
BaseYearResults
LinkID
Output from TDF
Volume
Output from TDF
FunctionClass
Output from TDF
AvgSpeed
Output from TDF
DCRatio
Output from TDF
Delay
Output from TDF
QueueLength
Output from TDF
TruckFraction
Output from TDF
C02 Equivalent
Output from emission estimation
EnergyConsumption
Output from emission estimation
BaseYearResults
LinkID
Outputted from TDF
Volume
Outputted from TDF
FunctionClass
Outputted from TDF
(1 =Off-Network;
2=Rural Restricted Access;
3=Rural Unrestricted Access;
4=Urban Restricted Access;
5=Urban Unrestricted Access)
214
-------
Table Name
Fields
Data Source
AvgSpeed
Output from TDF
DCRatio
Output from TDF
Delay
Output from TDF
QueueLength
Output from TDF
TruckFraction
Output from TDF
C02 Equivalent
Output from emission estimation
EnergyConsumption
Output from emission estimation
IDIRSIInfo
InitialYear
Specified by the user in AIR-SUSTAIN GUI
Folderpath
Specified by the user in AIR-SUSTAIN GUI
TDF, Travel Demand Forecasting
The data tables for the regional and project-level MOVES inputs in the databases are the
same. In addition, the data tables of regional and project-level MOVES outputs in the databases
are also the same. The data tables and fields of MOVES input database are listed in Table A2-2.
The data table and included items of the MOVES output database are illustrated in Table A2-3.
Table A2- 2 Tables and Items of the MOVES Input Database
Table Name
Items
Data Source
link
linkID
TDF
countylD
Specified by the user
zonelD
MOVES database
roadTypelD
TDF
linkLength
TDF based on geometry input
linkVolume
TDF or microscopic simulation data source
linkAvgSpeed
TDF or microscopic simulation data source
linkDescription
N/A
linkAvgGrade
Calculated by AIR-SUSTAIN based on geometry
input
linksource-
typehour
linkID
TDF
sourceTypelD
TDF
sourceTypeHourFraction
TDF or microscopic simulation data source
opmode-
distribution
sourceTypelD
MOVES database
hourDaylD
TDF or micro-simulation data source
linkID
TDF
polProcessID
MOVES database
215
-------
Table Name
Items
Data Source
opModelD
MOVES database
opModeFraction
Calculated by AIR-SUSTIAN based on traffic
input
opModeFractionCV
Null
isUserlnput
Null
countylD
From State and County
state ID
From State and County
CountyName
From State and County
County
altitude
MOVES database
GPAFract
MOVES database
barometricPressure
MOVES database
barometricPressureCV
Null
sourceTypelD
MOVES database
sourcetypeage-
yearlD
Specified by user
distribution
agelD
From Age Distribution
ageFraction
Non-traffic input
state ID
From State and County
state
stateName
From State and County
stateAbbr
From State and County
yearlD
From input base year or target year
year
isBaseYear
Calculated by AIR-SUSTAIN based on traffic
input
fuelYearlD
From file Fuel Supply
zonelD
MOVES database
countylD
From State and County
zone
startAllocFactor
1
idleAllocFactor
1
SHPAIIocFactor
1
monthID
Specified by the user
zonelD
MOVES database
hourlD
TDF
zone-
monthhour
temperature
From Meteorology
temperatureCV
Null
relHumidity
From Meteorology
heatlndex
Null
specificHumidity
Null
relativeHumidityCV
Null
zonelD
MOVES database
zoneroadtype
roadTypelD
TDF
SHOAIIocFactor
1
fuel-formulation
fuelFormulationID,
fuelSubtypelD, RVP, sulfurLevel,
ETOHVolume, MTBEVolume,
TAMEVolume, aromaticContnet,
From Fuel Formulation
216
-------
Table Name
Items
Data Source
olefinContent, benzeneContent,
e200,2300,
volT oWtPercentOxy,
BioDieselEsterVolume,
Cetanelndex, PAHContent, T50,
T90
fuelsupply
countylD, fuelYearlD,
monthGroupID,
fuelFormulationID, marketShare,
marketShareCV
From Fuel Supply
Table A2- 3 Tables and Items of the MOVES Output Database
Table Name
Items
Description
activitytype
activityTypelD, activityType, activityTypeDesc,
This table lists the activity types
that can be reported in the
movesactivityoutput table and
provides their activitytypeid
(1= distance traveled; 2=source
hours; 3= source hour idling; 4=
source hours operating; 5= source
hours parked; 6= population; 7=
starts).
bundletracking
hostType, MOVESRunID,
loopableClassName, workerVersion,
workerCompterlD, workerlD, bundleNumber,
isCleanUp, iterationID, processID,
roadTypelD, linkID, zonelD, countylD, statelD,
yearlD, monthID, daylD, HourlD,
executionGranularity
This table contains information
about data that is processed by the
MOVES master and workers.
movesactivityout
put
MOVESRunID, iterationID, yearlD, monthID,
daylD, hourlD, statelD, countylD, zonelD,
linkID, sourceTypelD, fuelTypelD,
modelYearlD, roadTypelD, SCC
This table provides information on
the vehicle activity generated and
run.
moveserror
MOVESError, MOVESRunID, yearlD,
monthID, daylD, hourlD, statelD, zonelD,
linkID, pollutantID, processID, errorMessage
This table contains any error
messages or diagnostic
information that might be
generated if the MOVES run is
unsuccessful.
moveseventlog
EventRecordID, MOVESRunID, EventName,
WhenStarted, WhenStoped, Durantion
This table stores diagnostic
results.
movesoutput
MOVESRunID, iterationID, yearlD, monthID,
daylD, hourlD, statelD, countylD, zonelD,
linkID, pollutantID, processID, sourceTypelD,
fuelTypelD, modelYearlD, roadTypelD, SCC
This table contains the inventory
emission results of the run
disaggregated by parameters,
such as Year, Month, etc.
movesrun
MOVESRunID, outputTimePeriod, timeUnits,
distanceUnits, massUnits, energyUnits,
runSpecFileName, runSpecDescription,
runSpecFileDateTime, runDataTime, scale,
minutesDuration, defaultDatabaseUsed,
masterVersionDate, masterComputerlD,
The table contains information
about the date and time of the run,
information about the run
specifications, and the name of the
units in which MOVES outputs are
represented.
217
-------
Table Name
Items
Description
masterlDNumber, domain, domainCountylD,
domainCountyName, domainDatabaseServer,
domainDatabaseName, expectedDONEFiles,
retrivedDONEFiles
movestableused
MOVESRunID, databaseServer,
databaseName, tableName, dataFileSize,
dataFileModificationDate, tableUseSequence
This table stores a list of the tables
used when executing MOVES.
movesworkerus
ed
MOVESRunID, wokerversion,
wokerComputerlD, workerrlD, bundleCount,
failedBundleCount
This table contains information as
to which copy of the MOVES
Worker Program processed
portions of the run.
rate perdista nee
MOVESScenariolD, MOVESRunID, yearlD,
monthID, daylD, hourlD, linkID, pollutantID,
processID, sourceTypelD, SCC, fuelTypelD,
modelYearlD, riadTypelD, avgSpeedBinID,
temperature, relHumidity, ratePerDistance
This table stores emissions as
rates per distance with the units
depending on those selected on
run specification.
rateperfrofile
MOVESScenariolD, MOVESRunID,
temperatureProfile, yearlD, daylD, hourlD,
pollutantID, processID, sourceTypelD, SCC,
fuelTypelD, modelYearlD, temperature,
rateperVehicle
This table stores vapor venting
emissions from parked vehicles as
rates per vehicle.
ratepervehicle
MOVESScenariolD, MOVESRunID, yearlD,
daylD, hourlD, pollutantID, processID,
sourceTypelD, SCC, fuelTypelD,
modelYearlD, temperature, rateperVehicle
This table stores vapor venting
emissions from starts and
extended idle, and some
evaporative emissions from parked
vehicle as rates per vehicle.
All feature classes, including TAZ, TargetYearTAZ, RoadNetwork, Incentiveboundary,
Congestion Area, and LanduseTarget, are stored in a Geodatabase. The feature class names and
items of each feature class are illustrated in Table A2-4.
Table A2- 4 Geodatabase
Feature class
Item
Source
FID
Generated by ArcGIS automatically
TAZ
Specified by the user
TAZ_N
Specified by the user
TAZ_Order
The key to link TAZs in feature class and TAZs in
VISUM
TAZ
POP
Specified by the user
HH
Specified by the user
EMP
Specified by the user
HI
Specified by the user
UN
Specified by the user
AREA_TYPE
Specified by the user
218
-------
Feature class
Item
Source
GEOCODE_1
Specified by the user
GEOCODE_2
Specified by the user
GEOCODE_3
Specified by the user
GEOCODE_4
Specified by the user
GEOCODE_5
Specified by the user
FID
Generated by ArcGIS automatically
Incentive boundary
Shape_Area
Generated by ArcGIS after the user specify the
boundary
FID
Generated by ArcGIS automatically
TAZ
Specified by the user
TAZ_N
Specified by the user
TAZ_Order
The key to link TAZs in feature class and VISUM
POP
From linkage model
HH
From linkage model
EMP
From linkage model
TargetYearTAZ
HI
From linkage model
UN
From linkage model
AREA_TYPE
From linkage model
GEOCODE_1
From land use projection
GEOCODE_2
From land use projection
GEOCODE_3
From land use projection
GEOCODE_4
From land use projection
GEOCODE_5
From land use projection
FID
Generated by ArcGIS automatically
Landuse_Target
GEOCODE
From survey data
Shape_Area
From survey data
FID
Generated by ArcGIS automatically
NO
Specified by users
Length
Specified by the user
RoadNetwork
Volume_Base
Output from TDF
Speed_Base
Output from TDF
DCRatio_Base
Output from TDF
Delay_Base
Output from TDF
QueueLength_Base
Output from TDF
219
-------
Feature class
Item
Source
C02_Equivalent_Base
Output from emission estimation
Energy_Consumption_Base
Output from emission estimation
Volume_Target
Output from TDF
Speed_Target
Output from TDF
DCRatio_Target
Output from TDF
Delay_Target
Output from TDF
Queuel_ength_Target
Output from TDF
C02_Equivalent_Target
Output from emission estimation
Energy_Consumption_Target
Output from emission estimation
C02_Equivalent_Update
Output from TDF
Energy_Consumption_Update
Output from TDF
FID
Generated in ArGIS
NO
From RoadNetwork
Length
From RoadNetwork
Volume_Base
From RoadNetwork
Speed_Base
From RoadNetwork
DCRatio_Base
From RoadNetwork
Delay_Base
From RoadNetwork
QueueLength_Base
From RoadNetwork
Congestion_Area
C02_Equivalent_Base
From RoadNetwork
Energy_Consumption_Base
From RoadNetwork
Volume_Target
From RoadNetwork
Speed_Target
From RoadNetwork
DCRatio_Target
From RoadNetwork
Delay_Target
From RoadNetwork
QueueLength_Target
From RoadNetwork
C02_Equivalent_Target
From RoadNetwork
Energy_Consumption_Target
From RoadNetwork
TDF, Travel Demand Forecasting
220
-------
A 1.2.2 MOVES Emission Lookup Tables
The MOVES model incorporates similar regression-based equations for mean and
variance model for braking/deceleration and uses similar approach of heavy-duty vehicles. The
vehicle activity mix is determined by the emission source type, age group, road type and
operating mode distribution. The lookup tables for emission source type, road type, and vehicle
age distribution are presented in Table A2-5 to A2-8.
Emission Source Type
Table A2- 5 The MOVES Source Types
Source Type
ID
Source Type Name
HPMS Vehicle Type
ID
HPMS Vehicle Type
Name
11
Motorcycle
10
Motorcycles
21
Passenger Car
20
Passenger Cars
31
Passenger Truck
30
Other 2 axle-4 tire vehicles
32
Light Commercial Truck
30
Other 2 axle-4 tire vehicles
41
Intercity Bus
40
Buses
42
Transit Bus
40
Buses
43
School Bus
40
Buses
51
Refuse Truck
50
Single Unit Trucks
52
Single Unit Short-haul Truck
50
Single Unit Trucks
53
Single Unit Long-haul Truck
50
Single Unit Trucks
54
Motor Home
50
Single Unit Trucks
61
Combination Short-haul Truck
60
Combination Trucks
62
Combination Long-haul Truck
60
Combination Trucks
Road Type
Table A2- 6 The MOVES Road Type
Road Type ID
Road Type Description
1
Off-Network
2
Rural Restricted Access
3
Rural Unrestricted Access
221
-------
Road Type ID
Road Type Description
4
Urban Restricted Access
5
Urban Unrestricted Access
Vehicle Age Distribution
Table A2- 7 MOVES Age Distribution Categories
agelD
ageCategoryName
agelD
ageCategoryName
0
new
21
21 years old
1
one year old
22
22 years old
2
two years old
23
23 years old
3
three years old
24
24 years old
4
four years old
25
25 years old
5
five years old
26
26 years old
6
six years old
27
27 years old
7
seven years old
28
28 years old
8
eight years old
29
29 years old
9
nine years old
30
30 or more years old
10
ten years old
11
eleven years old
12
twelve years old
13
13 years old
14
14 years old
15
15 years old
16
16 years old
17
17 years old
18
18 years old
19
19 years old
20
20 years old
222
-------
Operating Mode
Table A2- 8 MOVES Operating Modes
opModel
D
opModeName
VSP
Lower
VSP
Upper
Speed
Lower
Speed
Upper
0
Braking
0
0
0
0
1
Idling
0
0
-1
1
11
Low Speed Coasting; VSP< 0;
1<=Speed<25
0
0
1
25
12
Cruise/Acceleration; 0<=VSP< 3;
*\<= Speed<25
0
3
1
25
13
Cruise/Acceleration; 3<=VSP< 6;
1<=Speed<25
3
6
1
25
14
Cruise/Acceleration; 6<=VSP< 9;
1<=Speed<25
6
9
1
25
15
Cruise/Acceleration; 9<=VSP<12;
1<=Speed<25
9
12
1
25
16
Cruise/Acceleration; 12<=VSP;
1<=Speed<25
12
0
1
25
21
Moderate Speed Coasting; VSP< 0;
25<=Speed<50
0
0
25
50
22
Cruise/Acceleration; 0<=VSP< 3;
25<=Speed<50
0
3
25
50
23
Cruise/Acceleration; 3<=VSP< 6;
25<=Speed<50
3
6
25
50
24
Cruise/Acceleration; 6<=VSP< 9;
25<=Speed<50
6
9
25
50
25
Cruise/Acceleration; 9<=VSP<12;
25<=Speed<50
9
12
25
50
26
Cruise/Acceleration; 12<=VSP;
25<=Speed<50
12
0
25
50
27
Cruise/Acceleration; 12<=VSP<18;
25<=Speed<50
12
18
25
50
28
Cruise/Acceleration; 18<=VSP<24;
25<=Speed<50
18
24
25
50
29
Cruise/Acceleration; 24<=VSP<30;
25<=Speed<50
24
30
25
50
30
Cruise/Acceleration; 30<=VSP;
25<=Speed<50
30
0
25
50
223
-------
opModel
D
opModeName
VSP
Lower
VSP
Upper
Speed
Lower
Speed
Upper
33
Cruise/Acceleration; VSP< 6;
50<=Speed
0
6
50
0
35
Cruise/Acceleration; 6<=VSP<12;
50<=Speed
6
12
50
0
36
Cruise/Acceleration; 12 <= VSP;
50<=Speed
12
0
50
0
37
Cruise/Acceleration; 12<=VSP<18;
50<=Speed
12
18
50
0
38
Cruise/Acceleration; 18<=VSP<24;
50<=Speed
18
24
50
0
39
Cruise/Acceleration; 24<=VSP<30;
50<=Speed
24
30
50
0
40
Cruise/Acceleration; 30<=VSP;
50<=Speed
30
0
50
0
100
Starting (Used for all starts)
0
0
0
0
101
Soak Time < 6 minutes
0
0
0
0
102
6 minutes <= Soak Time < 30
minutes
0
0
0
0
103
30 minutes <= Soak Time < 60
minutes
0
0
0
0
104
60 minutes <= Soak Time < 90
minutes
0
0
0
0
105
90 minutes <= Soak Time < 120
minutes
0
0
0
0
106
120 minutes <= Soak Time < 360
minutes
0
0
0
0
107
360 minutes <= Soak Time < 720
minutes
0
0
0
0
108
720 minutes <= Soak Time
0
0
0
0
150
Hot Soaking
0
0
0
0
151
Cold Soaking
0
0
0
0
200
Extended Idling
0
0
0
0
201
Hotelling Diesel Aux
0
0
0
0
202
Hotelling Fuel Operated Heater
0
0
0
0
203
Hotelling Battery AC
0
0
0
0
204
Hotelling APU Off
0
0
0
0
300
All Running
0
0
0
0
301
running; speed < 2.5mph
0
0
0
2.5
224
-------
opModel
D
opModeName
VSP
Lower
VSP
Upper
Speed
Lower
Speed
Upper
302
running; 2.5mph <= speed <
7.5mph
0
0
2.5
7.5
303
running; 7.5mph <= speed <
12.5mph
0
0
7.5
12.5
304
running; 12.5mph <= speed <
17.5mph
0
0
12.5
17.5
305
running; 17.5mph <= speed
<22.5mph
0
0
17.5
22.5
306
running; 22.5mph <= speed <
27.5mph
0
0
22.5
27.5
307
running; 27.5mph <= speed <
32.5mph
0
0
27.5
32.5
308
running; 32.5mph <= speed <
37.5mph
0
0
32.5
37.5
309
running; 37.5mph <= speed <
42.5mph
0
0
37.5
42.5
310
running; 42.5mph <= speed <
47.5mph
0
0
42.5
47.5
311
running; 47.5mph <= speed <
52.5mph
0
0
47.5
52.5
312
running; 52.5mph <= speed <
57.5mph
0
0
52.5
57.5
313
running; 57.5mph <= speed <
62.5mph
0
0
57.5
62.5
314
running; 62.5mph <= speed <
67.5mph
0
0
62.5
67.5
315
running; 67.5mph <= speed <
72.5mph
0
0
67.5
72.5
316
running; 72.5mph <= speed
0
0
72.5
0
400
tirewear; idle
0
0
0
0
401
tirewear; speed < 2.5mph
0
0
0
2.5
402
tirewear; 2.5mph <= speed <
7.5mph
0
0
2.5
7.5
403
tirewear; 7.5mph <= speed <
12.5mph
0
0
7.5
12.5
404
tirewear; 12.5mph <= speed <
17.5mph
0
0
12.5
17.5
405
tirewear; 17.5mph <= speed
<22.5mph
0
0
17.5
22.5
225
-------
opModel
D
opModeName
VSP
Lower
VSP
Upper
Speed
Lower
Speed
Upper
406
tirewear; 22.5mph <= speed <
27.5mph
0
0
22.5
27.5
407
tirewear; 27.5mph <= speed <
32.5mph
0
0
27.5
32.5
408
tirewear; 32.5mph <= speed <
37.5mph
0
0
32.5
37.5
409
tirewear; 37.5mph <= speed <
42.5mph
0
0
37.5
42.5
410
tirewear; 42.5mph <= speed <
47.5mph
0
0
42.5
47.5
411
tirewear; 47.5mph <= speed <
52.5mph
0
0
47.5
52.5
412
tirewear; 52.5mph <= speed <
57.5mph
0
0
52.5
57.5
413
tirewear; 57.5mph <= speed <
62.5mph
0
0
57.5
62.5
414
tirewear; 62.5mph <= speed <
67.5mph
0
0
62.5
67.5
415
tirewear; 67.5mph <= speed <
72.5mph
0
0
67.5
72.5
416
tirewear; 72.5mph <= speed
0
0
72.5
0
500
Existing
0
0
0
0
501
brakewear; stopped
0
0
0
0
A 1.3 Transportation Analysis Examples Using the AIR-SUSTAIN Tool
To evaluate the three competing scenarios for Cincinnati metropolitan area, a 15%
increase of population and employment is assumed to occur from the base year 2010 to the target
year 2030. All increase of population and employment is allocated and distributed around the
center(s). The process for a scenario analysis is taken in Steps 1 through 18, as described below.
Finally the analysis results of those three scenarios are compared at Step 19.
Step 1: Create a new scenario by clicking the Scenario button followed by clicking New
Scenario button on the menu bar, and then, input the Scenario Name (e.g., "Example")
and other required information (as shown in Figure A3-1) in the New Scenario tab. Then,
click on the Save Scenario button to save scenario files in the specified scenario folder,
and create the AIR-SUSTAIN database in MySQL and ArcGIS. Then, go to step 2.
Step 2: Select the Base Year first, or the Target Year in the Modeling Year panel if the Base
Year data is already created.
226
-------
Step 3 Import three feature classes: TAZ, RoadNetwork, and Incentive boundary in the Base
Year Data panel. The data import panel and the sample feature classes are shown in
Figure 3 .44. If in Step 2, the Base Year is selected, go to step 7 to perform regional level
analysis; if the Target Year is checked, go to step 4.
Step 4. Define the assumed Population Change, Employment Change, University Enrollment
Change and High School Enrollment Change in Assumed Changes in Demographic
and Socioeconomic Factors panel (as shown in Figure A3-2). Population Change and
Employment Change can be specified within and without incentive boundaries separately.
They can also be specified for individual TAZs and imported from Excel files as shown
in Tables A3-1. Then, go to step 5.
Base Year Data
Please select the data type:
Import
Remove
Imported Base Data:
TAZ
RoadNetwork
Incentiveboundary
Figure A3-1 Import base year data in example.
rqei Year Scenario Design
1. Assumed Changes in Demographic and Socioeconomic Factors
a Population Q Edit 0 Load File c. University Enrollment
b. Employment O Edit O [ Load File d. High School Enrollment
View
Load File
Load File
Base Year Data
Target Year Scenario Design
Travel Demand Forecasting
(A)
® Population Data Editor | o || a [| £3
Incentive Area Population Percentage Change
Non-incentive Area Population Percentage Change
l° I*
Save
Ml Employment Data Editor a ,| 0 ][ S3
Incentive Area Household Percentage Change
15 ] %
Non-incentive Area Household Percentage Change
° ] */.
Save
(B) (C)
Figure A3- 2 Program interface for (A) importing the Base Year data; (B) assigning
population change; and (C) assigning employment changes at TAZ levels.
227
-------
Table A3- 1 Sample of Population Change
TAZ
Population
330
156
338
191
318
268
249
274
261
383
336
822
337
1249
349
3571
208
7980
332
36784
Step 5: Specify the Initial Year, i.e., 2000, and load Land Use Inputs, and then project the target
year land use by clicking the Run button on the Land Use Projection panel. The target
year land use (shown as Figure A3-3) can be visualized in ArcGIS by clicking View
Results in the Land Use Projection panel. Go to Step 6.
Step 7: Select a TDFModel (only
VISUM is supported by
the current version of AIR-
SUSTAIN) and activate
the Travel Demand
Forecasting panel, then go
to step 8.
Step 6 Set up the Maximum Population Density in the incentive area, i.e., 15000
(person/mile2), then generate target year demographic and socioeconomic data by the
linkage model based on base year data and assumed demographic and socioeconomic
changes. Results can be
viewed by the user by
selecting the
corresponding data type
and displaying it (as shown
in Figure A3-4). Then, go
to Step 7.
Target Year Land Use
Figure A3- 3 Target year land use in example.
Legend
GRIDCODE
¦j Residential
228
-------
3. Socioeconomic Data Update Base on Assumed Data
a. Maximum Population Density 115000) |
b. Linkage Model I Run
c. Target Year Demographic and Socioeconomic Data 1. Population
I ° fa! £2 |
Table Of Contents
a
QiE3ia&
3»b
I IS M >S»SY ® ,
Figure A3- 4 Target year land use in example.
Step 8: Import a VISUM File containing TAZs, road network, and transit network, and four step
model parameters (the steps in create a VISUM file are briefly introduced by Step 8.1
through 8.6, and details can be found in VISUM user manual (PTV VISUM, 2013)).
When it is finished, go to step 9.
Step 8.1: Set up the travel demand model and travel demand segment in VISUM (shown
as Figure A3-5). Go to step 8.2.
Step 8.2: Load TAZ and road network shapefiles, or draw them in VISUM directly
(shown as Figure A3-6). For transit lines and stops, the user can only set them up
manually in VISUM. Go to step 8.3.
Step 8.3: Generate connectors to connect TAZs with road network, then go to step 8.4.
Step 8.4: Create the required fields in zonelist and linklist in VISUM (see in Table 3.7 in Section
3.2), then go to step 8.5.
229
-------
TSys/Modes/DSeg
Transport systems Modes Demand segments
Code
Name
Type Modes DSeg
im
Bus
PuT X X
_Jc
Car
PrT C C
Truck
PrT T T
4 jw
Wak
PuTWak X X
Create
Edrt
Delete
OK Cancel
e
SantodWKnH joemandttaesenes | Denard segments |
Select decani mode!
Bass | Person | Ace-nty pars | Demand strala | Mode w 'mafH ntum
Figure A3- 6 Example of road network from the input function.
Step 8.6: Set up the calculation procedure, and parameters for each step (as shown in Figure A3-
8), then stop.
230
-------
Figure A3- 7 Example of VISUM matrix.
Step 9 Import BomehoMFraction and Trip Rate (Excel file, as shown in Table A3-2), and
Employment Fraction (Excel file, as shown in Table A3-3). Then go to Step 10.
PTV Visum 64 Bit 13.00-16 - Network; BaseTDF.ver* - [Procedure sequence]
File Edit View Lists FiK
V Main turns
0 T Main zones
O ~ Territories
«• OD pairs
Y Main OD par
,¦ PrT paths
fO V Count locations
J— ? Detectors
(D Toll systems
-3HI1
| Procedure sequence
Count: 18 Execution Active
1 i n
2 >
l~l Linear combination of attributes for active zones only
Q Matrix balancing for active zones only
O Sum up values
[ Count: 5 I Demand stratum Matrix balancing
'y
From Node No
Type No
TSysSet
Cap PrT
VOPrT
VoIVehPrT(AP]
VolPere PuT(AP)
Cf Quick view X
HBO_G01
HBSC_G01
HBU_G01
HBW_G01
1 NHB_G0i
:h totals 1
totals 1
Reference object(s)
Comment Success Start Time EndTune
12:06:12 PM 12:06
12:06
12:06
12:06
M 12:06
M 12:06
M 12:06
Al HC Model de
31 Storage
Al HC Mode} demand strata
Al HC Model demand strata
31 Storage
12:42:13 PM 12:46:0E
H_county_OKU>ar.v7.vei
Parameters: Trip generation
0.3949S"HBA_HBO + 0.39495"HBA_HBW + 0.39005"HBA_HI
J 1.062'HH + 0.23'LEMP + 4.2"MEMP + 12.744*HEMP
1 0.436286"HIENR90
0.676444"UENROLL95
1.506*EMP
i 0.3925*HH + 0.3925*HBA_HBO + 0.39495"HBA_HBW + 0.39005"HBA_
Figure A3- 8 VISUM procedure set up.
231
-------
Table A3- 2 Sample of Household Fraction and Trip Rate
HH_ID
Fraction
HBO
HBSC
HBU
HBW
47
0.014699
0.0122
0.0122
0.0278
0.8797
48
0.009301
0.0122
0.0122
0.0278
0.8797
49
0.004951
0.0122
0.0122
0.0278
0.8797
50
0.002039
0.0122
0.0122
0.0278
0.8797
51
0.002277
0.0011
0.0011
0.0278
0.8797
52
0.197834
0.0011
0.0011
0.0278
0.8797
53
0.030421
0.0011
0.0011
0.0762
1.2348
54
0.007077
0.0011
0.0011
0.0762
1.2348
55
0.007276
0.0049
0.0049
0.0762
1.2348
56
0.008519
0.0049
0.0049
0.0762
1.2348
57
0.003471
0.0049
0.0049
0.0762
1.2348
58
0.019597
0.0049
0.0049
0.0762
1.2348
Table A3- 3 Sample of Employment Fraction
TAZ
Low Trip Rate
Employment
Medium Trip Rate
Employment
High Trip Rate
Employment
151
0.18
0.55
0.27
162
0
0.94
0.06
163
0.13
0.37
0.5
156
0.06
0.44
0.5
155
0.03
0.67
0.29
161
0.14
0.79
0.07
159
0.36
0.49
0.16
160
0.01
0.86
0.13
165
0.14
0.5
0.37
164
0.05
0.42
0.53
168
0.15
0.52
0.34
170
0.04
0.66
0.3
171
0.06
0.74
0.2
Step 10. Run VISUM model, and go to Step 11 when VISUM model is finished. TDF outputs
contain VISUM from four-step model, which can be viewed in VISUM (shown as
232
-------
Figure A3-9), and vehicle composition (Excel file, as shown as Table A3-4), which will
be used as an input for emission estimation.
Table A3- 4 Sample of Vehicle Composition
LinkID
Car Fraction
3
0.82
4
0.96
6
0.85
8
0.94
10
0.93
12
0.98
14
0.87
16
0.99
Legend
Trip generation Trips Between Centers
Figure A3- 9 Example of trip distribution result from VISUM (Trips between two centers).
Step 11: Select the Emission Estimation Model. In current version of the AIR-SUSTArN, only
MOVES is currently available. Go to step 12.
233
-------
Step 12. Import MOVES Inputs including Age Distribution (Excel file), Fuel Formulation
(Excel file), Fuel Supply (Excel file), Meteorology (Excel file), and State and County
(Excel file). Samples of those files are shown in Tables A3-5 through A3-9. Then go to
step 13.
Table A3- 5 Sample of Age Distribution
sourceTypelD
yearlD
agelD
ageFraction
21
2000
0
0.0798
21
2000
1
0.0847
21
2000
2
0.0749
21
2000
3
0.0799
21
2000
4
0.0735
21
2000
5
0.0754
Table A3- 6 Sample of Fuel Formulation
Table A3- 7 Sample of Fuel Supply
countylD
fuelYearlD
monthGroupID
fuelFormulationID
marketShare
markets hareCV
39061
2010
7
9309
1
0
39061
2010
7
20011
1
0
Table A3- 8 Sample of Meteorology
monthID
zonelD
hourlD
temperature
relHumidity
7
390610
10
63.93
41.42
234
-------
Table A3- 9 Sample of State and County
State Name
County Name
Ohio
Hamilton
icmrcJ.mjp - *«V#
% r o :::: * »• ^ 9
!*»»(<
;;3 0 *
2=
H n
: mi
- ro«c wrat
• 3335 = >•*
1 — H 4 A X, «—•
imrun nap*
Figure A3-10 Emission results displayed in ArcGIS.
Step 13 Run MOVES model. Sample
result is shown by Figure A3-10.
Then go to Step 14.
Step 14 Set up criteria in Congestion
Identification window (Figure
A3-11), and identify the traffic
congestion links by clicking OK
button (as shown in Figure A3-
12). Then go to Stepl5.
Congestion Identification
Please specify tdentAcafcon entens
Default
V 0»ly link volume >• 125000
gj Truck (rsctton >•
Optional
0 £3
] C02 equvaler; >•
OK
Cancel
Figure A3-11 Traffic congestion
identification criteria.
235
-------
-32s
Legend
Hotspots
RoadNetwork
s> Congestion Identification
Please specify Kfcntfcafcon cntena
g Daily l«* vcJuit* >• 125000
@ Truck frsctxxi >• 008
Opbonal
a S3
1C02 «awv«l«rt >•
OK
Caned
Table
a- m- «|§|
Hotspots
5694.536464
5723.360928
5627.114562
6457.288842
5729.889945
7986.814954
8859.022653
TruckFraction
2017.9174- v
B-
(0 out of 23 Selected)
Figure A3-12 Identified traffic congestion links in simulation.
236
-------
Step 15. Based on Congestion Identification results, select study road links and apply TCMs in
the microscopic simulation model (note: this step is processed in the VISSIM model).
Step 15.1 Select the traffic links (an example is shown in Table A3-10) from the
identified congestion areas (see Figure 3-12). Go to Step 15.2.
Table A3- 10 Sample of Selected Traffic congestion Links for Corridor Level Impact Analysis
LinkNO
Length (Mile)
Car Volume
Truck Volume
23540
0.477
5055
457
23523
0.385
5489
497
26830
0.316
5489
497
23876
0.991
6125
554
17830
0.193
435
39
16938
0.341
636
58
Step 15.2 Load a base
map to the VISSIM
model. Then the
VISSIM links are
drawn on top of the
base map. In this
example, ramp
metering is selected as
a traffic control
measure with the
purpose of analyzing its
impact on traffic
congestion area and
associated emissions.
Selected traffic
congestion road
network is built up in
the VISSIM
environment as shown
in Figure A3-13 (where
the base map with
VISSIM links is
superimposed). Figure
A3-14 shows the sketch
of the congestion area
Figure A3-13 VISSIM links over the base map.
237
-------
to be divided into six segments for scenario comparison. The segment in the red box
is the study area, and red dash line illustrates the locations of data collection points
for model calibration and validation.
Figure A3-14 Schematic for traffic congestion microscopic analysis.
Step 15.3. Prepare real vehicle volume input and truck percentage for microscopic simulation
analysis. Table A3-11 shows a sample of the traffic volume and truck percentage for
all links in the traffic congestion analysis corridor. Then go to Step 15.4.
Step 15.4. Set up the Desired Speed Distribution. They are used to model the changes of traffic
flow speed within VISSIM network. The desired speed changes are determined
based on the study site specifications and speed limits. An example is shown by
Figure A3-15. Then go to Step 15.5.
238
-------
Table A3- 11 Vehicle Volume
Link Number in VISSIM
Peak Hour Traffic Volume
Truck Percentage
1
4791
8%
2
5344
8%
3
5344
8%
4
5344
8%
5
6130
8%
6
6130
8%
On-ramp #12
553
8%
On-ramp #11
782
8%
No.: I Name: Truck
75.0 mph
-12. Then go
65.00
Undo
50.00
Undo
Desired Speed Distribution
OK | [ Cancel
OK ] | Cancel
El Desired Speed Distribution
Figure A3-15 An example of desired speed distribution for cars and trucks.
Step 15.5: Specify vehicle type, class, and category. Example is shown in Table A3
to Step 15.6.
-------
Table A3- 12 Vehicle Type Class and Category
No.
Name
Width (m)
Relative Flow
Desired Speed Range (mph)
100
Car
1.5
0.917
65, 75
200
HGV
2.5
0.083
50, 60
Step 15.6. Specify driving behavior. Car following and lane change are two main aspects of
driving behavior. Tables A3-13 and A3-14 illustrate the car following and lane
change parameters respectively in the VISSIM model for highway. Then go to Step
15.7.
Table A3- 13 Car Following Behavior Parameters
Parameters
Value
Unit
cco
Standstill Distance
4.99
ft
CC1
Headway Time
0.91
s
CC2
'Following' Variation
13.12
ft
CC3
Threshold for Entering 'Following'
-8
-
CC4
Negative 'Following' Threshold
-0.35
-
CC5
Positive 'Following' Threshold
0.35
-
CC6
Speed Dependency of Oscillation
11.44
-
CC7
Oscillation Acceleration
0.82
ft/s2
CC8
Standstill Acceleration
11.48
ft/s2
CC9
Acceleration at 50 mph
4.92
ft/s2
Table A3- 14 Lane Change Parameters
General Behavior
Freeway
Unit
Free Lane
Selection
Trailing Vehicle
Maximum deceleration
-4
-3
ft/s2
-1 ft/s2 per distance
200
200
ft
Accepted deceleration
-1
-0.5
ft/s2
Waiting time before diffusion
N/A
60
s
Min. headway (front/rear)
N/A
0.5
ft
240
-------
To slower lane if collision time above
N/A
0
s
Safety distance reduction factor
N/A
0.6
N/A
Maximum deceleration for cooperative braking
N/A
-3
ft/s2
Overtake reduced speed areas
N/A
Leave box un-
checked
N/A
Step 15.7: Set up the signal control at the metered ramps. Table A3-15 is the design criteria for
metering rate and signal cycle of Federal Highway Administration. An example of
fixed time signal is shown in Table A3-16 as the type of ramp metering for on-ramps.
Then go to Step 15.8.
Table A3- 15 Ramp metering design criteria of the Federal Highway Administration
Flow Control Scheme
No. of
Lanes
Cycle
Length
Approximate Range of Metering Rates
(veh/h)
One vehicle Per Green
1
4-4.5 sec.
240-900
Three Vehicles Per Green (Bulk)
1
6-6.5 sec.
240-1200
Dual-lane
2
6-6.5 sec.
400-1700
Table A3- 16 Signal Control Parameters
Signal Group
Type
Cycle (s)
Green Time (s)
Metering Rate
1
Fixed time
4
2
240-900
Step 15.8: Calibrate and validate microscopic model (more details are provided in Section 3.3.5).
Table A3-17 shows an example of VIS SIM calibration results, and Table A3-18
shows an example of validation results. Then go to Step 15.9.
Table A3- 17 VISSIM Calibration Final Parameter Values
Parameters
Value
Unit
cco
Standstill Distance
4.99
ft
CC1
Headway Time
0.91
s
CC2
'Following' Variation
13.12
ft
241
-------
CC3
Threshold for Entering 'Following'
-8
-
CC4
Negative 'Following' Threshold
-0.35
-
CC5
Positive 'Following' Threshold
0.35
-
CC6
Speed Dependency of Oscillation
11.44
-
CC7
Oscillation Acceleration
0.82
ft/s2
CC8
Standstill Acceleration
11.48
ft/s2
CC9
Acceleration at 50 mph
4.92
ft/s2
Table A3- 18 VISSIM Validation Results
Traffic Volume
LinkID
Real volume
Simulated
volume
Difference
Criteria
(Oregon)
Result
1
4791
4832
0.35
GEH<5
pass
2
5344
5431
1.40
GEH<5
pass
3
5344
5423
1.16
GEH<5
pass
4
5344
5425
1.22
GEH<5
pass
5
6130
6157
0.12
GEH<5
pass
6
6130
6154
0.09
GEH<5
pass
Travel Time
Range
Real travel
time (s)
Simulated
travel time (s)
Difference
Criteria
(<10%)
Result
Link 1-6
124.5
130
5.5
12.45
pass
Spot Speed
Data
collectionID
Real speed
(mph)
Simulated
speed (mph)
Difference
Criteria
(10% real
speed)
Result
1
59.08
64
4.92
5.908
pass
2
58
60.2
2.2
5.8
pass
3
58.73
63.1
4.37
5.873
pass
4
61.2
63.4
2.2
6.12
pass
5
56.08
56.6
0.52
5.608
pass
6
60.64
62.9
2.26
6.064
pass
242
-------
Step 15.9. After using real site traffic volume, spot speed and travel time to calibration and
validation the model, Traffic Volume and truck percentage (8.0% for all links) from
the regional-level results of VISUM for each link are applied for two project-level
analysis scenarios:
¦ Scenario 1 without any traffic control measure
¦ Scenario 2 with 4-sec cycle length ramp metering
Then, run VISSIM and export VISSIM outputs. Then go to Step 16.
Step 16. Import Microscopic Simulation Link ID in sheets 1 and 2 (Excel file; an example is
shown in Tables A3-19 and A3-2, which are constructed in microscopic simulation
software like VISSIM. The VISSIM model simulates traffic for one hour and finally
produces second-by-second vehicle speed, queue length, and delay (as shown in Tables
A3-21 and A3-22). Load Microscopic Simulation Results in sheets 1 and 2 (Excel file;
examples are shown in Tables A3-21 and A3-22) under different scenarios, separately.
Compare those imported results, and then go to step 17.
Table A3- 19 Microscopic Simulation Link ID (Sheetl)
GISLinkID
VissimLinkID
23540
1
23523
2
23523
3
26830
4
23876
5
23876
6
17830
12
16938
11
Table A3- 20 Microscopic Simulation Link ID (Sheet2)
GISLinkID
RoadType
LinkLength
LinkGrade
23540
4
0.477
0
23523
4
0.385
0
26830
4
0.316
0
23876
4
0.991
0
17830
4
0.193
0
16938
4
0.341
0
243
-------
Table A3- 21 Microscopic Simulation Results (Sheetl)
VissimLinkID
Time (sim
sec)
Car #
Car v
(m/s)
Car a
(m/s2)
Truck #
Truck v
(m/s)
Truck a
(m/s2)
1
1
0
0
0
0
0
0
2
1
1
22.6
0
0
0
0
3
1
0
0
0
0
0
0
4
1
0
0
0
0
0
0
5
1
3
29.7
-0.2
0
0
0
6
1
3
30.5
0.1
0
0
0
7
1
3
30
0
0
0
0
8
1
0
0
0
1
27.2
0.8
1
2
2
26.3
1.5
1
25.4
0
2
2
2
29
0
0
0
0
3
2
0
0
0
0
0
0
4
2
0
0
0
0
0
0
5
2
1
32.6
0
0
0
0
6
2
0
0
0
0
0
0
7
2
0
0
0
0
0
0
8
2
3
29.7
-0.2
0
0
0
1
3
3
30.5
0.1
0
0
0
2
3
3
30
0
0
0
0
3
3
0
0
0
0
0
0
Table A3- 22 Microscopic Simulation Results (Sheet2)
Link
Scenario 1
Scenario 2
Average
Speed (mph)
Delay
(s/veh)
Average
Queue Length
(vehs)
Average
Speed
(mph)
Delay
(s/veh)
Average
Queue Length
(vehs)
1
63.72
1.8
0
63.79
1.7
0
2
59.21
1.5
0
60.07
1.4
0
3
59.41
3.2
3
59.61
2
0
4
46.01
13.6
31
59.65
2.5
2
244
-------
Link
Scenario 1
Scenario 2
Average
Speed (mph)
Delay
(s/veh)
Average
Queue Length
(vehs)
Average
Speed
(mph)
Delay
(s/veh)
Average
Queue Length
(vehs)
5
46.93
11.7
28
56.21
3.3
3
6
56.92
2.2
0
57.03
2.2
0
12
46.28
0.6
0
46.38
0.6
0
11
49.92
0.4
0
17.17
5.1
8
Note: veh(s) - vehicle(s)
Step 17: Select the Emission Estimation Model (only MOVES is currently available in the
current version of AIR-SUSTAIN and will be added with more options in the future),
then run MOVES (an example is shown in Table A3-23). After it is finished, go to Step
18.
Table A3- 23 An Example of Traffic Congestion Link Emissions by Different Scenarios
LinkID
S1_C02 (kg)
S1_Energy (kJ)
S2_C02 (kg)
S2_Energy (kJ)
23532
1406.47
19,409,200
1406.46
19,409,190
23540
1073.49
14,812,350
1056.59
14,579,600
23876
944.21
13,029,600
811.57
11,198,300
26830
1933.76
26681,300
1778.17
24,535,100
17830
31.95
441,704
31.92
440,866
16938
58.43
808,036
80.97
1,120,090
Note: S1 represents Scenario 1, and S2 represents Scenario 2.
Step 18. Update regional emission results by project level analysis emission results. The
specification of variables and parameters in the program is shown in Table A3-24. If
both target year and base year have been analyzed, then go to step 19; else go back to
Step 2.
Step 19. Compare results between base year and target year. Then stop.
245
-------
Table A3- 24 Variable and parameter coding type in AIR-SUSTAIN program
Name
Field
Description
Type
TAZ
TAZ name, the
format is
TAZ_TAZ
Number, i.e.
TAZ_151
String
TAZ_N
TAZ number
Integer
TAZ_Order
The field to link
TAZs in ArcGIS
and TAZs in
VISUM
Integer
POP
Target year
population
Integer
HH
Target year
household
Integer
EMP
Target year
employment
Integer
HI
Target year high
school enrollment
Integer
UN
Target year
university
enrollment
Integer
AREA_TYPE
1=CBD&Urban; 2=
suburban; 3= rural
Integer
GEOCODE_Base_1
Base year
residential area
(ft2)
Double
GEOCODE_Base_2
Base year
employment area
(ft2)
Double
GEOCODE_Base_3
Base year
institutional area
(ft2)
Double
GEOCODE_Base_4
Base year
undeveloped area
(ft2)
Double
GEOCODE_Base_5
Base year other
area (ft2)
Double
GEOCODE_1
Target year
residential area
(ft2)
Double
GEOCODE_2
Target year
employment area
(ft2)
Double
246
-------
GEOCODE_3
Target year
institutional area
(ft2)
Double
GEOCODE_4
Target year
undeveloped area
(ft2)
Double
GEOCODE_5
Target year other
area (ft2)
Double
Landuse_Target
GEOCODE
Land use type,
Presidential,
2=employment,
3=institutional,
4=undeveloped,
5=other
Integer
Shape_Area
Area (ft2)
Double
RoadNetwork
C02_Equivalent_Updated
C02 equivalent
(kg)
Double
Energy_Consumption_Updated
Energy
consumption (kJ)
Double
247
-------
Appendix B
Water Treatment Plant-Climate
Change Adaption Model (WTP-
CAM) User's Manual
248
-------
Table of Contents
List of Tables 251
List of Figures 252
List of Abbreviations and Notations 253
B1.0 Program Overview 255
Bl.l WTP-CAM Setup 255
B1.2 WTP-CAM Workspace 256
B1.2.1 Main Menu Bar 256
Bl.2.2 Tool Bars 258
Bl.2.3 Status Bar 261
Bl.2.4 Cursor Menu 261
Bl.2.5 Processing Train Window 261
B.l.2.6 Property Editor 261
B1.3 Setting up a Processing Train 262
B1.3.1 Building a Physical Processing Train 262
B1.3.2 Editing Non-physical Settings 262
B1.4 Saving and Opening Projects 263
B1.5 Running WTP-CAM 263
B2.0 Understanding the Input Data 264
B.2.1. Introduction to the Example Processing Train 264
B 2.2. Inputs for Monte Carlo Simulation 264
B 2.2.1 Overview 264
B 2.2.2 Inputs for Monte Carlo Setting 268
B2.3 Customization of GAC Unit Process Model 272
B2.3.1 Overview 272
B2.3.2 Inputs for GAC Model Customization 273
B3.0 Understanding the Output Data 274
B3.1 Standard Output Tables for a One-time Run 274
B3.2 Tabular Outputs for Monte Carlo Simulation 280
B3.2.1 Samples/statistics of Raw Water Qualities 280
B3.2.2 Samples/Statistics of Effluent Water Quality 280
B3.2.3 Samples/Statistics of Adaptation Costs 281
B3.2.4 Samples/Statistics for Compliance/Non-compliance Realizations 282
B3.2.5 Running Log 282
B3.3.Graphic Outputs for Monte Carlo Simulation 283
B3.3.1. Sample Chart (to be developed) 283
B3.3.2. Frequency Chart (to be developed) 283
B3.3.3. Cumulative Frequency Chart (To be developed) 283
B4.0 Models and Algorithms in WTP-CAM 284
249
-------
B4.1 Monte Carlo Methods 284
B4.1.1 Seasonal Multivariate Analysis 284
B4.1.2 Simulation of quarterly running average (TOC compliance) 287
B4.1.3 Adaptation of Unit Process 288
B4.2 Customization of Unit Process 289
B4.2.1 Customization of GAC Unit Process 292
B4.3 Economics 292
B4.3.1 Adaptation Costs for GAC Processing 292
B5.0 References 294
Attachment A: Confirmation Tests 296
A-l Seasonal multivariate analysis 296
A-2 Customization of GAC model 298
Attachment B: Error and Warning Messages 303
250
-------
List of Tables
Table Bl- 1 List of Standard Toolbars 259
Table Bl- 2 List of Unit Process Toolbars 260
Table B2- 1 Options for Monte Carlo Analysis 269
Table B4- 1 Illustration of calculating running annual average for finished water TOC 288
Table B4- 2 GAC Contactor Cost 293
Table B4- 3 GAC Reactivation Cost 293
Table A- 1 Comparison of the mean and standard deviation in summer 297
Table A- 2 Comparison of cross correlation matrix in summer 297
Table A- 3 Comparison of the mean and standard deviation in winter 298
Table A- 4 Comparison of cross correlation matrix in winter 298
Table A- 5 Parameters estimated for TOC breakthrough model 299
Table A- 6 Summary of field data sets and estimated parameters 301
Table A- 7 Comparison of sum of least square for customized GAC models 301
Table B- 1 Error message (to be developed) 303
Table B- 2 Warning message (to be developed) 303
251
-------
List of Figures
Figure Bl- 1 WTP-CAM workspace 256
Figure Bl - 2 Three edit functions for unit process menu 261
Figure Bl- 3 The property editor 261
Figure Bl - 4 Setting dialogue box for Monte Carlo analysis 263
Figure B2- 1 Schematic diagram TWP-CAM program flow in the example simulation 265
Figure B2- 2 Schematic diagram for treatment unit process at the GCWW Richard Miller
Treatment Plant 265
Figure B2- 3 Original input data for the GCWW example processing train 266
Figure B2- 4 Illustration diagram for Monte Carlo analysis 267
Figure B2- 5 Monte Carlo inputs for the example processing train 269
Figure B2- 6 Manual input window for influent water quality statistics 270
Figure B2- 7 Dialogue window for name of data files 271
Figure B2- 8 Example format of influent water quality data file 271
Figure B2- 9 GAC unit process property window 272
Figure B2- 10 Dialogue window for TOC breakthrough customization 273
Figure B2- 11 Example format for TOC breakthrough data file 273
Figure B3- 1 Standard output "Table 1" for the example processing train 274
Figure B3- 2 Standard output "Table 2" for the example processing train 275
Figure B3- 3 Standard output "Table 3" for the example processing train 276
Figure B3- 4 Standard output "Table 4" for the example processing train 276
Figure B3- 5 Standard output "Table 5" for the example processing train 277
Figure B3- 6 Standard output "Table 6" for the example processing train 277
Figure B3- 7 Standard output "Table 7" for the example processing train 278
Figure B3- 8 Standard output "Table 8" for the example processing train 278
Figure B3- 9 Standard output "Table 9" for the example processing train 279
Figure B3- 10 Standard output "Table 10" for the example processing train 279
Figure B3- 11 Sample outputs of raw water quality 280
Figure B3- 12 Selected sample outputs of effluent water qualities at finished water 281
Figure B3- 13 Sample outputs for adaptation costs 281
Figure B3- 14 Sample outputs of raw water quality and adaptation cost for non-compliance
events 282
Figure B3- 15 Example format of the log file 282
Figure B3- 16 Example sample chart for raw water TOC 283
Figure B3- 17 Example frequency chart for raw water TOC 283
Figure B3- 18 Example cumulative frequency chart for effluent TOC at finished water 284
Figure B4 -1 Cost curve for annual cost of GAC unit process 294
Attachments
Figure A- 1 Comparison of GAC models with RSSCT dataset 1 299
Figure A- 2 Comparison of sum of error square for GAC models 300
Figure A- 3 Comparison of GAC models with RSSCT dataset 2 300
Figure A- 4 Validation of GAC model with field data 302
252
-------
List of Abbreviations and Notations
Abbreviations
AR(1)
autoregressive model of order one
Ch
chlorine
CAM
climate change adaptation model
DBP
disinfection by-products
GAC
granular activated carbon
GCWW
Greater Cincinnati Water Works
HAAs
haloacetic acids (nine individual species and the total of five
(HAAs), six (HAAe) and nine (HAA9) species)
ICR
information collection rule
ID
identification
MC
Monte Carlo
nh3
ammonia
O&M
operation and maintenance
PPI
Producers Price Index
RSSCT
rapid small-scale column test
TOC
total organic carbon
TTHM
sum of four individual species of trihalomethanes
USEPA
U.S. Environmental Protection Agency
UV
ultraviolet
UVA
ultraviolet absorbance at 254 nm
WTP
water treatment plant
253
-------
a vector of parameters to be estimated
initial parameter vector
corrected parameter vector
a variable between 0 and 1
a (9x 1) vector of standard normal deviates for season j
GAC model parameter, [-]
the Gauss-Newton coefficient matrix
(9x9) parameter matrix for season j
GAC model parameter, [-]
(9x9) parameter matrix for season j
GAC model parameter
GAC model parameter, [1/day]
a positive definite matrix, defined by I) = Bfl]
empty bed contact time, [min]
mean of random variable Y
TOC
TOC fraction remaining, defined by f{t) = "S-, [-]
TOCin
a mean of variable xt
a lag-zero covariance matrix of (jr. ->«.) for season j
a lag-one covariance matrix of {Xj-m^ for season j
sample size, [-]
the lag-zero correlation between x, and a-, , [-]
the lag-one correlation between variables y, and x}, [-]
a right-hand-side vector of Gauss-Newton equation
standard deviation of variable x,
covariance matrix, defined by sxx =e[xxt~\
GAC service time, [day]
field measurement of GAC service time, [day]
effluent TOC concentrations at the GAC unit, [mg/L]
influent TOC concentrations at the GAC unit, [mg/L]
process design or operating variable
log-normal distributed variable, defined by, = ln(x,)
defined by, X = - ntj_7
a (9x 1) vector of nine raw water quality parameters for season j
the capital, operational or maintenance cost, [US $]
a field measurement of TOC fraction remaining, [-]
defined by, Y = Xj-mj
a parameter is either 0 or 1 for adjusting cost functions for a range of
USRTvalues, [-]
254
-------
The Water Treatment Plant-Climate Change Adaptation Model (WTP-CAM) program
was developed from the climate adaptation models (CAMs) published in Li et al.(2012; 2014)
and Clark et al.(2010). The computer program is developed on the basis of Water Treatment
Plant (WTP) model that was proposed originally for the U.S. Environmental Protection Agency
(USEPA) in 1992 (USEPA, 2005). WTP software exists in two versions. The Version WTP v2.2
is further improved from original 1992 WTP v. 1.0. Appendix C contains a copy of the WTP v2.2
user manual.
This manual is intended to provide guidance to the WTP-CAM user with the new
features:
Utilizing the WTP-CAM program — navigating the user interface.
Selecting inputs and interpreting outputs related to Monte Carlo analysis and adaptation.
Understanding the algorithms applied in WTP-CAM relating multivariate analysis,
customization of processing units, and cost analysis.
In addition to this introductory Section, this User's Manual contains four other Sections:
¦ Section 2 explains menu components and describes how to set up and run WTP-CAM.
¦ Section 3 describes how to input the required data for the new features. A model of a
typical treatment plant is developed as an example; data input options are outlined.
¦ Section 4 provides guidance for interpretation of the output from the WTP-CAM based on
the example developed in Section 3.
¦ Section 5 describes the new algorithms used in the WTP-CAM program.
The user manual also contains two appendices. Appendix A shows confirmation tests to
verify the new algorithms introduced in WTP-CAM, including seasonal multivariate analysis, and
Granular Activated Carbon (GAC) model customization. Appendix B provides tables of error and
warning messages, including error or warning message identification (ID), their meanings, and
recommended actions for error correction.
B1.0 Program Overview
Bl.l WTP-CAM Setup
WTP-CAM Version 1.0 is designed as a Windows-based program that can be run under
the Windows 7 or newer operating system. It tested successfully on an Intel CPU 1.90 GHz
computer with 2 GB memory. The disk space requirement is mainly used to save simulation
results; 500 Mb minimal disk space is recommended. The program also allows users to use either
free or professional version of the SQL database. The file formats and designations remain the
same as those in the folder.
To run and use the WTP-CAM program, the user should set up a single folder in which to
place three files:
¦ WTP-CAMl.exe — the main executable file
¦ WTP-CAM User Manual— this support document
¦ WO example input.txt — a sample input file to illustrate the format of raw water quality
parameters
255
-------
When the executable file has been placed in a single directory, the user needs to double click
on the WTP-CAM.exe icon to launch the program. To remove WTP-CAM from your computer,
delete the file folder.
B1.2 WTP-CAM Workspace
The basic workspace for WTP-CAM consists of the following user interface elements: main
Menu Bar, Cursor Menu, Tool Bars, Status Bars, Property Editor, and Processing Train window
as shown in Figure Bl-1.
Mam menu bar
Processing tram window
Raw Water
n
Alum
Rapid Mixing
Flocculation
Lime
Settling
Filtration
GAC
H
Chlorine (Gas)
T.
Cleaiv/e
I
WTP Effluent
Ready
Toolbars
|r«l) WTP (
CAM - ITrainl]
I a I 0 83
D rii<
:.y;ion rdit View Prciect WwUpw I-or-
I -J* III.
QsSB & ..
** H[ err «t -H t|-*-
GAC Property
BCT, Minutes
GAC Reactivation Interval, days
How does GAC Contact System Operate,Single or Blended?
What is TOC Breakthrough Used for Single Unit, Max or Avg?
Oypco. Credit as 2nd Stage, logs
default Example
Bail
OK
For Advanced user, GAC Model customization...
Cancel
Status bar Property Editor
Figure B1-1 WTP-CAM workspace.
BL2.1 Main Menu Bar
The Menu Bar located across the top of the WTP-CAM workspace contains a collection of
menus used to control the program, including File Menu, Design Menu, Edit Menu, View Menu,
Project Menu, Window Menu, and Help Menu.
256
-------
File Menu: Contains commands for opening and saving data files and for printing.
Command
Description
New
Creates a new WTP project
Open
Opens an existing project
Save
Saves the current project
Save As
Saves the current project under a different name
Print
Prints the current view
Print Preview
Previews a printout of the current view
Print Setup
Sets page margins, headers, and footers for printing
Exit
Exits WTP-CAM
Design Menu: Contains commands to select unit processes, chemical feeds, and sampling points.
Unit Processes
Chemical Feeds
Sampling Points
Raw Water
Alum
WTP Effluent
Pre-settling Basin
Ammonia Sulfate
Average Tap
Rapid Mix
Ammonia
End of System
Flocculation
Carbon Dioxide
Settling Basin
Chlorine (Gas)
Filtration
Chlorine Dioxide
GAC
Iron
MF/UF
Lime
Nano-filtration
Ozone
Slow Sand Filtration
Permanganate
UV Disinfection
Sodium Hydroxide
Ozone Chamber
Sodium Hypochlorite
Contact Tank
Soda Ash
Reservoir
Sulfur Dioxide
Bank Filtration
Sulfuric Acid
DE Filtration
Bag Filtration
Cartridge Filtration
Edit Menu: Contains a control for copying.
Command
Description
Copy To
Copies the currently active view (processing train, graph, or table) to clipboard.
257
-------
View Menu: Contains controls for the user interface and commands for reporting results in
different formats.
Command
Description
Tool Bar
Status Bar
Graph
Table
Options
Toggles the tool bars on/off.
Toggles the status bars on/off.
Creates frequency/cumulative frequency chart of selected parameters.
Creates a tabular display of selected parameters.
Controls the display style of a graph, or table.
Window Menu: Contains commands for displaying open windows.
Command
Description
New Window
Cascade
Window List
Open another window for the active document.
Arrange windows so they overlap.
Lists all open windows; selected window currently with highlight.
Project Menu: Contains commands to define modeling conditions and to set up simulations for the
current project being analyzed.
Command
Description
Monte Carlo Setting
Cost Analysis
Optimization Analysis
One Time Run
Multiple Runs
Define conditions and inputs for Monte Carlo analysis.
Define the conditions for cost analysis.
Define the conditions for unit process optimization analysis.
Simulate water treatment at a defined condition.
Make Monte Carlo simulations.
Help Menu: Contains commands for identifying problems and solutions during simulation.
Command
Description
Error Message
About
Identifies problems and suggested solutions during simulation.
Lists information about current version of WTP-CAM.
Bl.2.2 Tool Bars
Toolbars provide shortcuts to commonly used operations. These operations are also
available at the Main Menu Bar. The toolbars can be docked underneath the Main Menu bar or
dragged to any location on the WTP-CAM workspace. The toolbars can be made visible or
invisible by selecting View » Toolbar. There are two types of toolbars:
• Standard Toolbars: contain speed buttons for commonly used commands (see Table Bl-1).
258
-------
• Unit Processes Toolbars: contain buttons for working with processing train (see Table B1-2).
Table B1-1 List of Standard Toolbars
Toolhiir Icon
Kquiviilcnl coiiiiiisHid ;il (lie
111:1 in in oil ii hiir
n|
File » New
1
j .fuHiiji 1
File » Open
IS
File » Save
a
File » Print
Edit» Copy to
_2_J
Project» MC Simulation
M
View » Graph
B
View » Table
259
-------
Table B1- 2 List of Unit Process Toolbars
Toolbar Icon
Equivalent command at the main menu bar
J§j
Design » Pointer (Deactivate selection)
-#
Design » Raw Water
Design » Presettling Basin
i^i;
Design » Rapid Mixing
DEI
Design » Flocculation
f:
Design » Sedimentation
m\
Design » Filtration
m
Design » Slow Sand Filter
|#§|
Design » GAC
Sf
Design » Micro/Ultra Filter
MF
Design » Nano-Filter
BKF ;
Design » Bank Filtration
DEF
Design » D.E. Filtration
BGF
Design » Bag Filtration
I— i
Design » Cartridge Filtration
fiffil
Design » UV Disinfection
(% ;
Design » Ozone Chamber
Toolbar Icon
Equivalent command at the main menu bar
I|||
Design » Reservoir
B
Design » Contact Tank
eff ;
Design » Effluent
Design » Average Tap
|h]
Design » End of System
CHM I
Design » Chemical Feed
-J-
Design » Connection
260
-------
Bl.2.3 Status Bar
The Status Bar appears at the bottom of the WTP-CAM workspace to show information
explaining the selected command in the Main Menu Bar or the Tool Bars.
Bl.2.4 Cursor Menu
There are two status options for the Cursor Menu in the processing train window. When
the mouse cursor does not point to any unit process in the processing train window, right clicking
the mouse will show the same Design Menu in the Main Menu Bar (refer to the introduction to
the Design Menu in Section B 1.2.1). When the mouse cursor points to a unit process in the
processing train window, right clicking the mouse leads
to a new cursor menu with three commands (Figure Bl-
2) as following:
• Move - move the selected unit process box to any
user desired location in the train window.
• Delete - delete the selected unit process from the
processing train.
• Property - show the Property Editor for the selected
unit process.
Bl.2.5 Processing Train Window
The Processing Train Window is the interface for users to build their own water treatment
processing train and input parameters for unit processes or Monte Carlo analysis. Section B1.3
describes how to build a processing train in this window.
B.l.2.6 Property Editor
The Property Editor (Figure Bl-3) is used to edit the properties of a unit process. It is
invoked when a unit process in the Processing Train Window is selected and double-clicked or
the property in the Cursor Menu is clicked. Following is an example Property Editor for
flocculation.
The following points help explain
how to use the Property Editor.
• The Editor usually consists of two
columns (one for the property's name
and the other for its value), an "OK"
button, a "Cancel" button and a
"WTP Example" button.
• The property value is initialized with
zero or the first element in the
dropdown list. Pressing the "WTP
Example" button will provide user
Move
Delete
Property
Figure B1-2 Three edit functions for
unit process menu.
Flocculation Property
Volume of Basin, MG
Ratio of T50/Detention Time
Ratio of TlO/Detention Time
1.94
1
jo.5
OK
Default Example
Cancel
Figure B1-3 The property editor
261
-------
example values for all properties in this opened Editor.
• Depending on the property, the value field can be entered either by typing in a value for a
text edit box or by selecting from a list of choices in a dropdown list box.
• You can use both the mouse and the tab key on the keyboard to move between properties.
• To have WTP-CAM accept what you have entered, press the "OK" button; to cancel, press
the "Cancel" button.
• The Editor window can be moved via the normal Windows procedures.
B1.3 Setting up a Processing Train
A processing train includes both physical objects that can appear on the Train Window,
and non-physical settings that encompass design and operational information as well as
simulation controls. The physical objects include water treatment unit processes, chemicals and
connection lines. Non-physical settings cover the properties of unit processes and settings for
Monte Carlo simulation, cost analysis, and optimization analysis.
Bl.3.1 Building a Physical Processing Train
To add a unit process to a processing train, the first step is to select the object unit
process from one of three methods (Design Menu, Cursor Menu or Toolbars), and then move the
mouse to a desired location on the Train Window, and click to finalize.
To add a chemical to the train, select the "Chemical" first from either Design
Menu/Cursor Menu, or the "Chemical" button in the Toolbars and then move the mouse to a
desired location on the Train Window and click. A dialogue box will appear, select the desired
chemical from the dropdown list and click "OK" button.
To add a connection line to the train, the first step is to select the "Connection" from
either Design Menu/Cursor Menu, or the Connection button in the Toolbars. The second step is
to move the mouse to the starting unit process and click; without releasing, continue to move the
mouse to the ending unit process and then release.
An object in the processing train can be deleted or moved using the Cursor Menu. To
delete or move an object, click on the object first, then click the right mouse key to invoke the
Cursor Menu, finally click Delete or Move from the Cursor Menu. For moving an object, move
the object to a desired location in the Train Window and click left mouse key to finish the
moving action.
Bl.3.2 Editing Non-physical Settings
The Property Editor (see Section B 1.2.6) is used to edit the properties of objects that can
appear in the Train Window. To edit one of these objects, select the object in the processing
train, then click the Property in the Cursor Menu or double-click the selected object. The
properties of objects usually consist of design and operational parameters for unit processes or
chemical feeds. A detailed explanation of such parameters can be found in Chapter C3 of the
WTP manual in Appendix C (U.S. EPA, 2005).
262
-------
Settings for Monte Carlo analysis are illustrated in Figure Bl-4, which includes control
parameters for computer simulation, data source for raw water quality statistics and cross
correlation matrix, and options available for the Monte Carlo simulation. Section 3 provides a
detailed explanation of the settings for a Monte Carlo analysis.
Monte Carlo Setting
Options—
p" Preserve Correlation
r Quarterly Running Average
p* Contamination Control
Controlled Contaminant
Controlled Processing Unit
[mc 3
Raw WQ Property Distn
Normal
"3
Default Example
r Control Parameters
Number of Runs, >1
Seed for Random Number, 1-50000
Regulation Standard, mg/L
Margin of Safety, mg/L
11000
1163
fox
r- Source of Influent WQ Statistics
Computed by Available Data File(s), Please Click Here
Or Input manually, Please Click Here
Correlation Matrix
Please Provide Data File(s) Here if Preserve Correlation is Checked
Figure B1-4 Setting dialogue box for Monte Carlo analysis.
B1.4 Saving and Opening Projects
Having completed the initial design and sequence of a processing train, it is a good idea to
save the project to a file.
• From the File menu, select the Save As option.
• In the Save As dialog that appears, select a folder and file name under which to save this
project. An extension of .wtp will be added to the file name.
• Click OK to save the project to file.
To open the project at some later time, select the Open command from the File menu.
B1.5 Running WTP-CAM
The WTP-CAM is designed to run under two modes:
• Single Case Run: make one-time run of the WTP analysis based on the deterministic influent
water quality entered from the Property Editor of Raw Water without use of Monte Carlo
setting.
• Monte Carlo Simulation: make multiple runs of the WTP analysis based on stochastic
influent water quality simulated with Monte Carlo setting.
When design of a processing train is complete, the WTP-CAM can be run by selecting either
263
-------
Project»Single Case Run
or
Project»MC Simulation (i.e., Monte Carlo Simulation).
If the run is successful, a notice window will appear indicating end of simulation. The
demonstrations and explanations of the outputs from WTP-CAM are described in Section 6.0 of
the main report.
B2.0 Understanding the Input Data
As introduced in Section B1.0, input data for WTP-CAM can be categorized into original
inputs and new inputs. The original inputs, including the design and operational parameters for
unit processes and information for chemical feeds, are used to make a traditional single case
simulation for a processing train and are introduced in detail in Chapter C3.0 of the original
WTP manual (USEPA, 2005). This WTP-CAM manual will not replicate the description for the
original inputs again. Instead, this manual focuses on definition and selection of the new inputs
added for the new features such as Monte Carlo analysis or customization of the GAC unit
process model. The description of the new inputs is illustrated through an example processing
train at Greater Cincinnati Water Works (GCWW) Richard Miller Treatment Plant ("Miller
plant") for drinking water.
B.2.1. Introduction to the Example Processing Train
WTP-CAM arranges the unit process components of a treatment train in a sequential block
diagram, as illustrated in Figure B2-1. In the Miller pant, the water treatment process is shown in
Figure B2-2. The plant treats the raw water through coagulation, sedimentation, rapid sand
filtration, followed by granular activated carbon (GAC) processing. The spent GAC is
reactivated in two large on-site furnaces. After chlorination disinfection, the treated water is
stored temporarily in a clearwell and then pumped into the distribution system. Figure B2-3
summarizes the original input data for the treatment train at the Miller plant.
B 2.2 Inputs for Monte Carlo Simulation
B 2.2.1 Overview
The ability to make Monte Carlo simulation is an important new feature not previously
available in the original WTP model. To understand the inputs for Monte Carlo analysis, it is
helpful to introduce the procedures involved in the analysis. Figure B2-4 outlines key steps of the
Monte Carlo analysis and the application of the new inputs (in bold). It can be seen that there are
three key options that govern the Monte Carlo analysis: Quarterly Running Average, Preserving
Correlation and Contamination Control.
264
-------
Figure B2-1 Schematic diagram WTP-CAM program
flow in the example simulation.
265
Figure B2-2 Schematic diagram for treatment unit process at
the GCWW Richard Miller Treatment Plant.
-------
influent
Alum
pH
influent Temperature
Minimum Temperature
Total organic carbon
uv Absoroance at 254nm
Bromide
Alkalinity
calcium Hardness
Total Hardness
Ammoni a
Turbidity
Peak Flow
Plant Flow
Surface water by SWTR
source water crypto, concentration
LT2 Rule Watershed control Prog, credit?
If GW System, is virus Disinfection Req
virus Disinfection for gw, if Req'd ....
7
8
18
6 (Celsius
_
0 icelsius
2
6 i,rg L^
0
0J0(ircn.)
0
U69^.ng L%
^ra L a
7"?
imq/L a
irq l a
0. 21 s^pq L
151.0 (N^U%
220.OQCUMGD^
120.onO(KGD
TRUE
0. 000
FALSE
FALSE
4. U
TUE FALSE)
(oocysts Liter)
v.TRUE FALSE)
(^RUE FALSE)
clogs'
Alur: Dose
Rapid Mix
1.1 (nrg/L as Al2(S04)3*14H2Q)
voluire of Basin 0. 005-1 ^"G^
Ratio of TO Detention Time 1.00 '..ratio-
Ratio of TLO Detention Time l.Gu (ratio"",
Flocculation
volure of Basin 1.9-
Ratio of T50 Detention Tin
Ratio of tio Detention Time
Presed. Basin
volume of Basin
Ratio of TO Detention Time
Ratio of T10-Detention Time
Eligible for LT2 Toolbox crypto, credit?
LT2 Toolbox crypto. Removal credit
Reservoir
volume of Basin
Ratio of T50/Detention Time
Ratio of TlO/Detenti on Time
1. ro
0. 50
l r at i o 1
;ratio)
Lin
2. 2300 i>ig *
I. 00 >„ratiu?
0.44 (^ratioN
FALSE TRUE FALSE)
0.4 oogs >
3~3.000GUHT
1.00 (ratio k
0.32 i ratio)
Softening (s) ...
Lire Dose
For pH adjustment (p)
settling Basin
voluire of Basin
Ratio of T50/'Detention Time
Ratio of TlO/Detenti on Time i
Filtration
Liqtid volure
Ratio of T50 Detention Tine ...........
Ratio of T10 Detention Tirre
Chlorinated Backwash w'ater"
Filter Media ^Anthracite sand or gac) .
Giardia Removal credit - conv. Filters
virus Removal credit - conv. Filters ..
crypto. Removal credit - conv. Filters
Giardia Reroval credit - Direct Filters
virus Reroval Credit - Direct Filters .
crypto. Reroval Credit - Direct Filters
CFE Turb. Meets LT2 Toolbox criteria" .
IFE Turb. Meets LT2 Toolbox Criteria7" .
crypto, credit as 2nd stage FiIt
5.0 (mg/L
PH_AC3. (P or
as Ca(OH)2)
.0000(^G)
1.00 (ratio1*
0.42 (ratio^
GAC
Erpty Bed contact Tire (at 'Plant Flow")
Gac Reactivation interval
gac contacting system (single Blended") .
TOC Breakthrough" for single Unit ;>"a\ A\ q)
crvpto. Rerroval credit as 2nd stage ....7
chlorine (Gas}
Chlorine Dose
contact Tank
volure of Basin
Ratio of T5Q/Detention Tine
Rati o of TlO/Detenti on Time .............
wtp Effluent
Average Tap
Average Residence Time (For Average Flow)
End of sister'
^a*inui Residence Time (For Average Flow)
2. 4"0S \J-°G ^
1. 00
0.1
FALSE
A/S
2
5
¦>
0
1
0
2
0
1
0
3
0
FALSE
FALSE
0. 5
Lrctio^
(ration
vTRUE FALSE'
's or g;
flogs"
i.logs 'i
Oogsj
(logs}
(logs)
(logs)
U"RUE FALSE)
(TRUE FALSE)
lloqs;
21 (nirutes)
ISO (ddySl
Blerded^s "or El
A\a_TOC W or A)
0. 5 {Jogs"
3.0 iwg L as Cl2)
28.3Q00(VG)
1.00 Cratio)
0.20 ^ratio)
1. 0
3.0
(Days)
(Days)
Figure B2-3 Original input data for the GCWW example processing train.
266
-------
Start
Stage 1: Parameter preparation
1. If "Quarterly Running Average" is checked, prepare four different sets
of parameters such as raw water statistics for spring, summer, autumn and
winter seasons. Otherwise, prepare one set of parameters.
2. If "Preserve Correlation" is checked, read corresponding data file(s) to
compute four/one set(s) of parameters for multivariate modeling.
3. If raw water quality statistics are provided by data file(s), read
corresponding data file(s) to compute four/one set(s) of raw water statistics.
4. Initialize the random number generator by Seed for Random Number.
5. Obtain Raw Water Probability Distribution.
Stage 2: Monte Carlo loop from 1 to Number of Runs.
1. Simulation of raw water quality.
a) If "Quarterly Running Average" is checked, compute raw water
qualities using raw water statistics and correlation matrixes in turn
from spring, summer, autumn and winter.
b) If "Preserve Correlation" is checked, compute raw water quality
based on multivariate modeling. Otherwise, simply based on raw
water probability distribution.
2. Performing a WTP run for this realization. If "Quarterly Running
Average" is checked, compute the quarterly running average using the
simulated water quality from this realization and previous three
realizations.
3. If "Contaminant Control" is checked and for a non-compliance
realization:
a) First to estimate the maximum permitted concentration of
"Controlled Contaminant" for this realization using "Regulation
Standard" and "Margin of Safety".
b) Second to seek a proper control variable for the "Controlled
Processing Unit" that make the "Controlled Contaminant" to be
the maximum permitted concentration.
c) Compute the adaptation cost with the current control variable.
4. Save outputs to files.
End
Figure B2-4 Illustration diagram for Monte Carlo analysis.
267
-------
The simulation option for Quarterly Running Average is specially designed for regulation of
contaminant total organic carbon (TOC). According to the USEPA disinfectant/disinfection
byproduct (D/DBP) rule, an important compliance criterion for TOC treatment of surface water
sources is that the treated water TOC level does not exceed 2.0 mg/L, calculated quarterly as a
running annual average. WTP-CAM applies four seasons to represent the four quarters per year.
As a result, this option affects the inputs of raw water quality (both statistics and correlation) and
simulation procedure for pursuing the quarterly running average. More details for simulation
related to Quarterly Running Average are introduced in Section B4.0.
The option for correlation is designed to preserve the joint correlation among raw water
quality parameters when simulating stochastic raw water quality variables in each realization. In
the presence of cross-correlation, concentrations of correlated reactants are possibly high or low
simultaneously. As a result, cross correlated raw water quality parameters might exert a strong
influence on DBP formation during water treatment and distribution. A multivariate seasonal
autoregressive model of order one (Bras and Rodriguez-Iturbe, 1984) was applied in WTP-CAM.
This seasonal model preserves all seasonal means and variance for all water quality parameters,
all cross correlation among all water quality parameters, and lag-one correlations between
adjacent seasons and between all water quality parameters. Section B5.0 describes the theoretical
basis for the multivariate analysis applied.
Contamination option is designed to modify the design and operation of the current
processing train when a non-compliance realization is simulated. For example, if a TOC
violation is detected, the WTP-CAM program will modify operation by increasing the frequency
of GAC regeneration in order to bring the TOC excursion within acceptable limits. The inputs
for this option are controlled contaminant, regulation standard, margin of safety, and unit process
to be controlled. So far, the option for contamination component has been developed only for
TOC contaminant and GAC unit process. More details are available in Section B5.0.
B 2.2.2 Inputs for Monte Carlo Setting
The input parameters for Monte Carlo analysis may be divided into three groups: options,
control parameters and source of influent water quality statistics/correlation. Figure B2-5
demonstrates these inputs for the example processing train shown in Figure B2-1.
Options: options are designed to govern the flow of Monte Carlo simulation. Table B2-1
provides the name of option, range of available values and description. Additional controlled
contaminant and controlled unit processes will be added with further development of WTP-
CAM.
268
-------
Monte Carlo Setting
mi
-Options—
p* Preserve Correlation
p" Quarterly Running Average
F Contamination Control
Controlled Contaminant
TOC
"3
Controlled Processing Unit
~~3
GAC
Raw WQ Probability Distn
|LogNormal
Control Parameters
Number of Runs, >1
Seed for Random Number, 1-50000
Regulation Standard, mg/L
Margin of Safety, mg/L
1000
168
0.05
SourceoflnfluentWQ Statistics
Computed by Available Data File(s), Please Click Here
Or Input manually, Please Click Here
Correlation Matrix
Please Provide Data File(s) Here if Preserve Correlation is Checked
Default Example
OK
Cancel
Figure B2-5 Monte Carlo inputs for the example processing train.
Table B2-1 Options for Monte Carlo Analysis
Control
Range of value
Description
Preserve Correlation
TRUE/FALSE
Multivariate analysis will be used to
simulate stochastic raw water quality if
TRUE (checked).
Quarterly Running Average
TRUE/FALSE
Simulation will be based on four seasons
if TRUE.
System Adaptation
TRUE/FALSE
Loading adaptation program for the non-
compliance realizations if TRUE.
Controlled Contaminant
TOC/None
Determining the contaminant to be
controlled by adaptation.
Controlled Unit Process
GAC/None
Determining the unit process that can be
adapted for controlled contaminant.
Raw Water Probability
Distribution
Normal/Lognormal
Determining the probability distribution
for all raw water quality parameters
269
-------
Control parameters: four control parameters are used in the Monte Carlo simulation:
• Number of Runs - a user defined integer to specify the number of runs required.
• Seed for Random Number - a positive number to initialize the random number generator in
the program. Monte Carlo simulation can be repeated using the same random number seed.
• Regulation standard - a value representing the compliance standard for the controlled
contaminant selected in Options.
• Margin of Safety - refers to the difference between the compliance standard and the real
controlled concentration that provides extra reliability for compliance. Margin of safety is
usually within 1%—10% of the regulation standard.
Source of influent water quality statistics/correlation: influent water quality statistics are
essential parameters to generate stochastic influent water quality parameters for each
realization. There are two methods provided by WTP-CAM to obtain these parameters. One
method is to input these parameters manually through clicking the manual input button. There
will be four dialogue windows appearing one at a time for the four seasons if Quarterly
Running Average is checked. Figure B2-6 illustrates an example of manual input window for
the spring of the example processing train.
Raw Water Quality Statistics Input Window
Time Horizon: Spring
[Parameter
Average
Standard Deviation
PH,-
7.7
0.17
Alkalinity, mg/L
55.5
18.2
Turbidity, NTU
43.4
38.0
Calcium Hardness, mg/L
63.5
23.3
Total Hardness, mg/L
110.4
18.4
TOC, mg/L
2.3
0.6
UVA, 1/cm
0.12
0.06
Bromide, mg/L
0.03
0.01
Ammonia, mg/L
0.29
0.41
Temperature, Celsius
12.4
0
Flov/ Rate, MGD
|l08.4|
I""
OK Cancel
Figure B2-6 Manual input window for influent water quality statistics.
270
-------
The other method is to compute the statistics using data file(s) provided by a user through
clicking the button of "Computed by Available Data file." Figure B2-7 demonstrates the input
window for the name (including the extension name) of data files prepared by user. The
following points ar e important:
• Location of file(s): the data file must
reside in the same folder as WTP-CAM
executive file.
• Format of data: as illustrated in Figure B2-
8, the data file consists of 11 columns. The
columns are pFI, alkalinity, turbidity,
calcium hardness, total hardness, TOC,
UVA, bromide, ammonia, temperature,
and inflow rate. The first two rows are
used to indicate the title and unit for each
column. There is no limit for the number
of data points. Each column needs to be
assigned a digit for correct reading of the
input file. Empty columns are not allowed
If the value in a column is not available,
fill the column with -100.
File Name Window
1^1
Enter Names of Water Quality Data File
C Annual Basis
<• Seasonal Basis
Spring
Summer
Autumn
Winter
Spring.txt
Summer.txt
Autumn.txt
Winter.txt
OK
Cancel
Figure B2-7 Dialogue window for name of data
files.
If "Preserve Correlation" is checked, users are required to provide the raw water data
file(s) for multivariate analysis through clicking the button in Correlation Matrix. The
requirements for location of file(s), the format of file and the file name input window are the
same as those for compute influent water quality statistics by file as described above.
PH
7. 72
7. 57
7.7
7.25
7. 74
7. 55
7. 83
7. 85
7.7
7. 65
7. 78
7. 71
7. 82
7. 98
7. 63
7. 81
7. 74
7. 72
7. 97
7.8
7. 82
Alk
mg/1
62. 5
68. 2
60. 94
19.12
70.12
36. 65
53.49
72. 55
84. 91
42. 93
89.18
49. 82
67.18
73. 69
37. 58
49. 56
68.48
40.19
89. 01
52. 82
56. 56
Turb
NTU
22. 51
56. 74
26. 72
19. 96
58. 77
64
26. 97
33. 61
24.4
37. 82
91. 96
39. 61
11. 59
26. 57
38. 38
12. 52
27. 88
35. 25
4. 31
12. 94
36. 99
Ca-H
mg/1
99.43
57. 08
77. 51
53. 73
56.11
70. 74
54. 36
75.43
53. 93
38. 34
36. 55
28.41
55. 91
113.48
58. 76
58. 76
31. 01
76. 38
77. 95
84. 92
73.16
Tt_H
mg/1
106. 72
111.41
128. 73
62. 22
105. 38
75. 75
103.91
106. 96
136.28
81. 68
96. 79
80.11
117.06
124.85
89. 53
97. 8
103.98
80. 51
128.13
96. 78
94. 35
TOC
mg/1
2. 77
4. 91
3. 04
1. 82
5. 02
2. 82
3. 7
4. 02
4. 25
3. 3
5.7
4. 21
3.13
3.79
2.14
2. 95
3. 82
3. 96
3. 53
2. 89
4.41
UVA
1/cm
0.0924
0.2106
0.0768
0.0376
0.1991
0. 061
0.1178
0.1536
0.2098
0. 0848
0. 267
0.1457
0.1153
0. 0863
0. 0655
0. 081
0.2052
0.0561
0. 0493
0. 0748
0. 0943
Bro
mg/1
0. 028
0. 024
0. 039
0. 027
0. 028
0. 028
0. 028
0. 028
0. 03
0. 026
0. 023
0. 029
0. 026
0. 036
0. 026
0. 028
0. 027
0. 033
0. 046
0. 027
0. 027
NH3
mg/1
0. 253
0. 231
0.199
0. 077
0. 091
0. 291
0. 256
0. 065
0. 035
0. 375
0. 3 51
0. 071
0. 064
0.127
0. 232
0.158
0. 089
0. 376
0. 23
0. 083
0. 662
Temp
Celsius
14.4
14.4
14.4
14.4
14.4
14.4
14.4
14.4
14.4
14.4
14.4
14.4
14.4
14.4
14.4
14.4
14.4
14.4
14.4
14.4
14.4
Qi n
MGD
108.4
108.4
108.4
108.4
108.4
108.4
108.4
108.4
108.4
108.4
108.4
108.4
108.4
108.4
108.4
108.4
108.4
108.4
108.4
108.4
108.4
Figure B2-8 Example format of influent water quality data file.
271
-------
B2.3 Customization of GAC Unit Process Model
B2.3.1 Overview
The performance of GAC for TOC removal has been studied using TOC breakthrough
experiments in GAC columns under various conditions to examine different raw water sources,
GAC size, pretreatment configuration, and bed depth/empty bed contact time (EBCT). In
developing the WTP model, a classic logistic function was used to represent the TOC
breakthrough curve for a single GAC contactor (USEPA, 2005), given by,
/ x TOC *
/(f)=—=£
a
TOC- l + be-
(B2.1)
Where, /(/) is TOC fraction remaining; roc., and TOCeff are TOC influent and effluent
concentrations at the GAC unit; t is GAC service time; a, b and d are model parameters
estimated by statistical regression.
GAC Property
EBCT, Minutes
GAC Reactivation Interval, days
How does GAC Contact System Operate,Single or Blended?
What is TOC Breakthrough Used for Single Unit, Max or Avg?
Crypto. Credit as 2nd Stage, logs
Default Example
OK
For Advanced User, GAC Model Customization...
Cancel
Figure B2-9 GAC unit process property window.
To improve the accuracy of GAC treatment modeling, WTP-CAM provides a new feature to
customize parameters a, b and d using non-linear regression method if users can provide site-
specific TOC treatment study data instead of the default statistical values. More details for the
TOC breakthrough model and non-linear regression are introduced in Section B4.0.
272
-------
B2.3.2 Inputs for GAC Model Customization
The GAC model customization is invoked by clicking the GAC model customization button
located at the bottom of the GAC property window as shown in Figure B2-9. A dialogue window
for TOC breakthrough customization will appear as shown in Figure B2-10. There are five edit
boxes for the user's input.
TOC Breakthrough Customization
Parameters in scaled up TOC breakthrough equation can be estimated using
site-specific GAC treatment data in a text file having three columns (Tab
delimited): GAC service time ft, days), Influent TOC (TOCJn, mg/L) and Effluent
TOC (TOC_eff, mg/L).
File name for TOC breakthrough data |TOC_breakthrough_sample.txt
Number of data points [53
TOC Breakthrough Eq.: TOC_eff/TOC_in = a/[l+b*exp(-d*t)]
Parameter: a h [-] d [1/day]
Initial value:
0.5
5
0.035
OK
Estimated:
0.604
9.444
0.036
Click to estimate the parameters
Figure B2-10 Dialogue window for TOC breakthrough customization.
• File name for TOC breakthrough data: provide the file name including the extension name in
the edit box and keep the data file in the same folder as the WTP-CAM program. The format
of data is illustrated in Figure B2-11. The data file consists of three columns: GAC service
time, influent and effluent TOC concentration to the GAC processing unit. The first two
rows are used to indicate the title and unit for each column. There is no limit on the number
of data points. Empty columns are not allowed.
• Number of data points: number
of valid data points in the data
file.
• Initial value for the parameter a:
a value between 0.6-0.9 (Roberts
and Summers, 1982).
• Initial value for parameter b\ a
value between 3-30 (Based on
USEPA [2005] and initial
studies) Figure B2-11 Example format for TOC breakthrough
data file.
• Initial value for parameter d\ a
value between 0.01-0. l(Based on USEPA [2005] and initial studies).
RunTi me
Tocin
TOCOUt
day
rug/1
mg/1
1. 5
1.8452
0.1034
5
1.8452
0.1268
9
1.8452
0. 084
13
1.8452
0.1103
17
1.8452
0.0812
21
1.8452
0.0772
25
1.8452
0.1372
29
1.8452
0.1223
33
1.8452
0.1867
273
-------
B3.0 Understanding the Output Data
This Section provides a brief overview of the outputs generated by the WTP-CAM Model.
Section B4.1 briefly describes the various output tables for a one-time run. Section B4.2 presents
tabular outputs for Monte Carlo analysis. Section B4.3 introduces the graphic outputs based on
Monte Carlo simulation. All outputs are based on the example processing train introduced in
Section B3.0 for various inputs.
The output module for WTP-CAM is still in development. Therefore, some results in this
Section are used for illustrative purposes to show program outputs expected in future.
B3.1 Standard Output Tables for a One-time Run
Based on the example treatment train shown in Figure B2-2, and input parameters
summarized in Figure B2-3, the WTP-CAM will generate full standard outputs contained in 10
output tables and save temporarily in a text file named "WTP-CAM stdout.txt" in the working
folder after "One Time Run" command. These 10 output tables are in fact replicated from the
outputs of original WTP model. The Tables 1-9 in WTP-CAM outputs are associated with the
(typical average) "Plant Flow" and "Influent Temperature" inputs. Outputs for "Table 10" are
associated with worst-case disinfection input parameters of "Peak Flow" and "Minimum
Temperature." Figure B3-1 through B3-10 demonstrated the standard outputs by one-time run.
For a detailed interpretation of these tables please refer to Chapter 4 of the WTP User Manual in
U.S. EPA (2005).
Table 1
water quality summary for Raw, Finished, and Distributed water
At Plant Flow (120', 6 mgd) and influent Temperature (18,6 c)
parameter
pH
Alkalinity
TOC
uv
(T)SUVA
ca Hardness
Hardness
Amtronia-N
Bror-ide
Free c!2 Res,
Chlorarcine Res.
TTHMS
HAAS
HAA6
HAA®
TOX
Brorate
chlorite
TOC Removal
C. not required
(mg/L
units
"co
as cacoB)
! r»g l ;
vl cr"!
(mg/L as CaCQ3)
(mg/L as catQ3)
fng L!
U
-------
Table 2
selected input Parameters
Parameter
TEMPERATURES
A\ erage
m'ni in. im-
plant FLO, RATES
a\erage
peak Hour ly
value units
18. 6
2.0
Cdeg.
(deg.
120.600 (rgd)
220.000 (rgd)
DISINFECTION INPUTS, CALCULATED VALUES
surface water Plant? true
Giardia: Total Disinfection Credit Required 3.0 (logs)
Giardia: Credit Achieved (other than by CT) 2.5 (logs)
Giardia: Inactivation Credit by CT Required 0.5 (logs)
virus: Total Disinfection Credit Required 4.0 (logs)
virus: credit Achieved (other than by CT) 2.0 (logs)
virus: inactivation Credit by CT Required 2.0 (logs)
crypto.: Total Disinfection Credit Required 3.0
crypto.: credit Achieved (other than by CT) 3,5
Crypto.: inactivation Credit by CT Required 0.0
(logs)
(logs)
(logs)
DISINFECT. CREDITS (not incl. CT) ; Giardia virus
(in order of appearance)
Filtration 2.5 2.0
GAC -2nd stage fiIt. 0.0 0.0
CHEMICAL DOSES
(in order of appearance)
Al uir
Liife
Chlorine (Gas)
crypto
3.0
0. 5
1.1 (rq L as Al 2 (504"; 3 14H2Q)
5.0 (rg L as ca(OH)2)
3.0 (rg L as Cl2)
PROCESS HYDRAULIC PARAMETERS:
(in order of appearance)
Rapid Fix
Flocculation
Presed. Basin
Reservoi r
Settling Basin
Filtration
contact Tank
GAC OPERATION INPUTS
Enpty Bed contact Tire
Reactivation Frequency
sys. config. ("s' = single, "e
T10/Tth
T50/Tth
VOL. (MG)
1. 0
1.0
0.0054
0. 5
1.0
1.9400
0.4
1.0
2.2300
0. 3
1.0
373.0000
0.4
1.0
26.0000
0.1
1.0
2.4708
0.2
1.0
26.3000
31. 0
(rinutes)
180.0
(days)
Blended)
B
Figure B3- 2 Standard output "Table 2" for the example processing train.
275
-------
Table 3
predicted water Quality profile
At Plant Flow (120.6 M6D) and
influent Temperature
(18.6 C)
i
Residence Time I
PH
TOC
UVA
(T)SUVA
Cl 2
NH2C1 1
process
¦ cum. I
Location
(-)
Cmg/L)
<1/cm)
(L/mg-ro)
(tr,g L)
(wg 'l> I
(hrs)
(hrs) i
influent
7,8
2,6
0,096
3.7
0,0
0.0
0.00
0.00
Alum
7,7
2.6
0.096
3.7
0.0
0.0
0.00
0.00
Rapid Mix
7,7
2.6
0.084
3.3
0.0
0.0
0.00
0.00
Flocculation
7,7
2.6
0.084
3.3
0.0
0.0
0. 39
0. 39
Presed. Basin
7.7
2,6
0.084
3,3
0.0
0.0
0.00
0. 39
Reservoir
7,7
2.6
0.084
3,3
0.0
0.0
74.23
74.62
Lime
9.0
2.6
0.0S4
3,3
0.0
0.0
0.00
74.62
Settling Basin
9.0
2.6
0.084
3.3
0.0
0.0
5.17
79.79
Filtration
9.0
2.6
0.0S4
3, 3
0.0
0.0
0.49
SO. 28
6A£
9.0
1.2
0.015
1.3
0.0
0.0
0. 52
80.80
chlorine (Gas)
8.1
1.2
0.010
0.9
1.4
0.0
0.00
80. SO
contact Tank
8.1
1.2
0.010
0.9
1.0
0.0
5.63
86.43
wtp Effluent
8.1
1.2
0,010
0.9
1.0
0.0
0.00
86.4 3
Additional Point
8.2
1.2
0.010
0.9
0.8
0.0
12.00
98.43
Average Tap
8.2
1.2
0.010
0.9
0.6
0.0
24.00
110.43
Additional Point
8,2
1,2
0.010
0.9
0.5
0.0
48.00
134.43
End of System
8.2
1.2
0.010
0.9
0.4
0.0
72.00
158.43
roc Removal (percent)
;
55
E.C, not required - raw TOC
, raw suva,
and ¦ or fi m" shed TOC <
- 2
E.C, Step 1 TOC
removal requirement ACHIEVED
Figure B3- 3 Standard output "Table 3" for the example processing train.
Table 4
predicted water
Quality Profile
At Plant
Flow (120.6 vgd) and I
nf luent Temperature (18. 6 C)
calcium
Magnesium
PH
Alk
Hardness
Hardness
solids
NH3-M
BroBii de
Location
( )
(rog/L)
(ir.g L)
(»g/L)
(mg/t)
(mg/L)
-------
Table
5
Predicted Trihalomethanes
and other
DBPS
At
Average
Flow (120.6 f»"GD)
and Temperature (18.6 C)
Bro3-
Cl02-
TOX CHCl 3
BDCM
DBCM
CHBr 3
TTHMS
Location
(ug/L)
C«i/L)
(ug t_) • (ug/L)
(ug/L)
(ug/L)
(ug/L)
(ug/L)
influent
0
0.0
0
0
0
0
0
0
Alum
0
0.0
0
0
0
0
0
0
Rapid Mix
0
0.0
0
0
0
0
0
0
Plocculation
0
0.0
0
0
0
0
0
0
Presed. Basin
0
0.0
0
0
0
0
0
0
Reservoir
0
0.0
0
0
0
0
0
0
L i me
0
0. 0
0
0
0
0
0
0
Settling Basin
0
0.0
0
0
0
0
0
0
Filtration
0
0.0
0
0
0
0
0
0
GAC
0
0.0
0
0
0
0
0
0
chlorine (Gas)
0
0.0
0
0
0
0
0
0
Contact Tank
0
0.0
33
3
4
4
3
14
WTP Effluent
0
0.0
33
3
4
4
3
14
Additional Point
0
0,0
43
4
6
?
5
22
Average Tap
0
0.0
49
5
?
8
6
26
Additional Point
0
0.0
57
7
8
10
7
33
End of system
0
0.0
62
8
9
12
8
37
— —-—
—.—.
—
—
-¦
Figure B3- 5 Standard output "Table 5" for the example processing train.
Table
6
Predicted Maloacetic
Acids
- through haa5
At Average Flow (120.6 MGD)
and
Terperature (18.6 c)
MCAA
DCAA
TCAA MBAA
DBA A
HAA5
Location
(ug/L)
(ug/L)
(ug/L) (ug/L)
(ug l)
(ug L)
Influent
0
0
0
0
0
0
a! um
0
0
0
0
0
0
Rapid rix
0
0
0
0
0
0
Flocculation
0
0
0
0
0
0
Presed. Basin
0
0
0
0
0
0
Reservoir
0
0
0
0
0
0
Lime
0
0
0
0
0
0
settling Basin
0
0
0
0
0
0
Filtration
0
0
0
0
0
0
GAC
0
0
0
0
0
0
Chlorine (Gas)
0
0
0
0
0
0
Contact Tank
1
2
1
0
1
5
wtp Effluent
1
2
1
0
1
5
Additional Point
1
2
1
0
2
7
Average Tap
1
3
1
1
2
8
Additional Point
1
3
1
1
3
9
End of system
1
4
2
1
3
10
— —
_
Figure B3-6 Standard output "Table 6" for the example processing train.
277
-------
Table 7
predicted Haloacetic Acids
(haas through HAA9)
At Average Flow (120. § mgd) and influent Temperature (18.6 c)
BCAA
BDC4A
PBCAA TBAA
HAAS HAA9
Location
(ug/L)
(ug L)
(ug/L) (ug/L)
(ug/L) (ug/L)
influent
0
0
0 0
0 0
Alum
0
0
0 0
0 0
Rapid Mix
0
0
0 0
0 0
Flocculation
0
0
0 0
0 0
preset!. Basin
0
0
0 0
0 0
Reservoir
0
0
0 0
0 0
Lime
0
0
0 0
0 0
settling Basin
0
0
0 0
0 0
FiItration
0
0
0 0
0 0
GAC
0
0
0 0
0 0
chlorine (Gas)
0
0
0 0
0 0
contact Tank
1
0
0 0
6 6
wtp Effluent
1
0
0 0
6 6
Additional Point
2
0
0 0
8 10
Average Tap
2
1
1 1
10 12
Additional Point
3
1
1 1
11 15
End of System
3
1
2 1
13 17
,
—,— —,— _______— __
Figure B3- 7 Standard output "Table 7" for the example processing train.
Table 8
predicted Disinfection Parameters - Residuals and ct Ratios
At Plant Flow (120,8 mgd) and influent Temperature (18.6 C)
CT Ratios
Temp
pH
Cl 2
NH2C1
Ozone
Cl02
——
Location
(c)
c~>
(isg/L)
(nig/L)
(Tig' L)
(mg/L) Giardia virus Crypto
influent
18.6
7.8
0.0
0.0
0.00
0.00
0.0
0.0
1.0
Al um
18.6
7,7
0.0
0.0
0.00
0.00
0.0
0.0
1.0
Rapid (>*ix
18.6
7.7
0.0
0.0
0.00
0.00
0.0
0.0
1.0
Flocculation
18.6
7.7
0.0
0.0
0.00
0.00
0.0
0.0
1.0
Presed. Basin
18.6
7.7
0.0
0.0
0.00
0.00
0.0
0.0
1.0
Reservoir
18.6
7.7
0.0
0.0
0.00
0.00
0.0
0.0
1.0
Lime
18.6
9.0
0.0
0.0
0.00
0.00
0.0
0.0
1.0
Settling Basin
18.6
9.0
0.0
0.0
0.00
0.00
0.0
0.0
1.0
FiItration
18.6
9.0
0.0
0.0
0.00
0.00
0.0
0.0
1.0
GAC
18.6
9.0
0.0
0.0
0.00
0.00
0.0
0.0
1.0
Chlorine (Gas)
18.6
8.1
1.4
0.0
0.00
0.00
0.0
0.0
1.0
contact Tank
18.6
8.1
1.0
0.0
0.00
0.00
4.2
33.7
1.0
wtp Effluent
18.6
8.1
1.0
0.0
0.00
0.00
4.2
33, ~
1.0
Additional Point
18.6
8.2
0.8
0.0
0.00
0.00
4.2
33. 7
1.0
Average Tap
18.6
8.2
0.6
0.0
0.00
0.00
4.2
33.7
1.0
Additional point
18.6
8.2
0. 5
0.0
0.00
0.00
4.2
33, ~
1.0
End of System
18.6
8.2
0.4
0.0
0.00
0.00
4.2
33.7
1.0
Crypto. CT Ratio = 1 because other credits met full disinfection requirements
Figure B3-8 Standard output "Table 8" for the example processing train.
278
-------
Table 9
Predicted
Disinfection parameters - CT values
At Plant Flow (120,§ mgd)
and influent Temperature (18.6 c)
Cl2
NH2C1
ozone
Cl02
Location
< •
--(rogL *
minutes)-
¦»
influent
0.0
0.0
0.0
0.0
Alum
0.0
0.0
0.0
0.0
Rapid Mix
0.0
0.0
0.0
0.0
Flocculation
0.0
0.0
0.0
0.0
Presed. Basin
0.0
0.0
0.0
0.0
Reservoir
0.0
0.0
0.0
0.0
time
0.0
0.0
0.0
0.0
Settling Basin
0,0
0.0
0.0
0.0
Filtration
0.0
0.0
0.0
0.0
GAC
0.0
0.0
0.0
0.0
Chlorine (Gas)
0.0
0.0
0.0
0.0
Contact Tank
67.4
0.0
0.0
0.0
wtp Effluent
6~.4
0.0
0.0
0.0
Additional Point
67.4
0.0
0.0
0.0
Average Tap
6". 4
0.0
0.0
0.0
Additional Point
6". 4
0.0
0.0
0.0
End of System
67.4
0.0
0.0
0.0
_
~~
» —,—,
— —
Figure B3- 9 Standard output "Table 9" for the example processing train.
Table 10
Predicted Disinfection Parameters
At Peak Flow (220.0 HGD)
and Minimum Temperature (2.0 c)
for surface water
Plant
•,ith coagulation and Filtration
CT Rat i os
Temp
pH
Cl 2
NH2C1
ozone
Cl02
_____
Location
CO
C-)
(rg L)
(iTjg. L)
Og.'L)
(mg L)
Giardia virus crypto
influent
2.0
7. S
0.0
0.0
0.00
0.00
0. 0
0.0
1.0
Alum
2.0
7. 7"
0.0
0.0
0. 00
0. 00
0.0
0.0
1.0
Rapid r»*ix
2.0
_ y
0.0
0.0
0.00
0.00
0.0
0.0
1.0
Flocculation
2.0
- f
0.0
0.0
0.00
0.00
0.0
0.0
1.0
Presed. Basin
2.0
' . '7
0,0
0.0
0.00
0.00
0.0
0.0
1.0
Reservoir
2.0
7 7
0.0
0.0
0.00
0.00
0.0
0.0
1.0
lime
2.0
$'.i
0.0
0.0
0.00
0.00
0.0
0.0
1.0
Settling Basin
2.0
9.1
0.0
0.0
0.00
0.00
0.0
0.0
1.0
Filtration
2.0
9.1
0.0
0.0
0.00
0.00
0.0
0.0
1.0
GAC
2,0
9.1
0.0
0.0
0.00
0.00
0.0
0.0
1.0
chlorine (Gas)
2.0
8.0
1.4
0.0
0.00
0.00
0.0
0.0
1.0
Contact Tank
2,0
8.0
1.0
0.0
0.00
0.00
0.8
6. 5
1.0
wtp Effluent
2.0
8.0
1.0
0.0
0.00
0.00
0.8
6. 5
1.0
Additional Point
2.0
8.0
0.9
0.0
0.00
0.00
0.8
6. 5
1.0
Average Tap
2.0
8.1
0. s
0.0
0.00
0.00
0. 8
6. 5
1.0
Additional Point
2.0
8.1
0.6
0.0
0.00
0.00
0.8
6. 5
1.0
End of system
2.0
8.1
0. 5
0.0
0.00
0.00
0.8
6. 5
1.0
crypto, ct Ratio
« 1 because
other
credits met full disinfection requirements
Figure B3-10 Standard output "Table 10" for the example processing train.
279
-------
B3.2 Tabular Outputs for Monte Carlo Simulation
Tabular outputs are saved in text format with extension name "txt" in the working folder to
allow viewing with any text editor. The tabular outputs will be managed through the main menu
"View"-> "Table..(To be developed). The tabular outputs for Monte Carlo simulations may be
classified into five types as described in the following sections.
B3.2.1 Samples/statistics of Raw Water Qualities
The samples of influent water qualities plus inflow rate and temperature for all
realizations can be collected as outputs to provide to users. At the bottom of results, the basic
statistics, including sample number, mean, standard deviation, minimum and maximum, are also
provided. Figure B3-11 illustrates a sample of influent water quality output.
Nurrber
Q1 n
Alk
Bro
Ca-H
Tt-H
NH3
Turb
PH
Temp
TOC
UVA
MGD
ing/L
iig/t
mg/L
iig/L
¦g/L
¦g/L
—
C
¦g/L
1/cm
1
120. 6
55. 52
0
035
49.4
100.8
0
061
12.0
7.60
18. 6
3. 06
0.061
2
120.6
63. 23
0
033
44.4
107. 3
0
124
19. 5
7.80
18. 6
3. 76
0.109
3
120. 6
61. 30
0
033
71.0
104. 9
0
191
28. 5
7. 57
18. 6
4. 74
0.110
4
120. 6
30. 96
0
027
62. 6
83. 5
0
141
26. 8
7.72
18. 6
1. 99
0.095
5
120. 6
82. 32
0
035
54. 8
117.0
0
171
21.1
7.77
18.6
4. 99
0.147
6
120. 6
59.12
0
031
95. 9
108.1
0
063
15. 9
7.93
18. 6
2. 77
0.061
7
120. 6
100. 52
0
026
89. 0
120. 5
0
265
281. 7
7.87
18. 6
5. 81
0.406
8
120. 6
45. 33
0
036
45.6
94. 3
0
291
32. 3
7. 39
18. 6
3.14
0.100
996
120. 6
47. 58
0
028
44. 5
90. 5
0
236
11. 9
7. 71
18. 6
2. 85
0. 042
997
120. 6
51. 37
0
031
56.6
93. 3
0
246
44. 9
7. 57
18. 6
3. 96
0.106
998
120. 6
41. 62
0
030
89.7
90.9
0
249
12. 5
7.44
18. 6
3.17
0.058
999
120. 6
58. 67
0
027
82.0
101. 6
1
558
146. 5
7.80
18. 6
5. 09
0.147
1000
120. 6
40. 39
0
031
72.4
94. 3
0
351
18.1
7.60
18. 6
2.14
0.041
Samp!es
1000
1000
1000
1000
1000
1000
1000
1000
1000
1000
1000
Mean
120. 6
58.18
0
030
62.6
98.9
0
3W
43. 7
7.71
18. 6
3. 83
0.113
St.dev
0.0
22. 36
0
006
23.2
18.0
0
446
40. 5
0.16
0. 0
1.11
0. 056
Mi rt
120. 6
15.48
0
014
23. 8
49. 5
0
003
2.1
7.14
18. 6
1. 36
0.024
Max
120. 6
232.32
0
053
183. 3
219. 6
4
17S
506. 9
8.13
18. 6
8. 82
0.406
Figure B3-11 Sample outputs of raw water quality.
B3.2.2 Samples/Statistics of Effluent Water Quality
Similar to the sample outputs of raw water quality, the sample and basic statistics of
effluent water qualities for all realizations can also be outputted. The difference is each unit
process has effluent water. Therefore, a location has to be designated for the sample outputs. In
addition, it may not be necessary to output all water quality parameters. Thus, an optional list
will be provided to users to select output parameters (to be developed). Figure B3-12
demonstrates an example output for selected water quality parameters, pH, TOC, chlorine,
TTHM (sum of four individual species of trihalomethanes)and HAAs (haloacetic acid, species
five), at finished water.
280
-------
Number
PH
TOC
rog/L
Cl 2
mg/L
TTHM
yg/L
HAAS
ug/L
1
8. 34
0. 67
1.70
7. 90
3.72
2
8. 35
1.07
1. OS
14. 2"
i. 69
3
8. 20
1.75
0. 39
22.54
10. 87
4
7. 86
0. 22
1. 52
1. 64
0. 94
5
8,37
1. 90
0.43
26. 90
12.08
6
8. 22
0. 55
1. 56
5. 06
2. 58
8,23
2.04
0. 00
2". 86
0.00
8
7. 94
0.74
0. 32
6. 78
2. 92
996
8. 08
0. 55
0. 77
4. 31
2.06
997
8.10
1.21
0. 34
13.63
6.40
998
7. 94
0.75
0. 55
7.00
3.45
999
8.12
1.97
0. 00
8. 20
5. 52
1000
7.79
0.28
0.19
1. 38
0. 55
Samp!es 1000
1000
1000
1000
1000
Mean
8.14
1.12
0. 66
10. 93
4. 08
St.Dev
0. 21
0. 57
0. 67
9. 82
3. 32
Mi rt
7,25
0. 09
0. 00
0. 00
0. 00
Max
8, 70
2.05
2. 27
41.41
15. 62
Figure B3-12 Selected sample outputs of effluent water qualities at finished water.
B3.2.3 Samples/Statistics of Adaptation Costs
Similarly, the sample and basic statistics of adaptation variable and adaptation costs for
all realizations can be outputted. Figure B3-13 demonstrates an example output for the
adaptation variable, GAC reactivation period, and adaptation cost.
Number
React
cost
—
days
MS
1
ISO
13.622
2
ISO
13.622
3
93
17.511
4
ISO
13.622
5
82
IS.760
6
ISO
13.622
7
56
22.663
8
ISO
13.622
996
ISO
13.622
997
177
13.706
998
180
13.622
993
78
19.284
1000
ISO
13.622
samples 1000
1000
Mean
146
15.480
St.Dev
46
3.142
Mi ii
24
13.622
Max
160
32.481
Figure B3-13 Sample outputs for adaptation costs.
281
-------
B3.2.4 Samples/Statistics for Compliance/Non-compliance Realizations
In certain sampling-based sensitivity analysis techniques used to identify important
dynamic input variables, each vector of input variables is classified behaviorally into two sample
sets: those that created simulation outputs above a threshold (regulated standard) as "non-
compliance" sample set and those that created outputs below the threshold as "compliance"
sample set. WTP-CAM provides similar outputs and their basic statistics for the compliance or
the non-compliance sample set for selected parameters from optional list (to be developed).
Figure B3-14 demonstrates an example for non-compliance samples to regulated TOC at finished
water with selected parameters: the raw water quality parameters and the adaptation cost.
Number
PH
Alk
|rg, 1
Turb
NTU
Ca-H
iig/1
Tt_H
irg. 1
TOC
p;g 1
UVA
1, CIT
Bro
ng 1
NH3
ng 1
Cost
MS
2
-.65
53. S
146. 9
56. 0
-"9. 2
7 S4
0.191
0. 020
0. 59
32.46
5
~ ¦ ~'™T
-15.3
34.7
5"*. 9
85.0
4.18
0. 093
0. 032
0.S1
15.41
8
56
55. 6
21.9
51.1
96. 5
4.65
0.122
0. 037
0. 50
1". 58
10
5. 01
63.0
88. 7
40.0
90. 5
5.27
0, 268
0. 023
0.21
20.46
14
5""
61. 3
28. 5
"O. 9
104. S
4. 74
0.110
0.033
0.19
18. 24
16
82. 3
21.1
54. 8
117. 0
4. Q9
0.147
0.035
0.17
19. 55
18
S"
100. 5
281. 7
89.0
120. 5
5. si
0.406
0 026
0. 27
23.43
20
"3
84. 6
13. 6
59. 9
123. 8
4.66
0.146
0. 035
0.18
if!
981
7.6?
78. 1
39. 2
57. 9
109. 5
6.19
0.147
0. 026
O. 36
24. 96
982
7. 64
"Q. 3
42. 8
70. 8
114. 0
4.47
0.115
0. 030
0. 93
16. 9~
983
7.99
101. 9
26. 8
43. 3
114. 3
4. 87
0. 227
0. 027
0. 01
18. 89
989
8.01
85. 8
45. 5
104. 5
120. 7
5.13
0.102
0. 033
0.45
20.46
998
7.97
112. 5
21. 3
136. 9
128. 2
S. 82
0.136
0. 035
0. 32
13. 62
999
7. 83
76. 2
111. 3
159.2
120. 5
4, 57
0.169
0. 030
0.10
17. 58
sarples 451
451
451
451
4 51
4 51
4 51
4 51
451
451
vear
,.,5
"1.1
48. 3
63. 5
104.6
4. 83
0.145
0.030
0. 36
18.47
St.Dev
0.16
22.5
41.0
25.0
19.1
0. SI
0. 060
0. 006
0.43
?. 62
i»'in
"T. 25
27. 5
5.3
25.9
66.9
3.85
0. 047
0.017
0. 01
13.62
yax
5. 12
232. 3
326.1
183.3
219. 6
S. 82
0.406
0. 053
3.06
32. 96
Figure B3-14 Sample outputs of raw water quality and adaptation cost for non-compliance events.
B3.2.5 Running Log
Running log (to be
developed) will be
generated automatically
when WTP-CAM is
executed. The log file
provides the status of
execution and messages of
error or warning occurred,
which will assist the user to
diagnose problems.
Appendix A provides
further information
concerning the error ID and
warning ID. Figure B3-15 shows the format of the log file.
wtp-ccam version 1.0
Latest update: August 12, 2010
Developed for U5EPA by the university of Cincinnati
Nare of project: sample Processing Train
Date of running: Fri Aug 13 13:03:56 2010
Reading inpyt file... successfully
Monte carlo simulation... successfully
simulation is completed, thank you.
Figure B3-15 Example format of the log file.
282
-------
B3.3. Graphic Outputs for Monte Carlo Simulation
The tabular results, after data processing, can be viewed using graphs. The graphs can be
further printed or saved as a data file. The results of Monte Carlo analysis can be illustrated
either with a sample versus realization chart, frequency chart, cumulative frequency chart or
sensitivity chart for the sample-based data.
B3.3.1. Sample Chart (to be developed)
A sample chart is used to
illustrate the changes of a sampled
random variable with realization.
The realization may represent the
time sequence or spatial sequence
depending on the circumstance
studied. Figure B3-16 demonstrates a
sample chart for raw water TOC.
The realization may present daily,
weekly, or monthly time horizon.
B3.3.2. Frequency Chart (to be
developed)
The frequency chart, a graphical
display of tabular frequencies, is
used to plot density of data and show the degree of uncertainty for a selected parameter. In other
words, a frequency chart illustrates how often they occur in the range of the selected parameter
values. Figure B3-17 shows an example frequency chart for raw water TOC.
B3.3.3. Cumulative Frequency
Chart (To be developed)
The cumulative frequency chart
provides another way to explain the
results from Monte Carlo simulation
and is often preferred. This chart
presents the probability that a value
falls within, above or below a given
range. Figure B3-18 illustrates an
example cumulative frequency chart
for effluent TOC at finished water.
It can be seen that only 62% of the
effluent TOC concentration is less Fi9ure B3"17 Example frequency chart for raw water TOC.
than 2 mg/L, the regulation
compliance. Conversely, the TOC compliance standard is violated in 38% of the samples.
£
U
o
100C
200 400 600
Realization number
Figure B3-16 Example sample chart for raw water TOC.
240
200
&• 160
O
§- 120
« so
40
0
JUL
4 6
TOC, mgL
10
283
-------
(X 1000)
TOC, mgL
Figure B3-18 Example cumulative frequency chart for effluent TOC at finished water.
B4.0 Models and Algorithms in WTP-CAM
B4.1 Monte Carlo Methods
Monte Carlo analysis is a practical tool that is widely used to obtain sample solutions by
repeating a simulation process for problems involving random variables with known probability
distributions. Monte Carlo methods are useful for modeling phenomena with significant
uncertainty in inputs such as climate change induced raw water qualities. Because Monte Carlo
simulation considers random sampling of probability distribution functions as ]model inputs to
produce hundreds or thousands of possible outcomes instead of a few discrete scenarios, the
results provide probabilities of different outcomes occurring. Monte Carlo methods usually
follow a particular procedure below:
• Define a domain of possible inputs.
• Generate inputs randomly from the domain using a specified probability distribution.
• Perform a deterministic computation using the inputs.
• Aggregate the results of the individual computations into the final result.
As briefly introduced in Section B3.1, the ability to conduct Monte Carlo simulation is an
important new feature of WTP-CAM. Three key options govern the Monte Carlo analysis:
Preserving Correlation, Quarterly Running Average and Contamination Control/Adaptation of
Unit Process. Sections B4.1.1 to B4.1.3 provide descriptions of these controls.
B4.1.1 Seasonal Multivariate Analysis
The control for preserving correlation is designed to preserve the joint correlation among
raw water quality parameters when simulating stochastic raw water quality inputs in each
realization. A multivariate seasonal autoregressive model of order one, AR(1), (Bras and
Rodriguez-Iturbe, 1984; Salas et al., 1980) was applied in WTP-CAM to simulate the raw water
quality since this seasonal model preserves all seasonal means and variance for all water quality
parameters, all cross correlation among all water quality parameters, and lag-one correlations
284
-------
between adjacent seasons and between all water quality parameters. According to Bras and
Rodriguez-Iturbe (1980), the lag-one multivariate seasonal autoregressive model is,
{xrmj) = AAxj-i-mj-i) + BjEj
(B4.1)
Where, Xj is the (9x 1) vector of nine raw water quality parameters for season j. ntj is the known
vector of the means for the nine parameters for season j. Sj is an (9x 1) vector of standard normal
deviates for season j. Aj and Bj are (9x9) parameter matrices for season j. Aj and Bj can be
estimated by the covariance matrices (Bras and Rodriguez-Iturbe, 1984),
T-l
BjBj = jM0-jMlj_1M-1jMj
(B4.2)
(B4.3)
where, }M0 is the lag-zero covariance matrix of (x. -mjfor season /; jM, is the lag-one
covariance matrix of (x} -m^ for season /; superscript (-/) refers 9 to the invertible matrix;
superscript (7) refers to the transpose matrix.
Let Y = Xj-ttij and X = Xj, - m -,, Equation B4.1 becomes,
Y = AjX + Bjsj
The covariance matrices are defined by,
,Mn=S =e[yyt1
J o yy |_ J
jM1=Syx=E[YXT]
j.1M0=Sxx=E[XXt]
(B4.1a)
(B4.4)
(B4.5)
(B4.6)
Matrices , Sy, and ^ can be represented in terms of variances, standard deviations
and correlations as,
Sx
x\
X2%i -'.'2 .V
-V .'.'2
sl
.V | Xij X Xij
rV9SX2SX9
Xij X Xij X Xij .'.'2 Xij .'.'2
St
x9 y
(B4.7)
285
-------
yy
S,
r S S
yiyi yi y\
r S S
y\y2 y\ yi
S2
yi
r S S r S S
V y*y\ y-) y\ y->yi y-> yi
yx
r S S
y\*\ xi yi
y2xi 'V, ^y2
y^x 2 -V2 ' 2 2 -v2 ^2
S* S* s* s*
^ J-jX,, x9 J! J2x9 x9 J>2
r V S
y\y9 y\ y->
r S S
y2y9 y2 y9
s
y9
y9x1 Sx ^ y9
y9x2 x2 y9
7* S S
y9x9 x9 y9 y
(B4.8)
(B4.9)
o y r
where, °x, is the standard deviation of variable /, xixj is the lag-zero correlation between
xi and Xj, ryiXj is the lag-one correlation between variables and xj. The sample means,
standard deviations and correlations are known parameters obtained from historical records.
Therefore, the matrix Aj can be computed directly with Equation B4.2. Matrix Bj can be
obtained by decomposition of B^ through taking matrix Bj as a lower triangular form,
B, =
(bn 0
b2l b22
\bgi bg2
0 ^
0
b99 j
(B4.10)
Let D~
d2 j d22
dn.
yd 91 d92 ¦ ¦ ¦ d99 j
= BJBJ
(B4.ll)
According to Salas et al.(1980), if D is a positive definite matrix, a unique solution for Bj
can be obtained when Bj is a lower triangular matrix. The non-zero elements of Bj are calculated
by,
?orj=l,bv=dtlbJI 7 = 1,...,9
(B4.12)
j-1
For j=2, 3,..., 9 and i=j, b0 ~Y.h.
(B4.13)
286
-------
( j~l ^
For / 2, 3,..., 8 and i=j+l, ¦¦¦,9, btJ = dtJ ~^bjkbik
\ k= 1
b
( JJ
(B4.14)
When matrices Aj and Bj are computed and vector Sj is simulated, the normally distributed
stochastic water quality parameters with preserved correlation, Xj, can be calculated with
Equation B4.1.
If the elements of vector X and F, xt and , are random variables following a two-
i (
parameter log-normal distribution, define the new variables, Xi and }'), as following,
x:=ln(xi)
(B4.15)
y] =ln(>0
(B4.16)
Thus, the transformed variables xi and }') are normally distributed with means mx and m
yt >
c* c* r . <
standard deviations ' X/ and °y. , and the correlation coefficient among them given by x, v, .
The sample means, standard deviations and correlations of the transformed variables xi and
can be also obtained from the transformed historical records through Equation B4.15 and B4.16.
The parameters of the transformed variables are then used to build the necessary auto-
covariance and cross-covariance matrices using the equations B4.7 to B4.9. Matrices Aj and Bj
can be obtained from the previous introduced Equations.
In order to get the original variables from results based on the transformed computation,
the inverse transformation must be performed as following,
X,. =c\p(x| +mx^j
(B4.17)
y, =exp [y\ +my,)
(B4.18)
B4.1.2 Simulation of quarterly running average (TOC compliance).
The simulation of Quarterly Running Average is specially designed for regulation of
contaminant TOC. According to the USEPA disinfectant/disinfection by-product (D/DBP) rule,
an important compliance criterion for TOC treatment for surface water as source is that the
treated water TOC concentration does not exceed 2.0 mg/L, calculated quarterly as a running
annual average. WTP-CAM applies four seasons to represent the four quarters per year.
Therefore, there are four running annual averages computed for each year. The running annual
average is defined as the arithmetic average of TOC concentrations at current season and
previous three seasons based on the USEPA D/DBP rule. Table B4-1 illustrates calculations of
running annual average for TOC in finished water.
287
-------
Since the means, variances and cross correlations of raw water parameters vary with
seasonal changes in most circumstances, it is necessary to prepare four sets of input parameters
for raw water qualities as shown in Figure 2-4. Therefore, there are four simulations each year,
corresponding to the four seasons. The TOC concentration is recomputed each season with TOC
values defined above.
Table B4-1 Illustration of calculating running annual average for finished water TOC
Year
Season
TOC concentration
Running annual average
Spring
1.3
--
2009
Summer
1.7
--
Autumn
2.2
--
Winter
1.7
1.7
Spring
1.2
1.7
2010
Summer
1.4
1.6
Autumn
2.4
1.7
Winter
1.5
1.6
B4.1.3 Adaptation of Unit Process
Adaptation refers to necessary changes of design and/or operation of the current water
treatment train when a non-compliance event is simulated. So far, the only adaptation module
that has been developed is for TOC treatment in the GAC unit process. More contaminant
controls and unit processes will be added with further development of WTP-CAM.
There are four parameters required from users: controlled contaminant, regulation standard,
margin of safety, and unit process to be adapted. For example, if TOC is selected as the
controlled contaminant, the regulation standard is 2.0 mg/L. In order to better ensure the
compliance, a margin of safety may be applied to adaptation. Margin of safety refers to the
difference between the compliance standard and the real controlled concentration that provides
extra reliability for compliance. For instance, if margin of safety is O.lmg/L, the controlled the
TOC concentration will be 2.0 mg/L - 0.1 mg/L =1.9 mg/L. In other words, the simulated
running annual average of TOC concentration will be less than 1.9 mg/L after adaptation. The
unit process to be adapted is where a change of a design or operation parameter happens. For
example, if a noncompliance event happens for TOC in finished water and GAC unit process is
available in the treatment train, an effective way to enhance TOC removal is to reduce the GAC
service time in GAC contactors (see Section 5.2.1 for detail). WTP-CAM will seek a GAC
service time so that the TOC concentration is right below the controlled concentration 1.9 mg/L.
The specific procedure of computation is as follows. The first step is to reduce current GAC
service time by one day. The second step is to use the new service time to re-compute the TOC
concentration for each of four seasons without change of other conditions in each season. The
third step is to calculate the new running annual average of TOC. The final step is compare the
new calculated TOC to the controlled concentration 1.9mg/L; if new TOC is less than 1.9 mg/L,
the new service time is adopted; otherwise, go back to the first step and repeat computation
again.
288
-------
B4.2 Customization of Unit Process
Chapter 5 of original WTP model user manual (USEPA, 2005) in Appendix C provides a
detail description of equations used to model various unit processes. The WTP Model primarily
uses empirical correlations to predict central tendencies of natural organic material removal,
disinfection, and DBP formation in a treatment plant. The algorithms were generally developed
using multiple linear regression techniques. As a result, the empirical correlations usually consist
of independent variables and empirical constants. These statistical models generally work well
for providing the central tendencies. However, they may not provide sufficiently accurate
predictions for a specific utility. As a new feature, therefore, WTP-CAM provides options to
customize the empirical constants in regression equations using site-specific treatment study
data. To date, only the GAC treatment unit process has modified to allow customization of the
TOC breakthrough model. Customizations for other unit processes will be added with the
development of WTP-CAM.
B4.2.1 Customization of GAC Unit Process
GAC treatment has been used as an alternative for reducing organic contamination in water
supplies since early 1970's (Roberts and Summers, 1982). The performance of GAC for TOC
removal has been studied using TOC breakthrough experiments in GAC columns under different
conditions, such as GAC sources or pretreatment configurations. Roberts and Summers (1982)
found that complete removal of TOC by GAC cannot be achieved under water treatment
conditions. An immediate, partial breakthrough of TOC can be observed, even using a column
filled with fresh GAC, which indicates that a portion of the influent TOC is not amenable to
removal by GAC treatment. With increased service time, the effluent TOC concentration rises
and eventually reaches a steady state value, which indicates that the GAC becomes saturated
with organics. They also observed that the effluent TOC seldom reaches the influent
concentration but is lower than the influent level. This constant steady-state removal usually is
attributed to biodegradation (USEPA, 1996). During early stages of operation, the ratio of
effluent to influent TOC concentration (called "fraction remaining") generally ranges from 0.1 to
0.5, depending on composition of the organic constituents and EBCT/bed depth. For steady-state
removal, the fraction remaining varies from 0.6 to 0.9 with corresponding range of service times
from 3,000 to 14,000 measured in bed volumes.
The TOC breakthrough curve in a single GAC contactor is often described
mathematically by a logistic functions in Eq.2.1. The model parameters a, b and d are developed
to reflect the impact of influent TOC and pH and EBCT. Based on statistical regression, these
parameters can be estimated by (USEPA, 2005),
a = 0.682
(B5.19)
b = 0.161pH2-O.SOSpH + 19.086
(B5.20)
289
-------
d = TOC„ |^//[-0.0000058£BCT2 +0.00011 IEBCT +0.00125] + 0.0001444EBCT2 + 0.005486£BCT +0.06005}
(B5.21)
To improve the accuracy of GAC treatment modeling, WTP-CAM provides a new feature to
estimate parameters a, b and d using a non-linear regression method if site-specific TOC
treatment study data are available instead of the statistical values estimated by Equations B5.19-
B5.21.
It may be time-consuming and expensive to obtain site-specific data from a pilot-plant or
full-scale study of GAC adsorption processes. Instead, the rapid small-scale column test
(RSSCT) may be used to generate the data required (Crittenden et al. 1991; Zachman and
Summers 2010). An RSSCT is a scaled-down version of a pilot or full-scale GAC column
contactor. The RSSCT method use mass transfer models to scale down the full-scale contactor to
a small column. Similarity of operation to that of large-scale contactors is assured by properly
selecting the GAC size, hydraulic loading and EBCT of the small contactor (Crittenden et al.
1991; Zachman and Summers 2010). USEPA (1996, 2000) provides standardized guidelines for
GAC treatment studies that help obtain quality assurance data of TOC breakthrough in a GAC
column. The USEPA's information collection rule (ICR) treatment studies database also provide
GAC treatment study data from 63 treatment studies nationwide (USEPA, 2000), including 44
RSSCT studies, 18 pilot studies and 1 full-scale study.
When f(t) versus t dataset are obtained from GAC treatment studies, WTP-CAM applies a
modified Gauss-Newton method to estimate model parameters a, b and d by fitting the non-
linear regression function (Equation B2.1) through least square analysis based on Hartley (1961).
The objective function is defined as,
2
Min Q(a,b,d)=YJ{yk- f {tk',a,b,d))
k= 1
(B5.22)
Where, f{t',a,b,d) = ^ + ; a, b and d are the model parameters to be estimated; h
and yk are the known field values representing GAC service time and TOC fraction remaining;
n is a known number of field samples.
As a widely used method, Gauss-Newton method seeks solutions through iteration.
Therefore, an important step is to correct model parameters during iteration using equation,
0, = 0o+vD
Where, 0 is a vector of parameters to be estimated, 6 =
fa\
b
d
(B5.23)
, 6o represent the initial parameter
vector and represent the corrected parameter vector; D is a correction vector to the initial
290
-------
parameters as a solution from the Gauss-Newton equations,
to minimize Q(a,b,d) during each iteration.
The Gauss-Newton equation is given by,
AD = R
Where, A is the Gauss-Newton coefficient matrix, defined by,
D =
fD^
A
vAy
; v is a value from 0 to 1
(B5.24)
f2±{^\ 2V-— 2V—
k~i\da) f~fdadb f^dadd
A =
iV—— 2 V f—
tidbda fXdb
2y¥¥_
tidbdd
2Y—— 2V(—
f~idd da i~idd db k~l\dd
(Herein, f ~ )
R is a right-hand-side vector of Gauss-Newton equation, defined by,
R =
( 8Q}
da
SQ
db
SQ
(Herein, Q=Q(a,b,d))
Vector D can then be solved by,
(B5.25)
(B5.26)
D = A'R
(B5.27)
Where, A'1 is inverse of matrix A .
In order to find an approximate minimum of Q(a,b,d), V value is estimated by the parabola
through Q (v = 0), Q \ v = I], and Q (v = 1), given by,
(B5.28)
y=^^[Q(y = o)-Q(v = i)]/^Q(v = i)-2Q^ = ^j+Q(v = o)
291
-------
Where, Q(v = o), g|V = ij,and Q(v = l) represent the Q\aiA>^i) values evaluated with
v = o, v = i and v = i through Equation B5.23.
The specific procedure can be summarized by the following steps. Step 1 is to provide
initial vector for @o . Step 2 is to solve for the vector I) using Equation B4.27. Step 3 is to solve
for v value with Equation 4.28. Step 4 is to check whether Q{a iA>^i) value meets the
precision requirement. If the answer is "no", the initial parameters are replaced by the values
calculated with equation B4.23 and iterated from step 2.
B4.3 Economics
WTP-CAM provides an economic analysis to estimate the costs associated with adaptation
made to design or operation of water treatment in order to provide a metric to assess impact of
climate change. The total costs considered in WTP-CAM include capital, operational and
management costs. To date, only the GAC treatment unit process has a cost analysis model. Cost
models for other unit processes will be added with the development of WTP-CAM.
B4.3.1 Adaptation Costs for GAC Processing
The costs for GAC processing consist of four types of costs: initial GAC cost, annual GAC
make-up cost, GAC contactor cost and GAC reactivation cost. The initial GAC cost is one-time
charge for GAC required to fill the contactors, which is calculated by the product of the total
volume of contactors, the density and unit cost of new GAC. The annual GAC make-up cost is
yearly cost for GAC loss during reactivation, which is calculated by the product of GAC loss rate
for reactivation, GAC reactivation rate and unit cost. The GAC contactor cost can be estimated
with a general form of the cost models by Adams and Clark (1988),
y = a + b(USRT)c dz
(B4.29)
where, y is the capital, operational or maintenance cost; USRTis the process design or operating
variable, which is usually the total surface area of the GAC filter for contactors (total hearth area
for GAC reactivation) or the total effective volume of the GAC unit for capital cost; a, b, c and d
are empirical parameters determined from nonlinear regression analysis, and z is either 0 or 1 for
adjusting the cost functions for a range of USRT values. The model parameters can be found
from Adams and Clark (1988), which was obtained based on the costs in 1983. For consistence
of comparison, all costs were converted to 2009 currency using the Producers Price Index (US
BLS, 2008). The contactor cost can be further categorized by the costs of capital, process energy,
building energy, maintenance material and operation and maintenance (O&M) labor. The
computational parameters for contactors are listed in Table B4-2.
292
-------
Table B4- 2 GAC Contactor Cost
Type of
Capital
Process
Building
Maintenance
O&M Labor
Cost
energy
energy
Material
USRT
volume
area
area
area
area
a
93700
0
15150
540
1160
b
1999.1
12
350
23.6
0.3
c
0.712
1
0.916
0.753
1.068
d
0.958
1
1
1
1.152
z
1
1
1
1
1
Unit cost
Construction Cost
0.08 $/kwh
0.08 $/kwh
—
9 $/hr
1.3y
(in 2009)
(in 2009)
(in 1983)
Ratio of
—
--
2009PPI/1983
2009
2009
PPI
PPI/1983
to1983
2009ENR/1983EN
= 2.56
PPI
cost
R=
= 2.56
R=2.16
O&M, operation and maintenance; PPI, producers price index
The GAC reactivation cost can be estimated using a similar algorithm used to calculate GAC
contactor cost based on Equation B4.29. However, the model parameters are different from those
for contactor cost. Table B4-3 lists the parameters used to estimate GAC reactivation cost.
If the capital recovery analysis is assumed a return period of 20 years with an interest rate of
5%, a cost curve can be developed to illustrate the total annual cost of the GAC system varies
with GAC service time (reactivation period). WTP-CAM takes the cost curve and uses the curve
to estimate the adaptation cost through interpolations based on GAC service time.
Table B4- 3 GAC Reactivation Cost
Type
of Cost
Capital
Process
energy
Building
energy
Maintenance
Material
O&M Labor
Natural Gas
USRT
area
area
area
area
area
area
a
144000
354600
12250
0
2920
648400
b
198300.4
6387
312.1
4456.6
282
287714.9
c
0.434
0.755
0.649
0.401
0.7
0.899
d
1
1
1
1
1
1
z
1
1
1
1
1
1
Unit
Construction Cost
0.08
0.08
--
9 $/hr
$0.0035 /scf
cost
1.3y
$/kwh
(in
2009)
$/kwh
(in
2009)
(in 1983)
(in 1983)
Ratio
--
—
2009
2009PPI/1983
2009PPI/1983
of
PPI/1983PPI
PPI
PPI
2009
= 2.56
= 2.56
= 2.56
to1983
2009ENR/1983ENR
cost
= R = 2.16
O&M, operation and maintenance; PPI, producers price index
293
-------
Figure B4-1 demonstrate an example cost curve developed for GCWW's Miller plant. The
Miller plant has 12 down flow gravity contactors and two multi-hearth furnaces for onsite
reactivation. Each of the Miller plant contactors has a volume of 595 m3 and a surface area of
181 m2. The overall GAC loss rate through the system is about eight percent. The carbon loading
rate is 482 kg/day of GAC per square meter of hearth area.
Figure B4-1 Cost curve for annual cost of GAC unit process.
B5.0 References
Adams, J. Q., and Clark, R. M., (1988). Development of cost equations for GAC treatment
systems. Journal of Environmental Engineering, 114 (3): 672-688.
Bras, R.L. and Rodriguez-Iturbe, I., (1984). Random functions and hydrology. Addison-Wesley
Publishing Company.
Clark, R.M., (1987a). Modeling TOC Removal by GAC: The general logistic function. J.
AWWA,79 (1): 33-37
Clark, R.M., (1987b). Evaluating the cost and performance of field-scale granular activated
carbon systems. Environmental Science and Technology, 21 (6): 573-580.
Clark, R. M. Yang, Y. J., Impellitteri, C. A., Haught, R. C., Schupp, D. A., Panguluri, S., and
Krishnan, E. R., (2010). Chlorine fate and transport in distribution systems:
Experimental and modeling studies. J. AWWA, 102:5
294
-------
Crittenden, J.C., Reddy, P.S., Arora, H., Trynoski, J., Hand, D.W., Perram, D.L., and Summers,
R.S., (1991). Predicting GAC performance with rapid small-scale column tests. J.
AWWA, 83 (1): 77-87.
Hartley, H.O., (1961). The modified Gauss Newton method for the fitting of nonlinear regression
functions by least squares. Technometrics, 3 (2): 269-280.
Li, Z., Clark, R.M., Buchberger, S.G., Yang, Y.J., (2014). Evaluation of climate change impact
on drinking water treatment plant operation. J Environ Engrg., DOI: 10.1061/EE. 1943-
7870.0000824
Li, Z., Clark, R.M., Buchberger, S.G., Yang, Y.J., (2012). Evaluation of logistic model for GAC
performance in water treatment. J. AWWA, DOI jawwa.2012.104.0120.
Oulman, C.S., (1980). The logistic curve as a model for carbon bed design. J. AWWA. 75 (1): 51.
Roberts, P.V. and Summers R.S., (1982). Granular activated carbon performance for organic
carbon removal. J. AWWA, 74:113-118.
Salas, J. D., Delleur, J. W., Yevjevich, V. and Lane, W. L., (1980). Applied modeling of
hydrologic time series. Water Resources Publications, Littleton, Colorado, 484 pages.
US BLS (Bureau of Labor Statistics), (2008). BLS Handbook of methods, Chapter 14 Producer
Prices-Background. http://www. bis.gov/opub/hom/homchl4 a.htm.
U.S. Environmental Protection Agency (U.S. EPA), (2005). Water treatment plant model version
2.2 user's manual. Office of Ground Water and Drinking Water, U.S. Environmental
Protection Agency, Cincinnati, Ohio.
U.S. EPA, (2000). ICR treatment study database, version 1.0. Rep. No. EPA 815-C-00-003,
Office of Water, Washington, D.C.
U.S. EPA, (1996). ICR manual for bench and pilot-scale treatment studies. Rep. No. EPA 814-
B-96-003, Office of Water, Washington, D.C.
Zachman, B. A. and Summers, R.S., (2010). Modeling TOC breakthrough in granular activated
carbon adsorbers. J. Environmental Engineering (ASCE). Pp 204-210. (February 2010)
295
-------
Attachment A: Confirmation Tests
Confirmation tests are designed to verify that the algorithms applied in WTP-CAM are
correctly coded and the modeling results with these algorithms are consistent to the
corresponding evaluation criteria. The confirmation tests validate the following two algorithms
in WTP-CAM: seasonal multivariate analysis, customization of GAC model.
A-l Seasonal multivariate analysis
The algorithm of seasonal multivariate analysis is described in section B4.1.1. The
purpose of incorporation of seasonal multivariate analysis is to preserve the means, variances,
and cross correlations of the raw water quality parameters. For this purpose, the sample means,
variances and cross correlations of raw water quality series from the Monte Carlo simulations are
compared to the corresponding given means, variances and cross correlations of the inputted raw
water quality parameters.
The comparisons are made in two seasons: summer and winter. The given means, standard
deviations, and cross correlation matrix of raw water parameters are calculated from the input
data files "summer_example_data.txt", and "winter_example_data.txt", which will be provided
with this user manual as sample input files. There are 500 rows (sample size) in each of the input
files. The sample means, standard deviation and cross correlation matrix of raw water quality are
computed from 1000 Monte Carlo simulations.
Table A-l compares the sample means and standard deviation in summer, it can be seen that
the multivariate analysis algorithm well replicate the given mean since the maximum of the
relative error between the simulated means and the given means is 6.3%. Reasonably good
agreements are also achieved between the modeled standard deviation and given standard
deviation as the maximum of relative error is 34.2%. Similar results are also obtained in winter
as shown in Table A-3 since the maximum relative errors of means and standard deviations are
7.7%) and 39.8%>. The simulation of turbidity has much larger relative errors in sample mean and
standard deviation than other water quality parameters owing to its large coefficient of variation
(1.21 for summer and 1.34 for winter). Increase of number of Monte Carlo runs may reduce
these relative errors.
Table A-2 compares the cross correlations between the given and modeled correlation
matrix in summer. It can be observed that reasonably good agreements are achieved. Among the
36 pairs of correlation coefficients, the errors of 32 pairs are less than 0.1, errors of 3 pairs are
greater than 0.1 but less than 0.2, and only 1 pair's errors are greater than 0.2. Reasonably good
agreements are also achieved for the comparisons in winter as shown in Table a.4. Among the 36
pairs of correlation coefficients, the errors of 30 pairs are less than 0.1, errors of the rest 6 pairs
are greater than 0.1 but less than 0.2.
296
-------
Table A- 1 Comparison of the mean and standard deviation in summer
Total
Calcium
hardn
Parameter
PH
Alkalinity
Turbidity
hardness
ess
TOC
UVA
Bromide
nh3_n
Unit
—
mg/L
NTU
mg/L
mg/L
mg/L
cm-1
mg/L
mg/L
A,
128.5
0.11
7.71
79.08
25.85
74.43
4
4.43
1
0.053
0.25
129.0
0.11
7.71
79.96
27.48
75.01
5
4.42
0
0.054
0.25
Relative Error
(%)
0.0
1.1
6.3
0.8
0.4
0.3
1.0
0.7
0.1
0.05
0.24
22.19
31.17
27.92
24.96
0.91
6
0.022
0.128
0.05
0.24
26.04
41.83
31.04
26.82
0.94
3
0.024
0.133
Relative Error
(%)
0.0
17.3
34.2
11.2
7.5
3.5
4.8
9.4
4.0
Note: Subscript "0" representing given values from input data.
Subscript "m" representing results from Monte Carlo analysis
Table A- 2 Comparison of cross correlation matrix in summer
Parameter
PH
Alkalinity
Turbidity
Calcium
hardness
Total
hardness
TOC
UVA
Bromide
nh3 n
pH0
pHm
Error
1
1
0.568
0.473
0.096
-0.145
-0.131
0.014
0.104
0.087
0.017
0.432
0.397
0.035
0.367
0.308
0.060
0.203
0.222
0.019
0.102
0.108
0.006
-0.171
-0.187
0.017
Alkalinityo
Alkalinityn
Error
1
1
-0.114
-0.138
0.024
0.136
0.180
0.044
0.798
0.883
0.085
0.737
0.643
0.094
0.515
0.467
0.049
0.304
0.364
0.060
-0.207
-0.300
0.093
Turbidityo
Turbidityn
Error
1
1
-0.120
-0.110
0.011
-0.256
-0.265
0.009
0.131
0.103
0.028
0.387
0.359
0.028
-0.429
-0.362
0.067
0.138
0.154
0.016
Ca hardnsso
Ca hardnssm
Error
1
1
0.381
0.419
0.038
0.040
0.019
0.022
-0.155
-0.215
0.061
0.296
0.421
0.125
-0.100
-0.134
0.034
Total hardesso
Total hardessm
Error
1
1
0.496
0.337
0.159
0.296
0.175
0.121
0.565
0.639
0.074
-0.276
-0.334
0.058
TOC0
TOCm
Error
1
1
0.698
0.654
0.044
0.128
-0.076
0.204
0.020
0.029
0.009
UVA0
UVAm
Error
1
1
-0.335
-0.335
0.000
-0.223
-0.178
0.045
Bromideo
Brornidem
Error
1
1
-0.053
-0.033
0.020
Note: Subscript "0" representing given values from input data.
Subscript "m" representing results from Monte Carlo analysis
The comparison results in the confirmation tests indicate that the multivariate analysis
algorithm in the WTP-CAM works reasonably well and is confirmed for further application.
297
-------
Table A- 3 Comparison of the mean and standard deviation in winter
Parameter
Unit
PH
Alkalinity
mg/L
Turbidity
NTU
Calcium
hardness
mg/L
Total
hardness
mg/L
TOC
mg/L
UVA
cm"1
Bromide
mg/L
nh3_n
mg/L
A,
7.78
64.14
39.64
72.19
121.14
4.06
0.089
0.071
0.23
Mm
7.78
65.38
42.71
72.87
121.99
4.05
0.088
0.072
0.23
Relative Error (%)
0.0
1.9
7.7
1.0
0.7
0.3
1.2
1.7
0.1
0.16
23.64
53.14
29.67
33.73
0.96
0.054
0.041
0.124
0.16
28.56
74.28
33.19
36.65
0.99
0.051
0.046
0.129
Relative Error (%)
0.0
20.8
39.8
11.9
8.7
3.1
6.2
10.1
4.0
Note: Subscript "0" representing given values from input data.
Subscript "m" representing results from Monte Carlo analysis
Table A- 4 Comparison of cross correlation matrix in winter
Calcium
Total
Parameter
PH
Alkalinity
Turbidity
hardness
hardness
TOC
UVA
Bromide
nh3 n
pH0
1
0.562
-0.140
0.103
0.426
0.365
0.199
0.099
-0.171
pHm
1
0.459
-0.128
0.086
0.390
0.305
0.219
0.106
-0.189
Absolute Error
0.103
0.013
0.016
0.036
0.060
0.020
0.006
0.018
Alkalinityo
1
-0.108
0.138
0.793
0.735
0.512
0.299
-0.205
Alkalinitym
1
-0.125
0.184
0.885
0.627
0.451
0.365
-0.290
Absolute Error
0.017
0.046
0.093
0.109
0.061
0.066
0.086
Turbidityo
1
-0.119
-0.243
0.121
0.363
-0.377
0.133
Turbiditym
1
-0.105
-0.240
0.096
0.343
-0.309
0.144
Absolute Error
0.014
0.003
0.025
0.020
0.068
0.012
Ca hardnsso
1
0.377
0.039
-0.151
0.286
-0.101
Ca hardnssm
1
0.417
0.020
-0.208
0.415
-0.132
Absolute Error
0.040
0.019
0.057
0.129
0.031
Total hardesso
1
0.490
0.286
0.558
-0.272
Total hardessm
1
0.332
0.168
0.630
-0.325
Absolute Error
0.158
0.117
0.072
0.052
TOC0
1
0.694
0.127
0.020
TOCm
1
0.646
-0.072
0.027
Absolute Error
0.048
0.199
0.006
UVA0
1
-0.304
-0.215
UVAm
1
-0.309
-0.174
Absolute Error
0.005
0.042
Bromideo
1
-0.054
Brornidem
1
-0.038
Absolute Error
0.016
Note: Subscript "0" representing given values from input data.
Subscript "m" representing results from Monte Carlo analysis
A-2 Customization of GAC model
Section B4.2.1 introduces the algorithm of GAC model customization, which is used to
provide users options to refine the empirical constants in GAC model using site-specific
treatment study data so that better prediction can be obtained in a specific utility. There are two
tasks in this confirmation tests for GAC model customization: one is to verify the improved
performance of the customized GAC model over the original GAC model in the WTP model; the
other is to validate the customized GAC model using field data from the GCWW's Richard
Miller treatment plant.
To compare the performance between the customized GAC model and original model in the
WTP model, two sets of RSSCT data from the Richard Miller Treatment Plant were used to
298
-------
estimate the customized GAC model parameters in Equation B2.1 using the non-linear
regression algorithm given by Equations B4.22-B4.28. The original GAC model parameters were
calculated using Equations B4.19-B4.21 when pH is 7.8, EBCT is 20 minutes and inflow TOC
concentration is 2.25 mg/L. Table A-5 summarized the model parameters for the two RSSCT
datasets. Then, the customized and original models were used to simulate the GAC logistic curve
and are compared to the corresponding RSSCT data sets. Obvious improvements can be
observed with the customized GAC model over the original model as shown in Figures A-l to A-
2. Figure A-3 quantifies the improvements by comparison the sum of error squares given by
Equation B5.22. It can be seen that the sum of error squares by the customized model is only
10.4% of that by the original model for RSSCT dataset 1 and 37.4% for the RSSCT dataset 2.
Table A- 5 Parameters estimated for TOC breakthrough model
Data source
GAC model
Parameter a
Parameter b
Parameter d
[day1]
RSSCT data 1
Customized
0.644
5.448
0.0314
Original
0.682
22.94
0.0388
RSSCT data 2
Customized
0.604
9.445
0.0359
Original
0.682
22.94
0.0388
Customized GAC model can be further validated with field data at the Miller plant. There
are eight episodes identified from field measurements for one of 12 contactors at the Miller plant
during January 2004 to May 2010. The TOC fraction remaining is obtained by calculation of the
ratio of the contactor effluent TOC concentration over inflow TOC concentration. Each of the
eight datasets were used to estimate the GAC model parameters the using the non-linear
regression algorithm given by Equations B5.22-B5.28. Table A-6 summarizes the minimum and
maximum TOC fraction remaining, GAC service period, and estimated parameters.
Figure A-4 exhibits the TOC breakthrough field measurements for the 8 datasets. TOC
breakthrough field curves in Figures A-4a, A-4d, A-4e, and A-4h do not achieve steady state of a
logistic curve. As a result, model parameters estimated with these datasets present great
fluctuation as parameter a varies
from 0.53 to 3.05 or parameter d
changes from 0.016 to 0.046.
Obviously, GAC models with
parameters estimated with these
datasets are not amenable to
represent the TOC breakthrough
in the Miller plant because of the
incomplete data. Thus, these
incomplete datasets should be
ignored. The averages of the
parameters estimated from the
rest four "complete" data sets are
used for the customized GAC
model, given by,
Figure A-1 Comparison of GAC models with RSSCT
dataset 1
ofi
E
O
o
~ RSSCT data 1
Customized model
Ordinal model
40
80
120
160
200
Figure A- 1 Comparison of GAC models with RSSCT
dataset 1.
299
-------
parameters estimated from the rest four "complete" data sets are used for the customized GAC
model, given by,
/(') =
0.759
l + 8.124e
-0.029-f
(A.l)
In addition to Equation A.l (represented with "average-based" in Figure A-4), a customized
GAC model using averages of parameters estimated with RSSCT data listed in Table A-5 is also
validated against the field data (represented with "RSSCT-based" in Figure A-4), given by,
AO
0.624
l + 7A47e
-0.034-f
(A.2)
For referen ce,
customized GAC models
with parameters
estimated for individual
datasets, as listed in
Table A-6, are provided
as well, represented with
"self-based" in Figure
A.4.
Table A-7 provides
the sums of error square
(defined by Equation
5.22) of "self-based,"
"RSSCT-based" and
"average-based"
customized GAC model
for all eight data sets. As
expected, the self-based
models provide the best
fitting for individual
datasets. Similar
performances are achieved
for both RSST-based and
average-based models
when GAC service time is
less than 100 days
(incomplete datasets).
However, the average-
based model presents a
much better simulation
00
.3
.Sh
'53
U
o
H
0.8
0.6
0.4
0.2
~ RSSCT data 2
Customized model
«™™»Orginal model
.
30
60
90
120
150
180
210
GAC service time, days
Figure A. 2 Comparison of GAC models with RSSCT dataset 2.
0.5
0 0.4
cu
D
cr
o
CD
o
E
D
CO
0.3
0.2
0.1
RSSCT data 1
RSSCT data 2
Figure A. 3
Comparison of sum of error square for GAC
models.
300
-------
Table A- 6 Summary of field data sets and estimated parameters
Data set #
Min. of
observed
m
Max. of
observed
m
GAC
service
period,
day
Parameter
a
Parameter
b
Parameter
d day1
Comment
1
0.110
0.519
88
0.809
11.271
0.035
Incomplete
2
0.088
0.808
256
0.767
11.134
0.035
Complete
3
0.127
0.841
312
0.783
8.391
0.025
Complete
4
0.080
0.500
102
0.527
7.322
0.046
Incomplete
5
0.134
0.452
116
0.732
6.065
0.021
Incomplete
6
0.097
0.844
291
0.725
8.210
0.035
Complete
7
0.083
0.849
275
0.760
4.762
0.022
Complete
8
0.110
0.477
109
3.048
35.221
0.016
Incomplete
Average
for
"Complete"
datasets
0.099
0.836
284
0.759
8.124
0.029
Table A- 7 Comparison of sum of least square for customized GAC models
Data set #
1
2
3
4
5
6
7
8
Self-based GAC model1
0.019
0.036
0.036
0.040
0.021
0.082
0.072
0.029
RSSCT-based GAC model2
0.037
0.434
0.417
0.052
0.054
0.300
0.232
0.107
Average-based GAC model3
0.034
0.068
0.078
0.054
0.080
0.112
0.106
0.125
Note:1 Self-based GAC model refers to the model using parameters estimated for individual
datasets given in Table A.6;
2 RSSCT-based GAC model refers to the model using parameters estimated by RSSCT tests
given by Equation A.2;
3 Average-based GAC model refers to the model using average of parameters based on
"complete datasets" given by Equation A.1.
301
-------
Figure A- 4 Validation of GAC model with field data.
302
-------
Attachment B: Error and Warning Messages
WTP-CAM has two types of messages: error message and warning message. The error
message is a fatal error and has to be corrected before the WTP-CAM can be executed
successfully. The warning message is either caused by uncommon parameters user specified or
used to provide user information for unusual running conditions of WTP-CAM. The warning
messages do not affect execution of WTP-CAM. The error and warning message will be
developed in subsequence refinement of WTP-CAM and summarized in Tables B-l and B-2.
Table B-1 Error message (to be developed)
II)
1 \pkiiuilion
( oncclion
1
Coil I open file "1'ilc ikinic"
Check llic existence oi file
Table B- 2 Warning message (to be developed)
II)
1 \pkiiuilion
RccommciKUilion
1
User-defined parameter is out of range
Stay within recommend
range
303
-------
vvEPA
United States
Environmental Protection
Agency
National Center for Environmental Assessment
Office of Research and Development
Washington, DC 20460
Official Business
Penalty for Private Use
$300
------- |