903R86003
Mot
Final Report of the
Philadelphia Integrated Environmental
Management Project
Regulatory Integration Division
Office of Policy Analysis
Office of Policy, Planning, and Evaluation
U.S. Environmental Protection Agency
TD
181
.P4
F351
1986
copy 2
October 1986
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
' L. / r -:[•'.•-..-' I"! Protection Agency
'•-'-'' • i;!-''OR'::i"l. c,.nation Resource
i; t '-••" Cvii- p.':?)
ACKNOWLEDGMENTS 8'.1 tt:,lr.-jt SLreet
ri.:iiic;^ia,PA 19107
We would like to give special thanks to staff members Elaine
Haemisegger, Alan Jones, Catherine Campbell, Hope Pillsbury, and
Forest Reinnardt for their crucial role in designing the
Philadelphia IEMP. For their managerial support, we thank
Michael Alford, Robert Currie, Deborah Taylor, Samuel Napolitano,
and Daniel Beardsley. We would also like to thank Thomas Walton,
William Reilly and Clemens Lazenka for their substantial tech-
nical assistance and continuing support.
While we have received generous and valuable assistance from
many people, too numerous to name, we would especially like to
thank the following individuals and organizations for their help
throughout the Project.
MEMBERS OF THE STEERING
COMMITTEE
Stuart Shapiro, Philadelphia Public Health Department
Richard Boardman, Pennsylvania Department of Environmental
Resources \
Gerald Hansler, Delaware River Basin Commission
William Marrazzo, Philadelphia Water Department
George Tyler, New Jersey Department of Environmental
Protection
Richard Morgenstern, US Environmental Protection Agency
Peter Bibko, US Environmental Protection Agency
MEMBER OF THE TECHNICAL
COMMITTEE
Robert Ostrowski, Philadelphia Air Management Services
Seymour D. Selzer, Delaware River Basin Commission
Steven Kurtz, state of New Jersey
Rao Kona, Pennsylvania Department of Environmental
Resources
Wayne Lynn, Pennsylvania Department of Environmental
Resources
Robert Kramer, US Environmental Protection Agency
William Schremp, US Environmental Protection Agency
Dennis Blair, Philadelphia Water Department
Thomas Burke, New Jersey Department of Environmental
Protection
Leon Gonshor, Pennsylvania Department of Environmental
Resources
Christian Beechwood, Pennsylvania Department of
Environmental Resources
-------
Robert Mitkus, US Environmental Protection Agency
Peter Ludzia, US Environmental Protection Agency
John Williams, US Environmental Protection Agency
While this report would not exist without the support of
these people and organizations, EPA and the authors are
responsible for its content.
John Williams—Site Manager
Stephen Horn—Project Staff
Technical Contractors
We would also like to acknowledge our technical contractors
for the Philadelphia IEMP.
American Management Systems, Inc.
Arthur D. Little, Inc.
Industrial Economics, inc.
Jellinek, Schwartz, Connolly & Freshman, Inc.
Sobotka and Company, Inc.
Temple, Barker & Sloane, Inc.
Versar, inc.
-------
Final Report of the
Philadelphia Integrated Environmental
Management Project
Regulatory Integration Division
Office of Policy Analysis
Office of Policy, Planning, and Evaluation
U.S. Environmental Protection Agency
October 1986
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
CONTENTS
Page
EXECUTIVE SUMMARY
I. INTRODUCTION
INTEGRATED ENVIRONMENTAL MANAGEMENT 1-1
• Traditional Approach i-l
• Integrated Approach: lEMPs 1-2
THE PHILADELPHIA IEMP 1-4
• Outline of the Report 1-5
II. PROJECT METHODOLOGY
RISK ASSESSMENT AND RISK MANAGEMENT II-l
• Risk Assessment II-l
• Risk Management II-'
• What Quantitative Risk Assessment is Not II-
GENERATION OP RISK ESTIMATES II-4
• Estimating Toxicological Potency II-5
• Cancer Potency Evaluations: Some
Simplifying Assumptions II-6
• Noncancer Effects 11-10
—No-Effect Thresholds 11-10
—Quantifying Risks of Noncancer Effects
Above the Threshold 11-12
• Risk Assessment as Applied in the
Philadelphia IEMP 11-12
• Evaluating Exposure to Toxic Pollutants 11-13
• Interpreting Risk Assessment Results 11-18
APPLICATION OP METHODOLOGY 11-19
• Phase I: initial Issue Selection and
Screening 11-21
—Establishing Institutional Arrangements 11-21
—Defining the Scope of the Project 11-21
—Developing a Database 11-21
—The Screening Process 11-22
• Step 1: Risk Screen 11-22
• Step 2: Qualitative Evaluation 11-23
• Risk Management Challenges in
Interpreting Results and Applying
the Process 11-24
—Phase I Products 11-25
-------
CONTENTS (continued)
• Phase II: Risk Management Control
Options 11-26
—Additional Data Gathering 11-26
—Developing Pollution Control Options 11-26
III. PHASE I OF THE PHILADELPHIA IBMP
REASONS FOR SELECTING THE PHILADELPHIA III-2
METROPOLITAN AREA
• Population and Industrial Base III-2
• Environmental Characteristics III-2
• Local Environmental Agencies and Toxics
Data III-6
INSTITUTIONAL FRAMEWORK AND SCOPE OF THE
PROJECT III-6
• Establishing the institutional
Arrangements III-7
• Defining the Scope of the Project III-8
—Quantitative Risk Assessment 111-10
—Geographic Boundary III-11
—Initial Pollutant Selection 111-14
—Relevant Pathways of Exposure III-15
• collecting, Organizing, and Automating
the Data 111-15
—Existing Data 111-15
—Engineering Estimates 111-16
• Area Air Sources 111-16
• Nonpoint Water Sources 111-18
• Intermedia Transfers 111-18
—The Automated Data System 111-20
PHASE I PRIORITY-SETTING 111-21
• Generating issue Papers III-21
—Pollutant Toxicity 111-23
—Sources, Releases, and Pathways 111-26
—Ambient Concentrations 111-26
—Receptor Population 111-27
—Exposure Levels and Adverse Effects 111-27
• The Risk Screen 111-30
—Primary Criteria 111-30
—Secondary Criteria 111-33
• Analytical Feasibility 111-33
11
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
CONTENTS (continued)
• EPA, State, and Local Program
Objectives 111-33
• Ability to Control Impacts 111-34
—Example: Pollutant Selection for
Control-Options Analysis 111-34
CONCLUSIONS FROM THE PHASE I SCREENING
ACTIVITIES 111-42
SETTING OBJECTIVES FOR PHASE II 111-44
• Monitoring Topics for Phase II 111-44
—Benzene Emissions 111-44
—Formaldehyde Releases to the Ambient Air 111-44
—Combustion of Used Oil 111-45
—Air Emissions from Landfills 111-45
• Control-Options Analysis Topics for
Phase II 111-45
LIMITATIONS TO PHASE I SCREENING III-46
IV. PHASE II MONITORING ACTIVITIES: BENZENE
FORMALDEHYDE, LANDFILLS, AND USED OIL
BENZENE IN THE AMBIENT AIR IV-1
• Overview of the Sampling Plan IV-2
—Traffic Intersection and Service
Station Study IV-2
• Site Selection IV-2
• Sampling Procedure IV-4
—Breathing Zone Study IV-4
* Site Selection IV-4
• Sampling Procedure IV-4
• Results IV-4
--Traffic Intersection and Service
Station Study IV-4
--Breathing Zone Study IV-6
• Implications for Exposure and Risk IV-7
—Traffic Intersection and Service
Station Study IV-7
—Breathing Zone Study IV-8
—Recent Changes in the Benzene Unit Risk
Factor IV-9
• Cancer IV-9
• Noncancer Effects IV-10
iii
-------
CONTENTS (continued)
FORMALDEHYDE IN THE AMBIENT AIR IV-10
• Overview of the Sampling Plan IV-10
• Results IV-12
• implications for Exposure and Risk IV-14
• Recent Changes in the Formaldehyde Unit
Risk Factor IV-16
AIR EMISSIONS FROM LANDFILLS IV-17
• Overview of the ROSE System Monitoring
Plan and Results IV-18
—The ROSE System IV-18
—Site Selection IV-18
—Results IV-18
• Overview of the NJDEP Landfill Monitoring
Plan and Results IV-20
—Sampling Plan IV-20
—Results IV-20
COMBUSTION OF USED OIL IV-22
• Overview of the Sampling Plan IV-22
• Results IV-24
V. CONTROL-OPTIONS ANALYSIS: AMBIENT AIR AND
WATER MONITORING PROGRAMS
AMBIENT AIR MONITORING PROGRAM V-3
• Overview of the Sampling Plan V-3
—Placement of Air Monitors V-3
—Sampling Procedures V-6
—Quality Assurance/Quality Control V-7
• Results from the Ambient Air Monitoring
Program V-8
• Evaluation of Model Performance and the
Emissions Inventory V-ll
--Analytical Approach V-ll
—General Findings from the Model
Performance Evaluation V-12
—Feedback on the Model Formulation V-19
• Changes to the Emissions Inventory V-20
• Caveats V-22
IV
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
CONTENTS (continued)
Page
• Implications for Exposure and Risk V-23
—Cancer V-23
—Noncancer Effects V-25
• Recent Changes in Unit Risk Factors V-26
—Cancer V-26
—Noncancer Effects V-27
AMBIENT WATER MONITORING PROGRAM V-29
• Overview of the Sampling Plan V-30
—Quality Assurance/Quality Control V-30
—Delaware River Monitoring V-31
—Sampling at the NEWPCP V-36
• Caveats V-38
• Results V-40
—Delaware River Transect Monitoring V-40
• Variation in Tidal Cycle (Control
Option 1) V-40
• Variation Among the Three Transect
Sites (Control Option 2) v-41
• Delaware Upstream Monitoring
(Control Option 3) V-41
—Monitoring Midway Between the NEWPCP
and Baxter, Station B V-43
—Baxter Finished Water V-44
—NEWPCP Influent Monitoring V-47
• Somerset-Frankford Low-Level
Interceptors V-47
• Delaware Low-Level Interceptor V-47
• Frankford High-Level interceptor V-47
—NEWPCP Effluent Monitoring V-48
—NEWPCP Influent-Effluent Comparisons V-49
—NEWPCP Sludge Monitoring V-50
—Volatile Air Concentrations at the
NEWPCP V-51
SUMMARY V-51
-------
CONTENTS (continued)
VI. RISK ASSESSMENT AND CONTROL-OPTIONS ANALYSIS
RESULTS
COST-EFFECTIVENESS ANALYSIS VI-2
SOURCES AND POLLUTANTS VI-4
• Background VI-4
• Pollutants VI-4
• Point Sources VI-7
• Urban Area Air Sources VI-9
DATA REQUIREMENTS AND MODELING VI-9
• Baseline Risk VI-11
—Unit Risk Values VI-12
• Control Options VI-15
RESULTS OF THE ANALYSIS VI-16
• Baseline Risks Under Current Controls VI-17
—Aggregate Excess Cancer Incidence VI-17
Air and Drinking Water VI-17
Drinking Water VI-17
Air VI-20
— isks to the Most Exposed Individual VI-22
Cancer VI-22
Noncancer Effects VI-24
Cost-Effective Control Strategies VI-25
—Aggregate Excess Cancer Incidence VI-25
• Air and Drinking Water VI-25
• Air VI-31
• Drinking Water VI-33
—Risks to the Most Exposed Individual VI-36
Risk Context VI-38
—Mortality Statistics for Philadelphia
County VI-39
—Ambient Air Versus indoor Air Exposures VI-40
—Comparison of Air Exposure Risks From
Several Studies VI-40
VI
-------
CONTENTS (continued)
RECENT CHANGES IN RISK ASSESSMENT DATA VI-42
• Unit Risk Values VI-43
—Cancer VI-43
—Noncancer Effects VI-44
• Discharges to the NEWPCP VI-44
• Implications for the Control-Options
Analysis VI-44
—Baseline Risk VI-46
• Cancer incidence VI-46
• Risks to the Most Exposed Individual VI-46
• Cost-Effective Control Strategies VI-50
—Aggregate Cancer Incidence VI-50
LIMITATIONS
r
t
\
VII. CONCLUSIONS
SUMMARY OF PHASE I OF THE PHILADELPHIA IEMP
PROCESS AND FINDINGS VII-1
PHASE II FINDINGS VII-4
• Baseline Risk and Control-Options Analysis VII-4
—Methodology VII-4
—Limitations and Uncertainties VII-5
• Limitations in Scope VII-5
• Limitations in Exposure Data VII-6
• Limitations in Toxicological Data VII-6
• Limitations in Estimating Control
Option Costs VII-7
• Risk Assessment and Control-Options
Analysis Results VII-8
• Monitoring Activities Vll-14
• Significant Substantive Findings VII-15
—Benzene in the Ambient Air VII-15
—Formaldehyde in the Ambient Air VII-16
—Combustion of Used Oil VII-16
—Air Emissions from Landfills VII-17
OBSERVATIONS FOR THE FUTURE VII-17
vii
-------
CONTENTS (continued)
APPENDIXES
Appendix A:
Appendix B:
Appendix C:
Appendix D:
Appendix E:
Appendix F:
Appendix G:
Appendix H:
Appendix I:
Appendix J:
Benefit-Cost Analysis for the Philadelphia
Region
Preliminary List of Pollutants in Phase I
Summary of Preliminary Data Collected in
Phase I
Issue Papers
Model Description
Characterization of Pollutant Releases
Estimation of Exposure and Risk
Description of Control Options by Source
Category
Pollution Control Costs and Equipment Life
Pollutant Concentrations and Health Effects
by MEI Location
viii
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
LIST OP TABLES
III-l Employment and Economic Statistics for III-4
Philadelphia County
III-2 Participants in the Study III-7
III-3 Members of Steering and Technical Committees IIT-9
III-4 Apportionment of Area-Source Emissions by Land 111-19
Use
III-5 sample Issue Paper 111-24
III-6 Categories of Pollutants and Status of Primary III-31
Criteria
III-7 Data Confidence Rating for Phase I Risk Screen 111-32
III-8 Pollutant Selection 111-35
III-9 Phase I Climatological Dispersion Model (COM) 111-36
Results
111-10 Emission Loads 111-38
III-ll Phase II Cost-Effectiveness Analysis: Sources 111-40
and Pollutants by Receiving Medium
111-12 Overview of Phase II Study Topics and III-43
Activities
\
IV-1 Average Ambient Air Benzene Concentrations: IV-5
AMS Monitoring
IV-2 Ambient Air Benzene Concentrations: Delaware IV-6
Avenue Incinerator Yard Gas Station: AMS
Monitoring
IV-3 Preliminary Approximations of Average Lifetime IV-8
Individual Upper-Bound Risk from Exposure to
Benzene Concentrations (1984)
IV-4 Preliminary Approximations of Average Lifetime IV-9
Individual Upper-Bound Risk from Exposure to
Benzene Concentrations (1986)
IV-5 Average Ambient Air Formaldehyde Concen- IV-13
trations: AMS Monitoring Data
IV-6 Preliminary Approximations of Average Lifetime IV-15
Individual Upper-Bound Risk from Exposure to
Formaldehyde Concentrations (1984)
IV-7 Preliminary Approximations of Average Lifetime IV-17
Individual Upper-Bound Risk from Exposure to
Formaldehyde Concentrations (1986)
IV-8 Average Concentrations of Volatile Organic IV-21
Compounds at New Jersey Landfill
IV-9 Levels of Metal Contaminants in Waste Oil IV-25
Samples Collected in New Jersey and
Pennsylvania
ix
-------
V-l Average Measured Concentrations for Air
Monitoring Sites
V-2 Specifications of the SHORTZ Model for Model
Performance Evaluation
V-3 Summary of Modeled and Measured Concentrations V-13
V-4 Carcinogenic Potencies (Inhalation) and EPA V-23
Weight of Evidence Grouping (1984)
V-5 Rough Approximations of Average Lifetime V-24
Individual Risks: Monitored Versus Modeled
Ambient Air Concentrations (1984)
V-6 Noncancer Health Effects and RID-Derived Human V-25
Thresholds for Air (1984)
V-7 Carcinogenic Potencies (Inhalation) and EPA V-26
Weight of Evidence Grouping (1986)
V-8 Rough Approximations of Average Lifetime V-27
Individual Risks: Monitored Versus Modeled
Ambient Air Concentrations (1986)
V-9 Noncancer Health Effects and RID Human V-28
Thresholds for Inhalation (1986)
V-10 Pollutants Analyzed in Delaware River Sampling V-34
V-ll Pollutant Concentrations in Delaware River V-41
Transect Station
V-12 Pollutant Concentrations in Delaware River, V-43
Station C, Upstream Control
V-13 Pollutant Concentrations in Delaware River, V-45
Station B
V-14 Pollutant Concentrations in NEWPCP Influent V-48
V-15 Pollutant Concentrations in Northeast Water V-49
Pollution Plant Effluent
V-16 Volatiles in NEWPCP Sludge Concentrations in V-50
ppb (ng/g)
V-17 Air Concentrations at NEWPCP V-52
VI-1 Cancer Unit Risk Factors (Upper-Bound) Used in VI-13
the Control-Options Analysis
VI-2 Noncancer Health Effects and Presumed Human VI-14
Thresholds for Inhalation and ingestion
VI-3 Upper-Bound Estimates of Excess Annual Cancer VI-18
Incidence by Source
VI-4 Upper-Bound Estimates of Annual Excess Cancer VI-19
Incidence at Current Control: Ingestion of
Philadelphia Drinking Water
VI-5 Average Chloroform Levels in Drinking Water for VI-20
Selected Cities
VI-6 Upper-Bound Estimates of Annual Excess Cancer VI-21
Incidence at Current Control: Inhalation
VI-7 Upper-Bound Estimates of Cumulative Lifetime VI-22
Cancer Risks to the Most Exposed Individual
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
VI-8 Percentage Contributions of Pollutants to VI-23
Cumulative Lifetime Cancer Risks in Air by
Facility
VI-9 Threshold Pollutant Hazard index for Each MEI VI-24
Location: Air and Drinking Water
VI-10 Schedule of Control Options for Reducing Annual VI-26
Cancer Incidence: Air and Drinking Water
Vl-11 Schedule of Control Options for Reducing Annual Vl-34
Cancer Incidence: Air
VI-12 Cost-Effective Reduction in Cancer Incidence VI-35
(Upper-Bound) for Drinking Water Plants
VI-13 Schedule of Control Options for Reducing Risk VI-37
to Most Exposed Individual: Air and
Drinking water
VI-14 Rates of Resident Deaths by Selected Causes by VI-39
Health District: All Races
VI-15 Comparison of Selected Cancer Risks from indoor VI-41
Air Exposures to Ambient Exposures in
Philadelphia
VI-16 Comparison of Annual Cancer incidence Estimates VI-42
for Air Toxics in Philadelphia County with
Estimates from Other Studies
VI-17 Cancer unit Risk Factors (Upper-Bound) Used in VI-43
the Control Options Analysis
VI-18 Noncancer Health Effects and RID-Derived Human VI-45
Thresholds: Inhalation and Ingestion
VI-19 Cancer Incidence at Current Control VI-47
VI-20 Upper-Bound Estimates of Excess Annual Cancer VI-48
Incidence by Source: Current Control
VI-21 Upper-Bound Estimates of Cumulative Lifetime VI-49
Cancer Risks to the Most Exposed Individual
VI-22 Threshold Pollutant Hazard index for Each MEI VI-50
Location: Air and Drinking Water. (Current
Control)
VII-1 Overview of Phase II Study Topics and VII-3
Activities
VII-2 Upper-Bound Estimates of Excess Cancer VII-9
Incidence by Source: Current Control
(1984)
VII-3 Upper-Bound Estimates of Cumulative Lifetime VII-11
Cancer Risks to the Most Exposed Individual:
1984 Analysis
XI
-------
LIST OF FIGURES
III-l Map of Philadelphia and Surrounding Counties III-3
III-2 Study Area Boundaries 111-13
III-3 Example of PIPQUIC Computer Graphics 111-22
III-4 Receptor Locations on Grid System 111-28
III-5 Study Scope: Location of Philadelphia 111-41
Pollutant Sources
IV-1 AMS Sampling Sites at the Intersection of IV-3
Bustleton Avenue and Red Lion Road
IV-2 Map of Formaldehyde Monitoring Sites 17-11
V-l Cost-Effectiveness Analysis of Control Options V-2
V-2 Ambient Air Monitoring Locations V-5
V-3 Daily Comparisons Between Predicted and V-14
Observed Concentrations for
1,2-D ichloropropane
V-4 Daily Comparisons Between Predicted and V-15
Observed Concentrations for
1,2-Dichloroethane
V-5 Daily Comparisons Between Predicted and V-16
Observed Concentrations for
Perchloroethylene
V-6 Daily Comparisons Between Predicted and V-17
Observed Concentrations for Chloroform
V-7 Daily Comparisons Between Predicted V-18
and Observed Concentrations for Carbon
Tetrachloride
V-8 Sampling Sites for Delaware River Study V-32
V-9 Sampling Sites for NEWPCP Study of Liquids and V-37
Sludge
V-10 Sampling Locations for Air Monitoring V-39
V-ll Baxter Mean Raw and Finished Water V-46
VI-1 Study Scope: Location of Philadelphia Vl-6
Pollutant Sources
VI-2 Coat-Effectiveness Analysis of Control Options VI-10
VI-3 Cost vs. Reduction in Cancer Incidence: Air VI-29
and Drinking Water
VI-4 Cost vs. Reduction in Cancer Incidence: Air VI-32
VII-1 Cost vs. Reduction in Cancer Incidence: Air VII-13
and Drinking Water
xn
-------
LIST OF EXHIBITS
Page
VI-1 Description of Control Options: Air and VI-53
Drinking Water
I
I
I
I
I
I
I
xiii
I
I
-------
Final Report of the
Philadelphia Integrated Environmental
Management Project
Executive Summary
Regulatory Integration Division
Office of Policy Analysis
Office of Policy, Planning, and Evaluation
U.S. Environmental Protection Agency
October 1986
-------
EXECUTIVE SUMMARY
Philadelphia Integrated
Environmental Management Project
INTRODUCTION
The Philadelphia Integrated Environmental Management Project
(IEMP) was an innovative study designed to analyze and address
the problems posed by toxic chemicals in the Philadelphia area.
This project became a prototype for a series of lEMPs that the
U.S. Environmental Protection Agency (EPA) has conducted over the
last four years. The Project's goals were, therefore, both to
provide insights on particular environmental issues and to
develop new general methods. Our objectives were:
• To develop a methodological approach to evaluating and
comparing the risks to human health caused by exposure
to toxic pollutants in the environment, as measured by
cancer and, to a lesser degree, other chronic health
effects
• To use this evaluation to help local officials in the
process of setting priorities for more detailed analy-
sis and, where appropriate, regulatory controls
• To involve local government agencies in the development
and review of the analysis
The concept of integrated environmental management developed
out of EPA's recognition that there are drawbacks to the tradi-
tional approach that EPA and the states have used to develop
environmental regulations. That approach has focused on indi-
vidual industries, pollutants, and media. While very useful, it
has nevertheless limited our ability to determine where among the
various media our resources are best employed to get the most
health protection. It does not ensure that pollution controls
are not merely shifting risk from one medium to another. In
addition, we have often used national standards that do not ade-
quately address all site-specific situations.
In contrast, the IEMP approach provides a multimedia analy-
sis of issues and accounts for transfers of toxics across media--
in land, air, surface water, and ground water, it is founded
principally on the concepts of risk assessment and risk manage-
ment. Risk assessment requires a systematic evaluation of the
potential for adverse effects to human health from exposure to
-------
pollution. Risk management is a process for evaluating pollution
controls in terms of their projected cost and the level of risk
reduction they afford. Finally, we use the IEMP approach to
focus on the issues of one community at a time, so we can develop
environmental management strategies tailored to each area's
unique problems and characteristics.
Risk Assessment
The IEMP uses risk assessment techniques to evaluate and
compare the potential problems from toxic pollutants in the air,
land, and water, in Philadelphia, we focused specifically on
assessing the risks in air, surface water, and drinking water.
At times this work was qualitative in nature. However, in other
situations, we were able to provide quantitative estimates of the
probability that an individual would contract cancer and of the
expected cancer incidence in the entire population. Our tech-
niques for estimating health effects are in accordance with stan-
dard EPA practices, as described in the proposed EPA risk assess-
ment guidelines. Despite these standards, our analysis has some
significant uncertainties, which we elaborate below. The goals
of our risk assessment in Phase I of the Philadelphia IEMP were
(1) to determine which toxics issues were suitable for a more
detailed examination of risks and control options in Phase II and
(2) to identify study topics for which we would initiate ambient
monitoring programs to better assess exposure levels and, in some
cases, potential risks.
Risk Management
Risk management is the process by which policymakers balance
programs to reduce human health risks against the available
resources to support those programs. In its simplest form, it
requires an examination of how large the risks are, how much the
risks can be reduced by various regulatory controls, and the
costs of these controls. However, it often entails a lot more.
It can involve consideration of the strength of evidence we have
that a health problem could exist or whether effective regulatory
controls can be enforced. Risk management is a process that
requires an assessment of all pertinent information before deci-
sions are made to control pollution. Sometimes the appropriate
decision may be to 'conduct further analysis of the problem. In
Phase II of this study, we dedicated a significant amount of
effort to providing the data and analysis that serve risk manage-
ment needs. This is especially true of areas where we carried
out further problem definition through ambient monitoring or
performed cost-effectiveness analyses to allow local policymakers
to examine risk reductions and costs of controls for general
policy evaluation.
-------
Choice of Location
Philadelphia was chosen as the site of this pioneering study
because it provided a good setting for the examination of multi-
media issues. The City offered the benefits of several rela-
tively good toxics databases and local officials who we believed
provided a strong technical resource for project implementation.
Philadelphia was not chosen because it had significant environ-
mental problems. Local officials had already demonstrated exem-
plary leadership in some areas. EPA and local officials wanted
to explore ways to identify, assess/ and manage human health
risks in an area where significant progress had been made in
understanding and controlling toxics issues.
Conclusions
In the following sections we discuss the methodological and
substantive conclusions from the Philadelphia IEMP. First, we
review our activities in Phase I, largely devoted to screening
pollution issues, and report our findings. Second, we present
findings from our Phase II risk assessment and control-options
analysis. We also discuss the limitations in conducting risk
assessments of toxics. Third, we summarize our methodological
findings from the ambient monitoring programs and the substantive
insights gained from these activities. Finally, we close with
observations on our application of the IEMP methodology in
Philadelphia.
SUMMARY OF PHASE I OF THE PHILADELPHIA
IEMP PROCESS AND FINDINGS
Phase I consisted of three major activities:
• Establishing two intergovernmental committees. The
,first was the Steering Committee, which included pri-
marily senior appointed officials from all participat-
ing^ jurisdictions and EPA. The other group, the Tech-
nical Committee, consisted of technical staff from the
environmental agencies. The Steering Committee
directed the study. The Technical Committee reviewed
and guided the technical and scientific activities.
• Gathering and reviewing data, we collected data to
assess the potential effects of human exposure to toxic
chemicals. In some instances, data directly useful for
measuring exposure were already available, e.g., data
-------
on the quality of finished drinking water. In other
cases, especially in the case of air, we estimated the
expected concentration of a compound in the environment
using data on pollutant sources and the manufacturing
processes, which were used to run fate and transport
models. After gathering available data, we organized
it to facilitate easy review by those on our project
committees who had considerable knowledge about the
quality of the data. We relied heavily on their expert
judgment.
• Developing a screening process for identifying critical
toxics issues for detailed assessment in Phase II"
Once we had collected, organized, and reviewed the
preliminary data, we narrowed the project's geographic
and analytical scope. We set project priorities on the
basis of the primary criteria: quantitative and quali-
tative measures of risk. Then we applied secondary
criteria. These criteria included EPA, state, and
local program objectives, the analytical feasibility of
examining the issues, and our ability to control
environmental impacts.
After applying the screening criteria, which narrowed down
our original list of 170 pollutants associated with over
376 point and area sources, we arrived at a subset of 17 sources
of nine toxic chemicals released to the ambient air and surface
water and contained in drinking water. These sources and chemi-
cals constituted our Phase II study topics, which, along with our
analytical activities and objectives, are set out in Table 1.
We emphasize that not all study topics had the same objec-
tives. We also note that we could not analyze each issue with
the same degree of technical rigor. Finally, the Phase I results
warrant two concluding remarks:
• We successfully identified a manageable set of multi-
media topics that could be usefully examined with
available EPA resources and analytical methods and that
would address issues of interest to the community. Our
Phase II efforts would focus on toxic chemicals In air
from point and area sources, toxics in finished drink-
ing water, and intermedia transfers of toxics from the
City's major POTW.
• We had to exclude some important environmental issues
early in Phase I of the Philadelphia IEMP, for several
reasons"First, limited resources required that we set
priorities among possible topics. Second, limited data
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
Study Topics
Table 1
PHILADELPHIA IEMP
OVERVIEW OF PHASE II STUDY TOPICS MO ACTIVITIES
Analytical Activitiaa
Objectives
Risk And Control-Option* Analysis
-Identified through screening of
available data
-Addreaaed 7 of the 11 initial
screening issue papers
- Benzene emissions
- Area sources of solvents
- Refinery, pipeline, and
terminal emissions
- Baxter drinking water
- Queen Lane/Belmont drinking
water
- Multimedia discharges of
1,2-OCE and 1,2-OCP
- Haloforna (chloroform)
—Monitor ambient air
—Monitor ambient water
—Validate emission estimates
—Recalculate exposures and risk*
using air dispersion models
—Identify feasible controls
—Calculate control costs and
removal efficiencies
—Develop model to evaluate and
rank coat-effective control
option strstegies
—Provide reeults in useful format
for review by decision makers
—Determine baseline expoaures
—Identify which sources, pollu-
tants, and exposure pathways
contribute most significantly to
eatinated human health risk
—Develop coat-effective strategies
for reducing risks to human
health:
- Aggregate excesa cancer
incidence
- Riaka to the Host Exposed
Individual
—Examine noncarcinogenic risks
—Identify limitations of analysis
Monitoring
1. Benzene Emission* (specifically
from gasoline marketing)
—Identified by Philadelphia as
a topic of intereat
2. Formaldehyde Release* to Uw
Ambient Air
—Identified by Philadelphia aa
a topic of interest
3. Combustion, of Itamd Oil
—Identified by all study
participants) as a topic of
intereat
-Short-term benzene anbient air
monitoring at selected intersec-
tions with and without service
stations
—Ambient air monitoring
—Sampling at points of distribu-
tion and uae
—Analysis of samples for metals
and organic*
—Determine significance of
observed concentrations
- Philadelphia air guidelines
- Cancer risk
—Determine significance of
observed concentrationa
- Philadelphia air guidelines
- Cancer risk
—Compare observed used oil concen-
trationa with EPA fuel specifica-
tions
4. Air Emissiona from Landfills
(focusing specifically on New
Jersey landfills)
—Identified by New Jersey and —Short-term monitoring at a select —Identify compounds and observed
EPA as a topic of interest number of sanitary and hazardous concentrations
waata landfills in New Jersey
—Satisfied EPA interest in test- (but within geographic boundary —Determine significance of
ing the* applicability of a of project) measured concentrations
mobile monitoring system for
use in future geographic
studies
-------
restricted the extent of our analysis. Third, certain
research and analytical methods now in practice were
not available at the time of the study. Changes in any
of these factors probably would have altered our selec-
tion of issues. We wish to emphasize that the absence
of an issue from our list of study topics does not mean
that the issue is insignificant.
PHASE II FINDINGS
In the,following sections we present the findings of the
analytical efforts to characterize risks to human health and •
examine the cost-effectiveness of control options to reduce these
potential hazards. We also discuss our conclusions from the
ambient monitoring programs. In each area, we summarize both
methodological and quantitative results. Since the Philadelphia
IEMP was a pilot study, we believe that both types of findings
are important for consideration.
Baseline Risk andxControl-Options
Analysis for Policy Development
Methodology
We were successful in designing an analytical approach that
could'be used by decision makers to identify which environmental
issues present the most significant risks, roughly quantify the
magnitude of these risks, and evaluate the cost-effectiveness of
various control strategies to reduce the risks that we could
quantify^We were successful for several reasons:
• We were able to enlist the critical participation of
the members of our project committees, most notably the
representatives of the Philadelphia Water Department
and Air Management Services (AMS).
• The AMS emissions inventory allowed us to select a
limited number of pollutants and sources that contrib-
uted most significantly to ambient air releases of the
toxic pollutants included in our analysis. Without
this inventory, it would have been extremely difficult
to narrow the scope of the project.
• We employed existing EPA analytical methods used in
Agency research and rulemaking activities to generate
much of the data and our analytic results. We tailored
their application to our site-specific needs.
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
Limitations and Uncertainties
The success of providing an analytical approach to evaluate
the risks to human health and the cost-*ffectiveness of alter-
native control options to reduce these risks must be appreciated
in the context of the uncertainties associated with this work,
We discuss below the limitations in the project scope, exposure
data, toxicological data, atsfl estimates of the cost of controls.
Limitations in Scope. The Philadelphia IEMP was not an
epidemiological atxj&y, we did not collect data on diseases that
occurred ia local populations or attempt to trace their causes,
environmental or otherwise. Instead, we combined local data and
engineering estimates of environmental exposure to toxic chem-
icals with toxicological data to estimate the risks to human
health.
The Philadelphia IEMP only attempted to estimate the health
risks from exposure to toxic chemicals in the ambient environ-
ment. For example, we did not estimate risks resulting from
occupational and indoor air exposures. We also did not include
exposures through the food chain. The omission, of these routes
of exposure in our study does not imply that they are unimpor-
tant. It is quite possible that risks from any of these exposure
pathways could exceed risks from the exposures we considered, we
decided not to assess these exposure routes because of resource
constraints and because these areas were generally outside of
EPA's traditional regulatory purview and area of expertise.
We chose not to analyze the exposure and risks from conven-
tional pollutants in air and water (such as ozone and sulfur
oxides in air, and oxygen-depleting substances and oil and grease
in water) because we believed that we could make a more signifi-
cant contribution by concentrating on toxic chemicals, which are
neither as well understood nor as well regulated. Future proj-
ects, however, may want to consider these pollutants.
Limitations in Exposure Data. Another important limitation
to our analysis is that we did not examine exposures associated
with all sources and pollutants. While we tried to identify and
assess the cancer risks from the most significant sources and
pollutants, some of those for which we were unable to estimate
exposure, such as combustion of used oil, may also pose health
risks.
Even where exposure data were available, those data varied
significantly in quality. As a result, the exposure estimates
vary in their reliability. Those based on extensive monitoring,
-------
such a£ rihal. methanes in drinking water and selected chlor-
inated soi/encs in air, are the best exposure estimates we e.
The exposure d«*--& "om :noit-term ambient air monitorina : ,„,:
benzene from service St., "ions and formaldehyde are much more
Exposure estima , derive:? from modeling also vary in their
reliability. Estima :> of exposure to toxic organic chemicals in
air derived from dispersion models are dependent primarily on the
quality o£ the emissions data and a tt. other actors, such as
meteorological data. In general, the unc " .ies introduced by
the exposure data are probably much smaller ^- n those ass^-iated
with the dose-response information used to estimate human health
risks.
Limitations in Toxicological Data. Estimates of health
effects are designed to be conservative in several ways.
evaluating potential health hazards from a chemical, EPA scien-
tists assume that health effects observed in laboratory animals
are a reasonable indicator of potential effects in humans. In
converting the animal ja to estimate predicted human responses,
and in extrapolating? .... om high doses to low doses, we use models
**uit yield a plausible, upper-bound estimate of potency rather
than a "best guess" estimate.
Many substances of potential concern have never been evalu-
ated scientifically, or have not been evaluated in sufficient
detail to allow estimation of effects on humans. For example,
lead (present in air, water, and dust) is thought to pose a
health risk to children at ambient levels; however, at the time
we conducted t- .^3 project, we had no way of estimating individual
risks or numbers of possible cases.
The addition of new toxicological data and revised scienti-
fic interpretations of previous animal studies often leads to
revised potency values. This makes risk assessment of various
chemicals subject to changes in scientific understanding. Since
the time- wet completed our Phase II analysis, several unit risk
factors uaeet have been revised by EPA's Carcinogen Assessment
Group (CAG3FC However, these revisions do not appear to change
our findings dramatically.
We relied on CAG unit risk factors for all chemicals except
1,2-dichloropropane. We developed the inhalation and ingestion
unit risk factors for 1,2-dichloropropane using the (potency)
q^* value found in EPA's Drinking Water Criteria Document on
1,2-dichloropropane (March 2, 1984). The Drinking Water Criteria
Document is nearing completion of external review and, as a
result, the unit risk factor for 1,2-dichloropropane could
change.
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
/ Despite these uncertainties, our risk estimates are useful
./policy analysis results for comparing issues with one another;
setting priorities among environmental issues and concerns that
we examined; and roughly assessing the potential magnitude of the
overall risks from particular pollutants, sources, and pathways.
Limitations In Estimates of the Cost of Controls. For a
variety of reasons, we were unable to obtain complete site-
specific control cost information for all plants in our control-
options analysis. We were able to gather, with the assistance of
AMS, detailed information about the manufacturing process and
variations in production levels for many plants. Many of the
estimates are based on best engineering judgments, using standard
cost estimation techniques employed in EPA regulatory activities.
Risk Assessment and Control-
Options Analysis Results
Within the limitations described earlier, we are able to
draw the following conclusions, for use in general policy
development, about the nine pollutants and 17 sources we
investigated:1 ^
1. Drinking water accounted for over four-fifths of the
estimated excess cancer risks we found in our analysis.
Our upper-bound estimate of aggregate excess cancer
incidence for the general population of about 1.7 mil-
lion in Philadelphia was close to three cases per year.
Table 2 shows the upper-bound excess cancer risks by
source category and exposure pathway. The compound
responsible for most of the estimated upper-bound
excess cancer incidence in drinking water was chloro-
form, a trihaloraethane. Chloroform's presence in
finished drinking water is a byproduct of the process
for disinfecting the water. Chloroform concentrations
in the Philadelphia drinking water are less than half-
ofe EPA's primary interim drinking water standard.
2. The, remaining cancer risks (less than one-fifth) were
attributable to exposures in air and are divided about
evenly between point and area air sources.Table 2
shows the different air source categories we analyzed
1Unless otherwise noted, all conclusions on risk and control
strategies are based on the results from our original (1984)
analysis using the environmental and health data available at
that time.
-------
10
Table 2
PHILADELPHIA IEMP
PHASE II RESULTS INTENDED FOR POLICY DEVELOPMENT1
UPPER-BOUND ESTIMATES OF EXCESS ANNUAL CANCER INCIDENCE BY SOURCE
CURRENT CONTROL
(198* analyaie)2
Sources
Air (point)
PherMceuticel Manufacturer
Cheaical Manufacturer
Garment Manufacturer
Plaatic Cabinet Manufacturer
Induetrial Dry Cleaner
Refinery B
Refinery A
Northeaat MPCP3
Subtotal
EstiMted Exceaa
Annual Cancer
Incidence5
(caaee/yeer)
.006
.007
.001
.000
.000
.06*
.007
.090
.177
Air (area) v
Degreaeing
Dry Cleaning
Other Industrial Usage
Gaaoline Marketing (excluding aelf-aervice)
Sewer Volatilization (NEWPCP)
Delaware River (receiving NEWPCP effluent)
Subtotal
Drinking Water
Baxter DWTP
Belnont OMTP
Queen Lane DWTP
Subtotal
Total4
Percent.
of Total
Caaee
.2
.2
.0
.0
.0
2.3
.2
3.2
••M^K
6.3
1.7
2.3
.1
1.9
.7
.8
7.6
43.1
15.8
27.2
86.2
100.0
J
NPCT • Water Pollution Control Plant.
DWTf s Drinking Water Treatment Plant.
Note* Numbers have three declMl pieces not ee en indication of precision, but to
identify source contribution to the riafce.
1THE UNIT RISK FACTORS USED IN THIS ANALYSIS ANE BASED CN CONSERVATIVE ASSUMPTIONS
THAT GENERALLY PRODUCE UPPER-BOUND ESTIMATES. BECAUSE OF LIMITATIONS IN DATA AND
METHODS IN SEVERAL AREAS OF THE ANALYSIS, SUCH AS EXPOSURE CALCULATIONS AND
POLLUTANT SELECTION, RISK ESTIMATES WERE CALCULATED AS AIDS TO POLICY DEVELOP-
MENT, NOT AS PREDICTIONS OF ACTUAL CANCER RISKS IN PHILADELPHIA. ACTUAL RISKS
MAY BE SIWIFICANTLY LOWER; IN FACT, THEY COULD BE ZERO. THE PROPER FUNCTION OF
THE ESTIMATES IS TO HELP LOCAL OFFICIALS SELECT UNO EVALUATE ISSUES, SET PRIORI-
TIES, AND DEVELOP CONTROL STRATEGIES FOR THE TOPICS EXAMINED.
2The riak eatiMteo preaented in this table *er» calculated ueing unit risk
factors froei 1984.
'Recent reductions in dischargee to the NEWPCP MX result In lower risk nunbers
than presented in this table.
*Colusns My not su»> due to rounding.
*IEc, Inc., Cost-Effectiveness Analyeis of Strategies to Reduce Hu«an Health Ri«k
in Philedelphia. U.S. EPA, May 1965.
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
11
and the estimated cancer incidence associated with
exposures to ambient air releases from each. The tra-
ditional point sources (i.e., smokestack industries)
were accountable for a modest percentage (nearly 3 per-
cent) of the risks we could quantify.
Roughly one-third of the air toxics risks are attribu-
table to intermedia transfers (from water to air)
resulting from industrial wastewater discharges to the
Northeast Water Pollution Control Plant (NEWPCP).
Roughly two-thirds of the estimated upper-bound excess
cancer risks from these intermedia transfers occur at
the sewage treatment plant itself. The remaining one-
third is split fairly evenly between volatilization
from the major sewer line into the NEWPCP and volatil-
ization from the discharge to the Delaware River.
We found that the cumulative upper-bound cancer risks
from multiple chemical exposures are greater than
single-chemical risks by a factor of 10 and/ in some
cases, by a factor of 100.EPA's regulatory actions
often focus on individual pollutants, such as the list-
ing and control decisions for hazardous air pollutants
under Section 112 of the Clean Air Act/ and do not
account for the cumulative risks that may occur from
exposure to multiple chemicals.
We found no concentrations of toxic chemicals that
warrant increased concern about noncarcinogenic
effects.However,independent monitoring information
(e.g., AMS's breathing zone study) indicated that ben-
zene concentrations could pose some concern. A better
determination of the significance of these concentra-
tions will depend on wher« th« "no-effect" thresholds
for benzene, which are currently under internal EPA
review, are finally set.
People could experience upper-bound individual lifetime
risks of cancer of around 1 in 10,000 from their drink-
ing water and in some cases of air exposure. However,
upper-bound risks to the maximum exposed individuals in
air generally are above 1 in 100,000.Table 3 shows
the upper-bound risks to the maximum exposed individual
(MED from air and drinking water. We assumed that all
MEIs are served drinking water by the Baxter facility.
-------
12
Table 3
PHILADELPHIA IEMP
PHASE II RESULTS INTENDED FOR POLICY DEVELOPMENT1
UPPER-BOUND ESTIMATES OF CUMULATIVE LIFETIME CANCER RISKS
TO THE MOST EXPOSED INDIVIDUAL
(1984 analysis)2
MEI Location
Northeast MPCP3
Rafinary B
Chemical Mfr.
Plaatic Cabinat Mfr.
Pharmaceutical Mfr.
Garment Manufacturer
Rafinary A
Induatrial Dry Cleaner
Cumulative Lifetime
Cancer Riak
(Upper-Bound) Inhalation
5.6 i
1.4
2.2
6.5
4.5
1.2
3.0
2.2 i
t ID'5
10-5
10-4
10-*
10-5
ID'5
10-5
< IDT5
Cumulative Lifetime
Cancer Riak
(Upper-Bound) Inqeation
Cumulative Lifetime
Total Cancer Riak
(Upper-Sound) Inhalation
and Inqaation
1.0 x
1.0 x 10"*
1.0 x 10-*
1.0 x 10~*
1.0 x 10-*
1.0 x 10-*
1.0 x 10-*
1.0 x 10-*
1.6 x 10-*
1.1 x 10-*
3.2 x 10-*
1.1 x 10-*
1.4 x 10-*
1.1 x 10-*
1.3 x 10-*
1.2 x 10-*
1THE UNIT RISK FACTORS USED IN THIS ANALYSIS ARE BASED ON CONSERVATIVE ASSUMPTIONS THAT GENERALLY
PRODUCE UPPER-BOUND ESTIMATES. BECAUSE OF LIMITATIONS IN DATA AND METHODS IN SEVERAL AREAS OF
THE ANALYSIS, SUCH AS'EXPOSURE CALCULATIONS AND POLLUTANT SELECTION, RISK ESTIMATES MERE
CALCULATED AS AIDS TO POLICY DEVELOPMENT, NOT AS PREDICTIONS OF ACTUAL CANCER RISKS IN PHILA-
DELPHIA. ACTUAL RISKS MAY BE SIGNIFICANTLY LOWER; IN FACT, THEY COULD BE ZERO. THE PROPER
FUNCTION OF THE ESTIMATES IS TO HELP LOCAL OFFICIALS SELECT AND EVALUATE ISSUES, SET PRIORI-
TIES, AND DEVELOP CONTROL STRATEGIES FOR THE TOPICS EXAMINED.
?The riak estimate* presented in this table ware calculated using unit riak factors from 1984.
^Recent reductions in dischargee to the NENPCP may result in lower riak numbers than presented
this table.
Source: lEc, Inc., Coat Effectiveneaa Analysis of Strategies to Reduce Human Health Riak in
Philadelphia. U.S. EPA, May 1985.
in
8
Decision makers wanting to achieve different levels of
risk reduction will need to employ different control
strategies.For example, to reduce risks by less than
20 percent, the most cost-effective control strategies
focus solely on lowering ambient air releases of
toxics. However, to achieve significantly greater
reductions in risk, some previously recommended air
controls may be replaced by controls at the drinking
water treatment plants. Controls at the drinking water
treatment plants can achieve large, discrete reductions
in risk,.as opposed to the smaller, incremental reduc-
tions achievable through air emission controls at
different sources.
Because drinking water risks account for over four-
fifths of the estimated baseline cancer risks,
Philadelphia would have to implement controls at the
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
13
drinking water treatment plants to obtain a very sig-
nificant risk reduction in the estimated upper-bound
excess cancer incidence for the sources and pollutants
considered in our analysis. It is important to recog-
nize, however, that the upper-bound estimated individ-
ual risk associated with ingestion of Philadelphia
drinking water is smaller than the current nationally
accepted risk for drinking water, based on the concen-
trations established by EPA's primary interim drinking
water standards.
9. A significant reduction of risk would require that the
City make significant expenditures for controls at the
drinking water treatment plants. This option presents
local decision makers with a difficult risk management
decision, involving the examination of the amount of
risk reduced, the control costs, and other factors. As
stated above, over four-fifths of the risks found by
the IBMP came from drinking water. Significant reduc-
tions in the upper-bound estimated cancer risks from
drinking water can be achieved only by reducing chloro-
form levels through the user of granular activated
carbonN(GAC) filters. These controls reduce the chlo-
roform that forms in drinking water following
disinfection.
In our cost-effectiveness analysis, we incrementally
added GAC controls to each drinking water treatment
plant to achieve increasing levels of risk reduction.
The results of our analysis suggest that implementing
GAC at the Belmont drinking water plant only could
reduce approximately one-fifth of the estimated drink-
ing water risks. Implementing GAC at all three drink-
ing water plants could reduce nearly all the estimated
cancer risk. The annual costs of these controls would
range from roughly $11 million to $58 million. The
average cost per cancer case avoided ranges from around
^ $24 million to $26 million, depending on the drinking
^ water treatment plant controlled. The value of the
reduction depends on what other risk reduction oppor-
tunities are possible for these same expenditures and
the value of other social services that could use these
resources.
10. For the range of control options we examined, we did
not see any dramatic increase in the cost per case
avoided as we pursued additional control strategies to
reduce further increments of risk. Figure 1 illus-
trates our results. Typically one would not expect
such a linear relationship; rather, one would expect
-------
14
O £ to
0)3 O
2 £O
S'll
s^
/?»- o
°- oo
en
*
11
•&3S
O
Is
CD 0)
5s
*•* * %
O.J?
So
UJ Q.
o —-2
I i-"§
m -s|
^ s
££
0)
(7)
O
OC
8 S
c §
CO >
O g
C 3E
— c
C X
1!
**
DC
(0
0)
= o
(D O
0
0)
CO
I
i!L
^- "5
(D
(O
O
(O
(0
O
-~ >-^
5 o o
III
if!
CD 13 Q.
. 2 O
to fl fC
o
» O O =
§^€1
« 5 . Q.I
« o *•*
flZs]
5§«'
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
15
that, as costs increase, a larger fraction of the
potential risk could be reduced for a significantly
smaller portion of the cost. We attribute this result
to two factors affecting the dominant sources of risk
(i.e., the drinking water treatment plants). First,
the available inexpensive controls, such as switching
from chlorine to chloramines as the primary disinfec-
tant, have already been implemented. Second, as noted
above, the increase in risk reduction would be achieved
by applying the same control technology (i.e., GAG
filters) to additional drinking water treatment plants.
11. Some controls to reduce air risks would save money.
These controls included product and solvent recovery.
These controls would reduce the estimated cancer risks
from inhalation exposures by close to a tenth of the
air risks we quantified and save over one hundred
thousand dollars annually.
12. From a different vantage point, decision makers who are
concerned about the risks to the MET could realize a
risk reduction of roughly an order of magnitude by
implemehting controls costing about $23 million
annually^Again,the value of that purchase depends on
the community's priorities and the importance of pro-
viding everyone with protection above the current
level.
13. Our evaluation of the risks and costs of control was
Initially completed in 1984. Our 1986 revision of the
risk estimates, using more current environmental and
health data, indicates that there would be little, Tf
any, change in the conclusions discussed above"! The
major changes noted in our analysis were in the cancer
unit risk factors and reductions in the discharges from
the chemical manufacturer to the NEWPCP occurring after
completion of our monitoring programs in Philadelphia;
The reduction in the industrial indirect discharges
have undoubtedly led to lower volatilization rates from
sewer lines, the sewage treatment plant, and the
Delaware River.
14. For the nine pollutants and 17 sources we investigated,
we estimated an upper-bound incidence of nearly three
cases per year of .cancer in a city that had about 4,500
die of cancer in 1984.This point helps place our
analysis in context.
-------
16
Readers of these conclusions should recognize that local
officials will be considering other information outside our anal-
ysis in using this study's findings. They will evaluate the
control-options analyses we performed in a broader context of
other health, safety, and social services that they provide their
citizens. We provide our findings to local policymakers to
assist them in their continuing efforts to set priorities among
environmental issues.
Monitoring Activities
Our completion of various ambient monitoring programs leads
us to the following conclusions about our approach and about the
substantive environmental issues we addressed in Philadelphia.
Insights for Future IEMP Studies
• Our ability to quantify differences between monitoring
and modeling results highlighted the advantages of
using both techniques together to improve exposures
assessments substantially. Modeling, which relates
source releases to ambient pollutant levels, is limited
by the available information on sources and pollutant
loads. Monitoring, on the other hand, provides infor-
mation about pollutants and sources for which one may
know very little, but is not necessarily well-suited to
pinpoint the sources of toxic releases. The two in
combination complement the strengths of each and reduce
the inherent weaknesses of both when used indepen-
dently.
• Difficult technical and scientific problems in deter-
mining ambient toxic pollutant concentrations can con-
strain attempts to quickly fill in missing data. In _
addition, one may be forced to use new and unproven
monitoring techniques to gather the needed information.
For example, our experience in Phase II showed that the
ROSE system is not well-suited for measuring ambient
air releases from landfills. In another example, the
monitoring technique used by AMS to measure formalde-
hyde levels in the ambient air has not yet been proven
to be reliable for quantifying formaldehyde concentra-
tions at low levels representative of ambient condi-
tions. Finally, in the two years since our use of
Tenax for ambient air monitoring, its use, effective-
ness, and reliability have been called into question by
scientists inside as well as outside EPA. EPA is cur-
"rently exploring alternatives to Tenax.
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
17
Monitoring Results
We gained important insights on environmental exposures from
our examination of benzene, formaldehyde, used oil, and landfill
air emissions. The most important findings are stated below.
Benzene in the Ambient Air
• Average benzene concentrations at the Philadelphia
intersections that we examined ranged from 5.4 micro-
grams per cubic meter (ug/m3) to 22.3 ug/m3. However,
we were unable to distinguish between the influence of
traffic versus gasoline marketing on ambient concentra-
tions. From benzene concentrations at busy traffic
intersections, we calculated rough approximations of
average individual cancer risks (upper-bound, lifetime)
that were in a range comparable to drinking water and
air risks we studied in more depth.
• A breathing zone study showed benzene concentrations
during 14 refueling episodes that ranged from a low of
1.5 ug/m3 to a high of more than 88,714 ug/m3. Because
of the small numbers of sampling events, these values
are only a limited indication of average benzene con-
centrations from refueling.
Formaldehyde in the Ambient Air
At the four sites monitored, average ambient formaldehyde
concentrations ranged 2.8 ug/m3 to 3.6 ug/m3. These concentra-
tions are significantly lower than those measured in several
major cities nationwide.
• The average ambient formaldehyde concentrations are
below the AMS ambient air quality guideline for formal-
dehyde. However, when we considered the individual
sampling events at each site, we found a few days on
which the ambient concentrations exceeded the guide-
line.
• Our 1984 rough approximations of average upper-bound
lifetime-individual cancer risks at each of the four
monitoring locations showed risks roughly comparable to
most of the air risks examined above in the control-
options work. If we employ recent changes (1986) in
the unit-risk factor for formaldehyde, we find the risk
to be in a range that is roughly twice to 30 times that
-------
18
high. Therefore, we now see that formaldehyde may be a
more significant health concern than many of the pollu-
tants we looked at in our control-options work.
Combustion of Used Oil
A short-term sampling program (45 samples collected and
analyzed) was conducted in coordination with Philadelphia, the
State of Pennsylvania, and the State of New Jersey. Because of
the limited data, we have not estimated exposure and risk. How-
ever, the data suggest that contaminated fuel oil contains lead
at levels that warrant further investigation. For all samples,
the lead concentrations exceeded the recently promulgated EPA
standards for burning used oil. However, the lead concentration.
in used oil will fall dramatically as the lead in gasoline is
phased out.
Air Emissions from Landfills
As noted above, the EPA field monitoring using the ROSE
system experienced serious equipment problems and resulted in
poor results. Better data were collected by New Jersey using
different monitoring equipment. While the New Jersey data are
too limited to assess exposure or risk adequately, the results do
indicate that a fairly large number of toxic pollutants can be
emitted from landfills and that concentrations of volatile
organic compounds (VOCs) around landfills can be above urban
background levels.
OBSERVATIONS FOR THE FUTURE
The Philadelphia IBMP has shown us that exercises to set
priorities across numerous environmental issues can be managed to
provide useful information to local decision makers. The project
has also revealed important practical limitations in our ability
to apply the IEMP methodology. Three of the most important les-
sons we learned are discussed below.
• It is essential to define the objectives and scope of
the project in its early stages. In Philadelphia we
learned that, because of methodological and resource
constraints, we cannot analyze every environmental
issue. It is therefore important to identify at the
beginning of an IEMP the areas of greatest concern to
the project participants and to set priorities among
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
1
19
this subset of issues for analysis. This allows the
limited resources available to be used well on a man-
ageable set of topics, rather than superficially on
every potential issue in a community.
The data and analytical methods available are very
significant influences on tne topics we analyze and how
we analyze them. Major advances in methods for assess-
ing noncarcinogenic risks and ecological damage could
:reatly increase the ability of lEMPs to aid local
ecision makers in priority setting. Some environ-
mental concerns may be given a low priority because we
either know little about them or do not currently have
the capability to analyze these issues. Users of this
analysis should not misconstrue our results as state-
ments about priorities based on a comprehensive
assessment.
When we have reasonable knowledge about an issue, we
can conduct monitoring or perform analyses that may
shed new light on a subject. Or, when data is the only
concern, we can collect some data even with our limited
resources if it appears that the increased base of
knowledge will improve local decision making. For the
lEMPs, methodological problems can be much more
intractable, especially on very complex topics.
Two new methods, beyond the scope of any particular
IEMP, would greatly enhance our assessments: (1) an
ability to quantify the noncarcinogenic risks for com-
pounds present in the environment at concentrations
above the threshold, in ways analogous to carcinogenic
risk assessment, and (2) an ability to assess ecosystem
effects. Estimates of noncancer effects would allow
comparisons between cancer and noncancer problems and
provide insightful information for those responsible
for public health protection policies. We then could
treat noncancer health concerns more equally with can-
cer concerns, which often seem more important because
they can be quantified. Methods for quantitatively
measuring risks to ecosystems currently do not exist in
any very sophisticated form, it is, therefore, diffi-
cult to assess very rigorously the cost-effectiveness
of control options that reduce ecological risks or to
clarify for local decision makers the implications of
environmental issues that span cancer, noncancer, and
ecological concerns.
-------
If
20
Finally, we found it essential to have representatives
from state and local public health and environmental
encies participate in the project,Local participa-
on in this IEMP greatly enhanced the quality of the
analysis performed. It also increased the amount of
available data, facilitated the communication of com-
plex issues (such as risk assessment), and built credi-
bility into the project. Also, since we cannot practi-
cally analyze all environmental issues, local involve-
ment ensures that our limited resources are used for
priority setting and issue evaluation in the areas most
critical to the host community. We believe direct
local involvement in this and future projects is vital
to achieving sound results, essential to gaining local
acceptance of the project findings, and critical for
using the IEMP approach to aid better environmental
policymaking in specific geographical locations.
-------
-------
I. INTRODUCTION
INTEGRATED ENVIRONMENTAL MANAGEMENT
This document reports the findings of the Philadelphia
Integrated Environmental Management Project (IEMP), conducted
from 1982 to 1986 by the Regulatory Integration Division (RID) of
the U.S. Environmental Protection Agency (EPA). The Philadelphia
IEMP was EPA's first large-scale study of the effects of toxic
pollutants on human health and the environment conducted by
focusing on all media within a single geographic area. It dif-
fered from earlier EPA efforts because it employed an integrated
approach to measuring the complex, interactive effects of pol-
lutants and identifying cost-effective methods of control. The
Philadelphia IEMP also reflected a shift in public interest by
focusing on the effects of toxic chemicals rather than "conven-
tional" pollutants. By briefly comparing it with traditional
approaches, this Introduction aims to clarify the rationale for
the integrated approach and therefore the objectives of the
Philadelphia IEMP.
Traditional Approach
Traditionally, the United States has responded to its envi-
ronmental problems by passing new laws, such as the Clean Air
Act, Clean Water Act, or Safe Drinking Water Act, each focused on
a single environmental medium. Accordingly, EPA has assumed a
medium-specific organizational structure that reflects the
statutes: offices that focus on controlling air and water pollu-
tion and on hazardous waste disposal practices. Most state and
local environmental agencies mirror this federal format.
The traditional EPA response to environmental problems has
therefore been to focus on one pollutant at a time in a single
environmental medium. When applied to toxics, this single-
pollutant^- single-medium approach has sometimes led to environ-
mental programs and regulations characterized by inefficient
employment of resources. In particular, there has been concern
that the traditional approach could contribute to a number of
unfortunate consequences:
(1) The solution to one single-pollutant, single-medium
problem might simply transfer the problem to another
medium (e.g., from water to air), perhaps incurring
greater risks and costs of control.
-------
1-2
(2) Problems involving several environmental media may not
be addressed sufficiently by an environmental agency
that generally examines each medium independently.
(3) Policymakers have no way of setting priorities across
sources, pollutants, and exposure pathways in different
media. Consequently, environmental policies and regu-
lations may not be the most efficient, spending too
little on some problems and too much on others.
(4) One-dimensional studies of pollutants and media may not
consider total or cumulative environmental exposure
either within a medium or across all media.
(5) Laws and regulations developed by separate EPA divi-
sions or analyses may involve different and sometimes
inconsistent objectives, methods, and standards.
For example, industrial wastewater treatment can shift pol-
lutants from water to land or air or both. The treatment process
produces not only a relatively clean discharge, but also a
residual sludge containing many of the original pollutants in
concentrated form, which must be disposed of on land. These more
highly concentrated pollutants can, if not properly managed,
leach from land disposal sites into drinking water supplies where
they may cause greater environmental risks than in their original
context. Alternatively, the more volatile compounds within the
sludge can evaporate into the air, thereby posing a threat to
human health through inhalation. Similarly, air pollution
control devices, such as scrubbers, create sludges that require
land disposal and can also pose environmental hazards.
EPA's traditional approach has also been directed toward
establishing policies and standards for the nation, while a
reasonable strategy on the whole, the national scale of attention
can lead to situations where particular areas receive inappro-
priate levels of control relative to their needs. Of special
concern are issues that have regional, but little national, sig-
nificance and may therefore be overlooked or placed low on EPA's
agenda.
Integrated Approach; IEMP3
EPA developed the integrated environmental management
program as a solution to the shortcomings of the traditional
approach to pollution control and as a response to the growing
challenge of toxic pollution. The approach seeks to account for
-------
1-3
multimedia phenomena, identify and track situations where pollu-
tion is merely pushed around to different sectors of the envi-
ronment, focus attention on the cost-effective allocation of
resources to programs of greatest concern, and address other
issues that once lacked an appropriate forum. An integrated
approach to environmental management also calls for a focus on a
specific geographic location. Thus, the geographic basis of an
IEMP enables it to consider all major sources and pathways of
pollutants, as well as local environmental, industrial, and
demographic patterns.
lEMPs are also intended to involve all responsible local
parties and agencies in the management and coordination of the
projects. Broad public involvement ensures that issues of
greatest local concern are adequately addressed regardless of
their perceived national importance. By pooling the informa-
tional and analytical resources of all levels of government
(federal, state, regional, and local), responsible agencies under
the auspices of EPA are able to undertake efforts that none would
be able to undertake alone.
An lEMP's shift from a largely uniform national policy
orientation to one that is particular and local also signals a
new emphasis with regard to risk assessment and risk management.
These concepts are discussed in Chapter II. But here we should
note a theme recurrent throughout the project and therefore
throughout this report. The Philadelphia IEMP concentrated on
measuring risks to the environment and human health in such a way
as to assist regional and local environmental managers in making
environmental decisions. The objective of EPA's IEMP has been to
provide a coherent methodology for improving decisions concerning
local environmental priorities and appropriate pollution control
options for priority issues. The decisions themselves are often
more properly the responsibility of regional and local
authorities.
Finally, the integrated environmental management approach
allows for greater consistency in the assessment, and greater
flexibility and efficiency in the management, of risk to the
environment and human health. As noted above, laws and regula-
tions developed in the early stages of EPA's efforts involved
such various criteria as technology standards, air or water
quality, health effects, or costs. To address problems involving
different pollutants, sources, and exposure pathways, and to
account for intermedia transfers, the Philadelphia IEMP used the
reduction of adverse effects on human health as a common measure
of effectiveness. When feasible, this common measure is a useful
way for managers to compare different environmental problems and
solutions and set priorities for further study or action.
-------
1-4
Comparisons and priorities in turn make possible the coordination
of efficient control strategies that can achieve the greatest
reduction of risk at a given cost of control.
THE PHILADELPHIA IEMP
EPA's Regulatory Integration Division has thus far initiated
six geographic projects. The first was a small-scale effort in
the Kanawha Valley of West Virginia. This was followed in 1982
by the full-scale project in Philadelphia, which is described in
this report. Later, in 1984, EPA initiated concurrent studies of
Baltimore and the Santa Clara Valley ("Silicon Valley") in Cali-
fornia. Two other projects include a second one in the Kanawha
Valley and another recently begun in Denver, Colorado.
We must strongly emphasize that the Philadelphia IEMP did
not attempt a comprehensive qualitative or quantitative assess-
ment of all risks from toxic chemicals in the Philadelphia metro-
politan area. To make such an assessment would have required far
more than the resources available to the project. This report,
then, is a review of the methods developed and approaches taken
in making an integrated risk assessment and risk management
analysis of a select, limited number of chemicals.
We also emphasize that Philadelphia was not chosen for this
study because EPA believed the Philadelphia metropolitan area had
a significant toxics problem. Philadelphia has already demon-
strated exemplary leadership in pollution control through its own
initiatives in developing ambient air guidelines and instituting
innovative approaches to improving the quality of its drinking
water. Philadelphia was chosen because EPA, together with local
officials, wanted to explore better ways to further identify,
assess, and manage human health risks of environmental pollutants
in the metropolitan area.
-------
1-5
Outline of the Report
The Philadelphia IEMP was conducted in two phases: Phase I,
primarily devoted to risk assessment, and Phase II, primarily
devoted to risk management. (An outline of the steps in each
phase is included in Chapter II.) The remainder of this report
describes how we conducted the project and presents its findings.
The chapters are organized as follows:
• Chapter II, Project Methodology, discusses the concepts
of risk assessment and risk management, the process of
risk assessment, and the organization of the project.
• Chapter III, Phase I of the Philadelphia IEMP, reports
on the activities and findings of Phase I, including a
detailed review of the screening process.
• Chapter IV, Phase II Monitoring Activities; Benzene^
Formaldehyde, Landfills, and Used Oil, describes the
monitoring performed to assess exposures for four
specific sources or pollutants identified in Phase I as
being of significant concern.
\
• Chapter V, Control-Options Analysis; Ambient Air and
Water Monitoring Programs, provides a technical discus-
sion of the procedures for and results from ambient air
and water monitoring programs conducted to support the
control-options analysis.
• Chapter VI, Risk and Control-Options Analysis Results,
presents an overview of the baseline risks and analyti-
cal structure and findings of the control-options
analysis.
• Chapter VII, Conclusions, summarizes the conclusions
from the Philadelphia IEMP, the specific findings from
the Phase II activities, and other related issues.
-------
J
-------
II. PROJECT METHODOLOGY
This chapter defines the concepts of risk assessment and
risk management; describes the analytical methods used for risk
assessment; and provides an overview of the whole project by
outlining the analytical steps included in Phase I and Phase II.
RISK ASSESSMENT AND RISK MANAGEMENT
The two key organizing concepts of an integrated environ-
mental management project are risk assessment and risk manage-
ment, and, as noted in Chapter I, they correspond to the two
major phases of the Philadelphia IEMP. Risk assessment is the
central task of Phase I; risk management is the central task of
Phase II. As will become clear, this conceptual organization is
largely a matter of emphasis, since assessment and management
tasks are interdependent, and the activities of each Phase over-
lap. These two concepts underlie the project's methodology,
however, and therefore form the basis for understanding and eval-
uating the remainder of this report.
It is important to emphasize that our discussion of the IEMP
methodology is presented at a conceptual level; our actual appli-
cation of the methodology to specific geographic projects may
vary from site-to-site. Furthermore, since Philadelphia was the
first full-scale IEMP, it was more difficult to implement all
aspects of the prescribed Phase I and Phase II activities than in
our most recent studies, especially the Santa Clara IEMP. In our
following discussions, we note where the Philadelphia IEMP
differed from the general methodology.
Risk Assessment
There- are always more environmental issues to study than
there are resources to address them. The key task of Phase I of
an IEMP is to compare risks attributable to different sources,
pollutants, and exposure pathways in order to set research and
management priorities. The most direct way to make these compar-
isons requires analysts to first identify issues and then esti-
mate their associated human health risks quantitatively. A quan-
titative risk assessment for carcinogens estimates the number of
cases or incidence of disease that may occur within a population.
It is statistically based on estimates used throughout EPA for
-------
II-2
regulatory decision making. In lEMPs, quantitative risk assess-
ment is the primary measure used for setting project priorities
and selecting the issues for further analysis in Phase II. Thus
Phase I involves a screening process for allocating project
resources to the most important issues.
During the Phase I of our first major projects in
Philadelphia and Baltimore, we relied heavily on priority-
setting methods other than quantitative risk assessment. These
other methods include reliance on expert judgment in a systematic
process for ordinal ranking of issues.
In our current projects in the Santa Clara Valley and
Denver, we rely more heavily on quantitative risk assessment to
determine the extent to which various environmental issues pose a
potential threat to human health. We then compare these issues
across environmental media, pollutants, and exposure pathways to
set priorities among them.
Another approach uses existing ambient pollutant guidelines,
criteria, or standards as benchmarks to evaluate the significance
or importance of ambient pollutant concentrations. Policy makers
generally consider ambient pollutant levels below these standards
as safe, warranting no further action. For noncarcinogens,
Acceptable Daily Intake (ADI) or Reference Dose (RfD) thresholds
are used to determine if an exposure to a pollutant may warrant
further analysis. This approach is most useful when ambient
levels are below the standard. However, there is often no quan-
titative way to measure the degree of risk created when concen-
trations exceed the standard thresholds. This is also the case
with most ecological measures, e.g., ambient water quality
criteria. Currently, there is no quantitative way to assess the
amount of ecological damage that may result from pollutant con-
centrations above water quality criteria levels.
We use quantitative methods, existing standards, criteria,
and Reference Doses to the extent that data and available method-
ologies allow. When we have determined which issues are rela-
tively more important, we often use other criteria to supplement
quantitative risk assessment. These criteria include analytical
feasibility, relevance to EPA, state, and local program objec-
tives, and the potential for effective response. These criteria
(which incorporate risk management perspectives) are discussed
later in the report. It should also be noted that the basic
objective of IEMP risk assessment procedures is to allow for
relative comparisons of potential problems rather than to make
definitive statements about the absolute risks posed by a parti-
cular substance, source, or pathway. Other qualifications about
the reliability of quantitative health risk assessments are also
discussed later in the report.
-------
II-3
Risk Management
In Phase II, we focus the project's resources on risk man-
agement considerations. Risk management is the process of evalu-
ating alternative controls to reduce the health risks identified
through risk assessment. Risk management considers not only the
level of risk posed by a particular pollutant or source of pollu-
tion but also the feasibility and cost of control, public prefer-
ences, and institutional capabilities. Setting priorities for
research is also a key aspect of risk management. The objectives
of this Phase are to investigate control alternatives, estimate
their efficiency in controlling pollution, develop estimates of
their cost, and then assess their cost-effectiveness. Where
necessary, we supplement the analysis by developing detailed
exposure estimates and other specific data required to refine the
quantification of risks assessed in Phase I.
Phase II also involves the continued adjustment of priori-
ties according to quantified risk assessment and other criteria.
These additional criteria include the risk management concerns
regarding the feasibility, cost, and effectiveness of control
alternatives. We prefer that the overall criterion of cost-
effectiveness be the amount of human health risk reduced by
alternative controls in relation to the cost of implementing
those controls. However, in some cases this is not possible and
other methods must be used. Also, we find that in cases where
the current level of control is deemed appropriate by the com-
munity, but the cost-effectiveness of the current control stra-
tegy is unclear, a cost-effectiveness analysis may be able to
identify control strategies that attain the same level of control
for less cost. In the Philadelphia IEMP we focused only on
assessing cost-effective control strategies to reduce risks to
human health.
What Quantitative Risk Assessment Is Not
It is important to realize that our analysis does not
examine disease incidence in the local population and then
attempt to link it with environmental exposure. For many of the
specific exposures assessed at the time of this study and for
health effects on which we have focused, such epidemiological
study is difficult because background disease incidence is high,
exposures to particular individuals are difficult to quantify,
and a variety of factors—genetics, occupational exposures, food,
exposures from other geographic locations, and many others—might
be the cause of an observed effect.
Because it is not an epidemiological study, the IEMP risk
assessment is not intended to and does not answer questions such
-------
II-4 |
as what caused a statistically higher rate of cancer in one
neighborhood or part of the community. Instead/ it attempts to
evaluate what health effects might result from current and future I
environmental exposures. I
While different in approach and interpretation, risk assess-
ments and epidemiological work are complementary. Risk assess-
ment can help to identify populations and geographic areas that
appear to be at risk and therefore might be appropriate subjects
for an in-depth epidemiological study. Epidemiological analysis
increases scientific understanding of the relationship between
exposure and health effects, thereby strengthening the basis of
risk assessment. Epidemiological analysis may also, in some
cases, be useful for confirming specific hypotheses suggested by
risk assessment.
The overall uncertainty in the risk assessment assumptions
is great enough that results should be considered as rough indi-
cators of the probable magnitude of effects, not as precise,
site-specific predictions of effects.
\
GENERATION OF RISK ESTIMATES
The following overview of the methods for quantifying health
risks applies in theory to both cancer and noncancer health
effects. In the lEMPs, we define risk to an individual as the
increased probability that an individual exposed to one or more
chemicals will experience a particular adverse health effect
during the course of his or her lifetime. (An average lifetime
is assumed to be 70 years.) Risk to the population is the
expected increased incidence (number of cases), above the back-
ground rate, of an adverse health effect in an exposed popula-
tion. For cancer risks, we present both types of quantitative
estimates in the Philadelphia IBMP. A more detailed discussion
of our assumptions and approach is provided later in this
chapter.
The IBMP risk assessment combines estimates of toxicological
potency, derived from laboratory and occupational studies, with
estimates or measurements of area-wide contamination (or expo-
sure) levels to estimate risk to the population. This approach
involves a relatively high degree of uncertainty, and, in most
cases, cannot easily be verified or disproven by observation. It
is useful, however, because it is analytically straightforward
and uses available methods for estimating the future health
effects of current and future exposures.
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
II-5
The two key elements in estimating risk are the estimated
toxicoloqical potency of a chemical and an individuals's (or a
population's) exposure to that chemical.
Estimating Toxicological Potency
The assessment of toxicological potency for carcinogens
involves (1) a qualitative "hazard identification" to determine
whether there is evidence that a chemical causes an adverse
health effect and, if so, (2) a quantitative estimate of the
"dose-response relationship" or potency of the chemical. For
cancer risk assessment, potency estimates relate dose (or the
specific quantity of a chemical) to the probability that an .indi-
vidual will experience the adverse health effect in his or her
lifetime. Our carcinogenic assessment methodology is consistent
with EPA's Proposed Guidelines for Carcinogenic Risk Assessment
as published in the Federal Register, November 23, 1984, which
the reader should consult for a more detailed discussion.
The qualitative evaluation involves a review of the relevant
literature to determine whether sufficient evidence exists to
establish a substance as a suspected carcinogen. This review
includes both human epidemiology studies and animal experiments.
The qualitative evaluation results in an estimate of the
strength of evidence that a substance is carcinogenic. To sum-
marize the strength of evidence, two classification schemes are
available: either the scheme developed by the International
Agency for Research on Cancer (IARC) or the scheme developed by
EPA (which is an adaptation of the IARC system). In the IARC
classification method, the evidence that an agent produces cancer
in humans is divided into three categories: known human carcino-
gen, probable human carcinogen, or insufficient evidence. The
EPA scheme stratifies the weight of evidence in similar manner
with five groupings as follows:
• Group A; Human Carcinogen. This category is used only
when there is sufficient evidence from epidemiological
studies to support a casual association between
exposure to the pollutant and cancer.
• Group B; Probable Human Carcinogen. This category
includes -agents for which the evidence from epidemi-
ological studies of carcinogenicity in humans ranges
from almost "sufficient" to inadequate. To reflect
this range, the category is divided into higher
(Group Bl) and lower (Group B2) degrees of evidence.
-------
II-6
• Group C; Possible Human Carcinogen. This category is
used for pollutants for which there is limited evidence
of carcinogenicity in animals and an absence of data
for humans.
• Group D; Not Classified. This category is used for
pollutants for which there is inadequate evidence of
carcinogenicity in animals.
• Group E; No Evidence of Carcinoqenicity for Humans.
This category is used for pollutants for which there is
no evidence of carcinogenicity in at least two adequate
animal tests in different species or in both
epidemiological and animal studies.
If there is sufficient qualitative evidence to establish a
substance as a suspected carcinogen/ the next step is a quantita-
tive analysis to estimate the degree of risk associated with
exposure to that chemical. This second step in toxicity evalua-
tion involves calculating a specific potency score, which is a
quantitative estimate of probability that the health effect will
occur at different pollutant concentrations.
In estimating the potency of a known or suspected carcino-
gen, we use, where available, chemical-specific, peer-reviewed
potency scores developed by EPA1s Carcinogen Assessment Group
(CAG). For carcinogens without GAG scores, other health scoring
methods are used.^
Cancer Potency Evaluations;
Some Simplifying Assumptions
In evaluating the potency of a substance, EPA's CAG relies
on a number of simplifying assumptions that are widely used in
this type of analysis. These assumptions allow risk assessors,
albeit with some uncertainty, to make extrapolations, using dose-
response relationships at occupational or laboratory exposures,
to estimate the likelihood of effect at the generally lower
environmental exposure levels. Important types of assumptions
include: (1) the interpretation of lab or animal data; (2) the
1For chemicals without CAG scores, other potency scores.have been
developed for use in projects sponsored by EPA1s Regulatory
Integration Division. These potency scores have been, or will
be, informally reviewed by EPA's Office of Research and Develop-
ment; the methods involved are being reviewed by EPA's Science
Advisory Board.
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
II-7
extrapolation of data to other species or populations; (3) the
extrapolation of high-dose effects data to low-dose exposure
situations; and (4) estimates of the effects of exposure to
combinations of toxic chemicals.
In estimating the potency of a suspected carcinogen/ EPA's
CAG assumes that cancer is a nonthreshold health effect. This
assumption means that any level of exposure to a carcinogen is
deemed to involve some risk. Because it is impossible to measure
directly the human risks associated with low levels of exposure
to carcinogens, scientists must rely on mathematical models to
extrapolate from high animal or human occupational exposures to
low environmental exposures. The extrapolation techniques are
different for human data and animal studies. In estimating
potency from animal studies, CAG uses the most sensitive species
observed and calculates the upper (95 percent) confidence limit
on the estimated slope of the low-dose range of the dose-response
curve; both assumptions tend to increase the estimated potency of
the substance. The mechanisms of carcinogenesis are not under-
stood well enough to develop precise predictive models. Although
several models may fit the observed data, different plausible
models may sometimes lead to large differences in projected risk
at low doses.
An important aspect of the procedure for extrapolating from
high-dose to low-dose effects is the assumption that the dose-
response relationship is linear in the low-dose range covering
most of the actual environmental exposures. EPA has adopted the
approach discussed below for estimating the linear dose-response
function at low exposures. This method is only one approach;
others also have merit.
According to general theories of how carcinogens act to
produce cancer (largely derived from experimental studies and
epidemiological data), all finite exposure levels will produce a
finite risk. The magnitude of the risk will decline as the mag-
nitude of exposure declines (this is even clear in the animal
data).2
If the-quantitative relationship between exposure and risk
were known* for all exposures, risk to rodents exposed at very low
levels could be predicted from the measured exposure-risk data.
Risks to humans could be predicted at these very low levels if
the relationship between rodent and human susceptibilities were
is important to note that these statements do not mean that
all finite exposures will cause cancer, but that all finite
exposures will increase the probability that cancer will occur.
-------
II-8
known. Although these relationships cannot be known with accu-
racy, a plausible upper limit on human health risk can be pre-
dicted with sufficient accuracy to be used as a guide to making
risk decisions. Actual human risk is not likely to exceed the
upper limit (although it may), and it is likely to be less.
EPA and the other agencies emphasize that the "predicted"
numerical risks are not known to be accurate, but, because of the
nature of the models used to predict them, they are likely to be
upper-bound estimates of human risk. An upper-bound estimate is
one that is not likely to be lower than the true risk, and may be
significantly higher.
We estimate low exposure risks using the model currently
used by EPA. A model in this context is a mathematical formula
that describes the relationships between various measures. Two
models are needed to predict low exposure risk:
• A high-exposure to low-exposure extrapolation model is
needed to predict low exposure risks to rodents from
the measured high-exposure, high-risk data. EPA cur-
rently uses a so-called linearized multistage model for
this purpose. This model is based on general (not
chemical-specific), widely held theories of the biolog-
ical processes underlying carcinogenesis. Application
of the model to the rodent exposure risk data produces
an estimate of the lifetime risk for each unit of expo-
sure in the low exposure region. This is called the
unit cancer risk. This "linearized" model is used to
ensure that the unit cancer risk is an upper bound on
risk.
• An interspecies extrapolation model is used to extrapo-
late from rodent unit risks to human unit risks. There
are empirical data and theory to support EPA's current
use of the assumption that rodents and humans are at
equal risk at the same exposure measured in milligrams
of carcinogens per square meter of body surface area
per day.
Alternative models are available for both these forms of
extrapolation and cannot be ruled out. In most cases, but not
always, use of plausible alternative models will yield lower
estimates of risk than those predicted by the two described here.
Differences can sometimes be very large, but in most cases dif-
ferences are relatively small, especially when the models are
limited to those which are linear at low exposure.
We emphasize that the linearized, multistage model used by
GAG leads to a plausible upper limit to the potency estimate that
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
II-9
is consistent with widely accepted mechanisms of carcinogenesis.
However, such an estimate does not necessarily give a realistic
prediction of the risk. EPA's Science Advisory Board (SAB), in
its preliminary review of EPA's proposed guidelines, noted that
the true value of the risk is uncertain/ and for many substances
the lower-bound estimate is zero. The SAB also noted, however,
that an established procedure applicable to a variety of sub-
stances for making most likely or best estimates of risk does not
yet exist.
The traditional approach to carcinogen risk assessment at
EPA has been to take the most conservative approach in developing
potency estimates. In so doing, EPA develops estimates that are
highly likely to overestimate the true potency of a chemical. We
feel that such an approach is appropriate, particularly for pub-
lic health protection and setting priorities, in order not to
underestimate potential human health impacts. Therefore, we
follow the traditional approach and use the plausible upper-bound
potency estimates that GAG develops based on the linearized
multistage model.
Another important simplification is the assumption that we
can add the cancer risks from various exposures to arrive at a
total cancer risk to the exposed individual. In some cases, the
risks of cancer that we are adding are for different types of
cancer. For example, an individual may be at risk of lung cancer
from air exposure to one substance and also be at risk of liver
cancer from exposure to another substance. We add these two
risks of cancer to arrive at an estimated risk of cancer for that
individual. We emphasize that when w« talk about an individual
we do not mean to imply that our analysis is precise enough to
identify specific individuals living in specific areas who may be
at risk. We use the term to convey the notion that each individ-
ual receives exposures from a wide variety of sources, pollu-
tants, and routes of exposure.
The assumption that cancer risks are additive is potentially
important because individuals are typically exposed to many toxic
chemicals. Additivity implies that these pollutants are neither
synergistie nor antagonistic in their combined effect (i.e., that
the presence of one chemical neither enhances nor inhibits the
effects of another). Synergism and antagonism probably do occur,
at least some of the time. Unfortunately, most data analysis and
experimentation on toxicological potency attempt to isolate the
effect of a single chemical. Analyzing the effects of a chemical
in combination with other chemicals is a far more complex task,
and such analysis has not been done for most pollutants. In the
absence of information on synergism and antagonism, the simple
additivity assumption seems most appropriate.
-------
11-10
Finally, it is important to remember that various subgroups
within any population exposed to a toxic chemical vary signifi-
cantly in their sensitivity to the substance. For example, preg-
nant women, young children, persons with genetic impairments, and
other groups may be especially susceptible to a particular chemi-
cal exposure. The potency values we use to estimate risk reflect
this variation; while some people may face higher risks (and
others lower risks), the incidence estimate is intended to
reflect the risk in a population with a typical distribution of
sensitive persons.
Noncancer Effects
Chemicals that give rise to toxic effects other than cancer
and genetic mutations are often referred to by EPA as "systemic
toxicants" because they affect the functioning of various organs
or systems, such as the liver, the kidney, and the cardiovascular
and respiratory systems. Generally, based on our understanding
of physiological mechanisms, systemic toxicants are treated as if
there is an exposure threshold (both for the individual and for
the population) below which "no effects" are observable. This
characteristic distinguishes systemic toxic effects from carcino-
genic and mutagenic effects, which are usually treated as
"nonthreshold" processes.
Because of the assumed thresholds, systemic effects tradi-
tionally have been evaluated by the Agency in terms of an
"acceptable daily intake" (ADI) and a "margin of safety."
Although there are limits to some of these approaches, EPA is
often called upon to apply these concepts when making and
explaining regulatory decisions concerning human health. Thus,
the threshold concept for noncancer effects is extremely impor-
tant to decisions about regulation and risk management.
More recently, the Agency has tried to address some of the
scientific and policy issues inherent in defining a "safe" expo-
sure level for systemic toxicants. This reevaluation has led to
the recommendation that the concept of the "reference dose" (RfD)
replace that: of the ADI. An RfD is calculated in the same manner
as an ADI but its interpretation is different. The RfD is under-
stood simply as a guideline, not a guarantee of certainty regard-
ing what is safe or acceptable, as tended to be the case with the
ADI. EPA now believes that exposures less than the RfD are
unlikely to be associated with noncancer health risks and are
therefore unlikely to be of regulatory concern. Conversely, as
the frequency and size of exposures exceeding the RfD increases,
the probability increases that adverse effects may be observed in
a human population. Nonetheless, a clear conclusion cannot
-------
11-11
always be categorically drawn that all doses below the RfD are
"acceptable" and that doses in excess of the RfD are
"unacceptable."
No-Effect Thresholds
In determining the threshold dose below which no observable
adverse effects are assumed to occur, we rely on EPA Reference
Doses (RfDs; also referred to as Acceptable Daily Intakes or
ADIs) and on thresholds computed by our toxicologists and con-
sultants. Our thresholds were derived using the same procedures
used to estimate RfDs.
Depending on the nature and extent of the specific database,
RID has estimated human inhalation thresholds: (1) directly from
animal or human inhalation data using the RfD methodology; (2) by
using route-to-route extrapolation procedures to convert an
existing RfD; or (3) by using route-to-route extrapolation to
convert thresholds that are based on oral exposure data. For
each chemical of concern, RID has prepared a profile report that
details the hazard identification and dose-response relationship,
including derivation of exposure thresholds. These profile
reports are currently under review by scientists within and
outside EPA.
To calculate an RfD, EPA scientists collect the available
animal and human data and note the various dose levels (in milli-
grams per day) at which different health effects are seen. The
scientists then compare the dose with effect information and
identify the NOEL (the No Observed Effect Level). The NOEL
represents the highest dose tested that did not produce observ-
able results. The scientists also try to define the LOEL (Lowest
Observed Effect Level), which is the lowest dose tested at which
some type of effect occurs. They also try to define the PEL
(Frank Effect Level), which is the dose that involves more
serious health problems.
Therfc will be many NOELs and LOELs for each chemical. These
levels depend on the doses selected and the health effects tested
for by the-researchers. Research is very expensive and each
experiment cannot be exhaustive. Since EPA is dependent on
available research, there is some uncertainty in the definition
of NOELs and LOELs.
To calculate the RfD, EPA scientists select the highest
reliable NOEL and divide it by safety factors. The selection of
appropriate safety factors is based on the nature of the study
from which the NOEL was derived. Most safety factors are mul-
tiples of ten, with each one representing an extra degree of
-------
11-12
uncertainty to account for extrapolation from animal data to the
average human, from the average human to the most sensitive sub-
group, from subacute effects to chronic effects, and the LOEL to
the NOEL. In general, the safety factor is larger than the
expected differences in the exposure levels producing these
effects, and is therefore a conservative estimate of the actual
threshold.
The IBMP analysis uses the RfD-derived thresholds as well as
thresholds developed by RID toxicologists and consultants (using
the same procedure as is used to estimate the RfD thresholds) to
examine possible noncancer health effects. This examination
involves comparing monitored or modeled concentration levels with
estimated human thresholds. Because a single substance may have
many different thresholds for the different health effects asso-
ciated with it, we examine each health effect category for each
substance separately.
If exposures to a substance are below the threshold for a
given effect, we have less concern for noncancer effects. As the
ambient concentration of the substance approaches or exceeds the
estimated threshold, there is an increased concern for noncancer
health effects Lit the exposed population. If exposures exceed
the estimated threshold for a particular health effect, we then
attempt to estimate the population exposed to such
concentrations.
We restrict our evaluation of mutagens to a qualitative
assessment. Most of the experimental data addressing mutagenicity
involve in vitro testing of mammalian cell cultures and non-
mammalian organisms such as bacteria. In most cases, it is
impossible to relate these mutagenicity data meaningfully to
quantified human thresholds. Evidence of mutagenicity is gener-
ally regarded as evidence of carcinogenic potential. Since we do
quantify cancer risk, our omission of a. separate calculation of
mutagenic effects is probably not a serious flaw.
Using, thresholds in the way discussed above allows us to
identify pollutant concentration levels of concern for noncancer
effects without estimating the probability of disease or the
possible number of cases. As such, the threshold alone provides
a means of screening risks. Exposures exceeding the no-effect
threshold, regardless of degree, are appropriate subjects for
further investigation.
-------
I
I
I
I
B
I
I
i
I
I
I
I
I
I
I
I
I
I
I
11-13
Quantifying Risks of Noncancer
Effects Above the Threshold
EPA has recently begun to develop methods to estimate the
probability of disease for noncancer effects once exposures
exceed the no-effect threshold. The calculation and use of dose-
response curves (the quantitative estimate of potency) for non-
cancer effects currently does not enjoy broad consensus within
EPA. In contrast, use of GAG potency estimates and threshold
values to estimate risks or levels of concern is relatively
straightforward; such indicators of risk have been used through-
out EPA for many years^
Risk Assessment as Applied
in the Philadelphia IBMP
The Philadelphia lEMP's Phase I (conducted in 1982 and 1983)
activities emphasized cancer as the primary effect of concern; in
fact. Phase II study topics considered only chemicals that were
carcinogens. We emphasized cancer in Phase I for several
reasons:
\
• There is better information for cancer than for other
chronic toxic health effects.
• Dose-response data are available for carcinogens.
Furthermore, the nonthreshold assumption permits quan-
titative risk assessment at the low concentrations
typical of most ambient conditions.
• The public has expressed frequent concern about the
possible link between environmental pollution and the
incidence of cancer.
As we neared completion of Phase II, we recognized that
Phase I should have done more to address noncancer health
effects. Because of the emphasis placed on quantifying cancer
risks, otfiawSthealth effects were dismissed too easily. While we
could not n**isit Phase I of the Philadelphia project, we did
evaluate the- noncancer health effects for the subset of pollu-
tants that were considered in Phase II.
Evaluating Exposure To Toxic Pollutants
Most of our exposure assessments attempt to identify ambient
concentrations of substances in the media of concern: air and
-------
11-14
drinking water. We then make certain assumptions about how ambi-
ent concentrations relate to actual human exposure (or dose
level).3 These assumptions—the exposure constants regarding how
much air a person breathes or water he or she drinks—are
described below.
We estimate ambient concentrations in two ways: direct moni-
toring or simulation modeling. There are a number of advantages
to modeling emissions in order to estimate ambient concentra-
tions:
• Modeling may provide the only way to estimate ambient
concentrations in situations where such conditions
cannot be observed (e.g., under conditions other than
those currently in existence).
• It can take into account the geographic variability of
a large area, whereas monitoring data are often from a
few specific points and must be extrapolated over the
area of concern.
• It is often less costly than extensive monitoring.
• It enables predictions of exposures in any location
(and, in particular, the location of the most exposed
individual), whereas monitoring can only provide an
indication of the exposures in the vicinity of the
sampling sites.
• It links concentration estimates, and hence exposures,
to sources. Such source information is important as a
risk management tool in that it allows us to estimate
the impact of various pollution control options.
On the other hand, constructing a model of pollutant
releases and resultant ambient concentrations involves making
assumptions about the important processes between pollutant
source and, human recipient. Building such a model thus requires
an understanding of those processes, which is not necessary if
one can simply monitor ambient concentrations directly.
In conducting an IEMP, we prefer to use monitoring data if
resources allow. The primary advantage of reliable, long-term
3For a more detailed discussion of total human exposure see Wayne
R. Ott, "Total Human Exposure," Environmental Science and Tech-
nology, vol. 19, No. 10, 1985.
-------
11-15
monitoring data is that they provide a more direct and often
simpler means of estimating ambient conditions. Even with direct
monitoring, however, interpretation of results can be difficult
if data are limited or if lab or sample contamination is
suspected.
The key steps in modeling pollution are (1) estimating the
number of sources and the frequency and magnitude of releases;
and (2) characterizing the processes and pathways by which a
pollutant is transported in air and water, including the speed of
transport, the extent of dilution cr dispersion, and any chemical
transformation the pollutant might undergo (such as degradation
to a nontoxic form). With such assumptions, one can estimate
resulting ambient concentrations at various distances from the
source. Estimating long-term average concentration levels, which
are of primary concern for evaluating chronic health impacts, is
simpler than short-term modeling, which must take greater account
of variations in meteorological conditions.
In the geographic studies, we calculate three measures of
risk: risk to the most exposed individual (MED, risk to the
average exposed individual (AEI), and the excess aggregate popu-
lation incidence 'of disease. We express individual risk as
either MEI risk or AEI risk. We define risk to an average indi-
vidual as the increased probability that an individual exposed to
one or more chemicals will experience a particular adverse health
effect during the course of his or her lifetime (an average of
70 years). We define risks to the most exposed individual as the
increased probability that an individual exposed to the greatest
amount of one or more chemicals will experience a particular
adverse health effect during the course of his or her lifetime
(an average of 70 years). Aggregate population risk is the total
estimated increased incidence (number of cases), above the back-
ground disease rate, of an adverse health effect (cancer) in an
exposed population. We assume that an average person weighs
70 kilograms (about 154 pounds), breathes 20 cubic meters of air
each day, and drinks 2 liters of water each day.
To e&t&nate the risk to the most exposed individual, we
typically need to know how far an individual is from the maximum
pollutant Concentration near a source. To estimate average indi-
vidual risk, we estimate an average pollutant concentration to
determine exposure. To estimate population incidence, we must
identify the number of people exposed to a given pollutant con-
centration. Our population exposure methodology implicitly
assumes that people are exposed to estimated outdoor ambient air
concentrations 24 hours a day. In fact, people spend most of
their time indoors, either at home or at work. While this sim-
plification overstates actual exposure to outdoor air, the bias
-------
11-16
introduced by this procedure may not be too great, since recent
EPA studies show that many outdoor air contaminants are also
found indoors at equal or greater concentrations.4 Assuming
exposures based on residential population demography also over-
simplifies actual exposure patterns, since many people breathe
air and drink water from a number of different locations in a
day. Because of these and other conservative assumptions, the
values generated by the risk assessment represent the highest
plausible estimates of risk. Actual risk is expected to be sub-
stantially lower. Under these assumptions, lifetime risk to the
exposed individual is simply the product of exposure and
potency:
R - / B \ x / P \(D
individual risk (exposure) (potency factor
As discussed above, exposure is the product of the ambient
concentration of the pollutant in the medium of concern (air or
drinking water) and exposure constraints (i.e., the standard
assumptions about body weight, intake, and absorption):
*
B - / Y \ x / Z \<2)
exposure I ambient concentration) [exposure constants]
in median of concern y y /
Therefore, individual lifetime risk can be calculated by multi-
plying ambient concentrations (Y) and exposure constraints (Z) by
potency (P):
R -/ Y \x/Z\x/ P \(3)
individual / ambient \ / exposure\ /potency factor
risk (concentration) [constants I [
lin medium of / \ I \
\ concern / \ / \
To illustrate how a simplified risk assessment might be
performed, let us suppose that we wish to calculate the lifetime
risk to the MEI, as well as the aggregate lifetime incidence,
associated with exposure to ten micrograms per cubic meter
(ug/m3) of benzene in the ambient air in a given geographic area,
We have divided the area of concern into sections which we will
call "grids."
In our example, we have four grid sections. For each grid
we have the following average ambient concentration values and
population data:
4Wayne R. Ott, op. cit.
-------
I
I
I
I
I
I
t
1
I
I
I
I
I
I
I
I
i
i
i
n-17
Benzene
Concentration Exposed
Grid t (ug/m3) Population
1 10 20,000
2 8 30,000
3 8 25,000
4 5 25,000
The MEI risk of cancer for an individual exposed to
10 micrograms per cubic meter of benzene in the air (the highest
benzene exposure) is calculated by filling in equation (3) pre-
sented above. The ambient concentration (Y) is 10 micrograms per
cubic meter (which is the same as 0.01 milligrams per cubic meter
or 0.01 mg/m3).
Our exposure constants, as discussed above, assume a typical
person who weighs 70 kilograms and breathes 20 cubic meters of
air (20 m3/day) each day. Our potency value for benzene through
inhalation, developed by EPA's Carcinogenic Assessment Group
(GAG), is 0.029 (mg/kg/day)"1. This value indicates that an
individual taking, in one milligram (mg) of benzene per kilogram
of weight per day for a lifetime (70 years) has an estimated
probability of contracting cancer of about three in one hundred
(upper-bound).
We calculate risk to the MEI using equation (3) presented
above and the maximum ambient concentration value (10 ug/m3):
[THIS IS ONLY AN EXAMPLE]
(
R - 0.01 mg/m3 x [20 m3/day x 1/70 kg] x 0.029 (mg/kg/day)"1
maximum
MB! * ambient X exposure X potency (4)
RISK concentration constants factor
Note: O.Oimg/m3 » 10 ug/m3 where 1 milligram (mg)
3 1*000 micrograms
Working through the algebra, the lifetime risk to the MEI (R) in
this example is 8.3 x 10~* (upper-bound) or roughly eight chances
in 100,000 of contracting cancer over a lifetime. Average indi-
vidual risk is calculated the same way, using an average exposure
determined from either monitoring data or air dispersion models.
To calculate the estimated increased incidence (upper-bound)
of disease over a lifetime in this example, we multiply the life-
time individual cancer risk for each grid—not just the grid with
-------
11-13
maximum concentration—by the number of individuals exposed in
each grid. We then sum across all grids to calculate the life-
time aggregate incidence of cancer for our area of concern:
n
Aggregate
Incidence
Where:
(Exposure
\
\
exposure
geographic grid i
(ambient concentration) x
x (population) L (5)
\ /
(exposure constants)
[THIS IS ONLY AN EXAMPLE]
I - (8 x 10~5 x 20,000) + (7 x 10~5 x 30,000)
(Aggregate
Incidence)
(Grid 1)
(7 x 10-5 x 25,000)
(Grid 3)
(Grid 2)
(4 x 10'5 x 25,000)
(Grid 4)
* 6. 5X lifetime cancer cases
We estimated that the actual risk is unlikely to exceed roughly
seven cancer cases over a 70-year period. Generally, EPA pre-
sents incidence as the expected number of excess cancer cases per
year. Dividing seven by 70, we arrive at 0.1 cases per year, or
about one case every ten years.
We estimate individual and population risks from ingestion
of drinking water in much the same way as for air. We first
estimate a chemical's potency through ingestion (not necessarily
the same as that for inhalation). concentrations of selected
chemicals in the drinking water are measured or estimated , and
then converted to doses by use of expoaure constants (the typical
individual drinking 2 liters of water a day). We calculate both
maximum individual risk and aggregate excess incidence in a popu-
lation.
Interpreting Risk Assessment Results
The estimates, of individual health risk and aggregate inci-
dence from exposure to toxics should not be interpreted as pre-
cise or absolute estimates of future health effects.The simpli-
fying assumptions and uncertaintiesin both the toxicology and
exposure components are simply too great to justify a high level
of confidence in the precision of the results.
-------
I
I
I
I
I
I
I
1
I
1
I
i
i
i
i
i
i
i
i
11-19
The potency and threshold estimates used in this study are
consistently conservative in the direction of overestimating
risk; they may overstate the likely effects of chemical exposure
but are unlikely to underestimate them. This leads to assess-
ments that have biases for health protection when the results are
used. By contrast, our exposure estimates are not as clearly
conservative; some assumptions are conservative while others are
our best guess of an actual value. Overall/ we have tried to be
somewhat conservative in our exposure assessments, as is appro-
priate in a priority-setting exercise. It is important to read
the following detailed chapters carefully to understand what
confidence to place in a particular risk estimate.
On the other hand, we may understate risks to the extent
that we do not estimate exposure and risks from all sources and
pollutants that may have toxic effects. We have tried to iden-
tify the sources and pollutants of greatest concern. We acknowl-
edge that we have not addressed all sources and pollutants.
Because of the uncertainties involved, the results should
not be interpreted too literally. For example, one should not
conclude that a source projected to cause three cases of cancer
per year is clearly worse than a source projected to cause two
cases; given the overall uncertainty of the analysis, these two
results are virtually indistinguishable. On the other hand, for
example, it is reasonable to conclude that a source projected to
result in one case per year represent* less risk to the popula-
tion than a source projected to result in ten cases per year.
Despite the uncertainties, our rick estimates are useful
policy analysis results for roughly assessing the potential
magnitude of the overall risks fro* particular pollutants,
sources, and pathways; comparing issues with one another; and
setting priorities among environmental issues and concerns.
APPLICATION OF METHODOLOGY
In title section, we present a comprehensive description of
our general methodology for conducting an IEMP. When we actually
apply this approach, inevitable practical limitations (such as
time and resource constraints, or limits on the state of knowl-
edge) and the characteristics of the particular site force us to
tailor our efforts to our conditions. As a result, we do not
necessarily apply and develop the full framework and its analyti-
cal tools in any one IEMP. Moreover, we also believe that we can
achieve the most progress if we are flexible in applying the
approach and if we adapt to new circumstances as information
-------
11-20
becomes available during the course of the analysis. Therefore,
the framework should not be considered to be a rigid blueprint:
its specific application can vary from one study to another,
although the general approach will remain unchanged.
There are significant limitations and uncertainties associ-
ated with the lEMP's methodology, which warrant consideration
before examining the actual procedures. First, the risk esti-
mates are based primarily on existing knowledge about pollutant
potency, releases, and ambient conditions; these data, however,
vary widely in quality and are almost always incomplete. Second,
the exposure estimates incorporate a series of simplifying
assumptions: although these assumptions are necessary, they
remain open to question and may be controversial. Third, the
potency estimates are necessarily based on current knowledge of
the toxicological effects of various substances. Considerable
controversy exists about the degree of hazard posed by different
pollutants, and about whether some are hazardous at all.
Finally, resource and time constraints and the breadth of our
focus prevent us from analyzing individual issues in as much
depth as might be possible. We have attempted to strike a
balance between the desire for exhaustive and definitive analysis
and the need for results at a reasonable cost.
As noted at the beginning of this chapter, the range of
potential environmental issues at any site is so large that it
would be impossible, given limited resources, to conduct the
in-depth analysis that would be required to study all of the more
complex problems. Therefore, to manage an integrated study
effectively it is necessary to focus on a small set of study
topics. Although the screening process is designated as a single
step in Phase I, screening actually takes place continually
throughout the project as new information becomes available.
In its simplest form, an IEMP consists of seven steps
organized into two phases. An outline of the steps is provided
below and is followed by a more detailed discussion.
Phase I
1. Establishing arrangements between institutions cooper-
ating on the project
2. Defining the scope of the project:
—Setting geographic boundaries
—Making an initial selection of pollutants and issues
for study
-------
I
I
I
I
1
I
I
I
1
I
I
I
I
I
I
i
I
i
i
11-21
—Establishing risk assessment approaches, e.g.,
selecting the health effects of concern
3. Collecting information on sources, pollutants, and
exposure pathways for entry into a computerized data-
base
4. Performing a screening analysis on the initial selec-
tion of pollutants and sources to determine which of
those should receive further attention in Phase II.
The screen involves two complementary approaches:
—Evaluating risks to determine which pollutants,
sources, and exposure pathways are most significant
—Qualitatively assessing analytical feasibility; rele-
vance to EPA, state, and local program objectives;
and potential for effective response
Phase II
5. Gathering additional data to confirm and refine the
risk assessments performed in Phase I, and to adjust
priorities accordingly
6. Analyzing and evaluating the cost-effectiveness of
alternative control options
7. Developing conclusions
These steps and related issues are discussed in the following
sections.
Phase I; Initial Issue Selection and Screening
Establishing Institutional Arrangements
The first step is to set up cooperative working relation-
ships with all responsible state, local, and federal groups with
direct responsibility for public health protection. Working
relationships with-industry, environmental groups, and o.ther
interested parties are also highly advisable.
-------
11-22
Defining the Scope of the Project
One of the first tasks is to reach an agreement with these
groups on such basic matters as the general scope, focus/ and
objectives of the effort, the responsibilities of the various
parties, and the study's geographic boundaries. The key activity
of this task is to identify and select chemicals and issues for
consideration in Phase I. Decisions must be made as to which of
the many thousands of potential environmental contaminants are
likely to be present in the media of concern, and which of these
warrant the expenditure of time and resources to detect. This is
because our ambient monitoring equipment cannot adequately iden-
tify all substances present; indeed, it is generally necessary to
specifically look for a pollutant in order to detect it.
There are many approaches for arriving at a list of sub-
stances which should be looked for, but they all face the same
constraints: the amount of resources available and the availa-
bility and sophistication of detection equipment. The approach
used should ideally allow a fair comparison of issues yet be
flexible enough to accommodate specific geographic constraints.
Developing a Database
The next step is to design a database that is appropriate to
the scope and objectives of the study. One approach is to gather
and inspect all readily available information on the issues
selected, i.e., sources, environmental releases, and exposed
populations for each medium. State, county, and city agencies,
permit writers, EPA, and local industrial facilities are the
primary sources of these data, which should be collected and
evaluated before new data is generated.
After the available data have been reviewed, engineering
estimates, in many cases, can be used to fill the gaps and pro-
vide the needed information to compare issues. In almost all
cases, engineering estimates are needed to calculate intermedia
transfers. For certain common sources, such as dry cleaners,
degreasers, mobile sources (air), and nonpoint run-off (water),
RID has developed algorithms to estimate pollutant loadings to
all relevant media that are easily adapted to different geo-
graphic areas. In some situations, some new monitoring may be
initiated at this stage if it is deemed necessary for setting
priorities.
-------
11-23
The Screening Process
The screening process selects and sets priorities for the
issues to be studied in Phase II. The process is emphatically
not one that can be performed in a mechanical way, but instead
relies on continuous evaluation of data, especially when compar-
ing potential risks.
The screening process involves two complementary approaches.
The first approach quantifies risks to human health to identify
the sources, pollutants, and exposure pathways of greatest con-
cern. This "risk screen" represents a preliminary attempt at
quantitative assessment, using available data, and therefore may
not generate results sufficiently complete to use in setting
priorities.
For that reason, we also used the second approach, which
focuses on issues that are not readily quantified, in the
Philadelphia IEMP. This approach qualitatively assesses an
issue's analytical feasibility; its relevance to EPA, state, and
local program objectives; and its potential for effective
response (i.e., controllability). We are emphasizing the use and
application of quantitative risk assessment to a much greater
extent in Phase I of our other projects in the Santa Clara
Valley, California, and in Denver, Colorado, than we did in our
first projects in Philadelphia and Baltimore, Maryland.
Step 1; Risk Screen. The risk screen is a preliminary
quantitative assessment of the risks to human health, performed
to identify the sources, pollutants, and exposure pathways of
greatest concern. This assessment can be made in a variety of
ways, depending on the topic of interest. For example, we can
examine either the variety of sources that contribute to pollu-
tion in one medium, or we can address the sources of pollution in
several media, each of which contributes to the same health
effect.
It rajjfetake several cuts at the entire set of issues to
develop faUltest "common denominator" or unifying principle to be
used in a^poating the overall level of health effects attribu-
table to e4c&issue. Once this common denominator is developed,
a subset of issues will become particularly significant. For
example, if the study participants determine that carcinogenic
effects should be the "common currency" of evaluation, those
effects that might be expected to cause greater levels of cancer
will become more important than those that might cause other
types of health consequences.
The choice of a common denominator therefore drives the
determination of which issues will receive the most attention
-------
11-24
during the remainder of the study. This common denominator could
be public health impacts, ecological impacts, some other concern,
or a combination of concerns. The goal of organizing the data
around some unifying concept is to arrive at the best tradeoff
between analytical simplicity and the need to identify high
risk.
Principles of affective management call for the expenditure
of as few study resources as possible to complete the database
necessary for a first cut at the issues at hand. The project
manager must reserve the bulk of the study resources for the
later steps. For that reason, the risk estimates generated for
the screen are not precise estimates but rather very rough
approximations that are just suitable for setting priorities', in
fact, the information available to drive the risk screen is not
likely to be adequate even to identify the relative ranking of
issues with much certainty; instead, it serves only to group the
issues into three broad categories: (1) those that are likely to
pose high risk, (2) those with the potential for high risk but
little substantive evidence, and (3) those that are likely to
pose relatively low risk. Discretion must be used in determining
when enough data have been gathered to support a successful risk
screen.
Step 2; Qualitative Evaluation. After the issues identi-
fied in Phase I as potential candidates for Phase II analyses are
separated into the three groups based on qualitative measures of
risk, the issues classified in the first two groups (high risks)
are further evaluated using nonrisk or secondary criteria. These
secondary criteria, applied to the first two groups of high risk
issues, provide broader perspectives for setting priorities than
those based solely on risk assessment criteria. As more informa-
tion becomes available, further reclassification of the issues
may be warranted. The secondary criteria are discussed below.
Like the rest of the priority-setting process, they should not be
regarded as being inflexible. Other groups conducting integrated
studies may want to modify or add to these criteria.
Analytical Feasibility
The primary elements determining analytical feasibility are
the amount of supporting data available, the availability of
analytical methods, and the level of effort required to generate
new data. These are basically program management considerations
in that they indicate how much effort would be needed for basic
data gathering before developing management alternatives (e.g.,
for control). There is an important tradeoff implicit in the
criterion of analytical feasibility: the greatest payoff in
terms of identifying and controlling toxics may be in studying
issues that have not been studied in the past or have been
-------
11-25
•-••*•-
-'
avoided bejpii^ae of the complexity or magnitude of the issue. On
the other &*•£> within the time and budget available for the
project, it may not be feasible to characterize these issues
adequately, and it may be necessary to curtail the analysis short
of developing complete risk management strategies.
Relevance to EPA, State, and
Local Program Objectives
The elements of this criterion involve the significance of
the issue for national EPA programs and state and local interest
in the issue. One of our objectives is to use the IEM projects
to indicate where shifts in EPA priorities and methods are appro-
priate, so the relationship of the issue to national EPA programs
is an important factor. State and local interest affects the
extent of support that local and state participants will give to
the detailed study effort, and also strongly affects the feasi-
bility of implementing control strategies, when warranted.
-^
Potential for Effective Response 3
The level of existing control provides a rough measure of
the likelihood that additional controls are cost-effective, since
cost-effectiveness generally declines with increasing control.
The feasibility of additional controls is also important; if no
technological, institutional, and political means are available
to reduce risk, the issue is probably not a viable candidate for
study.
Risk Management Challenges in Interpreting Results and
Applying the Process.In the process of establishing priorities,
study participants are likely to encounter several basic problems
involving the interpretation and application of environmental
risk information. Some specific issues are discussed below.
Maximum Exposed Individual Risks
Versus Total Population Risks
The IBHpement or project advisory group selecting study
topics wtaKiive- to confront one of the classic tradeoffs in risk
managemenfPetnat of maximum exposed individual versus cumulative
population risks. Inevitably, some of the potential study topics
will involve situations where a relatively small number of people
experience risks higher than those faced by the aggregate popula-
tion; these topics do not necessarily coincide with those where
there are widespread risks. Decisions may have to be made
whether to spend study resources to help a few people by a sub-
stantial margin or to spend them to help many people by a modest
margin. This issue is one of the most difficult issues in
setting priorities.
-------
11-26
Comparing Across Effects
Another difficult issue involves the comparison of risks
across different health effects (e.g., kidney damage versus
cancer). This is a purely judgmental issue. RID has developed
an approach to setting priorities that attempts to account for
the severity of different effects. There have been several other
attempts to scale different health effects (for instance, EPA's
Office of Solid Waste has developed such a scale), but this issue
may be most effectively resolved by considering local concern for
the severity of different effects.
Comparing Human Health Effects
to Environmental Effects
In cases where the scope of the study includes human health
as well as ecological issues, difficult value judgments must be
made. These judgments are not generally made within the same
policy decision analysis because the quantitative means of equat-
ing the two very different public policy objectives are lacking.
It is clearly important to protect both public health and our
ecosystems, but the value placed on human health is not the same
as that placed on^our other environmental goals. Decisions con-
cerning the setting of priorities between these two types of
issues, because the quantitative measures used to assess adverse
impacts for each are not scientifically or analytically compar-
able, involve the difficult and necessarily subjective balancing
of both public policy objectives.
Phase I Products
We did not write a Phase I report for the Philadelphia IEMP.
In our more recent projects, however, we have presented our risk
assessment results in a Phase I report. The reports contain
information about the risks we quantified, identifies priority
issues based on our risk assessments, and describes the methods
we used to assess exposure and risks.
Phase II; Riak Management Control Options
After the selecting Phase I issues warranting further analy-
sis, we begin Phase, II. Phase II consists of two major tasks:
additional data gathering to improve our exposure assessment
and analysis of control options.
-------
11-27
Data Gathering
By definition, in Phase I we identify the study topics for
further analysis using existing and special engineering esti-
mates. While this is reasonable for setting priorities, the
quality of the data becomes even more important at the later
stages of the study. Depending on the specific issue, in
Phase II we generally engage in further data gathering in order
to: (1) confirm whether an issue warrants further investment of
time and resources, (2) ensure that data used in later steps are
accurate, and (3) revise Phase I priorities where appropriate.
In some cases, the objective of additional data collection
may be to improve our understanding of the significance of parti-
cular pollutants and sources. This situation may only call for
limited monitoring.
In other cases, which are really at the core of the Phase II
activities, the objective will be to analyze the costs of poteni
tial control alternatives, where effectiveness is measured by tie
reduction in health risk—either MEI risk or incidence. In thefe
situations, we will need to ensure that our estimates of pollu-
tant loadings and\ambient concentrations, for example, are rea-
sonable and representative of average annual conditions. These
efforts may include consulting local environmental agencies and
plant managers to confirm underlying data on ambient releases, as
well as initiating an extensive monitoring program.
As additional information is gathered in Phase II, it is
helpful to periodically re-evaluate the Phase II issues in terms
of potential risk. This can often be done by conducting a sensi-
tivity analysis using the revised data. New information may
indicate that the risks, are not likely to be as great as once
thought, or they may be greater than originally anticipated.
Sensitivity analyses of new information provide a way to further
focus the use of project resources during Phase II.
Pollution Control Options
The nVttrstep in Phase II is to assess alternative control
strategies for the subset of environmental issues identified for
this work. A very significant objective is to provide analysis
of how to reduce health risks (the measure of effectiveness) to
the population, as well as to the most exposed individual, at the
minimum cost. This occurs through the lEMP's cost-effectiveness
analysis, which presents the tradeoffs of costs and risk reduc-
tions that decision makers take into consideration when formulat-
ing regulatory strategies. The analysis considers risks that
-------
11-28
result from exposure to primary as well as secondary (intermedia)
releases. It is our intention that community decision makers use
it to shape their general strategies for providing additional
environmental protection in the area.
There are three major components to an analysis of control
options: quantitative measures of ambient concentrations at the
point of exposure; estimates of exposure and risk; and the costs
and efficiencies of feasible control options. Data collected in
Phase I and supplemented in Phase II provide the necessary infor-
mation on pollutant loadings estimates. We generally employ EPA
fate and transport models to develop ambient concentrations at
the point of exposure. Exposure and risk calculations are then
made using standard EPA assumptions, which we discussed earlier
in this chapter.
We identify feasible control options and estimate their
associated costs and efficiencies by employing engineering
assessments and EPA technical documents developed to support
various regulatory activities.
The application of this process will inevitably vary from
one project to the next. In Philadelphia, we implemented the
basics of this methodology. However, because Philadelphia was
our first project, we did not have experience to guide us in
assembling the steps in an efficient manner. For example, we did
not attempt to make a rigorous distinction between the influence
of quantitative factors and nonquantitative factors in determin-
ing which issues to focus on in Phase II. As a result, when we
discuss the Phase I overview, it may not appear to follow exactly
the outline provided in this chapter. In addition, we found much
less data on toxics than we expected would be available; thus we
were unable to implement the methodology completely. Finally, we
used a limited risk methodology, focuting primarily on cancer.
-------
I
I
I
I
I
I
*' . »
! f
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
III. PHASE I OF THE PHILADELPHIA IBMP
This chapter describes the activities undertaken in Phase I
of the Philadelphia IBMP and the results of these efforts,
including the identification of issues for further analysis in
Phase II. Following this introduction, the chapter has four main
sections. The first section discusses the reasons for choosing
the Philadelphia metropolitan area for the study and describes
its pertinent demographic, industrial, and environmental charac-
teristics.
The second section describes the steps taken to define the
environmental and institutional foundations of the project:
establishing the institutional arrangements; defining the scope
of the project; and gathering, organizing, and automating all
available environmental data.
The third section describes the four activities involved in
the Phase I priority-setting process: defining the IBMP's geo-
graphic boundaries, developing issue papers on a wide range of
environmental tppics, establishing the priority-setting criteria
and selecting study topics from the overall screening process,
and defining the framework for analysis in Phase II.
\
The final section of this chapter discusses several limita-
tions of the priority-setting process and how they might be
reflected in the final list of topics studied under Phase II of
the Philadelphia IBMP. As noted in Chapter II, the screening
process involved narrowing down the list of potential topics to a
manageable size. In retrospect, we feel that the final list was
smaller than it could have been, given our current understanding
of environmental issues. This limitation stems in part from the
fact that the Philadelphia I BMP, which was the first of its kind,
focused on the development of methods 'and procedures at the cost
of comprehensiveness; this emphasis made for a relatively narrow
scope o%||fi|&« Other limiting factors were the paucity of avail-
able dsfBHfP-limited resources.
Rea^BIIshould also be aware that occasionally the screening
process hacTto rely on data that provided only rough approxima-
tions of risk to human health. Where quantitative data were
inadequate or inconclusive, the ranking of issues depended on
qualitative criteria as described in Chapter II.
-------
III-2
REASONS FOR SELECTING THE
PHILADELPHIA METROPOLITAN AREA
EPA selected Philadelphia as the site of its first major
geographic study in the spring of 1982. Several factors
influenced this choice: the size and diversity of the city's
industrial base and population; the existing environmental qual-
ity; the availability of data on toxic chemicals; and the
presence of active local environmental agencies. These topics
are discussed in the following sections.
Population and Industrial Base
Philadelphia is located at the confluence of the Delaware
and the Schuylkill Rivers, on the border between Pennsylvania and
New Jersey. A map of the eight-county Philadelphia metropolitan
area that formed the initial geographic boundaries for the proj-
ect can be found in Figure III-l (Philadelphia County is coter-
minus with the City). Philadelphia is the largest city in
Pennsylvania. With an estimated population of 1.6 million
(1980), it is also one of the most populous counties in the
nation.1
As a long-established industrial city, Philadelphia has a
broad cross section of manufacturing plants, including petroleum
refineries, chemical manufacturers, and other industrial facili-
ties that are potential sources of toxic releases. Table III-l
displays employment and value added statistics by major
industry.
Environmental Characteristics
Like most urban settings, Philadelphia faces numerous
environmental concerns in all media. In air, for example, the
city has relatively high annual emissions of hydrocarbons—over
130 thousand tons/year compared to a national county average of
roughly 10 thousand tons/year.2 similarly, Philadelphia County
has issued more major National Permit Discharge Elimination
System (NPDES) water permits than the average county nationwide
(13 permits compared to an average of 1.3).3
^Bureau of the Census, U.S. Department of Commerce.
^National Emission Data System, U.S. Environmental Protection
Agency.
•^Industrial Facilities Discharge File, U.S. Environmental Protec-
tion Agency.
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
i
I
I
III-3
Figure 111-1
Philadelphia IEMP
Map of Philadelphia and Surrounding Counties
BUCKS COUNTY
MONTGOMERY COUNTY
CHESTER COUNTY
PHILADELPHIA COUNTY
DELAWARE
COUNTY
BURLINGTON COUNTY
CAMDEN
COUNTY
GLOUCESTER
COUNTY
-------
III-4
Table III-l
PHILADELPHIA IEMP
EMPLOYMENT AND ECONOMIC STATISTICS
Industrie*
Service*
Manufacturing
Retail Trade
FOR PHILADELPHIA COUNTY
(dollars in milliona)
Employ itent
(1982) Percent
222,141 J7.2S
122,364 20.6
87,323 14.6
Financial, Insurance, Real Eatate 62,914 10.$
Wholesale Trade
38,708 6.$
Tranaportation/Public Utilities 36,00$ 6.4
Other
Total
Manufacturing Industrie*
Food
Printing and Publishing
Chemicals
Fabricated Metal*
Apparel
Electric & Electronic Equip.
Machine*
Textile Mill
Paper and Allied Product*
Primary Metals
Transportation Equipment
Petroleum & Coal Product*
Other Manufacturing
Total
2$, 222 4.2
$97,177
Employment
(1982) Percent
IS, 562 12.75
17,081 13.9
7,367 6.0
9,851 8.0
18,398 15.0
7,928 6.5
7,182 5.8
6,571 5.3
5,159 4.2
1,675 1.4
5,436 4.4
2,201 1.8
18,453 15.0
122,864
Sources: 1982 County Business Patterns, Pennsylvania, U.S.
1977 Census of Manufscturers, U.S. Department of C
Value Added
( 1977)
$789.2
502.7
391.6
358.5
353.1
322.8
287.3
185.7
144.8
49.2
NA
NA
NA
Department of Co
ommerce.
Percent
19. IS
12.2
9.5
8.7
8.6
7.8
7.0
4.5
3.5
1.2
nmerce.
Philadelphia also has one of the larger wastewater treatment
plants in the country, the Northeast Water Pollution control
Plant (NEWPCP), into which several major industrial facilities
discharge their wastewater for treatment, in 1983, the NEWPCP
treated an average of 180 million gallons of wastewater each day
(MGD); the plant is currently being expanded to increase its
treatment capacity to roughly 210
Although Philadelphia does not have any active hazardous
waste landfills, it has several active hazardous treatment/
storage tanks and waste piles. In addition, Philadelphia
Philadelphia Water Department, Annual Report Fiscal 1983,
p. 11. '
-------
I
I
I
I
I
I
I
I
1
I
I
I
I
I
I
I
I
I
I
III-5
received a Superfund grant in 1983 to assist in the cleanup of
the Enterprise Avenue landfill. The Enterprise site is a 40-acre
landfill adjacent to the Southwest Water Pollution Control Plant
leased for the disposal of incinerator residue. During the
1970s, several waste-hauling firms illegally dumped over
11,600 drums of toxic wastes on sections of this site.5
There are no sanitary landfills within Philadelphia. How-
ever, the city has two municipal incinerators, the Northeast and
Southeast facilities.
In addition, Philadelphia has relatively flat terrain, which
is favorable for conducting the types of air quality dispersion
modeling typically used to estimate ambient concentrations. Many
dispersion models assume flat terrain to simplify the calculation
of ambient concentrations.
Philadelphia was an attractive site for conducting a multi-
media analysis because of possible interactions among several
sources, pollutants, and exposure pathways. For example, inten-
sive industrial development in Philadelphia, Marcus Hook, and
Camden, New Jersey, has resulted in increased loadings of trace
hazardous substances into the Delaware River (both directly from
the facilities and indirectly via the sewage treatment plants).
Individuals can be exposed to effluent discharges to the
river in two ways: through drinking water and inhalation. Since
there is a tidal upsurge in the Delaware River, pollutants dis-
charged from point sources downstream can make their way up to
the intake point at the Baxter drinking water treatment plant.6
Furthermore, the drinking water treatment plant generally draws
most of its water during high tide. Many of the pollutants from
downstream sources have been found in finished drinking water,
albeit in very small concentrations, and therefore exposure can
occur through ingestion.
In addition, organic compounds can volatilize from the
Delaware River, as well as from the sewer system that transports
wastewater to the sewage treatment plant. As a result, there is
also a potential for exposure through inhalation of the chemicals
discharged into the Delaware River.
Slbid., p. 16.
6Linda Shelden and Ronald A. Hites, "Sources and Movement of
Organic Chemicals in the Delaware River," Environmental Science
and Technology, Vol. 13, No. 5, May 1979.
-------
III-6
Local Environmental Agencies
and Toxics Data
The City of Philadelphia has had a long-standing reputation
for concern about toxic substances, as manifested in such local
initiatives as the nation's first right-to-know ordinance. The
Philadelphia Air Management Services (AMS) has drafted ambient
air quality guidelines for 99 "noncriteria" substances and has
also developed a comprehensive emission inventory of most sources
emitting these compounds. This is one of the better air toxics
emission inventories in the nation.
The Philadelphia Water Department (PWD) has also been inno-
vative with regard to investigations of contaminants in drinking
water. PWD has completed a three-year investigation of granular
activated charcoal, ozone disinfection, and other advanced drink-
ing water treatment processes to assess the costs and effective-
ness of these techniques for removing trace organic compounds.
Finally, three of the surrounding counties (Gloucester,
Camden/ and Burlington) are located in New Jersey, which has long
been one of the most progressive states in the country in under-
standing and managing toxics problems. New Jersey has passed
several of its own regulations controlling hazardous substances,
such as its air toxics statute, which has been in effect since
1979. The New Jersey Department of Environmental Protection
(NJDEP) oversees the state's programs, is actively involved in
all research efforts, and maintains data on toxics in all media.
INSTITUTIONAL FRAMEWORK AND
SCOPE OF THE PROJECT
The first three steps in Phase I were to: (1) establish the
institutional arrangements with the federal, state, and local
environmental officials in the Philadelphia metropolitan area;
(2) define the scope of the project, i.e., the parameters of
quantitative risk assessment and the pollutants, sources, and
exposure pathways of concern; and (3) gather all available infor-
mation to characterize current environmental conditions in Phila-
delphia and its neighboring counties. The latter activity was
particularly important since these data helped identify the
majority of the Phase II study topics. Data collection was dif-
ficult, since no single agency typically has access to all the
detailed information necessary for an integrated environmental
evaluation. Furthermore, once located, the data were often
inconsistent in format and unequal in quality from one program
area to another.
-------
III-7
Establishing the Institutional Arrangements
During the Philadelphia IEMP, we coordinated with numerous
groups representing EPA, the City of Philadelphia, and other
state and local environmental agencies. As part of the institu-
tional arrangements, the City took responsibility for informing
the industrial and environmental communities and other public
interest groups of our activities. Each of these participants
contributed valuable perspectives, advice, and information that
greatly enriched the study. Table III-2 lists the main groups
participating in the project.
Table III-2
PHILADELPHIA IEMP
PARTICIPANTS IN THE STUDY
Local Public Health Department,
City of Philadelphia
Water Department,
City of Philadelphia
State Office of Environmental Resources,
Department of Natural Resources,
Commonwealth of Pennsylvania
Office of Toxic Substances Research,
Department of Environmental Protection,
State of New Jersey
Department of Air Pollution Control,
Department of Environmental Protection,
State of New Jersey
Department of Water Pollution Control,
Department of Environmental Protection,
State of New Jersey
Regional Delaware River Basin Commission
Federal Region II
U.S. Environmental Protection Agency
Region III
U.S. Environmental Protection Agency
Office of Policy Analysis,
Regulatory Integration Division
U.S. Environmental Protection Agency
-------
III-8
We also established two intergovernmental committees. The
first was the Steering Committee, which included senior appointed
officials from all participating jurisdictions, EPA, and any
special purpose organization that might have a strong influence
on environmental policy in the area. The other group, the Tech-
nical Committee, consisted of technical staff from the environ-
mental agencies; it reported to the members of the Steering Com-
mittee. Table III-3 lists the members of these two committees.
The committee structures ensured an equal voice for each partici-
pating government.
The responsibilities of the Steering Committee were to
direct the study; decide which environmental issues would be
analyzed; select the appropriate analytical methodologies; deter-
mine the need for further pollution abatement; and, depending on
the Committee's decision regarding additional control, assist in
identifying pollution reduction strategies that would be politi-
cally and economically feasible. All policy judgments were
referred to this Committee, which relied heavily on advice from,
the Technical Committee.
The Technical Committee was primarily responsible for
reviewing and guiding the technical and scientific elements of
all planned activities. In particular, the Technical Committee
was involved in defining the scope of the project; overseeing all
technical analyses, such as ambient monitoring; reviewing the
quality of available data; and assisting in the identification of
feasible control options. The Technical Committee kept the
Steering Committee informed of the project status and all areas
of deliberation, making formal recommendations to the Steering
Committee on courses of action for the unresolved issues.
Defining the Scope of the Project
At the beginning of the Philadelphia IEMP, we made several
important decisions regarding the scope of the project. These
decisions were made to facilitate the evaluation of the Philadel-
phia metropolitan area and to limit our focus to those topics
that*would,-be of concern to local decisionmakers. We recognized,
however, that these initial decisions might require adjustment
after we completed analysis of the available data. Consequently,
we tried throughout the project to remain as flexible as pos-
sible, time and resources permitting, in our definition of the
study's scope of work. We emphasized three major areas in our
initial characterization of the Philadelphia IEMP: quantitative
risk assessment, pollutant selection, and the relevant pathways
of exposure.
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
III-9
Table II1-3
PHILADELPHIA IEMP
MEMBERS OF STEERING AM) TECHNICAL COMMITTEES
Steering Committee
Stuart Shapiro
Commissioner, Public Health Department
City of Philadelphia
Richard Boardman
Associate Deputy Secretary
Office of Environmental Management
Pennsylvania Dept. of Environmental
Resources
Gerald Hanaler
Executive Director
Delaware River Basin Commission
Willis* Marrazzo
Commissioner, Water Department
City of Philadelphia
George Tyler
Assistant Commissioner
New Jersey Dept. of Environmental
Protection
Richard Morqenatem
Director, Office of Policy Analysis
U.S. Environmental Protection Agency
Peter Bioko
Regional Administrator for Region III
U.S. Environmental Protection Agency
Technical Committee
Robert Ostrowski
Director, Program Planning
Air Management Services
City of Philadelphia
Seymour D. Selzer
Head, Planning Branch
Delaware River Basin Commission
Steven Kurtz
Director, Division of Environmental
Quality
State of New Jersey
Rao fans
Manager, Regional Air Quality
Pennsylvania Dept. of Environmental
Resources
Wayne Lynn
Manager, Regional Solid Waste
Pennsylvania Dept. of Environmental
Resources
Robert Krsner
Chief, Air Quslity Monitoring Research
EPA Region III
Willis* Schremp
Environmental Engineer
EPA Region III
Dennis Blair
Chief, Planning Technical
Water Department
City of Philadelphia
Thomas Burke
Director, Office of Cancer and Toxic
Substance Research
New Jersey Oept. of Environmental Protection
Leon Gonshor
Regional Director
Pennsylvania Dept. of Environmental
Resources
Christian Beecnwood
Manager, Regional, Water Quality
Pennsylvania Dept. of Environmental
Resources
Robert Mitkus
Director, Environmental Services Division
EPA Region III
Peter Ludzia
Philadelphia Coordinator
EPA Region III
John Williams
Integrated Environmental Management Program
EPA Office of Policy Analysis
-------
111-10
Quantitative Risk Assessment
We designed the Philadelphia project to assess the potential
risks to human health posed by toxic chemicals once released into
the environment, and then to identify the most efficient strate-
gies that local officials could implement/ if needed, to reduce
risks. Because of limited data and resources, we chose not to
explore ecological effects.
We also did not assess the environmental impacts from either
conventional pollutants in water (e.g., biological oxygen demand,
total suspended solids, oil and grease) or criteria pollutants in
air (e.g., total particulates, sulfur and nitrogen oxides). We
felt that we could contribute more by focusing on toxic compounds
that are neither as well understood nor as stringently regulated
as conventional and criteria pollutants.
We decided to use quantitative estimates of risk to human
health as a consistent measure by which to set priorities across
different sources, pollutants, and media. For example, by using
human health calculations, we could compare the potential risks
from ingestion of drinking water to the risks from inhaling toxic
organic compounds in the ambient air.
We also decided to use a risk-based cost-effectiveness
analysis to evaluate alternative control strategies. We defined
"cost" as the annualized expenditures (both capital and operating
and maintenance costs) required for a specific control option,
and "effectiveness" as the reduction in risk to human health
afforded by the control. Using this analytical framework, we
could generate control strategies that meet specified levels of
health risk reduction at minimum cost by simultaneously comparing
the costs and effectiveness of controls across all sources and
exposure pathways.
We should note that cost-effectiveness is riot the only
approach that can be used to establish control strategies. At
the timer it seemed the most reasonable analytical tool given
available-data. In recent years, however, other approaches have
become more viable, such as benefit-cost analysis, which can
provide different insights on how and where to control pollutant
releases. A recent EPA case study of the benefits and cost of
reducing air pollutant levels in the Philadelphia region pre-
sented a list of control priorities that differed somewhat from
the risk-based findings from the Philadelphia IEMP (although some
-------
III-ll
of this difference was because of the broader set of pollutants
and effects considered).?
Geographic Boundary
Our initial definition of the study area included the eight
counties contained in the Philadelphia Standard Metropolitan
Statistical Area (SMSA):
• Chester County, PA
• Bucks County, PA
• Montgomery County, PA
• Delaware County, PA
• Philadelphia County, PA (coterminus with the City)
• Gloucester County, NJ
• Camdert- County, NJ
• Burlington County, NJ
We decided that we could not focus a priori on a smaller
area without first analyzing environmental information from each
of these counties. In addition, at the start of the project,
local participants preferred a more regional perspective rather
than a narrowly defined boundary. One* we had collected and
evaluated the necessary data, however, we felt comfortable in
allowing the available information to guide us in narrowing these
boundaries. We discovered that much of the area surrounding
Philadelphia is rural and that the majority of the regional
industry and population is located within and immediately around
the City of Philadelphia itself.
To assess the cumulative ambient exposures and the risk to a
given population, all sources of pollution that affect that popu-
lation should be considered. A "perfect* boundary, however, will
never exist: no matter where a line is drawn, there will always
be individuals residing outside it who are affected by sources
inside, and vice versa.
7Dr. Ralph A. Luken and Brett Snyder, "Comprehensive Environ-
mental Analysis for the Philadelphia Region," Office of Policy
Analysis, U.S. Environmental Protection Agency, Revised May 8,
1986, draft paper. We discuss this paper in more detail in
Appendix A.
-------
111-12
Since Philadelphia was the most highly industrialized and
heavily populated county in our study area, we selected it as the
core area of the study/ within which all sources and types of
pollution exposures would be systematically evaluated. We then
selectively defined expansions of the site boundary beyond the
core area, permitting consideration of sources and populations
outside Philadelphia County on a case-by-case basis, according to
the following criteria:
• Sources outside the county; We included any sources of
pollution that we believed affected the core area popu-
lation, even if they were located outside the city.
However, we considered emissions only to the environ-
ment medium for which we judged that exposures in the
core area could be significant. For example, if we
determined that air emissions from a facility in New
Jersey would affect Philadelphia populations, we would
consider the effects of those air emissions on all
populations. We would not, however, consider the water
discharges from that plant, unless we determined that
those discharges affected populations in Philadelphia.
• Populations outside the county; Using similar logic,
we considered all populations affected by sources with-
in Philadelphia, even if they were situated outside the
city.
Using these criteria, we decided to include three areas
adjacent to Philadelphia:
(1) The area extending southwest of Philadelphia through
Delaware County into and including Marcus Hook, predom-
inantly because air emissions from this area could add
to population exposures in Philadelphia County
(2) Parts of Bucks County that purchase drinking water from
the PWD
(3) The area across the Delaware River, east and south of
Philadelphia, including the northern parts of Camden
and Gloucester counties, New Jersey, because of the
potential for Philadelphia air emissions to contribute
to human health exposures in these areas
At first, we identified several other sources in New Jersey that
we thought might influence exposure levels in Philadelphia.
However, after careful review of the available data by Technical
Committee members from New Jersey, we revised our emissions esti-
mates and concluded that the New Jersey sources showed only a
negligible impact on Philadelphia. Figure III-2 shows the boun-
daries of the study area.
-------
111-13
Figure 111-2
Philadelphia IEMP
Study Area Boundaries
CHESTER COUNTY
GLOUCESTER
COUNTY
-------
111-14.
Initial Pollutant Selection
The project first focused on chemicals with chronic human
toxicity, especially compounds that were known or suspected human
carcinogens. We concentrated on cancer for several reasons:
• The public has expressed a growing concern about the
potential link between environmental releases and inci-
dence of cancer.
• More dose-response information with sufficient weight
of evidence was available for cancer than for other
chronic human health effects.
• The "nonthreshold" assumption permitted quantitative
risk assessment of carcinogenic effects even at the
very low concentrations typical of most ambient
conditions.
Using several sources, we compiled a preliminary list of
about 170 pollutants of potential concern in the Philadelphia
area. We first reviewed which chemicals EPA regulates or is
planning to regulate under its various statutory authorities,
such as the Clean Water Act's 129 Priority Pollutants, and toxics
regulated under Section 112 of the Clean Air Act (National Emis-
sion Standards for Hazardous Air Pollutants). We also evaluated
existing EPA information on pollutant toxicity, such as EPA and
National Academy of Science drinking water quality documents and
Health Assessment Documents prepared by EPA's Carcinogen Assess-
ment Group.
We then identified which specific toxic compounds the state
and local environmental agencies regulate, such as those listed
in AMS's emission inventory, which is used in issuing permits for
point-source releases of 99 toxic substances. Finally, we
decided to consider adding any other pollutants that new data
indicated were a compound of concern or that were found in either
the ambient air or surface water (we could not make these deter-
minations until we reviewed the available monitoring data). A
complete list of the 170 preliminary pollutants and data sources
can be found in Appendix B.
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
111-15
Relevant Pathways of Exposure
We considered two exposure pathways: inhalation of ambient
air and ingestion of contaminated drinking water. We decided
against evaluating human health exposures through the food chain
or through dermal contact. We also did not assess the risks from
radiation or indoor air pollution. While we believe that these
routes of exposure may be significant and deserve attention, we
could not address these pathways because of limited resources and
the paucity of site-specific data on these routes of exposure.
Collecting/ Organizing,
and Automating Data
We collected environmental data to assess potential human
exposures to toxic chemicals. In some instances, data directly
useful for measuring exposure were already available, e.g., data
on the quality of finished drinking water. In other cases, how-
ever, data for directly assessing exposure was very limited or
nonexistent. In these situations, especially in the case of air,
we estimated the expected presence of a compound in the ambient
environment using data on pollutant sources, the manufacturing
process, the magnitude of pollutant releases into the environ-
ment, and fate and transport models.
The data collection effort involved three separate tasks:
gathering all existing information on sources, pollutant loadings
to each media, and ambient concentrations; generating engineering
estimates to fill in data gaps; and developing an automated data
management system to store and manipulate the information.
Existing Data
As detailed in Appendix C, we obtained information on a wide
range of environmental topics. To characterize point-source
pollutant releases to air and water, we relied on a mixture of
EPA and local databases, such as the AMS's emissions inventory
for industrial air emissions in Philadelphia, EPA's Industrial
Facilities Discharge file (water), permit data, and the available
Discharge Monitoring Reports (DMRs) for direct water dischargers.
In addition, data from special reports and studies conducted in
the Philadelphia metropolitan area were helpful.
At the time we conducted this phase of the IEMP (summer
1982), only limited data on hazardous waste were available. We
did, however, gather available Part A permit information from
EPA's Hazardous Waste Data Management System (HWDMS). We also
reviewed Superfund site investigation reports that were available
for those sites in the eight-county metropolitan area.
-------
111-16
As a next step, we collected all ambient data for toxic
pollutants. For surface water, we obtained measured data for
both tne Delaware and Schuylkill Rivers, as well as for the
finished drinking water from Philadelphia Water Department's
three drinking water treatment plants (Queen Lane, Belmont, and
Baxter). For air, we found little information on ambient concen-
trations of airborne toxics at the time we completed our prelimi-
nary data collection.
Finally, we gathered data on the geophysical characteristics
of the study area. For example, we collected meteorological and
climatological data, and U.S. Geological Survey (USGS) informa-
tion on flow of all major rivers and boundaries of watersheds and
subwatersheds.
Engineering Estimates
After we analyzed the available data, we identified many
important areas where more information was needed for assessing
exposures. We used engineering estimates in this phase of the
study, rather than more expensive and time-consuming monitoring.
EPA has made extensive use of engineering estimates in its regu-
latory development, and we felt that developing pollutant release
rates by source category was the only way to generate the infor-
mation needed to complete Phase I and still remain within the
project's budget and schedule. We developed engineering esti-
mates for three broad categories: area air sources, nonpoint
water sources, and intermedia transfers.
Area Air Sources. The available information did not charac-
terize toxic emissions from small, widely dispersed air pollution
sources. These so-called "area" air sources include:
• Degreasing
• Dry cleaning
• Other solvent usage—surface coating, graphic arts
(printing and publishing), rubber and plastic, indus-
trial, and nonindustrial
• Mobile sources
Using an emissions factor approach and information provided by
the National Emission Data System (NEDS) "Area Source Report," we
generated emission estimates for numerous compounds from each of
these categories.
-------
111-17
For the degreasing, dry cleaning, and other solvent usage
categories, we gathered data from NEDS on the amount of solvent
(volatile organic compounds, or VOCs) consumed. We assumed con-
servatively that all solvent consumed volatilizes and is emitted
to the atmosphere. Using existing EPA reports, we developed
source-specific emission factors to partition the total VOC load-
ings so we could estimate pollutant-specific emissions. We gen-
erated emission factors for between two and 15 VOCs, depending on
the source category.**
For mobile sources, we concentrated on gasoline- and diesel-
fueled road vehicles (not off-road vehicles). Using existing
information, we developed emission factors based on vehicle miles
traveled. NEDS provided us with the vehicle miles traveled for
each type of mobile source by county.9 Finally, we calculated
emissions of seven substances from gasoline marketing (benzene,
ethylene dibromide, ethyl benzene, 1,2-dichloroethane, toluene,
trimethylbenzene, ethyl toluene) using total VOC loadings for gas
marketing from NEDS and emission factors.10
Once we calculated emissions from the different area-source
categories, we allocated these emissions by land usage in the
Philadelphia area. In contrast, most traditional analyses assume
uniform distribution of area-source releases (frequently referred
to as a "box model" approach). Using data maintained by the
USGS, we divided the Philadelphia metropolitan area into grid
sections with the following land uses:
• Residential
• Commercial and services (e.g., shopping centers)
• Industrial
• Industrial and commercial complexes (e.g., industrial/
warehouse parks)
• Transportation, communications, and utilities (e.g.,
highways and airports)
Table III-4 shows how the emission allocation scheme oper-
ated. For example, we apportioned all residential heating emis-
sions to those grids designated residential. Similarly, we
8Versar, The Estimation and Control of Area Source Emissions in
Philadelphia: A Methodology and Case Study, EPA Contract
#68-01-6715, April 30, 1984.
9Ibid.
10TEI3.
-------
111-18
concentrated roughly 90 percent of the dry cleaning releases in
the commercial and service zones. The remaining 10 percent was
split evenly between the industrial and industrial and commercial
sectors.
One caveat to note about the area source emission estimates
is that they are based on NEDS. While the quality of NEDS has
been questioned by many, we felt that this system provided us
with the best readily available information on area sources for
all counties in the study area. We also attempted to improve the
quality of the information by requesting review of the Philadel-
phia NEDS data by the project participants most familiar with VOC
levels in the area. Where better local data existed, we
requested them.
Nonpoint Water Sources. There is little information on
"nonpoint" discharges into water systems. Concerned that these
toxic loadings to the Delaware and the drinking water supply
could be significant, we developed estimates for the following
major nonpoint source (NFS) categories in Philadelphia: agricul-
tural runoff, silvicultural runoff, urban storm water runoff, and
combined sewer overflow
Intermedia Transfers. In our data collection, we looked
specifically for intermedia pollutant transfers. Given the size
of the NEWPCP in Philadelphia and the number of indirect indus-
trial discharges, we felt that emissions from wastewater treat-
ment could be a significant air source. As expected, however,
site-specific data characterizing cross-media shifts were also
missing. But our review of basic research on cross-media pol-
lutant shifts discovered several relevant studies that EPA's
research labs in Cincinnati had sponsored on emissions from
wastewater treatment facilities. *•*
EPft^ Philadelphia Demonstration Project, Integrated
Environmental Management Program; Methods for Characterizing
Releases of Toxics to Air and Water, January 21, 1983. Office
of Water Regulation and Standards and others have also devel-
oped pollutant-specific estimates of air emissions from
biological/activated sludge treatment plants.
An important source to review is E.D. Pellizzari, Volatile
Organics in Aeration Gases at Municipal Treatment Plants, Con-
tract No. 68-03-2780, U.S. EPA Municipal Environmental Research
Laboratory, Cincinnati, Ohio, 1981. This and several other
research project reports are a part of the background documents
and are available on request.
-------
111-19
I I I
* 2
id *
s !
: s S
If.
Il'
uxl
II •
M
11 '
il
il
i;
11
ill
S
e
8
a
S S
e a
a
o
ass
«A • I t
I t t
I t I
12 a - -
5 1§ S J| 1^ = =
• ^8 ^ 2" ~.l \ I - S
*t-5-S
li£i
i I* I
i a 2
i!
21
j!
If
in
III
SIS
111
s
to
W
in
a i
I
1
!
Ms
• — e
Hi
i?? ^
= — ^ 5
'I- S
!]! I
-oi "
•;! 2
ill J
?e^ s
!"! ;
il
K'li
1^1
ii 1
*.ss
-------
111-20
To predict the magnitude of air emissions at NEWPCP, we used
a crude mass-balance algorithm:
Mass volatilized » Loadings to NEWPCP -
Loadings out of NEWPCP
While this algorithm is a simple functional relationship, it is
driven by the common-sense principle of conservation of matter.
We gathered data from the PWD's files on loadings to the
NEWPCP from indirect dischargers. In assessing the amount of
each chemical discharged to the NEWPCP/ we noticed two volatile
organic compounds of interest: 1,2-dichloroethane (1,2-DCE), and
1,2-dichloropropane (1,2-DCP). The PWD's files recorded the
annual loadings of 1,2-DCE and 1,2-DCP as 177 kkg and 248 kkg,
respectively. We estimated (no data were available) the pollu-
tant discharges from the NEWPCP by calculating back from the
PWD's monitoring data at the Baxter drinking water treatment
plant, which was the best available information. On the basis of
an EPA study, we assumed that dilution between the NEWPCP and
Baxter would be 1 percent.13 Our calculations resulted in
average annual air emissions from the sewage treatment plant of
113 kkg and 146'kkg for 1,2-DCP and 1,2-DCE, respectively. Since
the time of this project, EPA's Monitoring and Data Support
Division, Office of Water Regulation and Standards, and others
have also developed pollutant-specific estimates of air emissions
from biological/activated sludge treatment plants.
Other sources of secondary releases for which we developed
engineering estimates were fugitive releases from petrochemical
storage tanks; releases from sewer systems en route to the water
pollutant control plant (estimated during Phase II); VOC
emissions from the Delaware River resulting from direct dis-
chargers (estimated during Phase II); and emissions from cooling
towers of chromium (used as a fungicide) and chloroform (a by-
product of drinking water disinfection).
The Automated Data System
To store and analyze the numerous pieces of collected infor-
mation efficiently, we found it helpful to merge the different
files into a single automated data system. We developed a user-
friendly system, known as PIPQUIC (Program Integration Project
s. EPA, Evaluation and Characterization—Philadelphia
Northeast Water Pollution Control Plant, National Enforcement
Investigations Center, Denver, Colorado, and Region III,
Philadelphia, Pennsylvania, 1977.
-------
111-21
Queries Using Interactive Commands), that can be used in any
geographic^ location to manage and array environmental data in
numerous ways, such as matrices, pie charts, and histograms.
Figure III-3 exhibits some of the computer graphics available
through PIPQUIC. The system ties directly into many of EPA's
major program databases and allows the user to sort the data by
pollutant source or medium without extensive computer expertise.
For Philadelphia, PIPQUIC contains information—primarily
from 1980 to 1982—on 345 separate industrial facilities and
31 nonpoint area-source categories. Of the 345 industrial
sources, 36 are major, separately permitted emitters to both air
and water, 28 are major water dischargers only, and 281 release
pollutants to the air. There are relatively few major water
data in PIPQUIC varies in quality, the system identifies each
data source and contains information on pertinent characteristics
of the database. The PIPQUIC user's handbook is part of the
background documents that are available on request.
PHASE I PRIORITY-SETTING
Once the preliminary data had been collected, organized, and
reviewed by the project participants (we requested review of the
collected data by the Technical Committee and EPA, and made all
necessary revisions to improve the quality of the data), we began
the difficult process of narrowing the project's geographic and
analytical scope. Reducing the numerous potential issues to a
manageable number of topics for study in Phase II was particu-
larly complicated, primarily because making judgments about the
potential significance of an issue wh«n data are either absent or
ambiguous is very difficult. Setting priorities involved careful
consideration of potential risks to human health based on local
expertise and experience, value judgments, and resource con-
straints. This section describes th«s« activities in more detail
and discusses the major considerations in selecting Phase II
study topics.
Genera tinq^Issue Papers
We organized the information on potential environmental
concerns by developing "issue papers." Each issue paper con-
sisted of a standard matrix format for presenting the summary
information for each issue. The objective was to combine all
information needed to conduct a qualitative and, where possible,
a quantitative risk assessment. Table III-5 shows an example of
the format we used. Appendix D lists the issue papers by subject
title.
-------
111-22
Figure 111-3
Philadelphia IEMP
Example of PIPQUIC Computer Graphics
Too!
413 E«tbn»i»o E»c»»» Ltt»itm« Incidence
B*M>6 on E minion* from »i: Sourc*»
R.tk
0.03
60 00
PelJtunt-MMhytont! CntorM*
S3 33
106.67N
Tool 1J1: Menth-ly-Momn Wf»e1 Walt* Mielura**
U"«t KMognmt p*r D*y
10S
Tool 450: Air Emlislons In Mttrte Tons MI Y««r
Pollutint-Mcthylin* ChtorW*
T*el 410: Air Cmltilent In M.lfle Tent M'
0*i*»r». PA illiniums*
VTY
103 CS
3St 70
121 K-
3n i:
371 77
71 1 74
1
I
I
I
I
I
I
I
-------
111-23
In conjunction with the Technical Committee, we developed
issue papers on 39 topics on the basis of the data and any ideas
put forth by the study participants. The issue papers were gen-
erated by source, pollutant, and exposure pathway and were
divided into the following groups: water, air, hazardous waste,
solid waste, and multimedia (i.e., sources of intermedia trans-
fers). The major data elements were pollutant toxicity, sources
and emissions, potential exposure pathways, ambient levels, popu-
lation exposed, exposure levels, and adverse effects.
For each data segment characterized, we attempted to detail
all quantitative findings. In many situations,, however, we found
that the information was insufficient to calculate risks. For
example, releases were unknown, or there were no potency factors.
In these situations, we used as much qualitative information as
necessary to help us interpret the significance of an issue. In
Chapter II we elaborated on methods to make qualitative risk
assessments.
We also made an appraisal of the quality of the data pre-
sented. It was important to be able to distinguish between well-
documented data ^references and reports and those with more
questionable data. We know of no hard and fast rules used to
appraise data quality and, therefore, we relied on expert
judgment.
Components of the issue papers are discussed below.
Pollutant Toxicity
We presented information for each cell of the matrix using
different approaches. First, for each pollutant associated with
or likely to be associated with an issue, we used available toxi-
cological data to assign a quantitative indication of the pollu-
tant's cancer potency or association with other adverse health
effects. In the absense of a quantitative measure, we assigned a
qualitative human toxicity effect value of high, medium, or low,
based on a brief review of readily available toxicological infor-
mation:
• High (1). Compound is either a suspected or known
carcinogen, or is strongly linked to other human health
effects, such as neurotoxicity or renal toxicity.
• Medium (2). Compound may cause harm to human health,
either mortality or other adverse effects, at rela-
tively low concentrations.
-------
111-24
I
3
0
S
b
•*4
0
5
*> y §
i &
I
X
JJ
^J
o
I
Quant
i1
l
fj
ll
J
I
Pollutant
Si-
12
•H X
0 a
.1
Is
1*8
8ii
-ll
ill
1
i
-8
!-'
0. i-4 Al
oog
s -
33§i
b e ••» 9
•s |-s I
11.s
c a b
•rt -> 5*J
0
sl-I
e fl a 0
8
i 8
i-8 "
Sc?
5
1
•= -u
3 S
* I
2 i
-o •
5 5
x x
I *
! i
b N
1 S
O N
i 2~
5 U §'
b
O
1
O .u O
> 0 *J
b +1
o a
.U »
a u
IS
0
5
§ §£
b
O
. 1
e 2
0 O -»
2? 2 S5
0
•s
a.
•§
i
f
•" 5
3l
-• 0
st
I
-
* S •"
000
_J ^O 0
O
*4
s
18 2 IH*
• — o o a** o
OAJAf 4J4JUUJ
• -i o o « 5 S
-< e 2 S0-ua
~ S
x i x
0-ua
i* o •—
-
X JJ
•11
JIB
•si .
!^S
i S
^•4 O
ll
1 «
!|
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
111-25
X
*•
3
=
&
^ O
^- * X W
^ i <
1 Ul •«
ox —
1 fc *
d
s
tf>
I
1
(^
5
i*.
o
B
B
5
|
<§
?
'«•]
Qualltat
a
c
1
8.
-gab
O ^ B B B
>* b a ** o
•B o> i a 3
C S e .u u
"a — 15 x c
— a 8 9
0 tl b 01
5. 5 *> jj a
b n 5 Is
U SlS'
a ••< c t- •o
28*3
M o ••> •*
£ 5.0 a a
O» • 9 ** £
-* -* o\ to 3
x -o «- -o S
f
^ t ^J
•2-S S^
Jbi2-|
MO ax
s . I* i •
b £J 5 B b b
a ON a H w B
•g f^ili S B §
3 9\ -i +i C
** "\ a *j a
a ^ e « i =
"* ™ 'j •* 2
o 8 ° » 1i
3 a •— ** "3. T3
CT> O. •» — i O — i O
a. g o.^
llliiii
t
(b* **
o «
>.— u •
•** a a 9
0 > S
•S o a
c|S.f
'"* "3
11 8'
a a B b
x a o
- £ 8 a
S-5 5
*J O > 1
^?. S
—•*«•«
£S.I
a
b ""^
* 5 i
s 2 ••«
II"
a
•o
b
a *j
£5
B ?
4 ^w
|*
85
b e
a 'a
b AJ
S 85
o S x
S &
S !
1 «
3 iJ
a
S S
£ 8
I I
fa •
3 II
ID 01
6 b 3
• 6
B >» *J (M
b a a «-
3-0 S
It1 g1-2
X -*N V0
B .Q ^
-Ik. b b
< 0 0 »-
s -05
S 2 S S
b «4 OB
3 .u a a
a a .0 c
oL ** <^ a 01
X C O b £
B s 3 B
o >. a
3 "J2-S
3 8,f "S g.
•o * o B c
B b b n *4
— B i C b
•S a *" "" 5
"3 *j e 8
x 3 a « §
- 2 5 i. s a
c a *j -H a B
S i a B b jj
jj i B •$ a a
3-> a a «
— > B B -« a •*
—(•OUM u
£1 If s 8
B
b
fa
1>
X B
UJ ^
i
1
a
6
b
1
B
•8
^
S
1
S
•°
•§>5
£•§
"S
la* 8
8 Sf
f si
•S 1*7
s ^<
s I?
S p. a
r ib
X «- ^
I x x
CD J% ^
1
|
b
a
a
1
u
H
•'i
a at
•4 g
x •«
u
4J b
i %
3 =
11
g «
B O
It
-------
111-26
( 3 ) . Compound is unlikely to have adverse effects
on most individuals at levels found in the environment.
Sources, Releases , and Pathways
We then identified sources, pollutant releases to the envi-
ronment by media, and any intermedia transfers using either
existing data or engineering estimates. We also ranked our con-
fidence in this type of data. We assigned a confidence ranking
of high if the release rates were based on empirical data, and
low if the releases were based on engineering estimates, we
identified exposure pathways according to either modeled or mea-
sured concentrations in ambient air, surface water, and/or
finished drinking water.
Ambient Concentrations
Next we presented information on ambient pollutant concen-
trations. Where available, we recorded measured data; however,
since ambient data were generally very limited and rarely source-
specific for toxics, we used mathematical fate and transport
models to estimate ambient concentrations in air and surface
water. The confidence ratings for ambient concentrations
depended on whether the information was based on empirical data
(high) or models alone (low).
We did not assess hazardous wastes or ground water contam-
ination issues in Philadelphia for several reasons. First, we
judged that the hazardous waste issues for the city — primarily
treatment and storage tanks — were not expected to affect surface
water quality and the drinking water supply in any significant
way. In addition, while the Enterprise Avenue site was of local
concern, it had already been investigated and excavated for
exploratory purposes by the PWD in 1979 and 1980 and was already
undergoing cleanup by the time we began Phase I. We decided that
we could not add much new information to this issue. Finally,
and most important, Philadelphia relies on surface water (the
Delaware and Schuylkill rivers), not ground water, for all its
drinking water supply.
Where an air model was required, we used the Climatological
Dispersion Model (COM). While there are numerous EPA-approved
dispersion models, and some that are better at point-source
modeling, we judged COM as best for a screening approach for
several reasons. First, CDM is designed as an urban-scale model
for areas with relatively flat terrain, such as Philadelphia.
Second, it can model multiple chemicals from up to 2,500 area
-------
111-27
sources and 200 point sources in one run. Finally, and most
important, COM can generate summary tables showing percentage
culpability from all sources, which can be used to identify
priorities among sources and pollutants.
To calculate ambient air concentrations, we established an
84-grid system that encompassed the geographic boundary of the
IEMP Philadelphia project (see Figure III-4). CDM calculated
concentrations and source culpability for each 5-kilometer
receptor square.
For surface water, we used engineering estimates and a model
of the Delaware Estuary developed by EPA's Athens, Georgia,
research laboratories, TOXIWASP.^4 This model was superior to
other available EPA models, such as EXAMS, since it was designed
specifically for the Delaware Estuary.
Receptor Population
The next step was to determine the exposed or "receptor"
populations. W£ estimated the population exposed through inges-
tion of drinking water by gathering data from the PWD on cus-
tomers served by each drinking water treatment plant. For expo-
sure to air contaminants, we matched Census data to our disper-
sion model grid system to estimate the number of people residing
within each grid square. We then rated our confidence in the
population data as high (known exposure pathways) or low (expo-
sure pathway uncertain).
Exposure Levels and Adverse Effects
The final pieces of information needed for each issue paper
were quantitative exposure levels and quantitative estimates of
adverse effects. To estimate exposure levels, we used air dis-
persion models and standard EPA assumptions regarding doses (2
liters ingested/day or 20 cubic meters inhaled/day). We then
multiplied these levels by potency factors, if available, to
estimate cancer effects. Confidence ratings for this category
were either high (based on actual dose-response data) or low
(based on subjective scoring).
14A more detailed discussion of TOXIWASP can be found in Robert
B. Ambrose, Jr., "Modeling Volatile Organics in the Delaware
Estuary," Environmental Research Laboratory, U.S. EPA, Athens,
Georgia, Draft Paper, May 1986.
-------
111-28
Figure 111-4
Philadelphia IEMP
Receptor Locations on Grid System
M
IM
•UOMIUM
-------
111-29
As mentioned earlier in this report, we decided to develop
pollution abatement schemes according to measurements of cost-
effectiveness, where effectiveness is defined as the degree to
which a set of controls reduces human health risks. It was
therefore essential to have dose-response information that could
be used to explore the impact of reductions in pollutant loadings
on the exposed population, particularly at low concentration
levels reflective of ambient conditions.
At the time of Phase I, we had dose-response information
only for carcinogenic effects. For pollutants associated with
noncancer effects, we had information on the no-effect threshold
levels (e.g., Reference Doses) but no dose-response relationships
to determine incidence; thus, it was not possible to analyze
these disease categories in our cost-effectiveness framework.
Furthermore, we did not expect that ambient concentrations would
approach the threshold levels. For these reasons, we decided to
focus only on carcinogens during Phase I.
Our decision to consider only carcinogens greatly reduced
the number of chemicals to be addressed in Phase II. The number
of pollutants for which EPA's Carcinogen Assessment Group has
developed unit risk factors is limited (approximately 54 com-
pounds, excluding the Reportable Quantity values calculated for
Superfund).I5 Our analysis of the IEMP data for Philadelphia
revealed yet a smaller number of carcinogenic substances with
releases and exposures in the study area.
We recognize that our decision not to further assess the
potential for noncancer health effects in Phase I was a limita-
tion. More recent projects, especially the Santa Clara IEMP,
have considered these effects more thoroughly. The results of
our work in Phase II, however, indicated that the levels of expo-
sure to pollutants selected for additional study do not exceed
the threshold values for noncancer health effects. The one
important exception could be for lead, which was screened out
during Phaser I because of low releases and exposures, as well as
limited dose-response data. Since Phase I was completed, we have
much better, dose-response information for lead at ambient concen-
tration levels, and the thresholds for some effects, such as
hypertension and blood FEP, are extremely low. Furthermore,
there is limited evidence that some lead compounds induce tumors
in experimental animals. CAG is reviewing these data, but has
not yet made a determination as to the potential carcinogenicity
of lead.
list of the 54 compounds and their unit risk factors can be
found in the EPA Health Assessment Documents.
-------
111-30
We should again note that the standard EPA method used for
calculating human health exposures in the Philadelphia TEMP
addresses only exposure to ambient pollutant levels; it does not
accurately reflect total personal exposures. EPA's Office of
Research and Development has, however, been field-testing differ-
ent methods to assess total human exposures under its Total Expo-
sure Assessment Methodology (TEAM) Program, which we referred to
in Chapter II.
The Risk Screen
After arraying the information for each issue paper, we
evaluated each issue qualitatively and, to the extent possible,
quantitatively, in terms of pollutant toxicity or potency and in
terms of the potential for high exposures. The purpose of the
evaluation was to reduce the number of possible issues to a few
manageable issues, given the scope of the project, for further
study in Phase II.
The participants in the IEMP shared the view that two
general types of\ topics should be investigated in Phase II:
• Issues that appeared, on the basis of existing data, to
warrant study because of the potential for both high
risks and feasibility of control options
• Issues that could not be adequately characterized in
Phase I because of major data gaps, but were of concern
to the project participants
Issues were selected on the basis of primary (largely quan-
titative) and secondary (largely qualitative) criteria. While
the quantitative risk assessment was an important element in the
screening process, it was not always the deciding factor, espe-
cially in those areas where data were sparse. In these situa-
tions, we set priorities by relying, for example, on subjective
hazard determinations and the degree of concern expressed bv
state and local officials. We discuss primary and secondary
criteria in more detail below.
Primary Criteria
Our first step was to rank the issues according to their
potential for high exposures. We assessed the potential for high
exposures by using information about pollutant loadings to the
environment, available monitoring data or COM modeling estimates,
and information about population distribution. We then placed
-------
111-31
issues into one of seven categories on the basis of these avail-
able data. Table III-6 details the priority categories.
Category
Priority I
Priority II
Priority III
Priority IV
Priority V
Priority VI
Table III-6
PHILADCIPHIA IEMP
CATEGORIES Of POLLUTANTS AND STATUS OF PRIMARY CRITERIA
Primary Criteria
Toxicity
High
High
Medium
Medium
High
Medium
Source
Ambient Concentration Data
Medium and high loadings Independent of known presence in
to air and/or water.
Low loadings to air
and/or water.
High loadings to air
and/or water.
High loadings.
Low loadings, to air
and/or water.
Low and medium loadings
to air and/or water.
ambient air or water.
Known presence in either ambient
air or water, or both.
Known presence in either ambient
air or water, or both.
No known presence in either
ambient air or water, or both.
Independent of known presence in
ambient air or water.
Includes all other chemicals, ragardleas of toxicity rating, that had
no source and ambient data, or xero loadings to air and water and no
known presence in ambient air and •mter, or any combination of these.
Also included in this category arc all chemicals of low toxicity.
These are deleted from initial Uat.
Priority VII Chemicals of unknown toxicity.
We -then developed a comprehensive confidence rating to
reflect the quality of the underlying data and the degree to
which the data suggest there may b« a human health hazard.
Table III-7 presents this matrix and shows the three classifica-
tions for confidence (strong, medium, and weak). While the
"strong" category.indicates a high level of confidence in the
data for ambient levels, exposure, and adverse health effects,
the other two categories permitted different combinations of high
and low confidence in the data. In general, we were most inter-
ested in those topics that were rated either strdng or moderate.
-------
111-32
Because each criterion seemed to suggest a different set of
study topics, we used a subjective weighting scheme to determine
the final list of topics. We selected those pollutants identi-
fied as Priority I through IV (as explained in Table III-6) for
further consideration. Since the degree of human health risk is
a function of exposure and toxicity, we felt that limiting our
consideration to these four categories, where toxicity or expo-
sure or both were relatively high, was sufficient to 'identify
pollutants for which the potentially significant human health
risks could be determined with some accuracy. Existence of ambi-
ent data provided a good check on our estimates, but we did not
want to preclude a pollutant from further consideration solely
because monitoring data were scarce (as in Priority IV), which is
generally the rule rather than the exception.
Table III-7
PHILADELPHIA IEMP
DATA CONFIDENCE RATING
FOR PHASE I RISK SCREEN1
Confidence
Rating
Strong
Moderate
Moderate
Moderate
Weak
Weak
Weak
Weak
^Confidence
Ambient
Concentrations
High
Low
High
High
High
Low
Low
Low
High = based on
empirical data
Low = based on models
default values,
or assumptions
Population Exposures Adverse Effects
High
High
Low
High
Low
High
Low
Low
High = known exposure High
pathway
, Low = exposure pathway Low
uncertain
in the quality of the underlying data and the analytical
High
High
High
Low
Low
Low
High
Low
= based on actual
dose-response
data
= based oh subjec-
tive scoring
findings.
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
111-33
Secondary Criteria
After selecting issues that fell into Priorities I through
IV, we evaluated those issues in terms of our secondary criteria.
These criteria, described below, included analytical feasibility,
relevance to EPA, state, and local program objectives, and the
potential for effective response and control.
Analytical Feasibility. The primary elements of analytical
feasibility are the amount of supporting data and the level of
effort required to generate new data. These are basically pro-
gram management considerations in that they indicate how much
effort will be needed for basic data gathering before developing
management alternatives (i.e., control options). There is an
important trade-off implicit in the criterion of analytical
feasibility: the greatest payoff for the purpose of identifying
or controlling toxics may be in studying issues that have not
been studied in the past or have been avoided because of the com-
plexity or magnitude of the issue. On the other hand, within the
time and budget available for the project, it may not be feasible
to characterize these issues adequately, and it may be necessary
to curtail the analysis short of developing complete risk manage-
ment strategies.1
EPA, State, and Local Program Objectives. The project was a
cooperative effort with state and local agencies, and their con-
cerns weighed heavily in the final choice of study topics. AMS,
for example, was interested in learning about benzene releases
from gas marketing at service stations and the potential impact
of used oil combustion on the city's ambient air quality. In
addition, AMS wanted to monitor ambient formaldehyde concentra-
tions to determine whether they exceeded its acceptable ambient
air guidelines. The state of New Jersey was interested in
assessing air emissions from landfills.
Several EPA offices were interested in researching and
developing methods for analyzing additional topics. For example,
EPA's Office of Research and Development (ORD) was interested in
field-testing a mobile monitoring system they had developed for
detecting air emissions from hazardous and sanitary landfills.
We also wanted to evaluate this monitoring technique for possible
application in future geographical studies. Although there were
no landfills in Philadelphia proper, there were several active
and abandoned landfills within the study area in New Jersey that
served as testing sites.
-------
111-34
Ability to Control Impacts. One of the important methodo-
logical objectives of the project was to identify control strat-
egies for reducing the potential threat to human health posed by
specific environmental situations. An important consideration
during the setting of priorities was, therefore, to focus on
issues for which we could develop feasible control strategies.
For example, we chose not to include automobile emissions in the
cost-effectiveness analysis. At the time, this was considered an
issue best managed at the national level since it may be imprac-
tical to require automobile manufacturers to install controls in
cars destined for particular geographic areas.
As we started screening the issues for control feasibility,
we found it convenient to evaluate environmental impacts and
risks separately, by source or pollutant. However, we detected
two sources of risk for which this approach was inappropriate:
indirect industrial discharges to the NEWPCP and volatilization
during wastewater treatment at the plant; and effluent discharges
from the NEWPCP to the Delaware River and the presence of trace
organic compounds in the drinking water from the Baxter plant.
As discussed earlier, the tidal surge in this section of the
river pushes the pollutants discharged from the NEWPCP upstream
past the Baxter intake point. The plant, in turn, withdraws the
raw water during the high tide because it is energy efficient for
the plant's operations to do so. If we had not considered these
interrelationships, we might have missed several sources and
exposure pathways, as well as possible additional control alter-
natives.
Example: Pollutant Selection
for Control-Options Analysis
We discuss below how we narrowed the list of pollutants (and
therefore sources and issues) to a manageable number for consid-
eration in our control-options analysis.
As we noted above, we eliminated issues that did not involve
pollutants in Priorities I through IV. We added for considera-
tion pollutants categorized as toxic by Schedule A of Philadel-
phia's Air Management Regulation VI. The resulting 26 pollutants
are listed in Table III-8. Using the pollutant release data and
an air dispersion model, we then calculated ambient concentra-
tions.
We noticed that the modeled ambient concentrations changed,
in some cases rather significantly, from one contiguous receptor
grid square to another. We attributed these changes to different
pollutant source distributions, reflecting differences in land
use patterns, (e.g., industrial neighborhoods, commercial and
business locations) across the city.
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
Twenty-Six High-
Toxicity Pollutants
Acrylonitrilo1
Antimony
Benzene
Beryllium
Carbon tetrachloride1
Celloeolve
Chloroform1
Chromium1
Cyelohexanone
1,2-dichloroethane1
1,2-dichloropropane
Dioxane
Ethylane dibromide1
Formaldehyde1 s
Lead1
Mercury1 ;
Methylethylketone
Methylene chloride1
Nickel
Pentachlorophenol1
Selenium
Perchloroethylane1
1,1,1-trichloroethane
Trichloroethylene1
Vanadium
Vinyl chloride1
111-35
Table III-8
PHILADELPHIA IEMP
POLLUTANT SELECTION
Eleven High-Toxicity,
High-Volume Pollutants
Benzene1
Carbon Tetrachloride1
Chloroform1
1,2-dichloroethane1
1,2-dichloropropane '
Formaldehyde1
Lead1
Methylene chloride1
Perchloroethylene1
1,1,1-trichloroethane
Trichloroethylene1
Eight Air Pollutants
Included in the
Control-Options
Analysis1
Benzene1
Carbon Tetrachloride1
Chloroform1
1,2-dichloroethane1
1,2-dichloropropene
Methylene chloride1
Perchloroethylene1
Trichloroethylene1
*Toxic air contaminants listed on Schedule A of Philadelphia's Air Management
Regulation VI.
We also found that numerous small sources (i.e., those con-
tributing less than 5 percent individually to the ambient con-
centration) can in total account for the majority of the total
ambient concentrations of the 26 toxic pollutants in certain
parts of the study area. This finding is of some significance
for control possibilities. For example, Table III-9 presents our
estimates of the relative contribution of point and area sources
-------
111-36
to total ambient concentrations by grid square. The small
sources (point plus area) account for more than half of the
estimated ambient concentrations in squares 45, 47, 48, 50, 51,
63, and 64, most of which are located around the industrialized
section of Philadelphia. Table I1I-9 also shows the dominance of
the larger area-source categories in other areas of the city.
However, as we near the industrial corridor aloag the Delaware
River (squares 45, 46, 47, 48, 49, 50, and 51), the point-source
contribution increases.
Table I I 1-9
PHILADELPHIA IEMP
PHASE I CLIMATOLOGICAL DISPERSION MODEL (CON) RESULTS
Receptor Grid
Square Number
6
20
21
32
33
34
- 35
36
37
38
45
46
47
48
49
50
51
63
64
SOURCES AFFECTING POPULATION
(26 toxic pollutants liated in
(COM Run 4/8/83)
Percent Contribution to
Ambient Concentration
Point Sources (3) Area Sourcea (5)
56.59
63.11
64.95
55.39
59.43
56.77
59.09
60.38
58.82
64.05
34.66
26.55 24.21
49.24
5.09 43.96
16.78
12.11 26.98
5.81
10.52 38.42
14.49 31.05
33.65 12.35
27.03 14.78
RECEPTORS
Table 111-8)
Percent
Indiv idual
Contributions
<5S (Pt. ••• Area)
43.41
36.89
35.05
44.61
40.57
43.23
40.81
39.62
41.18
35.95
65.34
49.24
50.76
50.95
38.32
51.06
54.46
54.0
58.19
Ambient
Concentration
(ug/m3)
13.2
24.8
26.5
21.7
27.2
30.2
29.7
28.3
27.2
25.8
15.5
29.2
30.4
34.4
31.2
20.0
20.4
30.7
20.8
-------
111-37
We then examined the contribution to ambient concentrations
from all sources (area and point) within Philadelphia and the
surrounding counties considered in the study area. First, we
found that the 26 compounds account for an estimated 3,053 kkg/
year of toxic pollutant emissions within the study area. Point
and area sources outside Philadelphia account for roughly 56 per-
cent of this total, while emissions from sources within the city
account for 44 percent (1,337 kkg/year). Second, we generally
found that the sources located within a 5 km by 5 km receptor
grid square account for almost all of the modeled ambient pollu-
tant concentration within the same receptor square. Thus, sour-
ces in Philadelphia affected receptor populations within the
city, and sources in Pennsylvania and New Jersey primarily
affected receptor populations in their respective areas.
There was, however, one exception to this source-receptor
relationship that required further analysis. We identified a New
Jersey source that appeared to contribute to a portion of the
ambient air pollution concentration in two receptor squares (49
and 50) located within the city (see Table III-9). We found that
for square 49, the point source located in New Jersey contributed
6 percent. For square 50, the point source located in New Jersey
contributed 10.5Npercent. It therefore appeared that control of
emissions to reduce methyl ethyl ketone from this source would
only have a small impact on exposures and risks in these two
squares. We reasoned that any control strategy for this point
source would only slightly reduce the risks to subpopulations in
Philadelphia. As a result, we decided to focus our resources
only on sources located within the city. This decision also
narrowed our list of 26 pollutants to 25.
The next step in our review involved assessing each of the
remaining 25 toxic pollutants and their potential contribution to
exposures. We first looked at sources with emissions greater
than 10 kkg/year. This decision was based on two judgments:
(1) the smaller point sources would probably contribute less to
the total population risk than the larger sources, and (2) small
sources t^feas than 10 kkg/year) would probably not offer the most
cost-effe||£ive control opportunities compared to the larger sour-
ces. Narrbwing our focus to larger sources had several effects:
the number of pollutants in our analysis dropped from 25 to 11;
the number of sources (point and area) decreased; and the total
emissions considered in our analysis dropped from 1,337 kkg/year
to 1,042 kkg/year.-
We then reassessed the control feasibility and health
effects of these 11 compounds. We eliminated formaldehyde from
further consideration in the control-options analysis because
most formaldehyde is formed in the ambient atmosphere as the
result of- photochemical oxidation of hydrocarbons, many of which
are emitted by mobile sources. Controlling the 43 kkg/year of
-------
111-38
formaldehyde released by several point sources would have little
effect in reducing the total exposure to this chemical. Although
we dismissed formaldehyde from further consideration in the
control-options analysis, AMS, as mentioned above, established an
air monitoring program to assess ambient concentrations of for-
maldehyde as part of the project in Phase II. The results from
this particular effort are discussed in Chapter IV.
For several reasons, we also chose not to assess sources
emitting 1,1,1-trichloroethylene and lead. At the time this
study was conducted, there was insufficient evidence that 1,1,1-
trichloroethane was a carcinogen. Today, there is a 1985 NTP
study currently undergoing review that may change this pollut-
ant's status as a noncarcinogen.
Similarly, lead was also not considered a human carcinogen,
although recent limited evidence under review may show otherwise.
In addition, the City was already taking legal alternatives to
require reductions at the one major point source of lead.
Finally, EPA was addressing lead emissions from mobile sources as
part of its national analysis to phase out the lead in gasoline.
We did not feel that a site-specific study with limited resources
could make a meaningful contribution to the national effort. The
eight remaining chemicals accounted for emissions in Philadelphia
of 897.8 kkg/year (see Table III-IO).
Table
III-IO
PHILADELPHIA IE*
EMISSION LOADS
Pollutant
1, 2-OichIoroethane
1, 2-Dlchloropropane
Trichloroa thy lane
Perchloroethylene
Benzene
Carbon Tetrachloride
Chloroform
Methylene Chloride
Total
AMS Emission
Source Inventory
Total
Emissions
kkq/yr
0.2
0
75.7
125.8
14.4
17.1
51.9
240.0
PMladelptua County
Source*1
rot«l
CaiMlone
fchQ/yr
124. 4
144.0
30.7
137.5
90.0
17.1
64.1
240.0
525.1 897.8
^Sources include estimated air emissions from the Northeast and
Pollution Control Plants, and a waate water treatment facility
refinery plant.
Philadelphia County
and New Jersey
Sources
Total
Emissions
kkq/yr
124.4
144.0
105.7
183.7
100.1
17.1
64.6
255.5
995.1
Southwest Water
at a petroleum
-------
111-39
At JPp* point in the analysis, we also gave careful consid-
eration vKSlf2-dichloropropane, but felt that it was too impor-
tant to dismiss at this point in the study. While there was
limited evidence that this pollutant is a carcinogen, a new NTP
study on 1,2-DCP had just been released. (We later used this
study to generate potency estimates, as did EPA's Office of
Drinking Water for its Drinking Water Criteria Document.) Fur-
thermore, we strongly suspected that this chemical was involved
in the intermedia transfers from water to air at the NEWPCP.
There were additional point source releases of 1,2-DCP. For
these reasons, we decided to leave 1,2-dichloropropane in our
analysis, subject to any new information that would indicate
otherwise.
As a final check, we compared our estimated emissions loads
of these eight chemicals from sources in Philadelphia with those
in New Jersey and with those in the AMS inventory. We knew that
our estimated emissions were higher than the City's inventory.
We also considered several additional sources from which AMS did
not have the legal authority to request information. These com-
parisons appear in Table 111-10. -
At the end, of this assessment, we felt confident that the
sources and pollutants on which we continued to focus provided a
firm basis from which to proceed with the Phase II control-
options analysis. We decided, therefore, to address in our
analysis eight chemicals (seven of which, excluding 1,2-DCP, are
listed by Philadelphia's Air Management Regulation VI) and the
sources that released these compounds.
The sources and pollutants that we selected for evaluation
in the control-options analysis are shown in Table III-ll. The
Phase II control-options analysis focused on eight organic com-
pounds (benzene, 1,2-dichloroethane, 1,2-dichloropropane, methy-
lene chloride, perchloroethylene, carbon tetrachloride, trichlo-
roethylene, and chloroform). Figure III-5 shows the general
location of the sources within the city and the multimedia inter-
act iona.|&Qtured in this study.
-------
111-40
Table III-ll
PHILADELPHIA IEMP
PHASE II COST-EFFECTIVENESS ANALYSIS: SOURCES AND POLLUTANTS1 8Y RECEIVING MEDIUM
Source
Refineries A and 8
Pharmaceutical
manufacturer
Industrial dry cleaner
Chemicel manufacturer
Garment manufacturer
Houaahold cabinet
Manufacturer
Induetrial degreaser^
Induatrial dry cleaner5
Organic chemical
manufacturer5
Northeaat Water Pollution
Control Plant
Baxter, Queen Lane, and
SelBont Drinking Water
Treatment PIanta
Dry cleaning
-area source
Decreasing
-are* source
Misc. aolvant usage
-area source
Air
DCE, benzene, gas vapor
Methylene chloride,
chlorofors, carbon
tatrachlorid*
PCE
OCE, DCP, PCE
TCE
Methylene chloride
Methylene chloride
PCE
Benzene
Methylene chloride,
OCE, chloroforei, DCP,
TCE, PCE, benzene (via
volatilization)
PCE
Methylene chloride,
PCE, TCE
Methylene chloride,
perchloroethylen*
Surface Water2
PCE (via NEWPCP)
DCE, DCP, PCE (via
NEWPCP)
Methylene chloride,
OCE, DCP, TCE, PCE,
benzene
Drinking Water
Trace organica (via
NEWPCP discharges to
Delaware)
Methylene chloride,
DCE, chloroform, DCP,
carbon tet., PCE (via
intake froa Delaware
and chloraminatioo of
finished water)
^Additional pollutants were considered but were omitted fro« further consideration due to negligible
release*, if any.
^Several of these sources have wastewater discharges, but were not included in the analysis because
of negligible impact.
'These sources were eliminated from the analysis in Phase II. The industrial degreeaer and dry
cleaner ceaaed operations; the organic chemical manufacturer hed very low ambient air releases.
DCE i 1,2-dichloroethane
DCP s 1,2-dichloropropene
TCE a trichloroethylene
PCE s perchloroethylene
-------
111-41
I
I
I
I
a
i
i
i
i
i
i
I
1
'it
-------
111-42
CONCLUSIONS FROM THE PHASE I
SCREENING ACTIVITIES
By synthesizing the various screening results, we selected
11 issues, from the original list of 39, to analyze in Phase II.
We decided to study each topic either to provide more exposure
data through monitoring or to assess cost-effective control
strategies. We emphasize that not all study topics had the same
objectives. We also note that we could not analyze each issue
with the same degree of technical rigor. The issues meriting
further analysis were:
• Point sources of chlorinated solvents
• Area sources of chlorinated solvents
• Refinery, pipeline, and terminal emissions of benzene
• Baxter drinking water quality
• Queen Lane and Belmont drinking water quality
• Discharges of 1,2-dichloroethane and 1,2-dichloropro-
pane to air and water
• Chloroform in the ambient air
• Benzene releases
• Formaldehyde concentrations in the ambient air from
photochemical oxidation
• Hazardous constituents in used oil burned in boilers
and furnaces
• Air emissions of volatile organic compounds from land-
fills (in New Jersey)
We grouped the first seven issues into the control-options
analysis that used a common analytical framework to evaluate
cost-effective controls for the system of interrelated sources,
pollutants, and exposure pathways. Our control-options analysis
was designed to provide local policymakers with information on
least-cost control, strategies for reducing different specific
levels of health risk. The remaining four issues were treated
separately. In Table 111-12 we list the issues as explained
above, identify the Phase II analytical activities, and present
the objectives of each.
-------
111-43
Study Topis*
Table 111-12
PHILADELPHIA 1EMP
OVERVIEW OF PHASE II STUDY TOPICS AM) ACTIVITIES
Analytical ActivitiM
Objectivee
ftiak And Control-Optione Analyaia
—Identified through screening of
available data
--Addressed 7 of the 11 initial
screening issue papers
- Benzene emissions
- Arsa sources of solvents
- Refinery, pipeline, and
terminal emisaione
- Baxtar drinking water
- Quaan Lana/Belaont drinking
water
- Multimedia dischargee of
1,2-OCC and 1,2-OCP
- Haloforna (chloroform)
—Monitor ambient air
—Monitor ambient wstsr
—Validata emission eatimatea
—Recalculate exposures and riaka
uaing air dispersion models
—Idantify faaaibla control*
—Caleulata control coat* and
removal a/flcianciaa
—Oavalop nodal to avaluata and
rank coat-affaetiva control
option atratagiaa
->0ataraina baaalina axpoauraa
—Idantify which aourcaa, pollu-
tant*, and exposure pathwaya
contributa aoat ligniflcantly to
aatiMtad Hunan health riak
—Oavalop coat-affactiva atratagia*
for reducing riaka to hua*n
health:
- Aggregate axcaaa cancer
incidence
- Uafca to the Moat Exposed
Individual
—Cxaeine noncarcinoganic risk*
—Provide reeulta in uaeful format —Identify limitationa of analysis
for review by decision acker*
Msnitoring
1. Benzene Eaiaaiono (specifically
fro*) gaaolina earfceting)
—Identified by Philadelphia aa
a topic of interest
2. Fonsaldehyde Haliaaia to the
Aabient Air
—Identified by Philadelphia aa
a topic of interest
—Short-tsrm benzene ambient air —Determine significance of
monitoring at selected intersec- observed concentrstions
tions with and without ssrvics
stations - Philadelphia air guideline*
- Cancer riak
—Aabient air annitoring
3. Coabuatlon
—Identifi
parti
intareat
OU
—Sampling at point* of diatribu-
tion and uae
—Analyaia of samples for metals
and organica
—Determine significance of
observed concentrations
- Philadelphia air guideline*
- Cancer risk
—Compare observed used oil concen-
trationa with EPA fuel specifica-
tion
4. Air Eaiaaione frosj Landfills'
(focusing specifically on New
Jersey landfill*)
—Identified by New Jaraey and —Snort-tars; aonitoring at a select —Identify coapounda and obaarved
EPA as a topic of intareat
—Satiafiad EPA intareat in teat-
ing the applicability of a
mobile monitoring *y*tea for
uaa in future geographic
studiaa
number of sanitary and hazardoua
waate landfills in New Jersey
(but within geographic boundary
of project)
concentrations
—Determine significance of
measured concentrations
-------
111-44
SETTING OBJECTIVES FOR PHASE II
In this section we first discuss the four issues for which
monitoring was identified as the primary Phase II activitv (Chap-
ter IV presents more detailed information and results for each of
the four monitoring activities). Then we briefly discuss the
control-options analysis (monitoring activities and results sup-
porting the control-options analysis are described in
Chapter V).
Monitoring Topics for Phase II
Unlike the control-options topics, some monitoring topics
were identified by the study participants simply to better under-
stand the potential exposure levels. As a result, we jointly
agreed to initiate monitoring programs to obtain ambient data in
these areas and to identify the need for further research on the
basis of our crude approximations of exposure and risk.
Benzene Emissions
Our risk estimation in Phase I suggested that benzene was
one of the most significant toxic air pollutants in the study
area. In addition, AMS expressed interest in assessing benzene
ambient air concentrations in the city. Both EPA and AMS were
especially interested in exploring the ambient contributions
resulting from gas station and gasoline motor vehicle releases.
The Phase II activities focused on monitoring benzene ambi-
ent levels at traffic intersections within the city and specif-
ically around service stations. AMS agreed to perform the
monitoring and chemical analyses.
Formaldehyde Releases to the Ambient Air
Formaldehyde is a fairly ubiquitous pollutant released by
combustion, processing of petroleum and organic chemicals, and
the photochemical reactions of other compounds. While formalde-
hyde releases from the major point sources in Philadelphia are
relatively insignificant, photo-oxidation of atmospheric hydro-
carbons (a process- not considered in the screen) may be .the
largest source of formaldehyde in the area, occurring during
daylight hours. Because of formaldehyde's carcinogenic and toxic
effects—at higher doses it is a strong respiratory irritant—AMS
was concerned that ambient concentrations were exceeding guide-
lines for formaldehyde. The ambient guideline concentration for
formaldehyde is one of the lowest AMS has established. AMS
agreed to perform the monitoring and chemical analyses.
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
1
I
I
I
I
I
111-45
Combostion of Used Oil
Prom studies undertaken by EPA and others, it appeared that
used oil may contain high concentrations of lead, other trace
metals, organic compounds (such as trichlorethylene, perchloro-
ethylene, and toluene), and PCBs. From our very general work in
this area and the experience of the project participants, there
was reason to suspect that there might be significant maximum
individual risks in certain areas of Philadelphia. Also, the
City expressed concern that the combustion of contaminated fuel
oil could present a significant human health hazard.
As a result, the four participants, Philadelphia, Pennsyl-
vania, New Jersey, and EPA agreed jointly to choose distributors
and users of fuel oil in the study area for sampling in Phase II.
The information generated from this work would assist the state,
local governments, and EPA in further defining the extent and
potential human health hazards of used oil combustion in the
Philadelphia study area.
Air Emissions from Landfills
The final topic we identified for limited analysis in
Phase II was air emissions from landfills. As we discussed
earlier, there are neither hazardous nor sanitary landflls in
Philadelphia, and the ones present in New Jersey did not appear
to pose a threat to Philadelphia residents. However, EPA and New
Jersey were generally interested in the topic of air emissions
from landfills because these releases could be significant.
Furthermore, EPA was interested in experimenting with a mobile
monitoring system—the Remote Optical Sensing of Emissions (ROSE)
system—for use at landfills in future IBMP studies. We decided
to test the ROSE system on both hazardous and sanitary landfills
in New Jersey, choosing the exact monitoring sites later in
Phase II.
ControlHifctipn Analysis
Topics fjjpr Phase n
While further data collection for certain topics was a rea-
sonable Phase II pursuit, it became clear to us that the core of
the Phase II activities should be the development of control
options. We developed an analytical scheme for simultaneously
evaluating cost-effective controls for a system of interrelated
sources, pollutants, and exposure pathways that encompassed seven
of the 11 selected study topics (see Table 111-12).
-------
111-46
Initially, 15 industry sources were included in the control-
options analysis as traditional point-source air emitters. We
also included air emissions from a sewage treatment plant as a
nontraditional source of air pollution. We considered air area-
source categories, specifically dry cleaners, degreasers, and
other miscellaneous solvent users, because they are the major
consumers, and therefore emitters* of solvents. Finally, we
included the three drinking water plants. Additional details are
presented in Chapter VI, which describes the cost-effectiveness
analysis and findings.
There were two reasons for evaluating controls for these
sources together rather than separately. First we expected to
find important interrelationships between several sources and the
resulting exposures, most notably the discharges sent by indus-
trial facilities to the NEWPCP and the resulting volatilization
and exposure (through air) from the NEWPCP. We could not deter-
mine an optimal mix of controls to achieve a fixed level of human
health risk reduction for the lowest cost without evaluating the
contributing sources and their relative costs of control
together.
Second, through our work in Phase I we recognized that urban
exposures are complex and cumulative. The population in
Philadelphia can be exposed through both air and drinking water
to mixtures of pollutants released from numerous sources. As a
result, we can only set control priorities for reducing cumula-
tive exposure in the study area by weighing the risks and costs
of control for multiple sources and routes of exposure simultane-
ously.
In Chapters IV and V, we discuss monitoring programs
designed to refine our exposure assessments initially made during
Phase I of the project. In Chapter VI, we explain our cost-
effectiveness analysis and how we use our exposure and risk
assessments as part of the control-options analysis.
LIMITATIONS TO PHASE I SCREENING
The Philadelphia study was our first major IEMP. The
Phase I screening activities—while not perfect—were successful
in setting a manageable yet interesting subset of priority topics
for analysis in Phase II. However, certain limitations to the
Phase I findings need to be fully understood if the results pre-
sented in the following chapters are to be seen in the proper
perspective.
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
111-47
Noncancer health effects were not adequately addressed
in the Phase I priority setting process. As discussed,
the Philadelphia IEMP emphasized carcinogenic sub-
stances. While it seemed a reasonable determination at
the time, and one supported by some of our Phase II
findings, future lEMPs should consider noncancer
effects more thoroughly.
Recent changes in the dose-response data for several
carcinogens have occurred. Since completing Phase I,
there have been changes in the dose-response informa-
tion for several compounds, most notably lead. We do
not believe this new information would have changed the
pollutant and source focus in Phase II.
The ability to analyze many sources of toxic releases
was constrained by the available data.The lack of
information on numerous topics was a large impediment
to the Phase I screening. With all IEMP studies, how-
ever, the choice of topics will depend heavily on the
quality and quantity of available environmental data
when the analysis is completed.
Phase I may not have evaluated metal emissions sources
sufficiently. The Phase I priority-setting process did
not identify any metals for further evaluation in
Phase II. Using all available source and emissions
data, it did not appear that metals contributed sig-
nificantly to human health risks. From recent national
studies, we know that trace uetals and products of
incomplete combustion are still pervasive in the envi-
ronment—although their concentrations have been
reduced over the years--and that these compounds are
associated with numerous chronic effects. Future lEMPs
may want to focus on better characterizing sources of
metals, such as boilers and power plants, as well as
performing ambient air monitoring for metals as part of
Phase II.
The IEMP did not consider ecological effects. While
human health risks are important, several major envi-
ronmental statutes are also designed to protect eco-
logical integrity. Consideration of ecological effects
should be an important element of future geographic
strategies to reduce pollutant releases to the environ-
ment.
-------
111-48
The IEMP did not consider conventional pollutants. We
chose not to analyze conventional pollutants (such as
ozone and particulates in air, and biological oxygen
demand and oil and grease in water)/ because the
Philadelphia IEMP was specifically designed to evaluate
and address the potential human health hazards associ-
ated with exposures to toxic chemicals. In addition,
we felt that we could make a more significant contribu-
tion by concentrating on toxic chemicals since they are
not as well studied or regulated.
Future geographic projects that consider ecological as
well as human health efforts will need to evaluate both
.conventional and toxic compounds. In addition, the
current availability of benefits data for conventional
pollutants, especially in air, allows us to identify
possible research and control priorities using a
benefit-cost, in addition to a risk-based cost-effec-
tiveness, analytical framework. In some situations, we
suspect different priorities would result from use of
these different approaches. This is an important fac-
tor to be considered by the decision maker in deter-
mining the need for and direction of additional control
requirements.
The IEMP did not attempt to estimate the exposure to
indoor air pollution. Recent studies have shown the
relative significance of indoor versus ambient expo-
sures. At the time we completed Phase I, the issue of
indoor air was not as widely documented or as highly
visible. Even if it had been, we could not have
addressed these routes of exposure due to resource
constraints and analytical limitations. Furthermore,
indoor air exposures are outside the traditional, and
possibly the regulatory and statuatory, purview of EPA
and state and local governments.
-------
I
1
-------
I
I
I
I
I
I
1
I
I
I
I
I
I
I
I
I
I
I
1
V. CONTROL-OPTIONS ANALYSIS: AMBIENT AIR AND
WATER MONITORING PROGRAMS
This chapter reports on two additional monitoring programs
designed for Phase II of the Philadelphia IEMP. These programs
gathered information that was used to confirm, adjust, or refine
data already established in Phase I and to support the control-
options analysis, which was an assessment of the cost-effective-
ness of control strategies for the set of sources, pollutants,
and exposure pathways selected by the priority-setting process of
Phase I. This chapter provides the interested reader with a
relatively technical summary of the air and water monitoring
programs conducted for these purposes. Chapter VI provides a
more detailed description of the analytical structure, the input
data (including information from the air and water monitoring
programs), and the results of the cost-effectiveness analysis.
Figure V-l presents a flow chart that outlines the major
steps of the cost-effectiveness analysis of control options and
indicates the points at which the monitoring programs provided
information used\ in the analysis. Following the left-hand flow,
we find that, for the point and area sources, the control-options
analysis required an assessment of pollutants released into the
media, the degree of exposure by pathway, and the consequent
human health effects. These relationships were modeled so they
could serve as a baseline (i.e., represent conditions under
current control) for evaluating the effectiveness of additional
control alternatives. The analysis of the costs and effective-
ness of these control options is represented by the right-hand
flow in Figure V-l.
Data from the monitoring programs contributed directly to
our cost-effectiveness analysis. As the figure shows, the air
monitoring program provided feedback on emissions estimates,
which enabled improved estimates of exposure. Because this was
the first time that EPA had monitored air toxics on a citywide
scale, this program also played an important role in our methods
development and thus represented a major research initiative. A
primary goal of the water monitoring program was to help identify
feasible control options for improving the quality of raw water
at the Baxter plant, such as changing the withdrawal schedule or
shifting the water- intake point within the river. The water
monitoring program was also designed to confirm that wastewater
treatment at the NEWPCP released volatile organic compounds
(VOCs).
-------
V-2
Figure V-1
Philadelphia IEMP
Cost-Effectiveness Analysis of Control Options1
Input from Air and Water Monitoring Programs
Sources
(1964 Currant Control Levels)
• Point
• Area
Pollutant Releases
e Air
• Surface water
Exposure Pathways
• Inhalation (dispersion
modeling)
• Ingestion (drinking
water)
Human Health Effects
(Baseline Cancer Risks)
Aggregate incidence
ME) risk
• Air monitoring
• Water monitoring
Water
monitoring
• Air monitoring
• Water monitoring
Computer Mod*
(Mixed Integer Program)
Cost-Effective
Control Strategies
• Cost of the reduction
in aggregate cancer
incidence
• Cost of the reduction
in MEI risk
Feasible Incremental
Control Options
Air
Water
Control Costs
• Air
• Water
Control Effectiveness
(Percent Removal Efficiency)
Air
Water
1
Cost-effectiveness Is defined as the cost of control per unit of health risk (cancer) reduced.
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
™
V-3
AMBIENT AIR MONITORING PROGRAM
This section describes the ambient air monitoring program in
detail. We provide an overview of the sampling plan, including a
discussion of the Quality Assurance/Quality Control (QA/QC)
specifications, and present the average pollutant concentrations
detected at each of the monitoring sites. We explain how the
information from our monitoring program was used to evaluate the
existing emissions inventory and model performance for use in the
control options analysis. Finally, we conclude with a section on
the implications of the measured pollutant concentrations for
exposure and risk—separate from the baseline considered in the
control-options analysis. This last section is important for
providing perspective on what we could and could not capture in
the control-options analysis.
Overview of the Sampling Plan
Placement of Air Monitors
From the Phase I screening activities and use of the
Climatological Dispersion Model, we found that most of the
sources associated with the control-options analysis were located
in a highly industrialized corridor, adjacent to the Delaware
River, 15 kilometers long and 5 kilometers wide. As a result, we
designed a "source-oriented" network that placed nine of the ten
monitors in this area.1 Another monitor was placed further away
from the industrial area where we expected more uniform concen-
trations.
We favored a source- rather than receptor-oriented approach
to monitoring because it would provide a way to check the quanti-
tative estimates of pollutant releases compiled in the emissions
inventory and would be helpful in identifying and locating
sources not in the emissions inventory. Using a source-oriented
approach, we could develop some of the key inputs for assessing
cost-effective control strategies, such as maximum ambient con-
centrations (used to calculate MEI risks) and individual source
contributions to exposure. In fact, the only practical way to
generate this information, relating pollutant releases with
I
I
iMore detailed information on the sampling plan can be found in:
Versar, Inc., Quality Assurance Plan for Monitoring and
Analytical Activities to Evaluate Selected Air Pollutants in
Philadelphia, Pennsylvania, U.S. EPA, October 19, 1983, revised
November 11, 1983. This report is part of the background docu-
ments and is available on request.
-------
V-4
ambient concentrations in a cost-effectiveness framework, is
through the use of dispersion modeling.
The monitoring program employed ten monitors and one meteor-
ological station. We decided on the number of monitors by
balancing our study needs with the available funds for this
activity. While there are no guidelines specifying the number of
monitors needed to carry out a citywide monitoring program, we
decided that ten sites provided reasonable coverage to assess the
general adequacy of the emissions data. The data would also be
sufficient to permit regression analyses to assess the correla-
tion between modeled and observed concentrations. We selected
locations that evenly distributed nine of the ten monitors
throughout the central industrial corridor. We placed the tenth
monitor in South Philadelphia near the oil refineries.
Figure V-2 shows the placement of these monitors. Where pos-
sible, we selected monitoring sites that were being used or had
been used for air quality sampling by AMS. We also considered
several additional criteria:
• Monitors should not be unduly affected by local emis-
sions, e.g., heavily traveled roadways or commercial/
industrial sources.
• Monitors should be in well-exposed and open areas.
• Monitors should take in air samples at heights of two
to six meters.
• Monitors should be secure against vandalism or tamper-
ing.
• Sites should currently have electric power to operate
the monitoring equipment.
Although not all of the monitoring locations met all of the above
criteria, we felt that the sites we selected would best represent
the industrial corridor and meet the project objectives.
To help interpret the air quality data, we also established
a 20-meter: meteorological tower equipped with instruments to
measure the horizontal and vertical wind speed, horizontal wind
direction, and turbulence of the wind. We located the tower at
the Roadway Express, Inc. truck terminal in an industrial section
of the city near the intersection of Church and Pearce Streets.
The meteorological station was operated by AMS.
-------
I
V-5
Figure V-2
Philadelphia IEMP
Ambient Air Monitoring Locations
PHILADELPHIA
COUNTY
1. Naval Hospital
2. Goodyear
3. Philadelphia Fire Department #16
4. Philadelphia Fire Department #7
5. St. John's Church
6. Lardner's Point
7. Philadelphia Fire Department #71
8. Philadelphia Fire Department #36
in' Dh?Hr PrKnki?S Water Treatment Plant
10. Ph.iadeiphia (Northeast) Airport
-------
V-6
Sampling Procedures
We designed a program to collect 24-hour integrated samples
at each monitoring location approximately every three days over a
four-month period. We were limited to a four-month period
because of resource constraints. Continuous monitoring at ten
sites for a one-year period (to record seasonal variations) would
have been prohibitively expensive. We also felt that this period
would provide a sufficient database for evaluating model perform-
ance.
We began the sampling on December 11, 1983, and continued to
sample through March 31, 1984. Soon after initiating the moni-
toring program, we experienced contamination problems while
transporting the first month's samples to the laboratory for
analysis. To remedy this problem, we added an additional 21 days
of monitoring. A total of 380 ambient air samples were col-
lected, of which a total of 300 sample sets were analyzed and
included in the data.
We monitored for 12 volatile organic compounds. Eight of
these 12 chemicals accounted for the majority of the toxic chem-
ical releases in the Philadelphia area identified in Phase I and
were included in our control-options analysis. We also included
1,1,1-trichloroethane because we were interested in considering
it for future geographic studies. As noted below, we added ethyl
benzene, toluene, and xylene because these compounds were rou-
tinely observed and were amenable to the laboratory methods used
for analysis. Thus the 12 pollutants were:
• 1,2-Dichloroethane (1,2-DCE)
/ • 1,2-Dichloropropane (1,2-DCP)
• Methylene chloride
i/ • Perchloroethylene
• 1,1,1-Trichloroethane
/ • Trichloroethylene
,, • Benzene
, • Chloroform
/ • Carbon tetrachloride
-------
V-7
/ • Ethyl benzene
- • Toluene
• Xylene (meta, para, and total)
We should emphasize that numerous other pollutants may be found
in the ambient air; the chemicals listed above were selected on
the basis of the limited findings from Phase I and the capabili-
ties of Tenax.
We modified a method for collecting gaseous organic com-
pounds originally developed for the U.S. EPA by Research Triangle
Institute and Northrop Services.2 our method involved collecting
the volatile organic compounds on Tenax GC at a low flow rate
(generally less than 100 ml/minute). At the time, EPA had no
standard ambient air sampling methods for these compounds. Tenax
appeared at the time to be the best absorption medium for our
purposes. It is important to note, however, that the reliability
of Tenax has recently been questioned. EPA is currently
researching alternative methods for sampling.
Approximately 40 percent of the samples were analyzed using
gas chromatography/mass spectroscopy (GC/MS). Using GC/MS, we
could add a nontarget pollutant to the list of compounds for
future analysis in the program, provided that this substance was
routinely observed in initial scans. After a preliminary review
of several sets of GC/MS analytical results, we found sharp peaks
for toluene, ethyl benzene, and xylene (meta, para, and total).
Since these compounds were easily quantified and of interest to
us, we decided to add these substances to the target pollutant
list. These compounds are associated with automobile emissions.
The remaining samples were analyzed using gas chromatography
(GC) with a Hall detector for chlorinated compounds and a photo-
ionization detector for aromatic pollutants. Sample analyses
were distributed between GC/MS (40 percent) and GC (60 percent)
techniques to reduce analytical costs. The laboratory work was
conducted under contract to EPA and performed by Versar, Inc. of
Springfield, Virginia.
Quality Assurance/Quality Control
Before monitoring began, a formal Quality Assurance Plan was
prepared and submitted to both the Geographic Branch of EPA and
2K.J. Krost, E.D. Pellizzari, S.G. Walbur, and S.A. Hubbard,
"Collection and Analysis of Hazardous Organic Emissions,"
Analytical Chemistry, Vol. 54, 981, pp. 810-817.
-------
V-8
the EPA Region III Quality Assurance Officer.3 Other reviewers »
included personnel from the City of Philadelphia and selected
individuals at EPA. The plan established a variety of quality
control analyses to be performed throughout the program, includ-
ing:
• Routine analysis of standards
• Field blanks for both GC and GC/MS with every sampling
event
• Spiking of every tube with an internal standard
(4-bromofluorobenzene) immediately before analysis
• Spiking of every tube with chlorobenzene before
sampling and analysis of chlorobenzene in every tube in
order to detect serious breakthrough of less volatile
compounds and major analytical and sampling errors
• Duplicate analysis and spiked tube analyses for every
sampling event
• Analysis of a limited number of samples collected at
two different flow rates from every sampling event
In addition, the contractor developed a chain-of -custody
program to prevent inadvertent contamination, misidentif ication,
and tampering of the samples. The program was designed to pro-
vide a good record of sample custody, from preparation of the
sampling cartridges in the field to the point of analysis at
Versar's laboratories. Finally, the contractor calibrated the
air sampling equipment following the guidelines detailed in EPA's
Quality Assurance Handbook for Air Pollution Measurement Systems,
Volume 2 .
Results from the Ambient
Air Monitoring Program
The average ambient concentrations measured for the target
pollutants at each monitoring location are presented in
^Versar, Inc., Quality Assurance Plan for Monitoring and
Analytical Activities to Evaluate Selected Air Pollutants in
Philadelphia, Pennsylvania, U.S. EPA, October 19, 1983, revised
November 11, 1983.
-------
V-9
Table V-l.4 Because we could not gain access to Site 6,
Lardner's Point, on numerous occasions (the security guard never
arrived), the data from that site were limited and consequently
excluded. In addition, we have not reported the results for
methylene chloride and 1,1,1-trichloroethane at any of the sites
because of limitations encountered in the laboratory methods for
analyzing these pollutants.
Table V-l exhibits the sampling data in two ways; the full
data set (December 11, 1983, through March 31, 1984), and the
data set excluding the first 12 sampling periods (February 6,
1984, through March 31, 1984). As noted above, sample contamina-
tion was traced to the airline transportation of the samples to
the laboratory. The caps on the containers were not sufficiently
tight to prevent inflow of air during depressurization in flight.
Beginning with the February 6, 1984, sample, we employed differ-
ent procedures and containers and extended the sampling period.
Subsequent analysis of the field blanks showed a significant drop
in contamination to within tolerance levels. With these improve-
ments in the quality control, we have more confidence in the data
for the shortened sampling period than for the full period. This
table also presents a comparison with AMS's Ambient Air Guide-
lines. These guidelines are based primarily on occupational
threshold limit^values (TLVs) and safety factors.
We should note that average ambient benzene levels appear
somewhat low relative to other major industrial cities. For
comparison, we present benzene concentrations in other major
cities below:5
Mean Benzene Concentration
City (ug/m3)
Los Angeles 19.2
Phoenix 15.1
Oakland 4.9
Houston 18.4
St. Louis 4*5
Denver 14.0
Riverside, NY 12.6
Staten Island 13.4
Pittsburgh 15.9
Chicago 8.2
4A more detailed discussion of the results from the monitoring
program can be found in: versar, Inc., Draft Data Report for
Monitoring and Analytical Activities to Determine Ambient Air
Concentrations of Selected Toxic Pollutants in Philadelphia,
Pennsylvania, U.S. EPA, May 18, 1984.
5H. B. Singh, L. J. Salas, R. Stiles, and H. Shigeishi, Measure-
ment of Hazardous Organic Chemicals in the Ambient Atmosphere,
Project Summary, U.S. EPA, Office of Research and Development,
RTP, NC (EPA 600/S3-83-002), 1983.
-------
V-10
Table V-1
PHILADELPHIA IEMP
AVERAGE MEASURED CONCENTRATIONS FOR AIR MONITORING SITES1
(ug/m3)
Compound
12/11/83-03/31/842
Chloroform
1,2-Oichloroethane
Carbon Tetrachloride
Benzene
Tnchl oroethylene
1,2-Oichloropropane
Toluene
Perchloroethylene
Ethyl Benzene
Xylene
02/06/84-03/31/84
Chloroform
1,2-Oichloroethane
Carbon Tetrachloride
Benzene
Trichloroethylene
1,2-Dichloropropane
Toluene
Perchloroethylene
Ethyl Benzene
Xylene
Monitoring Sites
4578
AMS
Guidelines3
^2 ug/m3 ppb
3.4
1.0
2.8
7.1
1.5
0.2
16.9
4.1
6.0
26.4
0.1
0.1
1.4
4.6
0.5
0.1
4.9
2.8
3.0
7.9
1.4
0.3
2.6
7.9
1.3
0.5
15.8
3.6
7.0
14.0
0.5
0.4
2.7
7.7
1.3
0.6
12.4
3.7
7.2
12.7
2.1
0.4
2.7
7.4
2.0
2.0
14.0
5.3
6.2
20.1
0.9
0.5
3.0
8.1
1.8
2.8
16.1
6.2
5.8
21.1
3.8
0.2
2.8
6.0
8.3
0.9
13.6
10.2
11.3
25.3
0.4
0.3
2.5
6.1
6.8
0.7
12.2
10.5
7.8
17.8
2.8
3.4
2.1
6.1
1.4
6.1
18.1
6.1
7.2
13.7
0.3
1.1
1.4
5.5
0.9
5.6
12.2
6.9
6.5
15.4
4.0
0.3
1.3
9.2
0.8
0.5
24.4
3.6
5.2
18.1
0.1
0.2
1.0
7.7
0.5
0.2
16.7
3.3
2.9
.12.1
2.0
0.4
1.9
6.7
1.8
0.8
16.4
5.6
7.7
20.2
0.1
0.3
1.9
6.1
1.0
0.3
14.5
4.4
3.2
10.9
5.7
0.5
1.1
3.1
0.9
0.1
5.4
5.1
3.0
11.5
0.5
0.2
1.4
4.0
0.8
0.1
8.1
2.2
2.7
8.5
2.2
0.6
1.8
3.5
1.2
0.2
9.2
2.9
3.8
10.8
0.1
0.2
1.0
4.3
1.2
0.2
10.7
2.2
3.0
4.3
117
150
75
77
6,445
N/A
N/A
8,133
N/A
N/A
117
150
75
77
6,445
N/A
N/A
8,133
N/A
N/A
24
37
12
24
1,200
N/A
N/A
1,200
N/A
N/A
24
37
12
24
1,200
N/A
N/A
1,200
N/A
N/A
1The data from Site 6 were excluded because of poor recovery.
^This data set includes the values from the first 12 sampling events which had high field blanks.
3The Philadelphia. Air Management Services Ambient Air Guidelines are presented for comparison
with the measured data (N/A s not available).
Source: Veraar, Inc., Draft Data Report for Monitoring and Analytical Activities to Determine
Ambient Air Concentrations of Selected Toxic Pollutants in Philadelphia, Pennsylvania,
U.S. EPA, May 18, 1984.
Versar, Inc., Evaluation of the Performance of the Dispersion Model SHORTZ for
Predicting Concentrations of Air Toxics in the U.S. EPA'a Philadelphia Geographic Study,
February 8, 1985.
-------
V-ll
Evaluation of Model Performance
and the Emissions Inventory
As discussed above, we had several objectives in gathering
ambient air data. By comparing measured and modeled values at
each of the monitoring sites, the quality of the emissions data
and model results could be assessed. In particular, we would be
able to discover anomalies in the inventory (for example, missing
sources) and whether we should use different formulations of the
dispersion model to account for local meterological conditions
and other parameters.
Analytical Approach
We compared the observed and modeled values using several
techniques. First, we evaluated the degree and significance of
model bias for each pollutant. We defined model bias as the
average of the discrepancy between the observed and modeled
pollutant level across all sites and over the entire sampling
period.
Second, wev evaluated the model's precision, i.e., the amount
of scatter in the average discrepancies at each site. In par-
ticular, we measured precision by evaluating the level of
"noise," or the error scatter around the mean error. Finally,
using a simple linear regression analysis, we attempted to assess
the accuracy of the dispersion model in predicting the observed
value. The basic regression equation was:
(Observed » a + BQ Model + error)
where a is the intercept and BQ represents the slope of the
regression line. We tested each of the estimated correlation
coefficients at the 80 percent confidence limit; this was a sub-
jective determination.
We performed the analysis using four different specifica-
tions of EPA1 s SHORTZ dispersion model (used for estimating
short-tern* transport) as shown in Table V-2. The key parameters
we varied" were stability, mixing heights, and emissions vari-
ability. It is important to note that the evaluation of model
performance was limited to varying several input parameters; the
analysis was not .designed as an in-depth assessment or validation
of the SHORTZ model.
-------
V-12
Table V-2
PHILADELPHIA IEMP
SPECIFICATIONS OF THE SHORTZ MODEL
FOR MODEL PERFORMANCE EVALUATION
Model Basis for Dispersion
1 Default Turner
2 Turbulent intensity
3 Default Turner
4 Turbulent intensity
Mixing Height Emissions
Holzworth Climatologies! Constant
Daily (150m minimum) Constant
Daily (150m minimum) Constant
Daily (100m minimum) Variable
Source: David A. Sullivan, Evaulation of the Performance of the
Dispersion Model SHORTZ for Predicting Concentrations of
Air Toxics in the U.S. EPA's Philadelphia Geographic Study.
Versar, Inc. for U.S. EPA, January 1985.
We partitioned the data from the monitoring program in
several ways and compared them to the results from the four
different formulations of the SHORTZ model. The results were
used to assess the emissions inventory and to select the best
model formulation for this application. We evaluated the data
for both the entire data set and for the sampling period from
February 6, 1984, through March 31, 1984, and eliminated the set
of samples that were contaminated during shipment to the labora-
tory. We also partitioned the data by wind direction, precipita-
tion, mixing height, and weekend versus weekday.
General Findings from the
Model Performance Evaluation
For all pollutants, each dispersion model formulation pre-
dicted ambient concentrations lower than monitored ambient
measurements. We found good correlations between observed and
modeled data for several compounds, specifically 1,2-dichloropro-
pane, 1,2-dichloroethane, and perchloroethylene (see Table V-3).
Figures V-3, V-4, and V-5 present examples of daily compari-
sons of predicted versus observed concentrations based on Site 7
and Model 2. These figures clearly show a better model perform-
ance for 1,2-dichloropropane, 1,2-dichloroethane, and perchloro-
ethylene than for other compounds. The results from the other
three dispersion model specifications agreed with Model 2 predic-
tions very closely for all pollutants. The different model
formulations did not affect model prediction significantly.
-------
V-13
Table V-3
PHILADELPHIA IEMP
SUMMARY OF MODELED AND MEASURED CONCENTRATIONS
Observed Modeled Concentrations (ug/n3) Least Squares Correlation Coefficient
Compound
12/11/83-03/31/84
Chloroform
1, 2-Oichloroethane
Carbon Tetrachloride
Benzene*
Trichloroethylene
l,2-0ichloroprop«ne
Toluene
Perchloroethylene
Ethyl Benzene
Xylene
02/06/84-03/31/84
Chloroform
1 , 2-Dichloroethane
Carbon Tetrachloride
Trichlorethylene
Benzene*
1 , 2-Oichloropropane
Toluene
Perchloroethylene
Ethyl Benzene
Xylene
^Benzene correlations
at Site 7, with winds
Site 7 increase to 0.
(uq/m3) Model
3.0
0.8
2.1
6.3
2.1
1.2
14.9
5.2
6.4
17.8
0.3
0.4
1.8
1.6
6.0
1.2
12.0
4.7
4.7
12.3
Mere substantially
0.2
0.6
0.1
1.8
0.8
0.8
6.3
3.5
0.4
1.2
0.2
0.5
0.1
0.7
1.7
0.7
5.7
3.1
0.4
1.0
1 Model 2
0.3
0.4
0.1
2.6
1.1
0.6
9.9
4.0
0.5
3.1
0.2
0.4
0.1
1.0
2.3
0.5
8.B
3.5
0.4
2.4
Model 3
0.3
0.7
0.1
2.9
1.3
0.9
10.8
4.9
0.5
3.2
0.2
0.6
0.1
1.1
2.6
0.8
9.6
4.3
0.5
2.5
affected by the systematic
out of the southwest.
83 for 12/11/83 through
Sources -David A. Sullivan. Evaluation
Using Model
03/31/84 and
of the Performance
Model 4 Model 1
0.3
0.4
0.1
2.4
1.0
0.5
9.0
3.4
0.4
3.1
0.2
0.3
0.1
0.8
2.0
0.9
7.6
2.8
0.4
2.3
failure
2 as an indicator,
0.69 for
02/06/84
of the Dispersion
-0.49
0.88
0.28
0.60
0.00
0.94
0.35
0.87
0.33
0.56
0.27
0.73
0.46
0.05
0.61
0.82
0.62
0.91
0.54
0.41
of the model
Model 2
-0.49
0.92
0.28
0.62
0.02
0.99
0.53
0.80
0.38
0.51
0.27
0.89
0.47
0.00
0.58
0.95
0.76
0.88
0.64
0.23
to predict
Model 3
-0.49
0.88
0.29
0.62
0.00'
0.95
0.49
0.86
0.37
0.28
0.27
0.74
0.47
0.04
0.62
0.83
0.76
0.91
0.59
0.38
Model 4
-0.49
0.89
0.28
0.62
0.03
0.99
0.50
0.68
0.36
0.63
0.27
0.86
0.47
0.00
0.54
0.94
0.80
0.77
0.62
0.25
high concentrations
the correlation coefficients without
through 03/31/84.
Model SHORTZ
. . ., Versar, Inc.
for
U.S. EPA, January 1985.
In contrast, there were poor correlations for chloroform,
trichloroethylene, and carbon tetrachloride. Figures V-6 and V-7
compare daily modeled and observed concentrations for chloroform
and carbon tetrachloride at Site 7. These results indicated
problems with either the emissions inventory, the sampling and
analytical methods, or the inherent limitations in using disper-
sion models to predict ambient concentrations. The long half-
life of carbon-tetrachloride {about 35 years) may have explained
-------
V-14
Figure V-3
Philadelphia IEMP
Daily Comparisons Between Predicted and
Observed Concentrations for 1,2-Dichloropropane
(Monitoring Site 7)
I I I I I I I I I I I T I I
I I I I I I T I I I I I I T
DAY
MODEL- * MONITOR- SQUARE
CONCENTRATION IS IN MICROORAHS X CUBIC METER
Source: Sullivan, David A.'. Evaluation of the Performance of
the Dispersion Model 5HORTZ January 1985, p. xvii.
I
I
I
I
I
-------
c
o
N
C
E
N
T
R
A
T
I
o
N
1.5-
i.a-
0.6-
0.3-
0.0-
V-15
Figure V-4
Philadelphia IEMP
Daily Comparisons Between Predicted and
Observed Concentrations for 1,2-Dichloroethane
(Monitoring Site 7)
i i
010303040506*708*9111314151617181981882384863788893331
DAV
MODEL- * MONITOR- SQUARE
CONCENTRATION IS IN MICROGRAMS ' CUBIC METER
Source: Sullivan, David A.. Evaluation of the Performance of
the Dispersion Model SHORTZ January 1985, p. xviii.
-------
V-16
Figure V-5
Philadelphia IEMP
Daily Comparisons Between Predicted and
Observed Concentrations for Perchloroethylene
(Monitoring Site 7)
15-
C 12-
0
C
E
N
T
R
A
T
I
0
N
9-
6-
3H
d d
i i i i i i r i i i r iriiiiiiii i i i r i i i i
6162^304950607080811131-4151617181531828334262738233031
DAV
nODEL- X nCMITOR- SQUARE
CONCENTRATION 13 IN MICROORAflS / CUBIC METER
Source: Sullivan, David A.. Evaluation of the Performance of
the Dispersion Model SHORT2 January 1985, p. xix.
-------
V-17
Figure V-6
Philadelphia IEMP
Daily Comparisons Between Predicted and
Observed Concentrations for Chloroform
(Monitoring Site 7)
100-
C75~
0
N
C
E
N50-
T
R
A
T
IBS'
0
N
0~
d'\ m
i
i
i
i
DAV
nODEL- » HOHITOR- SQUARE
CONCENTRATION IS IN niCROORAMS X CUBIC METER
Source: Sullivan, David A.. Evaluation of the Per'c^ance of
the Dispersion Model SHORTZ January 1985. p.
-------
V-18
Figure V-7
Philadelphia IEMP
Daily Comparisons Between Predicted and
Observed Concentrations for Carbon Tetrachloride
(Monitoring Site 7)
6-
5-
C
0
N 4-
C
E
N 3-
T
R
A 2-
T
I
o i-
N
1!
1 t
!
i
j
f!
i j
/ \ 1 \ _
• '» / %A
» i : '
a I • \
Q
it
i
^
i
i
H '
/I r, '
t .' • 'VI
1 J\ J \ ( / \ \
l^vC" s°' ti.p'V i
I10DEL- *
CONCENTRATION IS
DAV
nONITOR-
IN niCROGRAflS
SQUARE
x CUBIC METER
Source: Sullivan. David A.. Evaluation of the Performance of
the Dispersion Model SHORTZ January 1985, p. xx.
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
V-19
part of the higher ambient concentrations, but we still investi-
gated the possibility of missing emissions and sources, and
necessary revisions to the emissions inventory.
Finally, the remaining compounds, ethyl benzene, benzene,
toluene, and to some degree, xylene, showed ambient concentra-
tions that were modestly well predicted by the model. Measured
values ranged from three to ten times higher than the predicted
results, suggesting that the emissions data for these compounds
may be underestimated.
We were able to make the following observations on the basis
of partitioning of the data in various ways:
• Monitoring days from February 6, 1984, on generally had
higher correlation coefficients than for the first
month of monitoring (days 1-12). This is not surpris-
ing since field blank contamination was substantially
reduced after monitoring day 12.
• In general, days with measurable precipitation had a
marginally poorer correlation between predicted and
observed values. Given the scope of the monitoring
exercise, however, we did not have sufficient data to
address definitively the importance of precipitation.
• Partitioning the results by wind direction quadrant was
useful in identifying source regions where unreported
or underestimated emissions may occur. In some cases,
quite pronounced and consistent model underestimates
were found to be a function of wind direction, thus
suggesting that improvements could be made in the emis-
sions inventory. We found these partitioning results
to be the most useful in the performance evaluation.
Feedback on the Model Formulation
While the main objective in gathering the ambient air data
and evaluating model performance was to provide feedback on the
emissions inventory, we also designed the overall effort to gen-
erate some information on the adequacy of some fundamental
assumptions used in the dispersion model. As discussed, we
employed four slightly different formulations of the SHORTZ
model, varying only stability, mixing heights, and variable emis-
sions (see Table V-2). In general, we found that Model 2 per-
formed the best, although we did not find a dramatic difference
in the predictive capabilities of all four model formulations.
We also observed the following:
-------
V-20
We found a better correlation between the observed and
modeled values using Models 2 and 4 than between those
values using Models 1 and 3. This is because Models 2
and 4 used the turbulent intensity data collected at
the meteorological station rather than the default
(Turner) values.
There were marginally higher correlation coefficients
when the models used lower mixing heights; however, it
was not obvious that gathering such data, given the
cost, would improve model performance significantly.
Finally, it appeared that more detailed modeling tech-
niques would not necessarily improve predictions of
concentrations in the ambient air.
Changes to the Emissions Inventory
Using our findings from the model performance evaluation, we
began the job of determining where sources were missing from the
emissions inventory and where existing release estimates needed
to be revised. We coordinated this activity with AMS personnel,
who provided not only their time but important insights to this
effort. For some pollutants, we uncovered additional sources and
identified needed changes to the existing release rates in the
inventory. All changes that could be quantified were included in
the baseline of the control-options analysis. For others, we
were unable to explain definitively any of the noted discrepan-
cies between the modeled and measured values. The major findings
regarding the emissions data for each pollutant are described
below:
• 1,2-Dichloropropane and 1,2-Dichloroethane. While the
correlation between the average observed and modeled
ambient concentrations were high, the modeled values
were still systematically low. The major existing
sources were a chemical manufacturer and volatilization
from the NEWPCP. Further analysis revealed that sub-
stantial volatilization was most probably occurring
enroute to the NEWPCP via the sewer lines, and from the
Delaware River. We estimated these emissions and
included them in the model with the result being an
improvement in model performance. Because of these
findings/ we added the two new source categories—river
and sewer volatilization—to the area source components
of the cost-effectiveness analysis.
-------
V-21
• Trichloroethylene. Fairly uniform concentrations were
found throughout the monitoring network, with the
exception of Site 4, which had an average observed con-
centration four to five times greater than any other
site. Follow-up verification to the west of this site
revealed that emissions from a garment manufacturing
plant, included in the control-options analysis, were
underestimated. The initial reported emissions were
4 kkg/year; however, upon further inquiry, total emis-
sions were found to be roughly 25 kkg/year owing to
previously excluded releases from a recovery still.
The emission rate adjustments were made in the AMS
inventory after AMS contacted the plant. Baseline
emissions for the plant were included in the control-
options analysis accordingly. We also explored several
other facilities in the area that were known users of
trichloroethylene, but we were unable to confirm any
changes in release rates from these sources.
• Perchloroethylene. We found reasonably good correla-
tions for PCE, which is primarily emitted by dry
cleaners. However, the model underpredicted substanti-
ally for Sites 8 and 9. The only significant source we
could identify was a prison dry cleaning school, but
the releases were too low to account for the ambient
concentrations. Again, we were unable to locate any
missing PCE sources in the relevant study area.
• Benzene. The model systematically underpredicted ben-
zene concentrations at Site 7. AMS and the project's
technical contractors evaluated the upwind source
region, but could not identify any industrial sources,
nearby roads/ or gas stations. The only possible
source located was a heavily traveled 12-lane highway,
Roosevelt Boulevard. We speculated that under condi-
tions of wind flow parallel to the roadway, benzene was
transported toward Site 7 during the rush hour traffic,
but could not pursue this hypothesis further with the
available information and resources.
• Chloroform and Carbon Tetrachloride. Model performance
greatly underpredicted both compounds. We were unsuc-
cessful in identifying any missing sources. Area
source releases of these compounds may have been partly
responsible. Also, background concentrations for car-
bon tetrachloride, resulting from a long half-life
(35 years), may explain a significant portion of the
observed ambient concentrations. We could only con-
clude that further efforts to characterize these
-------
V-22
emissions would ultimately be needed to resolve this
issue; the scope of the project did not permit any
additional analysis.
Toluene, Ethyl Benzene/ and Xylene. All of these
releases are related to mobile sources. Since neither
mobile sources nor these particular compounds were
considered in the control-options analysis, we did not
devote our resources to an in-depth analysis. The
relatively poor correlation between the modeled and
observed values, however, suggested that additional
work to refine the release estimates for mobile sources
may be warranted in future projects.
Caveats
It is important to emphasize a few caveats that help place
our results from the ambient air monitoring program and the model
performance evaluation in perspective. First, there are problems
using Tenax GC as the absorbent. Although Tenax was considered
the best sampling medium at the time for our purposes, in the
last two years its reliability has been called into question.
EPA is currently exploring an alternative to Tenax.
Second, our use of single-tuba Tenax sampling instead of
a distributive volume approach is currently controversial. Some
recent EPA criticisms of Tenax suggest that scientifically valid
measurements of pollutant concentrations in the ambient air can
be obtained only by using a distributive volume approach, i.e.,
four tubes plus blanks. Other EPA researchers do not support
this view and continue to use single-tube Tenax sampling in field
monitoring activities. We recognized this controversy early on,
but decided that it was better, given our objectives and budget,
to gather more samples using a single-tub* sampling approach than
fewer samples using a different approach. A distributive volume
sampling approach would have quadrupled the costs of our proposed
monitoring program and would have forced us to collect far fewer
samples.
Third, while it is likely that the incompleteness of and
the inaccuracies in the emissions inventory were major contribu-
tions to poor model performance, some of the discrepancies may
have been caused by sampling and measurement errors and inherent
limitations in dispersion models.
Fourth, the model performance evaluation considered only
several alternative formulations of the SHORTZ model. A more
thorough analysis, with larger resources, could have considered
-------
V-23
assessing other EPA-approved air dispersion models. We are not
aware of any other EPA efforts that have attempted to rigorously
compare the performance of different air dispersion models.
Finally, the GC method is more limited than GC/MS in identi-
fying and quantifying pollutants. Our usage of GC was tied to
resource constraints, and though we believe this method did not
present serious problems for this particular monitoring effort,
we would like to improve our sampling efforts at future sites to
permit analysis of all samples using GC/MS.
Implications for Exposure and Risk
Cancer
We calculated the average lifetime individual risk associ-
ated with the measured and modeled (using the preferred Model 2
formulation) ambient air concentrations for seven pollutants
presented in Table V-3: chloroform; 1,2-dichloroethane; carbon
tetrachloride; benzene; trichloroethylene; 1,2-dichloropropane;
and perchloroethylene. We selected these specific chemicals
because the weight of evidence suggests they are potential car-
cinogens. Table V-4 details the cancer potency values used in
this analysis and their EPA level of evidence classification. To
assist in the interpretation of all risk estimates, all cancer
potency values presented in this report will be accompanied by
the corresponding EPA weight of evidence classification.
-•*..:
Oapooond
^•••••••^^"^^
Chlorofont
1 ,2-Oichloroethane
Carbon Tetrachloride
Benzine
Trichloroethylene
1 ,2-Oichloropropane
Perchloroethylene
1The cancer potency values
SuMar 1984.
*EPA grouping: A s Hunan
Carcinogen.
*The cancer potency value
Table V-4
PHILADELPHIA IEHP
CARCINOGENIC POTENCIES (INHALATION) AND EPA
WEIGHT OF EVIDENCE GROUPING
(1984 values)1
Cancer Potency Value Grouping Baaed on
(ug/«J)-1 EPA Criteria2
2.0x10-* 82
6.6x10-* 82
1 .5x10-5 82
6.9x10-* A
4.0x10-* 82
1 .8x10-5 C
1.7x10-* 82
Source
CAG
CAG
CAG
CAG
CAG
RID3
CAG
shown in this table represent those values that were available as of
Carcinogen; 82 z Probable Hunan Carcinogen; C * Possible
for 1,2-OCP waa developed by RID using the potency (qi»)
CPA'a Drinking Water Criteria Document, March 2, 1984. Thia document ia currently
Huaan
value from
undergoing
external review and, aa a reault, our potency value could change.
-------
V-24
All cancer potency values were developed by EPA's Carcinogen
Assessment Group, with the exception of 1,2-dichloropropane
(1,2-DCP). We generated a unit risk factor (inhalation) for
1,2-DCP based on the potency (qi*) value developed for EPA's
draft Drinking Water Criteria Document (March 2, 1984), from a
1983 NTP study. The Drinking Water Criteria document is nearing
completion of external review and, as a result, the potency value
could change. Compared with the other pollutants considered in
this analysis, 1,2-DCP also has the lowest level of evidence
classification (Group C: Possible Human Carcinogen). Nonethe-
less, there was enough dose-response information to warrant
further evaluation of 1,2-DCP in our study.
The estimated average lifetime individual risks for the
seven pollutants are detailed in Table V-5. The average risk
calculations range from a low of 4.0 x 10"' for chloroform
(modeled) to a high of 4.1 x 10~5 for benzene (measured). As
discussed in our review of the model performance evaluation, the
modeled ambient concentrations were consistently lower than the
measured ambient levels. This was especially true for carbon
tetrachloride; the difference between the average modeled and
measured ambient air concentrations was almost an order of
magnitude.
Table V-5
PHILADELPHIA IEMP
PHASE II RESULTS INTENDED FOR POLICY DEVELOPMENT1
ROUGH APPROXIMATIONS OF AVERAGE LIFETIME INDIVIDUAL RISKS:
MONITORED VERSUS MODELED AMBIENT AIR CONCENTRATIONS
(1984 analysis)
Average Ambient Concantntion
(uoV)
Average Lifetime Individual
Risks*
Compound
Chloroform
1,2-Oichloroethane
Carbon Tatrachlorid*
Benzene
Trichloroathylene
1,2-Oichloropropane
Parchlorosthylsn*
Cumulative
0.3
0.4
1.8
6.0
1.6
1.2
4.7
Modeled-*
0.2
0.4
0.1
2.3
1.0
0.5
3.5
Monitored
6.0xlO-7
2.6x10-*
2.7xlO-5
4.1xlO-5
6.4x10-*
2.2xlO~5
8.0x10"*
1.1x10-*
Modeled
4.0xlO-7
2.6x10-'
1.5x10-'
1.6xlO-J
4.0x10-'
9.0x10-'
5.9x10-'
3.9xlO-5
MHE UNIT RISK FACTORS USED IN THIS ANALYSIS ARE BASED ON CONSERVATIVE ASSUMPTIONS THAT GENERALLY
PRODUCE UPPER-BOUND ESTIMATES. BECAUSE OF LIMITATIONS IN DATA AND METHODS IN SEVERAL AREAS OF
THE ANALYSIS, SUCH AS EXPOSURE CALCULATIONS AND POLLUTANT SELECTION, RISK ESTIMATES MERE
CALCULATED AS AIDS TO POLICY DEVELOPMENT, NOT AS PREDICTIONS OF ACTUAL CANCER RISKS IN
PHILADELPHIA. ACTUAL RISKS MAY BE SIGNIFICANTLY LONER; IN FACT, THEY COULD BE ZERO. THE PROPER
FUNCTION OF THE ESTIMATES IS TO HELP LOCAL OFFICIALS SELECT AND EVALUATE ISSUES, SET PRIORITIES,
AND DEVELOP CONTROL STRATEGIES FOR THE TOPICS EXAMINED.
2Average aonitored valuee are fro» the sampling period 2/6/84 to 3/31/84 (see text).
^Modeled values were obtained using Model 2.
*See Table V-4 for a Hating of the pollutant potency estinstea end the weight of evidence
classification. The potency aatimataa vere current as of Suamer 1984.
-------
V-25
The individual lifetime risk estimates for each pollutant,
particularly for chloroform, 1,2-dichloroethane, trichloroethy-
lene, and perchloroethylene, may not necessarily suggest an ele-
vated level of concern about ambient air concentrations. How-
ever, the cumulative lifetime individual risks for the average
person breathing all of these contaminants is 1.1 x 10"^ (mea-
sured data) and 4.0 x 10~5 (modeled data).
Noncancer Effects
Table V-6 presents thresholds developed by RID toxicologists
by disease category for the seven pollutants previously analyzed.
We did not include methylene chloride in this comparison because
of monitoring and laboratory problems we encountered. These
Table V-6
PHILADELPHIA IEHP
NONCANCER HEALTH EFFECTS AND RID-OERIVED
v
Ctmoound
Chloroform
1,2-0 icnloroethane
Carbon Tetrechloride
1,2-Oictiloropropene
Trichlofoethylene
Per chloroethy lane
Benzene
lThe threshold valuae presented
*This value has been reviaed in
N.A. « Not available.
(1984 analyaia)1
Health Effect
Liver
Kidney (renal)
Reproductive
N.A.
Liver
Neurobehavloral
Kidney
Reproductive
Fetal developmental
effects
Liver
Kidney
Liver
Neurobehavloral
Kidney
Liver
Kidney
Fetal developmental
effects
HUWN THRESHOLDS FOR
RIO-Oerived Human
Threshold (ua/ej3)
8.5
22.5
2.4
10
10
108
430
24.2
308
308
769
769
3,770
294
294
909
Fetal developmental 4.1*
affects
in this table were developed in Summer 1984.
1986 to 41.3 ug/«3.
AIR
Source
DeSelve at al 1973;
Heywood at al 1979
HeywMd et al 1979
Schwetz et al 1974
1984 RID analyaia
1984 RID analysis
EPA 1980
Adams at si 1952
Schwetz at al 1974
8eau et al 1984;
NTP 1983
NTP 1983
1984 RID analyaia
1984 RID analyaia
1984 RID analyaia
1984 RID analysis
1984 RID analysis
1984 RID analysis
1984 RID analysis
-------
V-26
values were current as of mid-1984 when we conducted this analy-
sis. As exposures exceed the estimated threshold for a particu-
lar health effect, there is greater concern that there may be
noncancer health effects in the exposed population. Our compar-
ison of the measured ambient air concentrations to the RID-
derived human threshold shows ambient benzene concentrations
above the 1984 threshold but below the revised (following Envi-
ronmental Criteria and Assessment Office review) 1986 threshold
for fetal developmental effects.
Recent Changes in Unit Risk Factors
As existing information is reviewed and new toxicological
information becomes available, it is inevitable that potency
values will change to reflect the new data. Since we conducted
this study in 1984, there have been several significant changes
in the potency values for cancer and noncancer effects. Also, we
have added threshold values for additional pollutants and non-
cancer health effects. We discuss each separately.
Cancer
Table V-7 presents the revised (January 1986) cancer potency
values. The most notable change is the order of magnitude
increase in the unit risk factor for chloroform. The potency
values for perchloroethylene and trichloroethylene each show a
decrease of roughly 70 percent.
Compound
Chloroform
1 ,2-Dichloroethane
Carbon Tetrachloride
Benzene
Trichloroethylene
1 ,2-Oichloropropane
Perchl oroethylene
1The cancer potency values
January 1986.
2EPA grouping: A s Hunan
Carcinogen.
'ihe cancer potency value
Table V-7
PHILADELPHIA IEMP
CARCINOGENIC POTENCIES (INHALATION) AND EPA
HEIGHT OF EVIDENCE GROUPING
(1986 values)1
Cancer Potency Value Grouping Based on
(ug/W)-1 EPA Criteria2
2.3x10-* B2
6.6x1Q-6 82
1.5x10-5 B2
8.3x10-6 A
1.3x10-6 B2
1.8x10-* C
4.8x10-7 82
Source
CAG
CAG
CAG
CAG
.CAG
RIDJ
CAG
shown in this table represent those values that were available as of
Carcinogen; 82 s Probable Hunan Carcinogen; C s Possible
for 1,2-DCP was developed by RIO using the potency (q^*)
EPA 'a Drinking Hater Criteria Document, March 2, 1984. This document is currently
external peer review and.
aa a result, the potency value could change.
Hunan
value fro»
undergoing
-------
V-27
The revised average lifetime individual risk estimates for
each pollutant and the cumulative mixture of pollutants are shown
in Table V-8. As expected, the average lifetime individual risks
for chloroform rise from 6.0 x 10"~7 to 6.9 x 10~6 (measured) and
from 4.0 to 10"7 to 4.6 x 10~6 (modeled). We also note the fall
in our risk predictions for perchloroethylene and trichloro-
ethylene. The risks for the remaining pollutants are basically
unchanged, as are the cumulative lifetime individual risks.
Table V-8
PHILADELPHIA IEMP
PHASE II RESULTS INTENDED FOR POLICY DEVELOPMENT1
ROUGH APPROXIMATIONS Of AVERAGE LIFETIME INDIVIDUAL RISKS*
MONITORED VERSUS MODELED AMBIENT AIR CONCENTRATIONS
(1986 reviaed)
Averoge Ambient Concentration
(ug/ei3)
Average Lifetime Individual
Rieka4
Compound
Chloroform
1,2-Oichloroethane
Carbon Tetrachloride
Benzene
Trichloroethylene
1,2-Oichloropropane
Perchloroethylene
Cumulative
Monitored2
0.3
0.4
1.8
6.0
1.6
1.2
4.7
Modeled3
0.2
0.4
0.1
2.3
1.0
O.J
J.5
Monitored
6.9x10'*
2.7xlO-5
S.OxlO"5
2.1x10-*
2.2xlO~5
2.3x10-*
1.1x10-*
Modeled
4.6x10"*
2.6x10-*
1.5x10-*
1.9xlO~5
1.3x10-*
9.0x10-*
1.7x10-*
4.0xlO~5
1THE UNIT RISK FACTORS USED IN THIS ANALYSIS ARE BASED ON CONSCmTIVC ASSUMPTIONS THAT GENERALLY
PRODUCE UPPER-SOUND ESTIMATES. BECAUSE OF LIMITATIONS 01 OAT A «tt METHODS IN SEVERAL AREAS OF
THE ANALYSIS, SUCH AS EXPOSURE CALCULATIONS AND POLLUtwT KLCCTION, RISK ESTIMATES MERE
CALCULATED AS AIDS TO POLICY DEVELOPMENT, NOT AS POCDtCtlM V ACTUAL CANCER RISKS IN PHILA-
DELPHIA. ACTUAL RISKS MAY BE SIGNIFICANTLY LOMERi IM '*CT, T»CY COULD BE ZERO. THE PROPER
FUNCTION OF THE ESTIMATES IS TO HELP LOCAL OFFICIALS KUCt *O EVALUATE ISSUES, SET PRIORITIES,
AND DEVELOP1 CONTROL STRATEGIES FOR THE TOPICS EXAMINED.
'Average monitat»A value* ere fro» the aampllng period 2/4/tA u 3/31/84.
'Modeled vatupMtaia obtained ueing Model 2.
*See Table VSJffitm liating of the pollutant potency eetlaoteo end the weight of evidence
cleeeificafcfl(Bjsc The? potency eatia«tea are current ea of Jenwery 19M.
Noncancer Effects •
Table V-9 presents our revised list of thresholds developed
by RID toxicologists for noncancer health effects. This table
includes thresholds for additional health effects and pollutants
(toluene and xylene). Two changes affect our initial findings.
-------
V-28
Table ¥-9
PHILAOaPHIA IEMP
NONCANCER HEALTH EFFECTS AND RIO HUMAN
Compound
Chloroform
1 ,2-Dichlorosthane
(1986 value*)1
Health Effect
Liver
Naurooehsvioral
Kidney
Reproductive
Fetal developmental effects
Liver
Neurobshavioral
Gastrointestinal
Kidney
Carbon Tetrschloride Liver
Benzene
Trichloroethylena
Perchloroethy lens
1 ,2-Oichloropropane
Toluene
.-
Xylenm
-
Fetal developmental effects
Neurobshavioral
Kidney
Reproductive
, Blood effects (noncancer)
Fetal developmental effects
Liver
Nsurobshmvioral
Kidney
Hver
Fetal developmental effecta
Kidney
Liver
Kidney
Respiratory
Kidney
Reproductive
Fetal developmental effecta
Liver
Neurobehavioral
Fetal dsvelopsantal effecta
Liver
. Neurobehavioral
Respiratory
Cardiovascular
Slight anemia and leukopeme
Kidney
Reproductive
THRESHOLDS FOR INHALATION
RIO Derived Human
Threshold (ug/«3)
8.5
11.7
22. 5
2.4
2.4
26.0
26.0
26.0
26.0
2.5
24.2
2.5
103
430
*
41.3
25.9
25.9
3,770
69.9
909
69.9
308
308
1,010
1,010
500
476
1,010
580
52.8
215
215
215
215
215
215
52.8
Source
OeSalvs at al. 1975;
Hey wood et al. 1979
Challen et al. 1958;
U.S. EPA 1985
Heywood et al. 1979
Schwetz et al. 1974
Schwetz et al. 1974
Kozik 1957
Kozik 1957
U.S. EPA 1982
Heppel et al. 1946;
Hoffman et al. 1971
EPA 1980 and 1984;
Smyth 19)5 and 1936;
Schwetz et al. 1974
Holler 1973
EPA 1980
Adass et al. 1952
Snyder et al. 1978
and 1980
Kuna and Kapp 1981
EPA 1984
Grandjean et al. 1955
Tucker 1982
Caler and Roasmillsr
1953
Nelson et al. 1979
NCI 1977
Beau et al. 1984;
NTP 1983
NIP 1983
Von Oettingen et al.
1942; Bruckner and
Peterson 1981
U.S. EPA 1983
Mstsumoto et al. 1971
Hudak and Ungvary
1978
U.S. EPA 1983
Haraunen et si. 1976;
Seppelainen et el.
1978
Ungvsry et al. 1980
Bowers et al. 1982;
Tatrai et al. 1981
U.S. EPA 1984;
Ssvolainen et al. '
1979
Hipolito 1980
Hipolito 1980|
Hirach 1932
Browning 1965;
Hipolito 1980
U.S. EPA 1984
Ungvary et al. 1980
Jlhs threshold values presented in this table are from January 1986.
This threshold is
and 11 subject to
currently undergoing review by EPA's Environmental Criteria
change pending this review.
and Assessment Office
-------
V-29
First, the RID threshold for fetal developmental effects
(benzene) changed from 4.1 ug/m^ to 41.3 ug/m^. Concentrations
of benzene in the ambient air are well below this revised
threshold. Second, an additional noncancer effect for
benzene—blood effects—has been quantified. This threshold is
currently undergoing review by EPA's Environmental Criteria and
Assessment Office and is subject to change pending this review.
AMBIENT WATER MONITORING PROGRAM
Similar to the ambient air monitoring program, the ambient
water monitoring program was initiated to provide support for the
control-options analysis. We had two objectives in designing the
program. The first and primary goal was to design a program that
would give feedback on the potential effectiveness of several
control options for improving the quality of the finished drink-
ing water at Baxter. The second objective was to assess the
degree to which volatile organic compounds (VOCs) were being
emitted during wastewater treatment at the NEWPCP. Because of
the limited scale and duration of the monitoring program, the air
monitoring data collected at the NEWPCP were used only to confirm
volatilization at the NEWPCP.
As discussed in Chapter III, the interaction between the
NEWPCP and the Baxter drinking water treatment plant was a major
cross-media concern of the control-options analysis. During the
tidal cycle, when Baxter draws in water from the Delaware River,
the tidal action causes the NEWPCP effluent discharge to move
upstream on the flood tide past the intake point for the Baxter
plant. The water monitoring program was designed to explore the
effectiveness of three control options for reducing pollutant
concentrations in the drinking water that result from this
problem. We identified these options for possible inclusion in
the analysis on the basis of the monitoring.
The control options were as follows:
• Control Option 1; Change the time of withdrawal from
high tide to another point during the tidal cycle?
We monitored water quality over several cycles to
determine whether there is a point during the tidal
cycle when river water quality is significantly
better.
• Control Option 2; Move the Baxter intake point to a
location farther out in the Delaware River.Results
-------
V-30
from a 1977 EPA study suggest that water is not
uniformly mixed in the Delaware River.6 we therefore
hypothesized that moving the intake point one-half mile
or more farther out in the river might result in the
withdrawal of "cleaner" water. We used ambient moni-
toring data to test for a significant difference
between the current and proposed intake points.
• Control Option 3; Move the Baxter intake point up
stream.Since the tidal action pushes the pollutants
discharged by the NEWPCP almost eight miles upstream,
we monitored to determine whether constructing an
intake point one to two miles above the Baxter plant
would result in the withdrawal of "cleaner" water.
A second objective of the ambient water monitor effort was
to confirm that VOCs are emitted during wastewater treatment at
the NEWPCP/ and to assess, if possible, their contribution to
ambient air concentrations. We designed the program to gather
information that would help develop a rough mass balance for VOCs
at the NEWPCP, with respect to the influent/effluent wastewater,
air emissions, and sludge. As part of this overall effort, we
also coordinated"these monitoring activities with sampling around
the Baxter plant.
We describe the sampling plans for the Delaware River and
the NEWPCP below. We present the results and discuss how the
findings provided us with important insights into the feasibility
and effectiveness of the proposed control options as well as the
issue of volatilization from the NEWPCP.
Overview of the Sampling Plan
Quality Assurance/Quality Control
Before beginning the proposed ambient water monitoring, we
developed a monitoring and Quality Assurance Plan that was
reviewed by EPA1s Region III Quality Assurance Officer as well as
selected individuals from the City of Philadelphia, EPA
*>U.S. EPA, Evaluation and Characterization—Northeast
Philadelphia Water Pollution Control Plant, National Enforcement
Investigations Center, Denver, CO, and Region III, Philadelphia,
PA, 1977.
-------
V-31
headquart€irs, and EPA's Athens, Georgia, laboratory.7 some of
the major elements of the QA/QC plan were:
• Development of accurate recordkeeping procedures and a
sample chain-of-custody program
• Use of internal standards, blanks, and spiked samples;
duplicate and triplicate sampling
• Analysis of samples using different laboratories to
check the results (EPA's Annapolis laboratory and
Versar, Inc.)
The study was undertaken with the knowledge, consent, and assis-
tance of the Philadelphia Water Department (PWD) and with the
knowledge of at least one of the indirect industrial dischargers
included in the control-options analysis.
Delaware River Monitoring \
To collect ambient water data for use in evaluating the
three control options for the Baxter plant, we established sam-
pling sites at three points both in the river system and at the
finished water outlet at Baxter. A total of six river locations
were sampled. First, we selected three monitoring stations dir-
ectly off the Baxter intake (see Figure V-8). This transect of
sites was chosen to achieve two of the water monitoring objec-
tives:
• To relate variations in water quality to changes in the
tidal cycle (Control Option 1)
• To examine the differences between the three sites to
determine the extent of mixing of NEWPCP effluent with
the Delaware River (Control Option 2)
Along the|||gan3ect, station "a" was approximately 200 feet off
the PennaKf^nia bank in the vicinity of the current Baxter
intake, tJBpliid-river station "A" was approximately 500 feet from
the Pennsylvania bank, and station "aa" was 1,200 feet from the
Baxter facility in a water depth ranging from 12 to 20 feet.
7Versar, Inc., Quality Assurance Plan for Monitoring and
Analytical Activities to Evaluate Selected Pollutants in the
Delaware River in the Vicinity of Philadelphia, Pennsylvania,
U.S. EPA, September 9, 1983.
-------
V-32
jgjo
u. .2*-
•= (A
D>
"5.
E
to
CO
I
I
I
i
i
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
V-33
The second monitor, station B, was situated between the
NEWPCP and the Baxter plant to provide data on the upstream move-
ment of pollution during the tidal cycle (all control options).
The third monitor, station C, was placed upstream from the Baxter
plant out of the range of the tidal influence on the NEWPCP dis-
charge (Control Option 3). Finally, we sampled the finished
water at the Baxter drinking water treatment plant to supplement
the PWD monitoring data for compounds not previously sampled. We
collected samples at the continuously running tap used by the PWD
to sample the finished water.
Grab samples were collected at each site throughout a two-
and-a-half.-tidal-cycle period, with a major emphasis around low
tide, commencing at low tide on October 5, 1983, and ending, at
high tide on October 7, 1983. We identified the tidal cycles
using tide tables published by the U.S. Department of Commerce,
National Ocean Survey. While the sampling period appeared very
short given our objectives, we felt that the information gathered
over this time would help us evaluate the relationships between .
the different intake points and water quality. We did not design
the monitoring program to generate estimates of annual average
concentrations; such a goal would have required a long-term
monitoring effort and a significantly higher monitoring budget.
We should also note that sampling took place during the
spring tide, i.e., a tide of greater-than-average range around
the time of the new and full moons. As a result, the tidal surge
was most extreme during the period of monitoring, potentially
carrying pollutants emitted downstream (for example, from the
NEWPCP) farther upstream (to the Baxter plant and beyond) than
normal. Sampling during the fall months introduced yet another
conservative layer to our monitoring program. Fall is tradition-
ally a low-flow period, thus the measured pollutant concentra-
tions detected by our monitoring program may overstate average
annual ambient levels.
Tabig V-10 lists the organic and metal compounds for which
we samplMRL^Tbe specific chemicals included in the control-
options JKlJIyais are designated by an asterisk. The remaining
pollutantpjpitere included to permit their evaluation as potential
additions to the control-options analysis. In addition, for the
transect monitoring location, in situ measurements at the subsur-
face, bottom, and five-foot intervals in between were taken to
evaluate the degree of mixing in the entire water column. The
specific parameters we measured were water temperature, pH,
dissolved oxygen, and conductivity.
-------
V-34
Table V-10
PHILADELPHIA IEMP
POLLUTANTS ANALYZED IN DELAWARE RIVER SAMPLING
Parameter
Organlea
1,2-dichlocoethane*
1,2-dichloropropane*
Methylene chloride*
1,1,2,2-tetrachlorethane
\
trans 1,2-dichlorethylene
Perchloroethylene*
Trichlorethylene*
1,1,1-Trichlorethane
Aerylonitrile
Rationale
Probable human carcinogen known to be released
from NEWPCP and confirmed'in Baxter water
Possible human carcinogen known to be released
from NEWPCP and confirmed in Baxter water
Probable human carcinogen; commonly used
industrial and commercial solvent detected in
river and finished water
Possible human carcinogen; detected in river;
source(s) unknown
Possible human carcinogen detected once at
Baxter during 1982; monitoring to confirm
presence/absence and evaluate NEWPCP as
source
Probable human carcinogen detected in river
and at Baxter; commonly used industrial and
commercial solvent
Probable human carcinogen detected in river
and at Baxter; commonly used industrial and
commercial solvent
Suspected human carcinogen not detected in PWD
routine monitoring but reported present at
lower analytical detection limits. Commonly
used solvent
Probable human carcinogen detected in river
but not confirmed at Baxter with 5 ug/1
detection limits; monitoring to confirm
presence/absence and evaluate NEWPCP as
source
(continued)
-------
V-35
Table V-1Q (continued)
PHILAOaPHIA IEMP
POLLUTANTS ANALYZED IN DELAWARE RIVER SAMPLING
P«
>ter
Orqenica (continued)
bis(2-chloroethyl)ether (BCEE)
Rationale
Probable humen carcinogen known to be releaaed
fro* NEWPCPj Monitoring with improved
detection Unite to eveluete current level•
1,2 bis(2-chloroethoxy)ethane (TGDC) Related to BCEE; similar rationale
Chloroform*
Carbon Tatrachloride*
Benzene*
Metals v
Arsenic
Beryllium
Tri- and hexavalent chromium
Conventional (in-eitu)
Temperature, pH, diaaolved oxygen,
conductivity
Carcinogene under study ea part of overall
Philadelphia control-options
Probable human carcinogen cannonly detected in
river end at Baxter et relatively high
concentrations; source evaluation
Probable hunan carcinogen detected in river at
relatively high concentretions; source
evaluation
Monitoring to quantify thie probable human
carcinogen by valence etete (related to
toxicity)
Water quality data useful in analyzing
significance of other pollutants end possibly
to determine extent of NEWPCP/Delaware River
mixing
•Pol
considered in the control-options analysis.
-------
7-36
The organic samples taken from the river and the Baxter
outlet were analyzed using standard approved EPA protocols. In
analyzing the sample for arsenic, we employed an atomic absorp-
tion spectrophotometer; for beryllium and trivalent and hexava-
lent chromium, we used the Inductively Coupled Argon Plasma
direct reading emission spectrophotometer (ICP).
A total of 34 samples were taken at the river transect moni-
toring location: ten at the midway station A; eight on the New
Jersey bank, station aa; 16 around the Baxter intake, station a.
At station B, half way between the NEWPCP and the Baxter plant,
we collected ten samples. At the upstream point we gathered
three samples. Finally, we collected three triplicate grab
samples at the Baxter outlet point.
Sampling at the NEWPCP
To explore volatilization from the NEWPCP in more detail, we
established sampling locations throughout the NEWPCP to monitor
influent and effluent wastewater, sludge, and air. For waste-
water, the existing data on the type and amount of industrial
waste received at the NEWPCP was helpful though limited. As a
result, we decided to establish three monitoring points on the
interceptor system for both flow and chemical analysis (see
Figure V-9):
• The Somerset-Frankford low-level influent interceptors,
which receive little industrial wastewater flow
• The Frankford high-level influent interceptor, which
receives minimal industrial wastewater
• The Delaware low-level interceptor, which receives
wastewater from at least two major industries, one of
which was included in the control-options analysis
A monitoring site was also established at the Delaware River
discharge location. The flow and chemical sampling were per-
formed using grab samples that were collected three times over a
24-hour period at each site. We monitored for most of the same
organic compounds identified in Table V-10. The samples were
analyzed using GC/MS.
We measured sludge-mixed liquor samples, both digested and
undigested, using a grab apparatus. We collected about ten grab
samples, which were analyzed for organic content using GC.
Figure V-9 shows the sludge sampling locations.
-------
I
I
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
V-37
Figure V-9
Philadelphia IEMP
Sampling Sites for NEWPCP Study of Liquids and Sludge
frr-,~
",;i3«b'
J i
1
1 •
* I
i
<
"""" 1
i
nuCUCTLOn
1 «AMWO«0«
i eeu«
vnrn 5
LCVB.
.on a
AMU
fUMWATAKT 2
A4UTIOM TJMC
>7\— «i iijtuocz i
W I I otstwamoN
H1-u L-Sfj-111^
UC9lf ICO
IU PAUU. tUCOf^lACl. 2 STABUZATIONI
SAMf LINO SITI - LIQUID
SAMPLING SITE - SLUDGE
-------
V-38
Finally, we gathered on-site measurements of air concentra-
tions upwind and downwind from the rotating biological disks and
settling basins in an effort to demonstrate that volatiles were,
in fact, being emitted during wastewater treatment and to roughly
gauge the extent of these releases. We recognize that this
sampling effort was far too limited to be used alone to quantify
actual emissions from the plant.
We collected four grab samples on Tenax and Florisil at each
of the two downwind sites and one upwind site (see Figure V-10).
Each set of samples was taken over a different four-hour period
between noon on October 6 and 11:00 a.m. on October 7. A dupli-
cate sample and a field blank were taken once during each moni-
toring period. Thus, the total number of samples was 24, plus
field blanks. Samples were analyzed using GC/MS for the
following compounds:
• 1,2-dichloroethane
• 1,2-dichloropropane
• t-l,2-dichloroethylene
\
• Benzene
• 1,1,2,2-tetrachloroethane
• Trichloroethylene
• Perchloroethylene
• Carbon tetrachloride
• Chloroform
Caveats
A careful review of the water monitoring data will reveal
the limitations of the program. Although we mention many of the
limitations throughout this section, we summarize the major
issues below:
• The water monitoring program was of short duration and
relatively few samples were taken.
• The 1,2-DCP episodic release from one of the indirect
dischargers affected the results, making them less
representative of typical conditions.
-------
1
I - V-39
1
Figure V-10
1 Philadelphia IEMP
Sampling Locations for Air Monitoring
1
i
i
i
i
i
i *
•
•
i ^
\
i ^
'
i
i
. —
^
•
o
u
•
^
>. *
*
\
i ,l
l(WI« *t
. . •___ . ___ .. __._ . • — — — •• * • •— •— — — ••— - - — f
p^JMMttA
turioM
7^
r^T
nfe ^n
ffj J X^^^^V
t ^ -J V1
o3s a
..JO-IP — q
OASHCO ARC A OCOICATCO TO PLANT EXPANSION «
\ \ '
\ \ '
LAOOON -C \ \ LA4OON-A * •
' \ \ ' 1
«-..., ) \
\ ;
TO DISPOSAL AICAS
F^ "SS* j A
^^^••JB^^ ^^ 1 ^r |
VMCATSMCAT IAMC / *
^
U l-«
gL.JIIlJ
1 MIHMT 1
1 OAtiruts j
loU
\ unmo B
Ml ~ '
OVCRPASS . \v T* . J
\\\\ (•:
• '-X N -r1
,-*t , , If, _ |
** \ -v '
A£ RAT ION f HlAL |
~~ ..CtH< ' CLACIHERS 1
•A9in«f ^» i
"SlUOU ItfATIK SITE 1
OllllOIIIG
Ijlf
rrtKajirf a \° "aoo«-t
1 T'."J-aiDl ..._ u^.
•\ ' \ . 1 . /-. \ ••••»•••«• • -v , I r
\ \\ v r £-*- ^ "^ 1 f
1
• - - •— - -i
-•
!
* i
f LAGOON -C
! i
1 '
; t
1
{' DISPOSAL AHCAS
' DOWNWIND
!' SITE!
r
. 4 LAOOON- 0
• Scile: Appro*. 1
i
«
>
«
•
l
4
j
o
'l|
i
i
(
* • ?50f
Jill Courtesy of U>*
;1 City af Htnl«J«li''«i
j ' t, IXWIIICAST UAIFI ponuiinn COHTROI. PI AIM. puiMinruiA, PA. AUTJIST mi
i
t .
•
t
i
-------
V-40
The river and NEWPCP wastewater contain complex pol-
lutant mixtures, which are difficult to identify even
when using GC/MS.
The program was designed to further define the rela-
tionship between the NEWPCP and the Baxter water treat-
ment plant. Therefore, we sampled in the fall (low
river flow) and during the spring tide (a period of
greatest tidal influence on pollutant transport).
Results
Our overall findings for the Delaware River monitoring pro-
gram were insightful and provided an important basis for evaluat-
ing possible control options. While we generated interesting
data for the influent and effluent points for the NEWPCP, the
information could not be used to estimate volatile emissions.
The timing of the influent and effluent grab samples did not
accurately reflect the retention time of the NEWPCP system, as is
most often the case in monitoring programs of this type. Simi-
larly, the grab samples from the sludge were not sufficient to
calculate a mass balance. The limited air sampling data, how-
ever, did confirm that volatiles are being emitted during waste-
water treatment, but could not be used to generate estimates of
average annual conditions. We discuss each sampling activity
separately.
Delaware River Transect Monitoring
Table V-ll presents the measured pollutant concentrations at
the three monitoring stations, a, aa, and A. As discussed above,
these monitoring sites were established to generate data on river
quality over the tidal cycles (Control Option 1), as well as
variation in river quality across the transect (Control
Option 2). We discuss our findings on each separately.
Variation in Tidal Cycle (Control Option 1). The data in
Table V-11-clearly show variation in pollutant concentrations
over time. Concentrations of 1,2-dichloroethane, 1,2-dichloro-
propane, perchloroethylene, bis(2 chloroethyl)ether, and a com-
pound detected by the GC (l,2-bis(2 chloroethoxy)ethane) are
obviously the highest at high tide and lowest at low tide; how-
ever, we were unable to prove whether those differences were
-------
V-41
^ ! g >
-
?
*j
- *.
•s °
s i
£
<
s
1 £
1 !
3*
i
i
M
a J
"?
1
3
^
S
f-
3
s
£
(M
a
S
•s
I
*•
s
o
(si
a
S
s
S
S
<~
2
-
i
s
s
3
~
a
s
-
s
a
~
*
|
"
S
§
M
a
M
i
i
s
2
2
|
-*
w
-,
*•
y.
j
X
J
«y
X
Ul
w
X
—
UJ
M
J
...
X
J
!
I Pollutant
*
e
a
a
9
a
a
a
e
a
a
o
a
a
a
e
a
a
JO
a
a
a
a
e
e
e
*
°.
a
a
Hethylene Chloride
9
9
a
a
e
a
a
a
«
a
e
a
a
a
a
o
a
a
a
a
a
a
a
a
*
o
a
a
Chlorofum
—
•4
<**
•»
a
a
e
*•
-
a^
-
9
a
a
a
a
a
°.
a
~
a
•>
=
a
-
j
a
—
•^
"
a
e
a
«
0
e
a
a
e
a
a
a
a
e
a
°.
a
e
a
•«
a
a
a
1 .1 ,1- Irtchlaruethen*
9
3
9
e
a
a
e
a
e
o
e
o
a
a
a
a*
a
a
a
e
e
a
a
a
a
e
a
9
1
0
u
S
a
M
^
•9
-
a
Z
5
—
-
a
a
*
a*
*
Z
-
a
9
s
-
"
-
i
o
a
*
^
J
1 (2-Oicltloruprot>aiie
9
O
a
a
9
3
a
a
9
9
e
a
g
o
o
•
•
9
0
O
a
a
i
3
e
a
e
a
3
a
e
a
3
e
3
3
a
a
*
o
a
0
9
"
-
t
t
0
-
;
1
1 1
3
"•
•»
a
a
a
3
i--
(SI
-
3
a
o
9
c
9
a
a
o
a
-
•
•
-
t
a
•
t
i '
s
I
9
V*
+
a
^
~o
r*.
9
-
3
9
a
3
3
9
3
a
a
a
•
•
•
*
f
0
3
j
O
2
9^
P4
-
c
••
C4
"
-
9
9
C
9
a
3
*
-
*
*
i
~
•
•
-
-
>>
>
j
i
° 1
3
9
3
a
a
9
a
a
9
9
a
9
9
a
9
9
9
a
9
s
9
a
a
a
o
e
a
a*
i
i
5
X
-
a
3
a
a
a
a
S)
—
a
3
e
9
e
9
a*
a
a
«.
e
~
a
a
-
a
a
a
a
1*4
{
1
•4
a
•S,
!
*
i
>
*~
a
0.4 = Del m- led out not quaiit
0.0 - Not datoctad
!!*!-
ifil
-------
V-42
statistically significant for individual chemicals. Interest-
ingly, though, the results were significant for the summed
pollutants, as a whole.8
We also noted that the benzene concentrations were not cor-
related with the tide, confirming an earlier suspicion that there
was an upstream source of benzene. We were unable, however, to
identify the source within the scope and resources of the proj-
ect. Measured metals concentrations were negligible for this
site and for all other sites monitored in the water sampling
program.
In summary, we felt that the aggregate data supported fur-
ther consideration of Control Option 1 (to take in water at
Baxter during low tide and portions of the ebb and flood tides)
in the control-options analysis. Under this withdrawal scheme,
it appeared that concentrations of all measured pollutants, with
the exception of benzene, would be lower.
Variation among the Three Transect Sites (Control Option 2).
Although we observed some variability among concentrations at
each of the three sites, the differences were not sufficient
enough to consider Control Option 2 further. In fact, the only
statistically significant difference we observed was for
1,2-dichloropropane concentrations at the New Jersey shore site,
which were much higher during ebb tide. We suspect an ion
exchange resin manufacturer in New Jersey was the source respon-
sible for these concentrations; it discharges to the Rancocas
Creek. However, these discharges would still have to be
extremely large to account for the amount of 1,2-DCP measured in
the Delaware.
Delaware Upstream Monitoring (Control Option 3). Table V-12
presents the measured pollutant concentrations detected at the
upstream monitoring station. Our general results demonstrated
with statistical significance (95 percent) that pollutant levels,
with the exception of benzene, are lower upstream than at the
current intake point. The benzene anomaly was once again attrib-
uted to an unknown source upstream, possibly refineries.
Although these results encouraged further consideration of
Control Option 3, the estimated costs of constructing an intake
point upstream were so prohibitive that we no longer considered
this option as a feasible alternative.
A more detailed discussion of the statistical analyses can be
found in: Versar, Inc., "Discussion Paper," May 31, 1984. This
report is part of the background documentation and is available
on request.
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
V-43
Table V-12
PHILADELPHIA IEMP
POLLUTANT CONCENTRATIONS IN DELAWARE RIVER, STATION C, UPSTREAM CONTROL
(ug/1)
Time
Pollutant Tide
Methylene Chloride
Chloroform
1,2-Oichloroethane
1,1, 1-T r ichloroethane
Carbon Tetrachloride
1 , 2-Oichloropropane
Trans 1,2-Dichloroethylene
Trichloroethylene
Perchloroethy len e
Benzene
Dimethoxymethane
Bis( 2-chloroethyl) ether
1 , 2-bis( 2-chloroethoxy) ethane
1041
L
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
1545
H
0.0
0.0
0.0
0.0
0.0
1.0
0.0
0.0
0.0
1.1
0.4
0.0
0.0
2305
1
u
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.6
0.0
0.0
0.0
2806
H
0.0,
0.0
0.0
0.0
0.0
1.5
0.0
0.0
0.0
0.3
0.4
0.0
0.0
3529
L
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
4035
H
0.0
0.0
0.0
0.0
0.0
2.5
0.0
0.0
0.0
0.3
0.0
0.0
0.0
0.4 = Detected but not quantified at 1.0 ug/1 detection limit (except benzene).
0.0 = Not detected.
L = Low tide.
H = High tide.
Sourcei Versar, Inc., "Discussion Paper," May 31, 1984; samples analyzed by
Versar, Inc.
Monitoring Midway between the
NEWPCP and Baxter, Station B
We selected this sampling site to improve predictions of
upstream pollutant movements from the NEWPCP. The data (see
-------
V-44
Table V-13) demonstrated this trend for several pollutants,
including perchloroethylene, 1,2-dichloropropane, and 1,2-di-
chloroethane. Benzene concentrations, on the other hand, did not
vary with changes in the tide, thus reinforcing the suspected
existence of an upstream source. We attempted to confirm the
upstream discharger but were unsuccessful.
It is also important to note that significant levels of
methylene chloride, chloroform, and trichloroethylene found in
the NEWPCP effluent discussed below were not detected at
Station B. We attributed this phenomenon to two factors:
• Simple dilution of the effluent. Analytical results
for pollutants detected at relatively high levels,in
the NEWPCP effluent, primarily 1,2-DCP and 1,2-DCE,
indicate that the river dilution factor was between 100
and 1,000. Levels of other pollutants in the river
were, therefore, lowered to less than the detection
limit.
• Volatilization of chemicals from the river itself.
Results from the water monitoring program suggested
that volatilization of waterborne chemicals can occur
on the order of 10 percent per day from the Delaware
River. This observation confirmed similar findings
from the model performance evaluation work in air moni-
toring. As a result, we decided to consider volatile
releases from the Delaware as part of the toxic pollu-
tant flows tracked in the control-options analysis.
Baxter Finished Water
Figure V-ll shows the average measured concentrations for
raw (high tide on the Pennsylvania shore, Station a) and finished
water at Baxter. Most compounds appeared to be effectively
treated by the plant. Chloroform and other trihalomethanes are
formed during disinfection at the plant (PWD uses chloramines,
resulting in lower production of trihalomethanes than chlorine
alone). Another compound, most probably carbon tetrachloride (GC
could not positively identify the chemical), appeared to be added
during treatment at the plant. The presence of carbon
tetrachloride has, in the past, resulted from an impurity in
PWD's chlorine supply. PWD subsequently checked their records
and determined that there was no contamination at the time of the
monitoring activity. The presence of carbon tetrachloride was
attributed to chemical interference during laboratory analysis.
Finally, benzene concentrations appeared relatively unaffected by
treatment.
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
V-45
Table V-13
PHILADELPHIA IEHP
POLLUTANT CONCENTRATIONS IN DELAWARE RIVER, STATION B1
(ug/1)
Pollutant
Methylene Chloride
Tin*
Tide
Chloroform
1 , 2-Oichloroethane
l,l,l-Trichloroeth«ne
Carbon Tetrachloride
1 , 2-Oichloropropane
Trans 1,2-Olchloroathylena
T r ichloroethy lane
Perchloroethylene
Benzene
Dimethoxymethane
Bis( 2-chloroethyl) ether
1 , 2-bis( 2-chloroethoxy)ethane
0956
L
0.0
0.0
0.4
0.0
0.0
4.5
0.0
0.0
0.0
0.0
3.5
0.0
0.0
1209
F
0.0
0.0
0.4
0.0
0.0
10.9
0.0
0.0
1.8
0.0
10.4
0.0
1.4
1336
F
0.0
0.0
2.1
0.0
O.Q
23.2
0.0
0.0
5.8
0.0
18.0
0.0
2.0
1504
H
0.0
0.0
1.9
0.0
0.0
13.4
0.0
0.0
4.6
0.0
9.3
0.0
2.3
1757
E
0.0
0.0
0.0
0.4
0.0
100.0
0.0
0.4
2.9
0.6
7.6
0.0
4.0
2008
E
0.0
0.0
1.2
0.0
0.0
26.7
0.0
0.0
1.1
0.6
2.6
0.0
0.0
2220
L
0.0
0.0
0.4
0.0
0.0
14.7
0.0
0.0
0.4
0.3
2.0
0.0
1.2
2432
F
0.0
0.0
1.9
0.0
0.0
50.9
0.0
0.0
0.4
0.3
4.0
0.0
0.0
2558
F
0.0
0.4
2.6
1.1
0.0
110.0
0.0
0.0
2.2
0.2
7.8
0.0
3.4
2725
H
0.0
0.0
3.7
0.4
0.0
70.4
0.0
0.0
1.8
0.3
13.7
0.0
3.4
3019
H
0.0
0.0
1.8
0.4
0.0
65.0
0.0
0.0
2.0
0.4
10.4
0.0
2.1
3231
£
0.0
0.0
0.9
0.0
0.0
9.6
0.0
0.0
0.0
0.3
0.4
0.0
0.0
3444
L
0.0
0.0
0.0
0.0
0.0
8.1
0.0
0.0
0.0
0.4
1.3
0.0
Q.O
3657
F
0.0
0.0
1.5
0.0
0.0
27.0
0.0
0.0
1.2
0.0
5.7
0.0
1.5
3828
F
0.0
0.0
2.6
0.4
0.0
71.0
0.0
0.0
2.8
0.3
10.9
0.0
0.0
3944
H
0.0
0.0
2.9
1.2
0.0
53.0
0.0
0.0
2.1
0.3
14.1
0.0
3.0
^Station 8 is midway between NEWCP and Baxter.
0.4 = Detected but not quantified at 1.0 ug/1 detection limit (except benzene).
0.0 = Not detected
L : Low tide
H = High tide
E : Ebb tide
F - Flood tide
Source: Versar, Inc., "Discussion Paper," May 31, 1984: samples analyzed by Versar, Inc.
-------
V-46
Figure V-11
Philadelphia IEMP
Baxter Mean Raw and Finished Water
70.0
20 1
10
—Raw
—Finished
CHLDPOFORM 1,1,1-
TRICMIORO-
ETHANE
1.? OCP
BENZENE DIMCTII-
OXY
METHANE
1,2 OCE PERCW.ORO- TGOC
ETHYLENt
I ilher rtichlnrotir
irtrtil il y imrrrt iin
(r ir h Idrnflhy lrn«* nr
t* or mrtion telrirhloridc;
Source: Versar, Inc., "Discussion Paper," May 31. 1984.
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
V-47
The concentrations of compounds in the finished water as
quantified by our monitoring program were generally in agreement
with the PWD's past monitoring data. We therefore relied on the
PWD data to characterize most pollutant concentrations in the
Baxter finished drinking water for our control-options analysis.
For benzene and methylene chloride, which are not monitored by
PWD, we developed estimates of concentration in the finished
drinking water using our monitoring results from samples of
Baxter-treated water.
NEWPCP Influent Monitoring
Somerset-Frankford Low-Level Interceptors. Overall pollu-
tant concentrations were relatively low throughout the sampling,
with the exception of 1,2-dichloroethane and 1,2-dichloropropane
(see Table V-14). These high levels, however, were traced back
to recycling of the liquids from the NEWPCP digester.
Delaware Low-Level Interceptor. Concentrations were fairly
low for most compounds in each sample, except for 1,2-dichloro-
propane, 1,2-dichloroethane, and dimethoxymethane. The first two
compounds were found in unusually large amounts. It should be
noted that although we have reported these values as they
occurred, they are not indicative of annual average conditions.
We have been informed that operator error at one of the major
industrial indirect dischargers using this interceptor caused
large quantities of these compounds to pass through the treatment
plant.
The exact source of dimethoxymethane could not be verified,
but it is a widely used compound in the chemical industry.
Recent work by ORD's laboratory in Athens, Georgia, using
TOXIWASP and our water quality data, suggests that this compound
may be discharged by one of the industries included in our
control-options analysis.9 However, since this compound is
neither a carcinogen nor a high-toxicity pollutant, we did not
pursue this topic further.
Frankford High-Level Interceptor. This interceptor receives
minimal amounts of industrial wastewater, and most pollutant
levels were relatively low. We only detected methylene chloride,
chloroform, and 1,1,1-trichloroethane in all samples.
^Robert B. Ambrose, "Modeling Volatile Organic in the Delaware
Estuary," Draft Paper, U.S. EPA, Athens, Georgia, April 1986.
-------
V-48
Table V-14
PHILADELPHIA IEMP
POLLUTANT CONCENTRATIONS IN NEWPCP INFLUENT
(ug/1)
Pollutant
Hethylene Chloride
Inter-
ceptor
Tide
Chloroform
1 , 2-Dichloroathanel
1,1,1-Trichloroethaiw
Carbon Tetrachloride
1 , 2-Oichloropropene*
Trans 1,2-Oichloroethylene
Trichloroethylene
Perchloroethy lane
Benzene
DioethoxyMthane
8is(2-chloroethyl)ether
1 , 2-bis( 2-chloroethoxy ) ethane
Soueraet-Frankford
Low Level
1145
30.0
13.0
2.7
0.0
0.0
53.0
0.0
S 9.2
17.0
5.2
12.0
0.0
0.0
1945
36.0
15.0
260.0
8.1
0.0
575.0
0.0
5.4
50.0
6.7
745.0
0.0
0.0
2745
8.4
18.0
1.8
3.9
0.0
5.4
0.0
4.0
1.4
4.6
0.0
0.0
0.0
Delaware
Low Level
1145
0.0
9.2
230.0
14.0
0.0
2,000.0
0.0
5.J
41.0
16.0
i2va
3.J
a :
1945
39.0
11. 0
840.0
12.0
0.0
1,800.0
0.0
0.0
150.0
10.0
3,000.0
0.0
0.0
2745
0.0
16.0
326.0
8.9
0.0
>2, 290.0
0.0
0.0
48.0
15.0
2,500.0
0.0
0.0
Frankford
High Level
1145
32.0
6.2
0.0
15.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
1945
5.6
5.6
0.9
24.0
0.0
0.0
0.0
0.9
16.0
0.0
1.3
0.0
0.0
2745
0.0
4.5
0.0 j_
1.5
0.0
4.6
0.0
0.0
0.0
0.0
36.0
0.0
0.0
1,2-dichloropropane and 1,2-dichloroethane concentrations 'or INI :». •«•(•• low-level interceptor were anomalies in the
data set and are not representative of average annual condition*.
NEWPCP Effluent Monitoring
Table V-15 presents the measured pollutant levels in the
NEWPCP effluent. All pollutants detected in the effluent were
also detected in the influent. Again, we must note that the
reported 1,2-dichloropropane and 1,2-dichloroethane concentra-
tions are the result of an unscheduled dump from an indirect
discharger, and do not represent annual average conditions. The
dimethoxymethane levels, however, were also found to be very
high.
I
I
I
I
I
I
I
-------
V-49
Table V-15
PHILADELPHIA IEMP
POLLUTANT CONCENTRATIONS IN NORTHEAST WATER
POLLUTION PLANT EFFLUENT
(ug/1)
Pollutant
Methylene Chloride
Chloroform
1 , 2-Oichloroethane
1,1, 1-T r ichloroet hane
Carbon Tetrachloride
1,2-Oichloropropane \
Trans 1,2-Oichloroethylene
T r ichloroethy lene
Perchloroethylene
Benzene
Dimethoxymethane
Bis( 2-chloroethyl) ether
1 , 2-bis( 2-chloroethoxy ) ethane
Time
0942
14.0
4.8
110.0
0.0
0.0
2,500.0
0.0
16.0
40.0
2.0
2,600.0
0.0
0.0
1451
0.0
4.0
220.0
3.8
0.0
1,300.0
0.0
29.0
69.0
0.0
0.0
0.0
0.0
2206
12.0
3.1
166.0
0.0
0.0
1,400.0
0.0
40.0
24.0
1.7
702.0
0.0
0.0
2712
0.0
10.0
>7, 600.0
0.0
0.0
>6, 400.0
0.0
0.0
40.0
0.0
2,600.0
0.0
0.0
3430
0.0
6.1
430.0
0.0
0.0
>2, 500.0
0.0
0.0
95.0
0.0
480.0
0.0
0.0
3941
0.0
4.0
206.0
0.0
0.0
>11, 000.0
0.0
0.0
67.0
0.0
430.0
0.0
0.0
^The 1,2-dichloropropane and 1,2-dichloroethane concentrations are anomalies in the data set.
and are not representative of annual average conditions.
0.0 = Not detected.
Source: Veraary Inc., "Discussion Paper," May 31, 1984; samples analyzed by EPA Region III,
Annapolis.
NEWPCP Influent-Effluent Comparisons
The instantaneous nature of grab sampling requires a precise
knowledge of retention time in the treatment system, especially
-------
V-50
if we are to gather effluent samples representative of the influ-
ent samples previously taken. While we designed our monitoring
program at the NEWPCP to coordinate sampling times, the results
indicated that the attempt to gather such information was unsuc-
cessful. It was, therefore, impossible to compare mean influent
and effluent concentrations. As a result, we could not infer
anything about treatment efficiencies from the data.
NEWPCP Sludge Monitoring
Table V-16 presents our results from the sewage sludge grab
samples. The data were insufficient to calculate a mass balance;
however/ we found that mean sludge concentrations ranged from 0
to 30 percent of the influent concentrations for all compounds
except methylene chloride, trans-l,2-dichloroethylene, and
perchloroethylene (one high observation biased the mean).
Table V-16
PHILADELPHIA IEMP
VOLATILES IN NEWPCP SLUDGE CONCENTRATIONS IN ppb (ng/g)
Sample
Pollutant
Methylene Chloride
Chloroform
1 , 2-Oichloroe thane
1,1, 1-Trichloroe thane
Carbon Tetrachloride
1 , 2-Dichloropropane
Trans 1,2-Dichloroethylene
T richloroethy lene
Perchloroethylene '
^Detected but not quantified.
ND = Not detected.
Detection Limit: 1 ng/g.
Source: Veraar, Inc., "Discussion
Versar, Springfield, VA.
Sludge
Grab I
18
ND
ND
8
ND
18
240
ND
M>
Paper," Hay
Sludge
Grab II
16
ND
86
ND
ND
300
84
ND
510
31, 1984;
Sludge Digested
Grab III Sludge
15
ND
ND
2
ND
33
12
ND
samples
16
ND
110
ND
ND
470
150
120
10
analyzed by
-------
V-51
We also noted that volatile concentrations in the digested
sludge were higher than in the undigested sludge grab samples.
This result suggested that influent concentrations may often be
higher than what we observed during monitoring.
Volatile Air Concentrations
at the NEWPCP
Table V-17 summarizes the measured ambient air data for the
short-term monitoring period from October 6 to October 7. The
data show that concentrations downwind are generally higher than
those upwind. We interpret these results to mean that volatiles
were emitted during wastewater treatment at the NEWPCP. Further-
more, we observed measurable concentrations for the pollutants of
interest in the control options analysis, most notably 1,2-DCP
and 1,2-DCP.
SUMMARY
This chapter has presented an overview and the results of
the ambient air and water monitoring programs that EPA conducted
in Philadelphia during Phase II. The purpose of these efforts
was to gather data that would refine and clarify various elements
of the control-options analysis, specifically, the baseline
estimates of pollutant releases to air and the feasibility of
control options at the Baxter drin
-------
V-52
r>
^
e
•4
1
9t *
u •-
5 8^ "^
1 IJ
irt UJ
u
ac
<
S
-^
1
w
KN
S
a
o
X
I
t/>
1
rv
9
S
2
u
•w
£
°;
*
in
«
O
t
CD
&
I
s
1
*
jl
UN
t ••*
I i 5
! S"
i'2-
i -3
! i •"
: 25
1
1
!!
! 2"1
i Z *
i Is
1
1
1
' "5
1 *4
! I j
i a w
I Q
r
i
i 1
S |
I 2 N
! £ *
t
i
II-
i i^
1
11
I I~
i js
1
t
if-
i
i
i
i i
1
^
O ff>
= s
CM ^
« J
• s
m
ov a
O in
2
« »
s»
s
«
2 s
« KV
4 «
M 1*1
e CB
-J
1 N
•» CO
C3 -»
Hethylene Chloride
1,2-Oichloroethana
„.
P*
O 1
i •
§ '
£
0 1
>•> a
-H O
r.
2 ~
S '
2 '
S '
1 1
9 '
2 •
1, l-Oichlaroethane
Chlorofon
in co CO CM c*
-1 S * 3 *
P4
V0 CM < O Cf
•" S """ oJ —
* ""
"5 — >O 9> 9>
cs o CM jj oo
_• *> r» m o
CM 00 O OJ «
« en »•> o *
N S 3 S *
in 7
O r«- aj ^
•* •*
_ SO — », ^v
eo o m K> *
- s « s "
f- •» M> O
o o i 9s in
CM N
r*
~t ^ CM »N O
O O CD CM **
-* r* ^ O i«
-« •« CM g o>
1 §
i 1 i I
£ U -N -M
u a >• >»
* o j= r
b i* ** «j
^ 5 § §
_. S U b
e 5 5 5 S
5 9 € € S
~ fM -4 u e
« - u e o
u — >- a- OJ
n co
•a a.
e £
1 I
O V
b »
j <£
f«
^
^> ^
S «
a —
^
* JS
01
J i
= H
0 a
5 0
1 £
9 >.
«
! 1
i 1
1 :
1 !
U 0
S s
i s
i i
t •-
1 1
2m £
II §
•a jj -.
1 -^ CB
-T S u
u a
•Aabient concentretlo
rapreaent actual aaii
Sourcet Veraar, Inc.
-------
V-53
However, we believe that the differences between the moni-
toring data and our modeling results should not be viewed as
shortcomings of the project. In relation to other uncertainties
that are part of the dose-response information currently avail-
able, the overall half order of magnitude difference between our
monitoring and modeling results does not seem that substantial.
In addition, we believe our inability to capture all the
possible sources of pollutants does not reduce the value of the
control-options analysis, and in particular, the cost-effective-
ness results. Within the limited framework of the analysis, we
were able to quantify the major sources, pollutants, routes of
exposure, and control costs. Obviously, the analysis applies
only to conditions we could define and quantify. Those situa-
tions with very limited data or beyond the analytical measures
available were recognized but not included within the scope of
our assessments.
Our ability to quantify differences between monitoring and
modeling results highlights the advantages of using both tech-
niques together to improve exposure assessments substantially.
Modeling is limited by the available information on sources and
emission loads. \Monitoring, on the other hand, provides inform-
ation about pollutants and sources for which one may know very
little at first. The two in combination complement the strengths
of each and reduce the inherent weakness of both when used
i ndependently.
Our calculations of average lifetime individual cancer risks
using the ambient air data (monitored and modeled) showed risk
levels ranging from 10"' to 10~5, depending on the pollutant.
Cumulative lifetime individual risk was approximately 1 x 10~4.
Recent changes in several unit risk factors did not alter these
basic findings but did result in a tighter range, 10~^ to 10"^,
in the average lifetime individual cancer risks. The risk esti-
mates associated with the modeled ambient concentrations were
lower because of discrepancies in the emissions inventory and
emissions from unaccounted sources.
Recent^changes in the threshold for fetal developmental
effects from benzene altered our initial (1984) noncancer benzene
assessment. Monitored and modeled ambient benzene concentrations
are well below the revised threshold (41.3 ug/m^).
The ambient water monitoring program, although a more
limited undertaking, also provided useful information for the
control options analysis. First, the aggregate data from the
Delaware River monitoring supported further consideration of
control strategies for the Baxter plant, including a change in
-------
V-54
the time of withdrawal from high tide to low tide and portions of
the ebb and flood tides. Second, the limited air sampling around
the NEWPCP confirmed that VOCs are being emitted during waste-
water treatment. Third, the sampling of the Baxter finished
water provided data for exposure assessments and for control-
options analysis for benzene and methylene chloride, which are
not monitored by PWD. This information was used, in conjunction
with PWD's monitoring data, to characterize average water quality
levels in the Baxter finished drinking water.
-------
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
VI. RISK ASSESSMENT AND CONTROL-OPTIONS ANALYSIS RESULTS
This chapter reports on the cost-effectiveness analysis of
control options for the set of sources and pollutants identified
in Phase I. We were able to conduct this analysis because we had
sufficient data to assess the ambient exposures arid risks associ-
ated with these sources and pollutants.
We conducted the control-options analysis because we wanted
to generate information that would assist policymakers in risk
management, i.e., addressing the key policy questions that face
them in setting control priorities. To do so we needed to
develop and test an analytical approach for assessing alternative
control strategies in a multimedia setting; and we selected cost-
effectiveness as the most promising criterion for making the
necessary comparisons.
This chapter is divided into six sections/ as follows:
• Cost-Effectiveness Analysis defines what we mean by
cost-effectiveness and discusses how the analysis was
used.,
\
• Sources and Pollutants provides background information
on the topics selected for analysis in Phase II. We
summarize the pollutants and sources covered in the
analysis and describe how the findings from the
Phase II ambient air monitoring program and other
activities expanded the list of sources.
• Data Requirements and Modeling briefly describes the
basic inputs to the cost-effectiveness analysis and the
data sources.
• Results of the Analysis establishes the baseline risks
to human health (for the aggregate population and the
MED associated with the exposure pathways considered
in the control options analysis; reports our findings
on cost-effective control options for each risk
scenario; and puts our baseline risk estimates in
context to assist decision makers in assessing the
significance of these environmental issues.
• Recent Changes to Risk Assessment Factors reports two
important recent changes that have affected data used
in the analysis: revisions to unit risk factors and a
reduction in the pollutant flows from one of the prin-
cipal industrial facilities included in the analysis.
These changes have occurred since our analysis was
-------
VI-2
completed; because of resource and scheduling con-
straints/ we were unable to repeat the analysis.
However, we did explore how these changes might alter
the set of control priorities identified by the study.
Limitations presents a list of the limitations and
uncertainties associated with the cost-effectiveness
analysis that must be considered when interpreting the
results.
COST-EFFECTIVENESS ANALYSIS
For the purposes of our study, we defined cost-effectiveness
as the cost of control per unit of health risk reduced. We
developed estimates of cost for each control option, considering
both capital and operating and maintenance expenditures. We then
annualized these costs over the life of the control equipment
using a 10 percent discount rate recommended by Executive Order
12291. We defined effectiveness as the reduction in the annual
cases of cancer from the baseline (i.e., existing conditions)
afforded by each control option. We did not attempt to reassess
the cost-effectiveness of controls currently in use; we accepted
these as part of the baseline.
A cost-effectiveness analysis of control options can gener-
ate information highly useful to local policymakers in setting
control priorities. They can determine for any risk reduction
level the cheapest approach. Also, decision makers can get a
sense of the effectiveness of individual controls and total con-
trol strategies by seeing their costs to reduce a statistical
upper-bound cancer case. This allows them to allocate their
community's limited resources into activities that provide the
greatest environmental health protection. Policymakers can also
use the results from a cost-effectiveness analysis in conjunction
with additional information on local social problems to assist
them in their determination of an appropriate level of risk
reduction, to pursue, given the opportunity cost of diverting the
area's resources from other health, safety, and community pro-
grams to environmental protection. Specifically, we tailored our
control-options analysis to answer the following types of
questions that might be posed:
• What are the major sources, pollutants, and exposure
pathways of the risks?
• If one wanted to reduce the baseline level of overall
risk by a fixed amount (e.g., 10 percent, 25 percent,
-------
VI-3
or 50 percent), what control options would one need to
implement in various media?
• What are the average and incremental costs of pursuing
various control options? It is important for decision
makers to know the total amount of money that could be
spent on pollution controls and the benefits derived
from these expenditures. In addition, the marginal
analysis can show decision makers at which point addi-
tional controls become increasingly expensive relative
to the amount of risk reduction they can provide (i.e.,
the point of diminishing returns).
In addition, our control options analysis had the flexi-
bility to address other concerns raised by decision makers. For
example:
• What is the highest level of risk reduction that could
be achieved for a specific annual budget for control
expenditures (e.g., $10, $50, or $100 million per
year)?
• What cost-effective control strategies are available
for reducing the human health risks posed only by point
sources? And what are their costs?
• How do our cost-effective response strategies change if
we assume that controls must be installed at all three
drinking water treatment plants in the study area or
none at all?
We designed our initial cost-effectiveness analysis to
identify the optimal, i.e., least-cost bundle of controls for
achieving a specified level of risk reduction. It is also
designed to allow decision makers to see the cost-per-case
reduced for each control level considered. We developed control
strategies for two individual measures of risk: annual aggregate
excess cancer incidence and lifetime risk to the most exposed
individual. We also assessed control strategies for air and
drinking water separately, as well as for all media taken
together. Finally, we mapped out a set of cost-effective control
strategies available for increasing levels of risk reduction.
There are, however, alternative analytical approaches for
establishing control priorities. For example, a recent draft EPA
study explores the approaches that rely on the dollar quantifi-
cation of benefits and costs of controlling pollution in the
-------
VI-4
Philadelphia urban area.1 This benefit-cost analysis expands on
the IEMP risk assessment/risk management framework by assessing a
wider range of pollutants and effects. Appendix A describes the
EPA benefit-cost approach and its findings in more detail.
SOURCES AND POLLUTANTS
Background
In Phase I, we identified a set of major sources and pollu-
tants that accounted for most of the ambient pollutant exposures
in the Philadelphia area; we identified these sources and pollu-
tants as candidates for analysis in Phase II. In defining human
health risk for the control-options analysis, we considered only
exposures through drinking water (ingestion) and the ambient air
(inhalation). Furthermore, we focused primarily on exposures to
carcinogenic substances. Although noncancer effects were not
emphasized in Phase I, we evaluated the increased risk of non-
cancer effects for those pollutants and sources considered in the
Phase II analysis. Our analysis showed that the modeled ambient
air and drinking water concentrations used in our analysis did
not approach the no-effect thresholds for individual pollutants.2
As a result, we developed cost-effective control strategies only
for reducing cancer risks in the Philadelphia study area.
The Phase I findings pointed to several industrial plants
and the city's major industrial sewage treatment plant (the
NEWPCP) as the largest emitters of hazardous organic air pollu-
tants. Urban area sources, such as dry cleaners and degreasers,
were also of particular concern for solvent emissions. The
finished water at the three municipal drinking water treatment
plants proved to be a potential exposure pathway of risk to human
health.
The Phase I screening results also showed important linkages I
and interactions among several of these sources and the resulting •
Ralph Luken and Brett Snyder, "Comprehensive Environmental
Analysis for the Philadelphia Region," Revised May 8, 1986,
Draft Paper.
2We should note, however, that our monitored ambient air concen-
trations for all pollutants were found to be higher than our
modeled values. We were unable to fully capture these ambient
levels in our control-options analysis because of missing
sources and our selection of facilities to evaluate in the
study.
-------
VI-5
risks to human health. Specifically, a large portion of the
emissions from the NEWPCP could be related to the indirect dis-
charge of volatile organic compounds (VOCs) from one of the
industrial facilities, also identified as a major air emitter.
Similarly, the presence of several trace organics in the finished
water at the largest of the three drinking water plants, Baxter,
could be traced back to the effluent discharged by the NEWPCP,
and, in turn, to the indirect dischargers. The tidal surge in
this region of the Delaware River pushes the pollutant discharges
from the NEWPCP upstream past the Baxter plant during the period
when raw water is withdrawn from the river for treatment.
In Phase II, we initiated an ambient air monitoring program
to provide feedback on the emissions inventory (i.e., sources and
release rates) used in the cost-effectiveness analysis. On the
basis of this work, we expanded the analysis to consider two
additional area source categories. No new point-source emitters
of air toxics were located. We also deleted three facilities
from the original Phase I source list: two industries that
ceased operations—the industrial degreaser and one of the two
industrial dry cleaners—and a chemical manufacturer that emitted
only a small amount of benzene (0.1 kkg/year), which had a
negligible impact on exposure and risk.
Figure VI-1 presents a schematic depiction of all the
sources considered in the Phase II analysis and their linkages to
exposures in air and water. Below we describe the Phase II
sources (area and point) and the pollutants released by receiving
medium in more detail.
Pollutants
Eight pollutants we examined accounted for more than 80 per-
cent of the total point-source releases to air as documented in
the Philadelphia AMS emissions inventory. Of the risks we could
quantify, these eight high-toxicity pollutants accounted for most
of the estimated human health ris«s from air and drinking water.
These pollutants are:
• Benzene
• 1,2-Dichloroethane (1,2-DCS)
• Methylene chloride
• 1,2-Dichloropropane (1,2-DCP)
-------
VI-6
-------
VI-7
• Perchloroethylene
• Carbon tatrachloride
• Trichloroethylene
• Chloroform
We also added a ninth pollutant, gasoline vapors, because of its
release from service stations during gasoline marketing; gas
marketing from service stations was an issue of interest to AMS.
Consideration of the cancer risks from gasoline vapors was made
possible by a recent American Petroleum Institute chronic inhala-
tion study of unleaded gasoline vapors in rats and mice. The
cancer risk from exposure to gasoline vapors is additive to that
from benzene.
Point Sources
We identified the most significant point and area sources of
the eight organic substances through an evaluation of the data
for air and water collected during Phase I. A listing of the
major point sources and source categories and a description
follow:
• Refinery A. A major producer of such products as gaso-
line and benzene; benzene and gas vapor are released
from the storage tanks (air emissions only).
• Refinery B. A second major producer of gasoline,
benzene, and other petrochemical products; benzene and
gas vapors are released from the storage tanks (air
emissions only).
• Pharmaceutical Manufacturer. A plant that uses
methylene chloride, chloroform, and carbon tetra-
chloride in very pure grades in its pill and tablet
coating process (air emissions only).
• Industrial Dry Cleaner. An industrial plant that uses
perchloroethylene in its operations (air emissions and
water discharges to NEWPCP).
• Chemical Manufacturer. A major manufacturer of herbi-
cides and ion exchange resins that uses 1,2-dichloro-
ethane, perchloroethylene, and 1,2-dichloropropane in
its processing (air emissions and water discharges to
NEWPCP).
-------
VI-8
• Garment Manufacturer. A plant that uses trichloro-
ethylene in its dry cleaning unit (air emissions
only).
• Plastic Cabinet Fabricator. A maker of molded plastic
household cabinets that uses methylene chloride to
clean the molds between use (air emissions only).
• The Northeast Water Pollution Control Plant (NEWPCP).
The largest sewage treatment plant in Philadelphia,
receiving large amounts of industrial waste from
several major facilities. Indirect discharges of
several volatile organic compounds, primarily
1,2-dichloropropane and 1,2-dichloroethane from the
chemical manufacturer, are emitted to air during waste-
water treatment. Also, pollutants discharged from the
NEWPCP to the Delaware River are an additional source
of volatilization, as well as contamination of drinking
water at the Baxter facility upstream (air emissions,
water discharges to Delaware River, drinking water).
• The Baxter Drinking Water Treatment Plant. Philadel-
phia's\major drinking water treatment plant, which
draws water from the Delaware River roughly eight miles
upstream from the outfall of the NEWPCP during high
tide—a withdrawal schedule that is energy-efficient
for the plant's operation (drinking water only).
• The Queen Lane Drinking Water Treatment Plant. A
facility operated by the City of Philadelphia that
draws water from the Schuylkill River (drinking water
only).
• The Belmont Drinking Water Treatment Plant. A second
facility operated by the City of Philadelphia that
draws water from the Schuylkill River (drinking water
only).
Three of the plants—the chemical manufacturer, the plastic
cabinet fabricator, and the industrial dry cleaner—are indirect
process wastewater dischargers. Because of the extremely small
discharges from the plastic cabinet fabricator, however, we did
not track the pollutant flows from this plant to the NEWPCP. The
garment manufacturer discharges only sanitary wastewater to the
NEWPCP; therefore, these flows were not tracked in the analysis.
Refineries A and B and the pharmaceutical manufacturer connect to
a different sewage treatment plant, the Southwest Water Pollution
Control Plant (SWWPCP). Refineries A and B only discharge sani-
tary wastewater to the SWWPCP, which is not a concern for this
-------
VI-9
analysis. We assessed the marginal contribution from the pharma-
ceutical manufacturer's processed wastewater to volatilization at
the SWWPCP, but the emissions were extremely small and therefore
excluded from further consideration in the control-options
analysis.
Urban Area Air Sources
We also considered several area air source categories that
were identified by the Phase I priority-setting process. We
recognized that permitting and enforcing controls at these widely
dispersed sources would be difficult, but these sources accounted
for a large portion of the organic ambient air releases. To
evaluate controls, we also wanted to compare the point sources
releasing the same chemicals with area sources. These sources
included dry cleaning, degreasing, and miscellaneous industrial
solvent usage. Following the Phase II ambient air monitoring and
model performance evaluation, we added two additional sources:
volatilization from one of the major sewer lines leading to the
NEWPCP (the Delaware Low Level Intercept), and volatilization
from the Delaware River. We modeled these categories as area
sources to simplify our exposure calculations. Finally, we added
gasoline marketing from service stations because of the concern
expressed by the Philadelphia AMS.
We did not include other area sources for which we had
developed emission estimates because either the source categories
(e.g., residential heating) did not emit the pollutants con-
sidered in the control-options analysis, or the sources (e.g.,
mobile sources) could not be effectively controlled at the local
level. We also did not model the process wastewater releases
from areawide degreasing and dry cleaning operations to the
sewage treatment plants. We determined that these discharges
would have a negligible impact on volatilization at the waste-
water treatment plants, as well as on the quality of the effluent
from the sewage treatment plants.
DATA REQUIREMENTS AND MODELING
There were two major types of information required, for our
cost-effectiveness analysis (see Figure VI-2): (1) an assessment
of the baseline risk levels, including intermedia transfers, for
the sources, pollutants, and exposure pathways considered in the
analysis; and (2) the costs and removal efficiencies (i.e.,
effectiveness in reducing the baseline risks) of alternative
control options. We also developed a computer model to integrate
-------
VI-10
Rgure VI-2
Philadelphia IEMP
Cost-Effectiveness Analysis of Control Options1
Information Flow
Sources
(1984 Current Control Levels)
• Point
• Area
Pollutant Releases
• Air
• Surface water
Exposure Pathways
Inhalation (dispersion
modeling)
Ingestion (drinking
water)
Human Health Effects
(Baseline Cancer Risks)
• A
• Aggrega
• MET risk
ate incidence
• Air monitoring
• Water monitoring
Water •
monitoring
Air monitoring
• Water monitoring
Water
monitoring
Feasible Incremental
Control Options
• Air
• Water
Control Costs
• Air
• Water
Control Effectiveness
(Percent Removal Efficiency)
Air
Water
Computer Model
(Mixed Integer Program)
Cost-Effective
Control Strategies
• Cost of the reduction
in aggregate cancer
incidence
• Cost of the reduction
in MEI risk
1 Cost-effectiveness Is defined as the cost of control per urtt of hearth risk (cancer) reduced.
-------
VI-11
the risk and control data and identify cost-effective control
strategies. Below we briefly describe how we generated the prin-
cipal data elements of our control-options analysis, a
description of the computer model can be found in Appendix E.
Baseline Risk
As shown in Figure VI-2, we calculated baseline risks to
human health through inhalation and ingestion of drinking water
by first identifying pollutant releases from the sources included
in our analysis to each receiving medium, we combined the air
emissions data with the LONGZ^ dispersion model to estimate
ambient air concentrations. We should remind the reader that the
ambient air monitoring program was designed primarily to provide
feedback on our emissions inventory; we still had to rely on
dispersion modeling to relate the emissions from each source to
ambient air concentrations and exposure. We generally used the
Philadelphia Water Department's annual average concentration data
to estimate pollutant levels in finished drinking water at each
of the drinking water treatment plants.
\
Finally, we calculated baseline risks using our ambient
pollutant concentrations in air and drinking water and the appro-
priate cancer unit risk factor (exposure assumptions are already
contained in the unit risk factors), we employed the method-
ologies discussed in Chapter II to calculate aggregate cancer
incidence and MEI risks.
We summarize the information gathered to characterize base-
line pollutant releases in Appendix F. The steps involved in
characterizing exposures and risk are detailed in Appendix G. We
describe the potency values used to assess cancer risks in the
cost-effectiveness analysis below. We also discuss how we
assessed the potential for noncancer health effects at current
controls.
3SHORTZ was used to assess the comparability of measured data
with concurrent modeled data, which were based on the emissions
inventory. Subsequently, comparisons were made between SHORTZ
(using meteorological data associated with the monitoring pro-
gram) and LONGZ (based on climatological data). Comparisons
were also drawn between the LONGZ and SHORTZ models on the basis
of the meteorological data set associated with the air quality
monitoring program. This work confirmed the comparability of
the two models.
-------
VI-12
Unit Risk Values
The Philadelphia IEMP focused on cancer as the primary
indicator of human health risk. Table VI-1 shows the unit risk
factors that were used in the analysis to calculate human health
risks from inhalation and ingestion. It also presents the level
of evidence for each pollutant. All of these values were devel-
oped by EPA's Carcinogen Assessment Group, with the exception of
1,2-dichloropropane. RID toxicologists and consultants calcu-
lated a unit risk factor for 1,2-dichloropropane from a potency
(qi*) value developed for EPA's Drinking Water Criteria Document
on DCP (March 2, 1984). The potency (qi*) value was based on a
1983 National Toxicology Program (NTP) study.4 While the NTP
study can be used as a preliminary basis for evaluating risks to
human health, we should note that the Drinking Water Criteria
Document on DCP, and thus the NTP study, are currently undergoing
EPA review. As a result, our unit risk factor could change.
Although cancer was the primary health effect of concern in
our analysis, and particularly in the control-options analysis,
we also attempted to determine whether maximum ambient concentra-
tions exceeded thresholds for five other health effect cate-
gories: liver, neurobehavioral, kidney, reproductive, and fetal
developmental effects. Table Vl-2 shows the estimated threshold
values for these effects by pollutant and exposure pathway
(inhalation and ingestion).
We used these threshold values in two ways. First, we com-
pared predicted maximum ambient concentrations with the no-effect
threshold on a pollutant-by-pollutant basis. As ambient concen-
trations approach or exceed the estimated threshold, there is an
increased concern for the health of the public exposed to these
pollutant levels. Second, we attempted to assess the chronic and
subchronic effects from exposure to chemical mixtures found in
the ambient environment. While some potential health hazards may
result from exposure to only one substance, most environmental
exposures (especially in urban areas) involve concurrent or
sequential exposures to a complex mixture of compounds that may
induce similar or dissimilar effects over time. We calculated a
"health hazard index" to explore the potential hazards from the
cumulative exposure (air and drinking water) to those chemicals
4NTP Technical Report on the Carcinogenicity Bioassay of
1.2-Dichloropropane (CAS No. 78-87-5) in F344/N Rats and B6C3F1
Mice (Gavage Study), May 1983, Draft (NIH Publication
No. 83-2519).
-------
VI-13
with the same systemic effects included in our analysis. The
health hazard index is one of several approaches identified by
EPA to assess the health risks of chemical mixtures.5
Table VI-1
PHILADEPHIA IEMP
CANCER UNIT RISK FACTORS (UPPER-BOUND) USED IN THE CONTROL-OPTIONS ANALYSIS
(1984 analysis)1
Pollutant
Benzene
Carbon Tetrachloride
Chloroform
1,2-Dichloroethane
Gasoline Vapors
Methylene Chloride
Perchloroethylene
1,2-Dichloropropane
Trichloroethylene
Exposure by Inhalation (uq/nr)"*
1.50 x
2.00 x
6.58 x
•6
l5-5
lO-J
49 x lO'I
,-7
7
1
1
1
3.98 x ID"6
x 10'
x 10-*
81 x ID'5
Exposure by Ingestion (uq/1)"1
6.89 x 10'7
3.69 x 10~6
2.00 x ID'6
6.58 x ID'6
N/A
1.80 x ID"8
1.00 x 10~6
1.81 x ID'6
5.38 x ID'7
Grouping Based on
EPA Criteria2
A
B2
B2
B2
B2
B2
B2
C
B2
N/A r Not applicable.
•'•The unit risk factors presented in this table were developed in 1984.
2EPA weight-of-evidence classifications: A - human carcinogen; B - probable carcinogen; C - possible carcinogen.
See Chapter V for a more detailed discussion.
Source: All cancer unit risk factors were developed by EPA'3 Carcinogen Assessment Group, except for 1,2-dichloro-
propane. RID calculated the unit risk factor for 1,2-dichloropropane using the linearized, multi-stage
model and data from a 1977 NTP study. This study is currently undergoing EPA review, thus the potency value
could change.
The health hazard index (HI) is based on the assumption of
dose additivity and is defined by the following equation:
HI
E2/AL2
where:
?i
exposure level to toxicant i
maximum-acceptable level for toxicant i
5U.S. EPA, Environmental Criteria and Assessment Office, Guide-
lines for the Health Risk Assessment of Chemical Mixtures, Final
Report, September 1985, pp. 12-14.
-------
VI-14
Table VI-2
PHILADELPHIA IEMP
NONCANCER HEALTH EFFECTS AND PRESUMED
HUMAN THRESHOLDS
FOR INHALATION AND
INGESTION
(1984 analysis)1
Compound
Chloroform
1 ,2-Oichloroethane
Carbon Tetrachloride
1 ,2-Oichloropropane
Trichloroethylene
Perchloroethylene
Benzene
Methylene ChlotiO*-
N.A. = Not available
Health Effect
Liver
Kidney
Reproductive
N.A.
Liver
Neurobehavioral
Kidney
Reproductive
Fetal developmental
effects
Liver
Kidney
Liver
Neurobehavioral
Kidney
Liver
Kidney
Fetal developmental
effects
Fetal developmental
effects
Liver
Fetal developmental
effects
•
1The threshold values presented in this table
^This value has been revised in 1986 to 41.3
This value has been revised in 1986 to 412.6
Source: Regulatory
R ID-Derived
Threshold
Inhalation
(uq/m*)
8.5
22.5
2.4
10
10
108
430
24.2
308
308
769
769
3,770
294
294
909
4.1a
650
650 .
RID
Threshold
Ingest ion
(uq/1)
85
225
24
100
100
1,080
4,300
242
3,080
3,080
7,690
7,690
37,700
2,940
2,940
9,090
41 .3b
6,500
6,500
Source
DeSalva et al. 1975;
Heywood et al. 1979
Heywood et al. 1979
Schwetz et al. 1974
1984 RID analysis
1984 RID analysis
EPA 1980
Adams et al. 1952
Schwetz et al. 1974
Basu et al. 1984;
NTP 1983
NTP 1983
1984 RID analysis
1984 RID analysis
Tucker 1982
1984 RID analysis
1984 RID analysis
Nelson et al. 1979
1984 RID analysts
1984 RID analysis
1984 RID analysis
were developed in 1984.
ug/m3.
ug/1.
Integration Division, Office of Policy, Planning and Evaluation, U.S. EPA.
-------
VI-15
We first calculate the ambient concentration-to-threshold ratio
(Ej/ALj) for each pollutant. We then sum across all pollutants
with the same systemic effect.
The health hazard index is a numerical indication of the
level of concern associated with exposures to complex mixtures of
pollutants in the environment. As the index approaches unity,
concern for the potential hazard of the chemical mixture
increases. If the index exceeds 1, the concern is the same as if
a no-effect threshold were exceeded by the same amount by an
individual compound. Similar to the no-effect threshold, the
hazard index cannot be used to formulate a mathematical predic-
tion of disease incidence.
Consideration of additional or new noncancer health effects
and the calculation of a health hazard index became available
toward the end of the control-options analysis. While we were
unable to fully take advantage of these and other advancements in
the Philadelphia IEMP, future geographic studies will benefit
greatly by having such methods available from the start of the
project.
\
Control Options
For each source category in the cost-effectiveness analysis,
we identified feasible control options. We also developed esti-
mates of costs, both capital and operating and maintenance, and
removal efficiencies. The removal efficiencies were offset by
any intermedia shifts resulting from the selected control strat-
egies. For example, the pollutant removal rates for a carbon
adsorption system were lowered somewhat if the spent carbon was
landfilled, thus resulting in potential VOC releases at the
landfill.
We identified current and potential new controls using
information provided by the Philadelphia Air Management Services,
the Philadelphia Water Department, the industrial facilities
themselves, and EPA documents, primarily Background Information
Documents from Office of Air Quality Planning and Standards and
Development Documents developed by the Office of Water. In addi-
tion, the results from the water monitoring program were used to
develop a control option at the Baxter plant that considered
changes in the schedule for withdrawing raw water from the
Delaware River.
The list of control strategies identified for the sources in
the control-options analysis can be found in Appendix H. Infor-
mation on the control costs, removal efficiencies, and equipment
life are detailed in Appendix I.
-------
VI-16
RESULTS Or THE ANALYSIS
In this section, we present the results from our cost-
effectiveness analysis. As mentioned earlier/ the impetus for
conducting the control-options analysis came from two complemen-
tary objectives. First, we wanted to develop and test an ana-
lytical framework for assessing control priorities across both
media and exposure pathways. Second, we wanted to generate
information useful to policymakers for risk management.
In general, we felt that we were successful in achieving
both objectives. We demonstrated that it was possible to develop
a model to identify cost-effective control strategies considering
complex interactions among numerous sources and exposure path-
ways. Furthermore, we could expand the flexibility of the cost-
effectiveness model with minor modifications. With respect to
our second goal, we also felt that we were able to provide
important insights from which policymakers can evaluate possible
control options and their costs and identify risk reduction
priorities.
It is important to recognize that our control-options analy-
sis and results require careful review and consideration before
action can be taken on the basis of our findings. First, our
risk estimates, for reasons stated previously, are likely to be
overpredictions of the actual risk. As a consequence, it is
likely we have overstated the cost-effectiveness (on a cost-per-
case reduced basis) of various control strategies. Second, the
pollutant-release data collected for our analysis are representa-
tive of conditions from 1983 to 1984. Since this time, and the
completion of our control-options analysis, the Philadelphia
Water Department has informed us that the chemical manufacturer
has, through solvent recovery, significantly reduced its loadings
of 1,2-dichloropropane and 1,2-dichloroethane to the NEWPCP. In
addition, construction of a pilot carbon adsorption system to
further reduce 1,2-dichloropropane discharges is also underway..
As a result, current air emissions of these compounds from the
NEWPCP are undoubtedly less than what we modeled in our control-
options analysis, and future air releases may be even smaller.
Finally, there have been changes in the unit risk factors used in
our analysis. Because of resource and scheduling constraints,
we were unable to rerun the control-options analysis to address
these changes. However, we did assess how the revised data could
alter our findings from the control-options analysis. This
discussion follows our presentation of the results of the anal-
ysis. We do not feel that the new information dramatically
affects our results.
In this section, we summarize our baseline cancer risk
estimates (aggregate incidence and MEI risk) for the 15 sources
and 9 pollutants considered in the cost-effectiveness analysis.
-------
VI-17
We also asse*!* the level of concern for noncarcinogenic health
effects at the maximum estimated ambient air and drinking water
concentrations under current controls. We describe our findings
on cost-effective control strategies to reduce aggregate excess
cancer incidence and risks to the MEI in the Philadelphia metro-
politan area. Finally, we attempt to provide a context for
evaluating these baseline risk calculations.
Baseline Risks Under Current Controls
Aggregate Excess Cancer Incidence
Air and Drinking Water. Table VI-3 shows our estimates of
the excess annual cancer incidence resulting from exposure to
the sources and pollutants examined in our analysis under current
control. The total annual cancer incidence from all sources in
air and drinking water is roughly 2.8. Clearly, the estimated
excess cancer incidence associated with ingestion of drinking
water (2.4 annual cancer cases, or 86 percent of the total
2.8 cancer cases) is significantly larger than the excess cancer
incidence attributable to inhalation of the nine volatile organic
substances in the ambient air (.4 annual cancer cases, or 14 per-
cent of the total 2.8 cases).
Drinking Water. The compound responsible for most of the
estimated excess cancer incidence in drinking water is chloroform
(97 percent of the total estimated excess cancer incidence for
drinking water), as shown in Table VI-4. The concentration of
chloroform in the finished drinking water at each treatment
facility (see Appendix E, Table F-3) is one to two orders of mag-
nitude higher than any of the other contaminant levels, inter-
estingly, chloroform's presence in finished drinking water in
Philadelphia (as well as other cities) is primarily a by-product
of disinfecting the water, and these concentrations are well
within EPA'a Maximum Contaminant Level (MCL) for total trihalo-
methanes o£^,OQ ug/1. For comparison, Table VI-5 lists the aver-
age chloro£&gK concentration levels for a subset of cities,
including 9fff£adelphia, with populations in excess of 500,000.
While some Cities on our list have lower chloroform concentra-
tions than Philadelphia, others have higher concentrations. For
example, at the time these data were compiled, Houston, Texas,
and Columbus, Ohio,- showed contaminant levels that greatly
exceeded Philadelphia's chloroform levels and the MCL for
trihalomethanes.
-------
VI-18
Table VI-J
PHILADELPHIA IEMP
PHASE II RESULTS INTENDED FOR POLICY DEVELOPMENT1
UPPER-SOUND ESTIMATES OF EXCESS ANNUAL CANCER INCIDENCE BY SOURCE
CURRENT CONTROL
(1984 analyaia)2
Estimated Excasa
Annual Cancer
(caaaa/yaar)
Air (point)
Pharmaceutical Manufacturer
Cheaical Manufacturer
Qareant Manufacturer
Plaatic Cabinet Manufacturer
Industrial Dry Cleaner
Refinery B
Refinery A
Northeaat WPCP3
Subtotal
Air (area)
Degree* ing \
Dry Cleaning
Other Induatrial Uaaga
Gaaoline Marketing (excluding aelf-aervice)
Sewer Volatilization (NEWPCP)
Delaware River (receiving NEWPCP effluent)
Subtotal
.214
Drinking Water
Baxter DWTP
Belaont DWTP
Queen Lane DWTP
Subtotal
Total4
Percentage
of Total
Caaea
.2
.2
.0
.0
.0
2.3
.2
3.2
•••"•^•B
6.3
1.7
2.3
.1
1.9
.7
.8
7.6
43.2
15.8
27.2
86.2
100.0
WPCP > Water Pollution Control Plant.
DMT* > Drinking Water Treatment Plant.
Notat Nuibara have 3 decimal place* not aa an indication of precision but to
identify aource contribution to the risks.
1THE UNIT RISK FACTORS USED IN THIS ANALYSIS ARE BASED ON CONSERVATIVE ASSUMPTIONS
THAT GENERALLY PRODUCE UPPER-BOUND ESTIMATES. BECAUSE OF LIMITATIONS IN DATA AND
METHODS IN SEVERAL AREAS OF THE ANALYSIS, SUCH AS EXPOSURE CALCULATIONS AND
POLLUTANT SELECTION, RISK ESTIMATES WERE CALCULATED AS AIDS TO POLICY DEVELOP.
MENT, NOT AS PREDICTIONS OF ACTUAL CANCER RISKS IN PHILADELPHIA. ACTUAL RISKS
HAY BE SIGNIFICANTLY LOWER; IN FACT, THEY COULD BE 2ERO. THE PROPER FUNCTION OF
THE ESTIMATES IS TO HELP LOCAL OFFICIALS SELECT AND EVALUATE ISSUES, SET
PRIORITIES, AND DEVELOP CONTROL STRATEGIES FOR THE TOPICS EXAMINED.
2The riak eatiaatsa preeented in this table were calculated using unit riak
factors fro* 1984.
3Recent reductions in discharges to the NEWPCP asy result in lower risk numbers
than praaented in this table.
*Colu»ns My not urn due to rounding.
3IEc, inc., Coat-Effectiveneaa Analysis of Strategies to Reduce Hunan Health Riak
*n_ Philadelphia. U.S. EPA, May 1985.
-------
IV
-------
IV. PHASE II MONITORING ACTIVITIES:
BENZENE, FORMALDEHYDE, LANDFILLS, AND USED OIL
Phase I of the Philadelphia project resulted in the identi-
fication and selection of a subset of issues for further analysis
in Phase II. We directed much of our Phase II activities toward
improving our exposure assessments. We grouped the study areas
into two broad categories: monitoring activities and control-
option analysis. The monitoring activities focused on four
issues: benzene in the ambient air (gasoline-related emissions
only); formaldehyde in the ambient air; air emissions from land-
fills; and emissions from burning used oil. The purpose of this
chapter is to describe the various sampling efforts initiated by
EPA and the study participants for the four Phase II monitoring
activities. Our control-option analysis is discussed in Chap-
ters V and VI.
We begin this chapter by summarizing the findings from the
Philadelphia monitoring programs for benzene and formaldehyde in
the ambient air, as well as the potential implications presented
by the measured^ concentrations for exposure and risk. We then
describe EPA's experience using Remote Optical Sensing of Emis-
sions (ROSE) to measure the release of volatile gases from three
landfills in New Jersey. We also report the findings from an
independent monitoring effort completed by the New Jersey Depart-
ment of Environmental Protection at one of the three landfills
tested using the ROSE system. Finally, we present the pollutant
concentrations found in a finite number of used oil samples that
EPA analyzed in conjunction with officials from Philadelphia, the
State of Pennsylvania, and the State of New Jersey.
BENZENE IN THE AMBIENT AIR
Benzene is a constituent in gasoline. It has been long
recognized by most health professionals that automobiles and
service stations contribute to the existence of ambient benzene
in all urban environments.
During Phase I, the Philadelphia Air Management Services
(AMS) expressed interest in exploring the contribution.from
service stations to ambient air benzene concentrations. As a
participant in this project, AMS conducted a short-term benzene
monitoring program in Phase II to characterize the presence of
benzene in the vicinity of gasoline service stations and heavy
traffic intersections. AMS also conducted additional air
-------
IV-2
sampling at a city-owned gasoline filling station to determine
benzene concentration in the breathing zone during refueling.
This section provides a brief summary of AMS' s sampling plans
(ambient and breathing zone), the results, and the implications
for exposure and risk presented by these observed ambient
concentrations.
Overview of the Sampling Plan
Traffic Intersection and
Service Station study
Site Selection. AMS established 12 permanent sampling
points along perpendicular axes emanating from the center of
intersections of Bustleton Avenue and Red Lion Road in Northeast
Philadelphia (see Figure IV-1). The site is characterized by
three gasoline stations, a large residential area, two apartment
complexes, and a shopping center. On any given day, samples were
taken from 9 of the 12 monitoring points, depending on the local
wind direction. Wind speed and direction data were obtained from
two meteorological stations at Northeast Airport and the Roadway
Truck Terminal.
Since traffic at an intersection may account for a signifi-
cant amount of the ambient benzene concentrations, AMS also
established monitoring sites at five other locations without
service stations in order to assess the contribution of vehicle
traffic to ambient concentrations. The five traffic intersec-
tions are identified in Table IV-1.
Two of the sites are located in jncongested intersections in
Northeast Philadelphia. Similar to the service station sampling
plan, 12 fixed sampling points were established along perpendi-
cular axes originating from the center of the intersection; how-
ever, only nine samples were taken at any one sampling excursion,
depending on wind direction. The latter three sites are situated
in the center city area. Because of high building density and
pedestrian traffic, samples were taken on only the four corners
and in the center of the boulevards.^-
A more detailed description of the sampling plan can be found
in: D. Ellis, and R. Obendorfer, Survey of Benzene Concentra-
tions in the Ambient Air, (Final), May 1984.
-------
IV-3
Figure IV-1
Philadelphia IEMP
A MS Sampling Sites at the Intersection of
Bustleton Avenue and Red Lion Road
• N3
P2
u
Sampling Site
Source: K. Ellis, and R. Obendorfer, Survey of Benzene
Concentrations in Ambient Air, (Final). May 1984.
-------
IV-4
Sampling Procedure. AMS took grab samples of approximately
five-minute duration at each of these traffic intersections
several times each week from February through April of 1984. AMS
sampled using a one-liter gas collection tube with septum sampl-
ing port, fitted with Teflon stopcocks at both ends. A total of
three tubes were used for each sampling site. Two tubes were
used to collect samples at predetermined flows, while the third
was set aside as a field blank. The samples were analyzed at
AMS's laboratory using a Photovac Gas Chromatograph.
Breathing Zone Study
Site Selection. In addition to the ambient analysis, AMS
also monitored at a city-owned gasoline filling station (with two
pumps) at the East Central Incinerator yard located at the inter-
section of Delaware Avenue and Spring Garden. The objective of
this sampling was to determine concentrations of benzene present
in the breathing zone, i.e., the area between the pump nozzle and
the refueler's face and head.
\
Sampling Procedure. AMS gathered samples during the refuel-
ing process using a portable organic vapor analyzer (OVA-108).
All samples were analyzed using GC in the AMS laboratory. OVA
readings were taken every 20 seconds during refueling; the OVA
sensor was placed at distances ranging from 4 inches to 2.5 feet
away from the nozzle.
Results
Traffic Intersection and
Service Station Study
Table IV-1 presents the range of average concentrations that
AMS observed for the monitoring stations at each of the six
sites. Table IV-1 also presents an average for the observed
range of average concentrations at each site. There is a fairly
wide range in measured average concentrations for each site,
especially for the Haldeman, Southampton, and Bustleton intersec-
tions. AMS found that the average benzene concentrations were
generally highest at the corners of the intersections and
decreased outward.
-------
IV-5
Table IV-1
PHILADELPHIA IEMP
AVERAGE AMBIENT AIR BENZENE CONCENTRATIONS
AMS MONITORING
Range of Average
Average
Range
for Observed
of Average
Concentrations*,* Concentrations
Site uq/m3 ppb
Broad and Vine (traffic intersection only) 11.2 - 35.7 3.5 - 11.2
16th and Walnut (traffic intersection only) 5.4-24.9 1.7- 7.8
16th and JX Boulevard (traffic intersection
only) 6.1 - 27.1 1.9 - 8.5
Haldeman and Red Lion (traffic intersection
only) tO - 18.2 ND - 5.7
Southampton and Roosevelt (traffic intersection
only) tt> - 18.5 N> - 5.8
Bus tie ton Ave. and Red Lion (service station
site) 1.1 - 38.3 0.35 - 12.0
ND s Not detected.
^Benzene (ppb) x 3.19 s benzene (ug/m3).
2A11 values have been rounded to the first decimal point.
AMS = Air Management Services.
Source: Kathryn Ellis and Raymond Obendorfer. Survey of 9tnte"« Concentrations in
Air Management Services, Philadelphia Department of Public Health, Final
ug/m3
22.3
14.0
16.9
5.4
6.7
12.1
Ambient
Report ,
ppb
7.0
4.4
5.3
1.7
2.1
3.8
Air,
May 1984.
The average concentration for tha tarvice station site
(Bustleton Avenue and Red Lion) is higher--roughly by a factor of
two—than,tn* average ambient levels at the Haldeman and
Southampton intersections, but is somewhat lower than the average
observed benzene concentrations at th« remaining traffic inter-
sections. These results may suggest that an overwhelming contri-
bution of benzene ambient levels probably results from mobile
sources. Unfortunately, because of the limited number of samples
taken, AMS was unable to determine the influence of traffic ver-
sus gasoline marketing on ambient benzene concentrations. As a
result, we cannot say with any certainty to what extent emissions
from gasoline marketing influence ambient benzene concentrations.
-------
IV-6
Breathing Zone Study
Table IV-2 summarizes the benzene concentrations measured
during 14 refueling episodes from April 16 to April 18, 1984.
The observed benzene concentrations are highly variable, ranging
from a low of 5 ug/m3 (1.5 ppb) to a high of more than
88,714 ug/m3 (>27,810 ppb). It is important to note, however,
that, because of the relatively small number of sampling events,
these reported values are a limited indication of benzene concen-
trations. However, these data provide useful information on the
potential benzene exposures to those people pumping gasoline.
The significance of the risks associated with periodic short-term
high exposure events is difficult to interpret because our risk
assessments are based on assumptions of continuous exposures over
70 years.
Table IV-2
PHILADEPHIA IEMP
AMBIENT AIR BENZENE CONCENTRATIONS:
DELAWARE AVENUE INCINERATOR YARD GAS STATION
AMS MONITORING
Sampling Day
4/16/84
4/17/84
4/18/84
Duration
of Tests
2 rain
1 min 5 sec
2 min 5 see
1 min 10 sec
1 min 25 sec
1 min 30 sec
1 min 20 sec
1 min 5 sec
1 min 2 sec
1 min
1 min
3 min 40 sec
1 min
1 min 30 sec
Distance
from Nozzle
(feet)
1.0
1.0
1.5
2.0
1.5
2.5
1.0
1.5
2.0
N.A.
Benzene
Concentration^,
ppb
276
>27,810
29
32
4,100
3,568
484
612
500
7.8
6,180
140
5.8
1.5
880
>88,714
93
102
13,079
11,382
1,544
1,952
1,595
25
19,714
447
19
5
N.A. = Not available
^Benzene (ppb) x 3.19 = benzene (uo/ra^).
^Values converted from ppb to ug/ra* have been rounded to whole
numbers.
Source: Kathryn Ellis and Raymond Obendorfer, Survey of Benzene
Concentrationa in Ambient Air. Air Management Services,
Philadelphia Department of Health, Final Report,
May 1984.
-------
IV-7
Implications for Exposure and Risk
Traffic intersection and
Service Station Study
We evaluated the results from this activity in two ways: by
comparing the observed concentrations to the AMS ambient air
guideline for benzene, and by calculating an average lifetime
individual risk using standard EPA assumptions discussed in Chap-
ter II. We found that the average concentration at each site did
not exceed the AMS annual ambient air guideline of 24 ppb
(77 ug/m3) for benzene. Furthermore, we did not see excursions
above the guideline for any single sampling episode. On the
basis of these data, it did not appear likely that the AMS
benzene guidelines would be exceeded in any residential area in
the vicinity of service stations or urban traffic centers.
We calculated rough approximations of average lifetime
(70-year) individual risk for people living near busy intersec-
tions using the average estimated concentrations for each site
and the benzene unit risk factor developed by EPA's CAG,
6.9 x 10~6 (ug/m3)"1 in 1984 (the time when this work was
completed). The estimated average lifetime individual risk
(upper-bound) from exposure to average benzene concentrations at
each intersection ranged from 3.7 x 10"5 to almost 1.5 x 10~4
(see Table IV-3). This represents a range of risk from almost
four chances in 100,000 to almost two chances in 10,000. We
should emphasize, however, that these values do not represent
long-term averages since only short-term ambient data were
collected.
We did not estimate risks from inhalation of gasoline
vapors. However, EPA's exposure and risk assessment for the
gasoline marketing industry suggests that plausible upper-bound
risks from inhaling gasoline vapors are, in general, about an
order of magnitude lower than those predicted for benzene.2 we
also did not assess the potential for noncancer health effects at
this point in time.
2U.S. EPA, Evaluation of Air Pollution Regulatory Strategies for
Gasoline Marketing Industry, EPA 450/3-84-012a, July 1984.
-------
IV-8
Table IV-3
PHILADELPHIA IEMP
PHASE II RESULTS INTENDED FOR POLICY DEVELOPMENT1
PRELIMINARY APPROXIMATIONS OF AVERAGE
LIFETIME INDIVIDUAL UPPER-BOUND RISK FROM
EXPOSURE TO BENZENE CONCENTRATIONS
(1984 analysis)2
Site
Broad and Vine
16th and Walnut
16th and JFK Boulevard
Haldenan and Red Lion
Southampton and Roosevelt
Buatleton Avenue and Red Lion
Average
Ambient
Concentration
(ug/m3)3
22.3
14.0
16.9
5.4
6.7
12.1
Average
Lifetime
Individual
Risk
1.5 x
9.7 x
1.2 x
3.7 x
4.6 x
10-
ID'3
ID'5
8.3 x ID'5
1THE UNIT RISK FACTORS USED IN THIS ANALYSIS ARE BASED ON
CONSERVATIVE ASSUMPTIONS THAT GENERALLY PRODUCE UPPER-BOUND
ESTIMATES. BECAUSE OF LIMITATIONS IN DATA AND METHODS IN
SEVERAL AREAS OF THE ANALYSIS, SUCH AS EXPOSURE CALCULA-
TIONS AND POLLUTANT SELECTION, RISK ESTIMATES WERE CALCU-
LATED AS AIDS TO POLICY DEVELOPMENT, NOT AS PREDICTIONS OF
ACTUAL CANCER RISKS IN PHILADELPHIA. ACTUAL RISKS MAY BE
SIGNIFICANTLY LOWER; IN FACT, THEY COULD BE ZERO. THE
PROPER FUNCTION OF THE ESTIMATES IS TO HELP LOCAL OFFICIALS
SELECT AND EVALUATE ISSUES, SET PRIORITIES, AND DEVELOP
CONTROL STRATEGIES FOR THE TOPICS EXAMINED.
The risk calculations presented in this table were derived
in 1984 using the available unit risk value for benzene at
that time—6.9 x 10~6 (ug/ra3)-1, EPA level of evidence:
Group A.
^These values represent the mean of the average concentration
for each monitoring location at each site.
Ambient Data Source: Philadelphia Air Management Services.
Breathing Zone Study
Based on the breathing zone measured data, AMS found evi-
dence that the AMS guideline for benzene (24 ppb) and the occupa-
tional standard for gasoline vapor in the workplace (300 ppm;
8-hour time weighted average) may be exceeded for service station
attendants and others at high-volume filling stations. for many
of the sampling events/ the AMS benzene ambient air guideline was
exceeded.
-------
IV-9
1
I
I
I
I
I
i
I
I
I
I
1
Recent Changes in the Benzene
Unit Risk Factor
Cancer. Since the collection of the benzene monitoring data
in 1984, CAG has revised the-benzene potency value upward to
reflect the findings of more recent animal studies. The current
(1986) benzene unit risk factor for inhalation is 8.3 x 10~6
(ug/nH)-l, an increase of roughly 20 percent from the previous
estimate.
We recalculated our average lifetime individual risk esti-
mates for the traffic intersection and service station ambient
monitoring data (see Table IV-4) and found no significant
changes.
Table IV-4
PHILADELPHIA IEMP
PHASE II RESULTS INTENDED FOR POLICY DEVELOPMENT1
PRELIMINARY APPROXIMATIONS OF AVERAGE
LIFETIME INDIVIDUAL UPPER-BOUND RISK FROM
EXPOSURE TO BENZENE CONCENTRATIONS
(1986 revised)2
Site
Broad and Vine
16th and Walnut
16th and JFK Boulevard
Haldenan and Red Lion
Southampton and Roosevelt
Bustleton Avenue and Red Lion
Average
Ambient
Concentration
(uq/m3)3.*
22.3
14.0
16.9
5.4
6,7
12.1
Average
Lifetime
Individual
Risk
1.9 x
1.2 x
1.4 x
4.5 x
5.6 x
10-*
10-5
10
,-5
1.0 x 10-4
1THE UNIT RISK FACTORS USED IN THIS ANALYSIS ARE BASED ON
CONSERVATIVE ASSUMPTIONS THAT GENERALLY PRODUCE UPPER-BOUND
ESTIMATES. BECAUSE OF LIMITATIONS IN DATA AND METHODS IN
SEVERAL AREAS OF THE ANALYSIS, SUCH AS EXPOSURE CALCULATIONS
AND POLLUTANT SELECTION, RISK ESTIMATES WERE CALCULATED AS
AIDS TO POLICY DEVELOPMENT, NOT AS PREDICTIONS OF ACTUAL
CANCER RISKS IN PHILADELPHIA. ACTUAL RISKS MAY BE SIGNIFI-
CANTLY LOWER; IN FACT, THEY COULD BE ZERO. THE PROPER FUNC-
TION OF THE ESTIMATES IS TO HELP LOCAL OFFICIALS SELECT AND
EVALUATE ISSUES, SET PRIORITIES, AND DEVELOP CONTROL
STRATEGIES FOR THE TOPICS EXAMINED.
2The risk calculations presented in this table were derived
using the updated benzene unit risk value (January 1986)—
8.3 x 10"6 (ug/ra3)"!. EPA level of evidence for benzene:
Group A.
JThese values represent the mean of the average concentration
for each monitoring location at each site.
^Ambient air data from Philadelphia Air Management Services.
-------
IV-10
Noncancer Effects. Another factor to consider is the poten-
tial for noncancer health effects at the measured benzene ambient
levels. As discussed earlier in this report, the Philadelphia
project, our first project, concentrated primarily on carcinogens
in Phase II, which seemed a reasonable decision at the time. In
our more recent IBMPs, however, we have expanded our health
effects focus to consider both cancer and noncancer impacts. As
part of our most recent analysis of the project data, we have
included noncancer health effects. The RID-derived threshold for
noncancer blood effects, which uses a 1980 study by Snyder
et al., is currently under review by scientists within and out-
side of EPA; consequently, we could not assess the increased risk
of noncancer effects at the measured ambient air benzene levels.
FORMALDEHYDE IN THE AMBIENT AIR
We developed a point-source emissions profile for formalde-
hyde in Phase I from the AMS emissions inventory data, but
assessing the exposures and risks posed by formaldehyde proved to
be difficult. We were unable to predict from our models ambient
concentrations, and thus exposure, because of formaldehyde forma-
tion in the atmosphere through photo-oxidation of volatile
organic compounds.3 AMS was concerned about the potential for
high formaldehyde concentrations and exposures and was therefore
interested in better assessing ambient conditions. As a partici-
pant in this project, AMS conducted a. short-term formaldehyde
monitoring program in Philadelphia. This section describes the
monitoring activities and presents the results. It also dis-
cusses the potential impact of the measured ambient levels on
exposure risks to human health.
i
Overview of the Sampling Plan
AMS sited monitoring stations at the Naval Hospital,
Lardner's Point, the Northeast Airport, and the AMS laboratory
(see Figure IV-2). The latter two sites are in EPA1s State and
Local Air Monitoring Stations (SLAMS) network, and AMS routinely
monitors criteria pollutants at each of these locations. In
addition, the first three sites were also used by EPA for its VOC
ambient air monitoring program discussed later in this report.
^office of Policy Analysis, Office of Air and Radiation, The Air
Toxics Problem in the United States; An Analysis of Cancer
Risks for Selected Pollutants, U.S. EPA, May 1985, p. viii.
-------
I
I
I
I
I
i
i
i
i
i
i
t
i
i
i
i
i
i
i
IV-ll
Rgure IV-2
Philadelphia IEMP
Map of Formaldehyde Monitoring Sites
PHILADELPHIA
COUNTY
1. Naval Hospital
2. Lardner's Point
3. Philadelphia Airport
4. AMS Laboratory
-------
IV-12
AMS initiated the monitoring in early December of 1983 and
completed the sampling at each of these four sites at the end of
February in 1984. They collected 24-hour integrated samples
every six days. Although AMS decided to continue sampling at the
AMS laboratory site for its own use, the results in this section
are based only on the work completed from December through
February.
AMS collected air samples by using an impinger train with
water as the trapping solution. They analyzed the samples by the
chromotropic acid spectrophotometric method developed by the
National Institute of Occupational Safety and Health. This
method was developed primarily for monitoring workplace exposure
where concentrations are higher than those generally found in the
ambient environment. Nonetheless/ this analytic technique is
also commonly used in ambient settings. There is, however, con-
siderable imprecision at low concentrations. Detection limits
are inversely proportional to duration of sampling. Thus a
sampling time of 24 hours has a detection limitation of about
2.5 ppb (3.1 ug/m3).
Results x
Table IV-5 shows the average ambient concentrations of for-
maldehyde at each of the four monitoring stations. The average
ambient levels range from 2.8 to 3.7 ppb. The relatively uniform
average concentrations at the four locations suggest that formal-
dehyde is probably present in the air as a result of atmospheric
transformation, rather than from point source releases. The
monitoring results could also reflect fairly uniform formaldehyde
releases by area combustion sources, particularly automobiles.
The measured levels are lower than the formaldehyde concen-
trations found in many other urban ar*ts across the nation.
Singh documented average ambient concentrations at six major
cities as follows:4
• St. Louis—11.3 ppb
• Denver—12.3 ppb
• Riverside, CA—19.0 ppb
H. B. Singh, L. J. Salas, R. Stiles, and H. Shigeishi, Measure-
ments of Hazardous Organic Chemicals in the Ambient Atmosphere,
U.S. EPA, Office of Research and Development, EPA 600/53-83-002,
1983.
-------
IV-13
Table IV-5
PHILADELPHIA IEMP
AVERAGE AMBIENT AIR FORMALDEHYDE CONCENTRATIONS:
AMS MONITORING DATA1
(December 1983 to February )984)
Site
Lardner's Point
12/08/83
12/14/83
12/26/83
01/19/84
01/25/84
Mean
Northeast Airport
12/02/83
12/26/83
01/13/84
01/19/84
01/25/84
02/06/84
02/24/84
Mean
AMS laboratory
12/02/83
12/08/83
12/14/83
12/26/83
01/13/84
01/19/84
01/25/84
02/12/84
02/18/84
02/24/84
Mean
Naval Hospital
01/19/84
01/25/84
Mean
Ambient Concentration
ppb
2.2
2.1
3.1
3.2
7.2
3.6
4.0
3.1
2.8
2.2
4.7
2.7
3.1
3.2
2.9
4.9
2.6
3.2
3.3
2.7
7.0
4.2
3.2
3.0
3.7
2.3
3,2
2.8
1.8
1.6
2.4
2.5
5,7
2.8
3.2
2.5
2.3
2.1
3.8
2.2
2.5
2.7
2.3
3.9
2.0
2.5
2.6
2.2
5.6
3.3
2.5
2.4
2.9
1.8
2.7
2.3
^Missing sampling episodes represent either
days on which AMS decided not to sample (i.e.,
rainy days) or data that were deleted from the
data set because of technical difficulties.
Source: Philadelphia Air Management Services.
-------
IV-14
• Staten Island, NY—14.3 ppb
• Pittsburgh—20.6 ppb
• Chicago—12.8 ppb
An exposure assessment performed for EPA's Office of Toxic
Substances also documented relatively high ambient air concentra-
tions at numerous sites throughout the country. In the Los
Angeles basin, which is the most extensively monitored area, for-
maldehyde levels were typically found to be between 10 to 30 ppb.
At several other urban locations, concentrations averaged about
5 ppb. In rural areas, where concentrations are generally much
lower, formaldehyde levels were found to range between the back-
ground atmospheric level of 0.4 ppb and the urban average concen-
tration of 5 ppb.5
AMS has expressed some reservations about the analytical
method employed in the monitoring program and believes that the
analytical findings may understate true formaldehyde concentra-
tions. A recent study indicates that the chromotropic acid tech-
nique is subject to negative interference (i.e., biased toward
reporting low values) when phenol is present at even a fraction
of the formaldehyde concentration.6 Since phenol is likely to be
present in Philadelphia's atmosphere—there are several point
source emitters of this compound in the AMS emission inventory—
it is not unreasonable to expect that the formaldehyde monitoring
results could be lower than actual ambient levels.
Implications for Exposure and Risk
We determined in two ways whether exposures at the AMS
measured formaldehyde ambient levels were high enough to warrant
further analysis: (1) by comparing the measured ambient concen-
trations to the AMS ambient air quality guideline for formalde-
hyde; and (2) by calculating rough approximations of average
lifetime individual cancer risks using EPA standard estimation
techniques.
5Versar, Inc., Exposure Assessment for Formaldehyde U.S. EPA,
Office of Toxic Substances, 1982.
^S. E. Knox, and S. S. Quettee, "Phenol Interference in the
Mercury-Free Parao Saniline and the Chromotropic Acid Method for
Formaldehyde," American Industrial Hygiene Association Journal,
Vol. 45, p. 325.
-------
IV-15
The AMS ambient air guideline for formaldehyde is 4.8 ppb
(6 ug/m3). When we compared the AMS guideline with the observed
average ambient concentrations, we found that the average levels
were below the AMS threshold for action. When we considered the
individual sampling events at each point, however, we found a few
days on which ambient concentrations exceeded the guideline.
We calculated a rough approximation of average lifetime
individual risks at each of the four locations using the mean
concentration levels and plausible upper-bound unit risk factors
based on the multistage model and EPA exposure assumptions. The
CAG unit risk factor for formaldehyde was 6.1 x 10~6 (ug/m3)"1 in
1984 when this analysis was completed. Our results are shown in
Table IV-6.
Table IV-6
PHILADELPHIA IEMP
PHASE II RESULTS INTENDED FOR POLICY DEVELOPMENT1
PRELIMINARY APPROXIMATIONS OF AVERAGE
LIFETIME INDIVIDUAL UPPER-SOUND RISK
FROM EXPOSURE TO FORMALDEHYDE CONCENTRATIONS
(1984 analysis)2
Site (mean concentration)
Lardner's Point (3.6 ug/m3)
Northeast Airport (3.2 ug/m3)
AMS Laboratory (3.7 ug/m3)
Naval Hospital (2.8 ug/m3)
Average Lifetime
Individual Risk
2.2 x ID-5
2.0 x 1Q-5
2.3 x 10-5
1.7 x ID'5
*THE UNIT RISK FACTORS USED IN THIS ANALYSIS ARE BASED ON
CONSERVATIVE ASSUMPTIONS THAT GENERALLY PRODUCE UPPER-BOUND
ESTIMATES. BECAUSE OF LIMITATIONS IN DATA AND METHODS IN
SEVERAL AREAS OF THE ANALYSIS, SUCH AS EXPOSURE CALCULATIONS
AND POLLUTANT SELECTION, RISK ESTIMATES WERE CALCULATED AS
AIDS TO POLICY DEVELOPMENT, NOT AS PREDICTIONS OF ACTUAL
CANCER RISKS IN PHILADELPHIA. ACTUAL RISKS MAY BE SIGNIFI-
CANTLY LOWER; IN FACT, THEY COULD BE ZERO. THE PROPER FUNC-
TION OF THE ESTIMATES IS TO HELP LOCAL OFFICIALS SELECT AND
EVALUATE ISSUES, SET PRIORITIES, AND DEVELOP CONTROL
STRATEGIES FOR THE TOPICS EXAMINED.
2The risk estimates in this table were derived using the
formaldehyde unit risk factor for inhalation available from
CAG in 1984: 6.1 x KT6 (ug/ra3)-1, EPA level of evidence:
Group B-2.
Source: Philadelphia Air Management Services data.
-------
IV-16
It is important to note that the general dose-response rela-
tionship from which the GAG unit risk factor was developed is
very controversial. Several toxicologists argue that there is a
threshold for formaldehyde carcinogenesis, asserting that there
is a metabolic pathway that can neutralize the cancer-causing
effects. Others state that there is no proof that a threshold
exists. Furthermore, the results from several epidemiological
studies suggest that formaldehyde may not even be a human car-
cinogen. The debate is still an issue in EPA's current proposal
to regulate formaldehyde under the Toxic Substances Control Act.
Recent Changes in the Formaldehyde
Unit Risk Factor
Similar to benzene, the unit risk factor for formaldehyde
has been revised very recently to reflect new peer-reviewed
animal data. EPA developed two upper-bound unit risk factors
(inhalation) for formaldehyde based on these data: 1.3 x 10~5
(ug/m3)~l for malignant tumors only, and 1.8 x 10~4 (ug/m3)-l for
total tumors.7 The dose-response relationship was calculated
separately for malignant and total (including benign) tumors
because of an overwhelmingly large number of benign tumors.
Because of the potential for a benign tumor to progress to a
malignant tumor, it is generally standard GAG policy, as put
forth in the proposed risk assessment guidelines, to calculate
dose-response relationships using total tumors.
Table IV-7 presents our recalculated average lifetime indi-
vidual risks using the revised cancer unit risk factors for
formaldehyde. While our risk estimates do not change dramatical-
ly when using the malignant-only potency value (the lifetime
individual risks remain around 10~5), the average lifetime indi-
vidual risks using the total tumor unit risk factor are more than
an order of magnitude higher (roughly 10~4). We could not assess
the potential for noncancer health effects: the reference dose
for formaldehyde is currently undergoing internal review. AMS
has continued to monitor at its laboratory site with no signifi-
cant change in the measured average ambient formaldehyde level
(about 3 ppb).
Memorandum from Nancy Pate, OAQPS, to Richard Hefter, OTS, "Unit
Risk Estimation for Formaldehyde," March 20, 1986. I
-------
I
I
I
I
I
I
i
i
i
i
i
i
i
i
i
i
i
i
i
IV-17
Table IV-7
PHILADELPHIA IEMP
PHASE II RESULTS INTENDED FOR POLICY DEVELOPMENT*
PRELIMINARY APPROXIMATIONS OF AVERAGE
LIFETIME INDIVIDUAL UPPER-BOUND RISKS
FROM EXPOSURE TO FORMALDEHYDE CONCENTRATIONS
(1986 revised)2
Average Lifetime
Individual Riak
Site («ean concentration)3
Lardner'a Point (3.6 ug/n3)
Northeast Airport (3.2 ug/a3)
AMS Laboratory (3.7 ug/a3)
Naval Hoapital (2.8 ug/*3)
Hill grant
10-5
Total
4.7 x
4.2 x 10~3
n-5
4.8 x
3.6 x
-5
10-
10-5
6.5 x
5.8 x
6.7 x
5.0 x
10"
10-
10-
10-
1THE UNIT RISK FACTORS USED IN THIS ANALYSIS ARE BASED ON
CONSERVATIVE ASSUMPTIONS THAT GENERALLY PRODUCE UPPER-BOUND
ESTIMATES. BECAUSE OF LIMITATIONS IN DATA AND METHODS IN
SEVERAL AREAS OF THE ANALYSIS, SUCH AS EXPOSURE CALCULATIONS
AND POLLUTANT SELECTION, RISK ESHNA'tS <« CALCULATED AS
AIDS TO POLICY DEVELOPMENT, NOT AS PREDICTIONS OF ACTUAL
CANCER RISKS IN PHILADELPHIA. ACTUAL »IS«S NAY BE SIGNIFI-
CANTLY LOWER; IN FACT, THEY COULD BE S»0. THE PROPER FUNC-
TION OF THE ESTIMATES IS TO HELP LOCAL QflCIALS SELECT AND
EVALUATE ISSUES, SET PRIORITIES. *0 QtrtLOP CONTROL
STRATEGIES FOR THE TOPICS EXAMINED.
2The risk estimates in this tsble or* ieri>e4 using updated-
forMldehyde unit risk factors 'or .nxeiationi 1.3 x 10
(ug/*3)-1 for Malignant tuaors end 1.1 • 13~* lug/*3)"1 for
total tuanrs, EPA level of evident*' i/»* >-2.
3A«bient data fro* the Philadelphia ».r "e^eaeeent Services.
AIR EMISSIONS FROM LANDFILLS
Environmental studies of land disposal facilities, such as
landfills and surface impoundments, have generally emphasized
ground water contamination. However, toxic organic chemicals
disposed of in hazardous waste landfills can also volatilize over
time, thereby posing a potential threat to human health and the
environment.
While Phase I'did not identify air emissions from landfills
as a significant source of exposure for Philadelphia residents,
both the New Jersey Department of Environmental Protection
(NJDEP) and EPA's Office of Research and Development expressed
interest in this topic for several reasons. First, the issue of
air emissions from hazardous waste facilities had not been
studied in any great detail, thus very little ambient data were
-------
IV-18
available. Second, volatilization from landfills may be a signi-
ficant source of toxic chemical releases and exposure. Finally,
EPA wanted to explore the feasibility of using alternative moni-
toring techniques in future research activities. For these
reasons, the study participants agreed to perform limited moni-
toring to evaluate air emissions from selected landfills as a
Phase II study topic.
This section describes the Remote Optical Sensing of Emis-
sions (ROSE) system, the sampling plan, and the results. Since
the field operation of the ROSE system was disappointing, we also
report the findings from an independent monitoring program con-
ducted by.the NJDEP at one of the landfills also monitored by
EPA. Because of the unexpected limitations of the EPA effort, we
felt that it was important to summarize some information that
would provide perspective on the relative significance of air
emissions from landfills.
Overview of the ROSE System
Monitoring Plan and Results
The ROSE System
The ROSE system was originally developed at EPA's Office of
Research and Development in 1972 and was revised in 1977.8 The
system uses a long pathlength infrared monitor to measure air
pollutants that are absorbed in the infrared portion of the
spectrum. A pathlength of 1 km is typically achieved by having
the light source separated from the sensor by this spacing or the
line-of-sight.
The ROSE system has been demonstrated to measure in the ppb
range, and we therefore believed it to- be well-suited for our
needs. The ROSE system was also attractive because the infrared
spectroscopic techniques offer the potential of carrying out
analysis-of. source emissions without sample collection and
analysis and without prior knowledge of the specific compounds
emitted. , •
Site Selection
In conjunction with the NJDEP, we selected three landfills
to monitor using the ROSE system. We had three criteria. First,
^William F. Hergert, "Analysis of Gaseous Air Pollutants Using a
Mobile "FTIR System," American Laboratory, December 1982.
-------
IV-19
we wanted to achieve some variety among the sites. In particu-
lar, we wanted to sample at both sanitary and hazardous waste
landfills, and at active and closed facilities. We decided to
monitor only the closed sites because the dust and exhaust from
vehicles operated at active facilities would interfere with the
measurements. However, we recognized that an inactive site may
exhibit lower release rates than an operating facility.
Second, the layout of the site had to be suitable for the
ROSE system. Since the ROSE system was maintained in a mobile
van, the landfill had to be bounded or crossed by roads that the
van could traverse. In addition, there had to be a clear line-
of-sight from one side of the landfill to other; the spectrometer
would be placed on one end of the line-of-sight and the infrared
radiation source at the other. Finally, for optimal conditions,
the length between the two sources would have to be 500 to
1,000 meters, and downwind from the landfill emissions. The
third criterion was that the landfill should be close to or in
the eight-county study area (there were no hazardous waste or
sanitary landfills in Philadelphia County).
Based on these criteria, we selected the following sites:
• Lipari Landfill—Pitman, New Jersey. The Lipari site
was currently undergoing Superfund remedial action at
the time of our study (November 1983).
• Kramer Landfill—Mantua Township, Gloucester County.
Kramer was an inactive municipal landfill (about
60 acres) that had become a Superfund site because of
disposals of large amounts of chemical waste. The
NJDEP also sampled at this site the same week we con-
ducted the ROSE monitoring; however, a different moni-
toring technique was employed.
• Big Hill Landfill—(BEMS-Burlington Environmental
Management Services), Southampton Township, Burlington
County. BEMS was a closed sanitary landfill (about
60 acres) that had accepted only household wastes for a
period of eight to nine years.
Results
The ROSE system measured ambient concentrations at each of
the three sites during the week of October 31, 1983, to
November 4, 1983. The results collected for the three sites
were, unfortunately, very limited. The Lipari site data, for
example, were not valid because of intermittent malfunctions in
-------
IV-20
the electrical generator. Furthermore, at the Kramer and BEMS
sites, calibration problems precluded an evaluation of data for
low concentration chemicals, such as benzene, toluene, trichloro-
ethylene, and other hazardous substances. We were only able to
measure concentrations of carbon monoxide and methane, which were
not pollutants of major concern for the study.
Many of the problems encountered using the ROSE system dur-
ing the sampling period, e.g., electrical malfunctions and
improper calibration of the equipment, could be easily corrected
in future monitoring efforts. However, our other observations
during this brief monitoring exercise suggested that the ROSE
system may only have limited usefulness in the direct vicinity of
hazardous waste landfills. Variations in the wind directions and
difficulties in locating an appropriate line-of-sight across the
landfill appear to be the major impediments to a successful
application of the ROSE system at landfills.
Overview of the NJDEP Landfill
Monitoring Plan and Results'
^—^^^^HB^^—M^^^^.^^WW^M^B.^^^^H^^^
Given the poor performance of the ROSE system and our
inability to formulate any conclusions about hazardous air emis-
sions from landfills based on the data set, we are presenting the
results from an independent monitoring effort conducted by the
NJDEP at one of the same landfills that EPA tested.
Sampling at this landfill was part of a larger NJDEP program
to measure ambient VOC concentrations around six abandoned hazar-
dous waste sites and one sanitary landfill in New Jersey.9
Sampling Plan
The sampling period extended from September 1983 to early
January 1984. Five monitors were located at each of the land-
fills. Where possible, monitors were situated upwind and down-
wind. In addition, monitors were located near populated areas,
as well as specific sources at the site, e.g., leachate pools.
NJDEP took 24-hour integrated samples over three consecutive days
at most of the site. All samples were collected on Tenax-GC and
analyzed using GC/MS.
9LaRegina, J., Bozzelli, Joseph W., Harkov, R., Gianti, S.,
"Volatile Organic Compounds at Hazardous waste Sites and a
Sanitary Landfill in New Jersey," Environmental Progress,
Vol. 5, No. 1, February 1986, pp. 18-27.
-------
IV-21
Results
The observed "field blank-subtracted" concentrations for the
landfill are summarized in Table IV-8. The data shown in the
table are the means of the ambient concentrations measured at all
five points above and near the landfills; values measured upwind,
downwind, and directly above the landfill are all averaged
together.
Despite some limitations with the sampling, we can make the
following observations. First, a fairly large number of toxic
pollutants were measured. Second, concentrations of these com-
pounds suggest that VOC levels around landfills can be above
background urban VOC ambient conditions.
Table IV-S
PHILADELPHIA IEMP
AVERAGE CONCENTRATIONS OF VOLATILE ORGANIC
COMPOUNDS AT NEW JERSEY LANDFILL
Field Blank-Subtracted
Concentrations:
Compound
Vinylidene chloride
Methylene chloride
Chloroform
1,2-dichloroethane
1,1,1-trichloroethane
Benzene
Carbon tetrachloride
Trichloroethylene
Oioxane
1,1,2-trichloroethane
Toluene
1,2-dibromoethane
Perchloroethylene
Chlorobenzene
Ethyl benzene
n,p-xylene
o-xylene
Styrene
1, L, 2,2-tetrachloroethane
o-chlorotoluene
p-chlorotoluene
p-dichlorobenzene
o-dichlorobenzene
Nitrobenzene
Naphthalene
ppb
35.8
10.9
0.8
0.4
2.3
6.9
0.1
2.4
0.0
1.2
43.3
0.5
1.3
o.a
3.7
7.2
2.3
1.5
O.S
0.5
0.4
0.5
0.9
0.5
0.3
144
38.4
4.1
1.5
12.9
22.3
0.8
13.0
0.0
6.8
185.0
3.6
12.5
3.7
16.5
32.1
10.3
6.6
5.4
2.9
2.1
3.3
5.2
2.8
1.6
Source: J. LaRegina et al.. "Volatile Organic
Compounds at Hazardous Waste Sites and
a Sanitary Land Fill in New Jersey,"
Environmental Progress. Vol. j, No. 1,
February 1936.
-------
IV-2 2
COMBUSTION OF USED OIL
The burning of contaminated fuel oils, also known as usecl
oil, can result in air emissions of organic and inorganic toxic
pollutants, thereby increasing the potential for human health
risks. State and local study participants identified used oil as
an issue of concern during the Phase I screening process. As a
result, we decided to assess used oil in the Philadelphia metro-
politan area as a potential source of air toxic emissions.
Given the paucity of information on contaminant concentrs.-
tions in used-oil, we elected to conduct a limited sampling pro-
gram with officials from the State of Pennsylvania, Philadelphia,
and the NJDEP. We hoped to use the results from this effort to
characterize the contaminants found in the used oil samples, as
well as the range of concentrations observed. In addition, we;
hoped to compare the measured concentrations to the EPA proposed
burning rules (now final) which would limit the concentrations of
various constituents that could be burned as fuel.
Initially, we had planned to use the results from our
sampling program to perform a preliminary exposure and risk
assessment for the Philadelphia area. The method of sample
collection used by the participating jurisdictions, however, did
not allow for a statistically based selection of sources burning
used oil in the Philadelphia area. When we completed monitoring,
we felt that many of the underlying assumptions (for example,
regarding sources burning used oil and in what amounts) and thus
the emission releases were too uncertain to evaluate exposure.
Overview of the Sampling Plan
Because of the limited budget for this activity, we were
able to collect and analyze only 45 -i*ed oil samples. On the
basis of recommendations of the study participants, we collect.ed
from distributors in New Jersey and us«rs in Pennsylvania. Field
crews from Pennsylvania, Philadelphia, and New Jersey each
gathered one-third of the samples. Th« sampling occurred between
July and August 1983.
The samples were analyzed by EPA's Central Regional Labora-
tory and GCA Corporation. Because of budget constraints, we had
to strike a balance between a cursory analysis of a larger number
of samples, and a more detailed analysis of fewer samples. It
was decided to subject only a subset of the 45 samples that
appeared to be most contaminated, and thus the most interesting,
through the more rigorous tests. We analyzed six parameters
using the tests described below:
-------
IV-2 3
• Heat Value
—Purpose: Measures suitability for use as a fuel
—Method: Calorimeter (ASTM D240)
—Fuel Contamination Level: Less than or equal to
5,000 to 10,000 BTU/pound
• Sulfur
—Purpose: There is some evidence that high sulfur
concentrations may be correlated with the existence
of other contaminants.
—Method: Energy-dispersive x-ray fluorescence
—Fuel Contamination Level: Greater than state and
local limits on sulfur concentrations in fuel oils
• Total Chlorine
—Purpose: Total chlorine is a rough indicator of
chlorinated hydrocarbon content.
—Method: Energy-dispersive x-ray fluorescence
—Fuel Contamination Level: Total chlorine concentra-
tions more than .1 percent may indicate that contam-
ination is occurring.
• Metals
—Purpose: Sample for arsenic, barium, cadmium,
nickel, lead, and chromium. These metals are not
generally found in virgin heating oils at levels high
enough to warrant concern; therefore, high levels may
signify possible contamination.
—Method: Energy-dispersive x-ray fluorescence
—Fuel Contamination Level: Lead at more than 10 ppm
in #6 residual fuel oil and 1 to 2 ppm in #2 fuel
oil; the presence of the other metals at levels above
the detection limits.
• PCBs
—Purpose: PCBs are toxic and very stable; levels
greater than 50 ppm are illegal.
-------
IV-2 4
—Method: Gas chromatography
—Fuel Contamination Level: Concentrations greater
than 50 ppm
Solvents
—Purpose: Monitor for chlorinated solvents as
1,1,1-trichloroethane, trichloroethylene, methylene
chloride, and other solvents, such as toluene. The
presence of such compounds indicates contamination.
—Method: Gas chromatography/mass spectroscopy
—Fuel Contamination Level: Concentration levels
greater than the detection limits—generally 10 ppm,
although in some cases 100 to 200 ppm of solvents may
be acceptable depending on the specific chemical
solvent.
Results
Table IV-9 presents the median (50th percentile), 25th,
75th, and 90th percentile pollutant concentration data for the
potentially hazardous metals in the 45 used oil samples collected
in New Jersey and Pennsylvania. Metals in used oil were a more
serious concern because the removal efficiency of pollution con-
trol equipment for metals is much lower than the boiler destruc-
tion efficiency of 99 to 99.99 percent for organics. As a
result, this table also presents typical contaminant levels in
virgin oil and EPA's final standards on used oil fuel specific:a-
tions for comparison. The most pronounced violation of the
promulgated standards is for lead. The sample lead concentration
levels exceed the EPA standards in all cases. The lead concen-
trations at the 75th and 90th percentiles, in particular, exceed
the EPA lead standard by a factor of five and seven, respec-
tively.
For the other detected metals, the level of contamination
does not appear to be as serious as for lead. The samples fail-
ing into the 90th percentile exceed the arsenic and total chro-
mium standards by only a factor of 1.2 and 2.1, respectively.
Finally, none of the sample concentrations are higher than th«
established cadmium levels. EPA has not established a standard
for nickel; however, the measured nickel concentrations are s:.g-
nificantly higher than those found in virgin oil. The measured
-------
IV-2 5
values for the other parameters were, on average, either not
detected or below any level posing a threat to human health.10
Table IV-9
PHILADELPHIA IEMP
LEVELS OF METAL CONTAMINANTS IN WASTE OIL SAMPLES
COLLECTED IN NEW JERSEY AND PENNSYLVANIA
Typical metals levels in virgin oil, and regulated maximum levels in waste oils
proposed by EPA, are provided for comparison.
Measured Contaminant Levels (ppm)
Arsenic Cadmium
Total
Chromium
Lead
Nickel
Waste Oil Monitoring Data:1
Median
25th Percentile
75th Percentile
90th Percentile
Virgin Oil:2
Proposed RCRA regulations:3
1.2
<1.0
3.8
6.0
0.45
5.0
0.34
<0.5
<1.0
1.2
0.51
2.0
3.3
2.3
10.3
21.5
3.8
10.0
206.3
120.0
500.0
740.0
2.5
100.0
14.6
6.5
35.0
35.0
0.05
N.A.
1 Values calculated from analysis of 45 used oil samples taken in New Jersey and
Pennsylvania in July and August, 1983.
2Values for arsenic, cadmium, chromium, and lead are from PEDCO, A Risk
Assessment of Waste Oil Burning in Boilers and Space Heaters, EPA, Office of
Solid Waste (contract number 68-02-3173), 1984. Value for nickel from
McNamara et al., An Exposure and Risk Assessment for Nickel, 1981.
'See 50 Federal Register 49164-49211, November 29, 1985, "Hazardous Waste
Management System; Burning of Waste Fuel and Used Oil Fuel in Boilers and
Industrial Furnaces" (Final Rule). OSW is currently working on Phase II Burning
and Blending Rules, which may change the used oil contaminant specifications.
The results from our limited sampling suggested that combus-
tion of used oil may pose a human health concern based primarily
on the high measured levels of lead. For the other detected
contaminants, concentrations were below EPA's fuel specification
levels for the majority of the samples. We believe that concen-
tration of lead in used oil will fall dramatically as the lead in
gasoline is phased out; however, we should note that EPA's Office
of Solid Waste is currently working on Phase II of its Burning
and Blending Rules and considering a change in its current lead
specification.
10The sampling analyses performed by GCA and Region III are part
of the background documents and are available on request.
-------
I
1
n
2
C
8
I
I
I
I
I
I
I
-------
-------
APPENDIXES
PHILADELPHIA IEMP
Appendix A: Benefit-Cost Analysis for the Philadelphia Region
Appendix B: Preliminary List of Pollutants in Phase I
Appendix C: Summary of Preliminary Data Collected in Phase I
Appendix D: 39 Issue Papers
Appendix E: Model Description
Appendix F: Characterization of Pollutant Releases
Appendix G: Estimation of Exposure and Risk
Appendix H: Descriptions of Control Options by Source Category
Appendix I: Pollution Control Costs and Equipment Life
Appendix J: Pollutant Concentrations and Health Effects by MEI
Location
-------
Appendix A
BENEFIT-COST ANALYSIS
FOR THE PHILADELPHIA REGION
-------
Appendix A
BENEFIT-COST ANALYSIS
FOR THE PHILADELPHIA REGION
Benefit-cost analysis is an alternative analytic technique
for assessing economically efficient control technologies. This
appendix briefly summarizes the findings from a recent draft EPA
study that assesses the monetary benefits and costs of control-
ling pollution in the Philadelphia region.1 The reader is
ancouraged to consult the EPA study itself for a more detailed
discussion.
A benefit-cost analysis provides decision makers with the
information needed to directly weigh the costs of control with
the-monetary benefits accruing to society from these controls.
Furthermore, a benefit-cost framework uses the monetary valuation
of the benefit derived from different types of pollution control
as the common currency to permit comparison across pollutants,
sources, and effects. There are numerous uncertainties inherent
in the analyses performed to quantify monetary benefits, in
comparison, a cost-effectiveness analysis shows the decision
maker the least cost path for achieving a desired action, such as
reduction in human health risk, without quantifying the value
placed by society on the control strategy. Measures of risk to
human health serve as the common denominator to compare different
environmental issues. The limitations in assessing human health
risks are detailed in Chapter II.
The EPA benefit-cost analysis expands on the IEMP risk
assessment/risk management framework applied in Philadelphia by
evaluating a wider range of pollutants and effects (both health
and nonhealth). The analysis demonstrates how the failure to
consider additional health and nonhealth effects may under-
estimate the economic benefits of controlling a particular sub-
stance or source. Furthermore, the results from the study show
that a different ranking of environmental control priorities may
result when additional pollutants and effects are evaluated.
A subset of the air and water pollution control options
considered in the IEMP analysis of toxic VOCs, and the -economic
benefit and cost estimates of the reduced health and nonhealth
Ralph Luken and Brett Snyder, "Comprehensive Environmental
Analysis for the Philadelphia Region," Economic Analysis Branch,
Office of Policy Analysis, U.S. EPA, Draft Paper, Revised May 8,
1986. -
-------
A-2
risks are presented in Table A-l. The first three air options
(1, 2, and 3) correspond to the control strategies to reduce the
upper-bound aggregate cancer incidence from air exposures by
24.1 percent, 49.9 percent, and 77.1 percent, respectively
(points c, e, and g on Figure VI-4). The drinking water controls
are defined as follows:2
Water Option 1 = Baxter Control Option 1 +
Belmont Control Option 2 +
Queen Lane Control Option 2
(best benefit-cost ratios)
Water Option 2 = Baxter Control Option 5 +
Belmont Control Option 1 +
Queen Lane Control Option
1 -(the highest number of upper-bound
statistical lives saved)
The estimated benefits of reducing cancer risks attributable
to* the subset of VOC emissions and sources are consistently less
than the costs of the control options. For example, the benefits
from reducing carcinogenic risks under air option 1 ($0.2 mil-
lion) are less than the control costs ($0.6 million). However,
if we include the concurrent noncarcinogenic health and nonhealth
benefits of reduced ozone levels that result from these toxic VOC
emission controls ($0.9 million), the total economic benefits
exceed costs ($0.2 million + $0.9 million = ?!•"• million >
$0.6 million). As a result, the total benefits for air options 1
and 2 are greater than the control costs. However, consideration
of concurrent ozone reduction benefits are not sufficient to make
air option 3 look favorable, and the total benefits are insignif-
icant for both water control options. The results suggest that
the noncarcinogenic, nonhealth benefits from reduced ozone levels
are much greater than the carcinogenic benefits alone for the voc
air control options. The VOC water control options do not have
significant noncarcinogenic benefits, and the control costs are
significantly greater than the benefits from the reduced cancer
risks.
In addition to toxic VOCs in the ambient air and drinking
water, the benefit-cost analysis also addressed the economic
desirability of controlling three other pollutant categories:
total suspended particulate matter from all sources; diesel par-
ticulate matter from heavy-duty diesel sources; and lead in gaso-
line. A comparison of the benefit-cost results for controlling
A more complete description of the specific control options can
be found in Appencfix H. Tables VI-11 and VI-12 detail the cost-
effectiveness of these control strategies.
-------
A-3
Tebl* A-l
KONETARY BENEFITS MO COSTS OF POTENTIAL CONTROL OPTIONS IN TIC PHILADELPHIA REGION
TO REDUCE EMISSIONS AND EXPOSURES FROM SELECTED VOLATILE ORGANIC COMPOUND SOURCES1
(annual Billion* !»*» dollar*)
Monetary Benefita2
Pollutant
VOC Eaiaaione from Selected
Air and Water Sourcn
Eaiaaiona
Reduction
(percent)
Cancer
Non-Cancer
Morbidity
Non-
Health*
Total
Monetary
efita
Total
Coat*
Net
Monetary
Benefit*
N/A > Not applicable— no eanear riak* ara ateecleted Kith tha pollutant.
N.C. » No additional coat* baeauaa control eoata mult froai reduction of pollutant of prlaary eoneam.
Coet Per
Cancer
SOS.
Air Op. li
Total
Air Op. 2i
Total
Air Op. Jt
Total
water Op.
Total
water Op.
Total
VOC • 29*
Oxen*
VOC 2e*
Ozone
voc j*i
Ozone
It ME dollar benefit*).
Source i Or. Ralph Lukan and Brett Snyder, •Coaprehanalve Environaental Analyeia for tha Philadelphia Region,* Economic Analyaie Branch, Office of Policy
Analyaia, U.S. EPA, Draft ••par, Raviaed Nay I. 1984.
all four environmental issues is shown in Table A-2. The
greatest estimated net economic benefits result from reduced
environmental exposures to particulate matter. The relative net
benefits of VOC controls when examined against a broader spectrum
of pollutants and effects are not as economically attractive.
Several of the selected VOC air control options (options 1 and 2)
do, however, show positive net benefits. The social costs of
control are greater than the estimated monetary benefits for
other air options Coption 3) and drinking water options
(options 1 and 2). Thus, the results from the benefit-cost
analysis suggest a^different ranking of control priorities than
that identified by'our cost-effectiveness analysis, which
addressed only VOCs and cancer effects.
-------
A-4
Tib la A-2
SUMMARY OF TOTAL NET BENEFITS AND COST-EFFECTIVENESS ESTIMATES
FOR MANAGEfCNT OPTIONS IN PHILADELPHIA
(annual millions of 1984 dollars)
Coat Per
Net Reduced
Monetary Cancer
Pollutant Benefits1 Case
Total Suspended Particulate Matter Fro*. All Sources Establish a 60 ug/mj-150 ug/m3 standard $ 14.S N/A
Diesel Particulate Matter fro* Heavy-Outy Diesel Achieve 0.25 g/BHP-hr for HDOE and 0.1
g/BHP-hr for buses 13.4 $360.00
Lead in Gasoline Phase out leaded gaa 3.4 N/A
Selected VOC Emissions - VOC Air Option 1 0.6 6.8
Selected VOC Emissions VOC Air Option 2 0.1 6.9
Selected VOC Emissions VOC Air Option 3 (4.7)2 27.9
Selected VX Emissions VX Water Option 1 (22.9)2 24.6
Selected VX Emissions VX Mater Option 2 (52.2)2 24.7
Total Net Benefits
Top Three Net Benefit Options7 $31.6
Top Three Cost-Effective Options4 $(8.9}2
N/A = Not applicable—no cancer risks are associated with the pollutant.
^Cancer case estimates are based on the 95 percent upper-limit estimates developed by EPA's Cancer Assessment Group.
These estimates will be higher than maximum likelihood estimates (expected values). In addition, all estimated
cancer cases contracted are assumed to be fstsl. Together, these two assumptions lead to conservative (high)
measures of effects. Noncsncer mortality, morbidity, and nonhealth effects are estimated using maximum
likelihood estimates (expected values). These estimates will be lower than the 95 percent upper-limit estimates for
each effect. Therefore, comparisons between controls on cancer and noncancer risks will tend to favor the
controls for cancer risks.
2( ) = Negstive net monetary benefits (i.e., dollar costs > dollar benefits).
3Three options with largest net benefit estimates. Value includes TSP, Diesel PM, and Lead: [14.8 + 13.4 + 3.4] =
31.6.
*Three options with least cost per reduced cancer cases. Since only one VX option for air and water is feasible among
the examined aet of controls, we chose the lowest cost per reduced cancer case options for each media. Value includes
VX Air Option 1, VX Water Option 1, and Diesel PM: [0.6 + (22.9) + 13.4] = (8.9).
Source: Dr. Ralph Luken and Brett Snyder, "Comprehensive Environmental Analysis for the Philadelphia Region," Economic
Analysis Branch, Office of Policy Analysis, U.S. EPA, Draft Paper, Revised May 8, 1986.
In summary, the benefit-cost analysis shows that the failure
to consider the benefits of reducing a wider range of environ-
mental and human health effects can result in a significant
understatement of the economic benefits accruing to society from
a proposed control strategy. Furthermore, consideration of a
broader list of pollutants and effects for the Philadelphia area
results in a different set of control priorities for local deci-
sion makers.
-------
Appendix B
PRELIMINARY LIST OF POLLUTANTS IN PHASE 1
-------
I
I
I
-------
CD
V. VI
« ill KN
J3 31 I
•* zl K\
U •< Zl
X
X X
X X
I I
X i X X X
I ! I
xx xx
I I
x I x i x
X X XXX
I
I IJ
I I I I I I
X| X X X IX I X
i
°
44
3 2
I I
I
-chl
he
CD
1*1 i
! I
x: xi —
2| x|
-------
I
I I
X X
3 3
X X
I
3 3l3
X X 'X I S/J
I I I I
I I
3 3
3 3 <
3 3
X X
X X X|
1! i'
1
I
I
I
-------
-------
I
X
I
I I
X X
3
I
3 •< •<
I
CD
U
OL
X
c
•O
O
a
c
X
o
V
c
a
3
O
I
g
I
I
r—
\
-------
< 3
s. *
+J ~*
8 £
X X X X
3:3 313
a.
u
V
a
x
X
c
I
o
o
,o
1
u
M
I*
u
u
b
^
I
I
n
1
0)
•&•
u
o
4)
C
-s
c
:>
s
§
s
CM
CO
i
CM
s
ri
lA
^
r^
i
CM
I
CM
a
-------
VI-19
Table VI-4
PHILADELPHIA IEMP
PHASE II RESULTS INTENDED FOR POLICY DEVELOPMENT1
UPPER-BOUND ESTIMATES OF ANNUAL EXCESS CANCER INCIDENCE AT CURRENT CONTROL:
INGESTION OF PHILADELPHIA DRINKING WATER
(1984 analysis)2
Pollutant
Chloroform
1,2-Dichloroethane
Methylone Chloride
1,2-Dichloropropane
Perchloroethylene
Trichloroethylene
Carbon Tetrachloride
Benzene
Total'
Baxter
(cases/year)
Queen Lane
(cases/year)
Belmont
(cases/year)
1.22
Note: Numbers have three decimal places not as an indication of precision but to
identify source contribution to the risks.
1THE UNIT RISK FACTORS USED IN THIS ANALYSIS ARE BASED ON CONSERVATIVE ASSUMPTIONS
THAT GENERALLY PRODUCE UPPER-BOUND ESTIMATES. BECAUSE OF LIMITATIONS IN DATA AND
METHODS IN SEVERAL AREAS OF THE ANALYSIS, SUCH AS EXPOSURE CALCULATIONS AND
POLLUTANT SELECTION, RISK ESTIMATES WERE CALCULATED AS AIDS TO POLICY DEVELOP-
MENT, NOT AS PREDICTIONS OF ACTUAL CANCER RISKS IN PHILADELPHIA. ACTUAL RISKS
MAY BE SIGNIFICANTLY LOWER; IN FACT, THEY COULD BE ZERO. THE PROPER FUNCTION OF
THE ESTIMATES IS TO HELP LOCAL OFFICIALS SELECT AND EVALUATE ISSUES, SET
PRIORITIES, AND DEVELOP CONTROL STRATEGIES FOR THE TOPICS EXAMINED.
2The risk estimates presented in this table were calculated using cancer unit risk
factors from 1984.
•'Columns may not sum because of rounding.
Source: lEc, Inc., Cost-Effectiveness Analysis of Strategies to Reduce Human
Health Risk in Philadelphia, U.S. EPA, May 1985.
We should note that our assessment of drinking water expo-
sures may be limited in two ways. First, our assumptions regard-
ing uniform contaminant concentrations throughout the distribu-
tion system for each plant may not accurately estimate exposures
for residents in some areas. This is especially true for chloro-
form, which can increase in concentration with resonance in the
distribution system. Second/ we did not consider exposure to
chloroform that volatilizes in the home. There is a growing
speculation that chloroform volatilization through showering and
cooking may be an important contributor to indoor air exposures.
-------
VI-20
Table VI-5
PHILADELPHIA IEMP
AVERAGE CHLOROFORM LEVELS IN DRINKING WATER
FOR SELECTED CITIES
(population >500,000)1
Average Population
uq/1 1980
Baltimore, MD 36.5 786,775
Boston, MA 3.7 562,994
Chicago, IL 14.5 3,005,072
Cleveland, OH 15.0 573,822
Columbus, OH 171.0 564,871
Dallas, TX _ 18.0 904,078
Detroit, MI 10.5 1,203,339
Houston, TX 123.0 1,595,138
Los Angeles, CA 32.0 2,966,850
Memphis, TN 2.5 646,356
Milwaukee, WI 8.9 636,212
New York, NY 22.0 7,071,639
Philadelphia, PA2 47.7 1,656,300
San Antonio, TX 0.1 785,880
San Diego, CA 43.5 875,538
San Francisco, CA 58.5 678,974
Washington, DC 47.0 638,333
^These cities were selected from a sample of 137
cities studied in the source document.
2Reported chloroform concentrations are an average of
the levels at Baxter, Queen Lane, and Belmont (see
Table F-3). Population data as of July 1983.
Source: U.S. EPA, Office of Standards and Regula-
tions, Chemical Control Options Analysis for
Chloroform, (Contract Nos. 68-02-3166,
68-01-6775, and 68-01-6715), December 1984.
Air. Point and area sources contribute almost equally to
the estimated excess cancer incidence associated with inhalation
(48 percent and 52 percent of the .4 cases, respectively). Dry
cleaning accounts for roughly 30 percent of the incidence from
exposure to area source emissions, followed by gasoline marketing
(25 percent) and degreasing (23 percent). As shown in
Table VI-6, which displays our incidence by source and pollutant,
perchloroethylene explains all of the exposures from dry clean-
ing. Gasoline vapor accounts for more than 94 percent of the
incidence from gas marketing (.05 annual cases), and trichloro-
ethylene accounts fdr roughly 82 percent of the risks from
degreasing (.05 annual cases).
-------
VI-21
•ollytor*
MKhylene Oiloruo
1 ,2-OichlurMtheno
Oiloroforoi
tic.l oft.
..to-*
0
7.7e-«
CofMn TotrtOlloriM t.W-)
l.Z-OicMoroetaMn.
*irehlorMthrl.ni
•miene
GOMlr. reeor
Tot el*
Metei Colueno one
M«*)ero hero
0
0
0
0
a
roM My net MO
tnrM OMIMI pi
'THE WIT mw ration wu l» THIS
•Minis. SUCH M
mas wtr • sioii
STMitcils roii nc
'The nek lei IOMI M
Seureoi ICe, Int.,
CinBUK CW.CU.HT
ncwM LOetti i
ionics Owen.
OtMiCOl
MTr.
0
1.00-1
0
0
e.7e->
0
•••»-}
0
0
to totele ee
•CM net M
•MLrSIS ME
IOMJ «e m.
»r«t! , THEY
preoenteri in thio to* to were
Cott-crroctirow
M Anolroll
Teklo It-*
mn. tof
im* ii cuts IKTOOCO rat nuicr
uprcn-wue CSTIMICS or num. atca umi IKUC«
COFMnt CMinM Induotrlol Mftnory hrinorr loirtWoM
»rr. Mfr, on
0 ..»»-7
0
0
0
0
I.M. 0
0
0
0
eoyM of roundinf .
en ingieecien of erMi
enu ON coNscmMTiic
LUIMI SCUCIIO", «15«
COUtO « ZCIIO. THE >
eKlevlouri uoinf cone
or Strotoileo To IMue
Cloonot • • 22'
0 0 To.)
2.oo>) 2.*o-5 le-2
0 0 1^>
0 tt-t
0 H-l
0 Oe-t
2.H^ 0 le.)
0 7,*e* 7.«^4 )^4
0 '.4.- l.Oe-> 0
-
oien bye to identifr OMjree centfie«ti«»
asumioa nut eCNtiuit* '•ODUCS i»^«
CST1MTCS ft UtCUMICO B MM n| «•,
«vc» rwcTiai or THE CSTIMITC] is >• «L
o< unit riM 'octon rr. im.
• Omtf HMltn tlok in PhlleMltTm. • >
OC«U.Ooomil
•1 CUMOil n*TKOli INK
orr
Omieoino Cleemni
..70-. 0
0 0
0 0
0
0
«.0o. 0
I.Ze- A...-2
0
0 0
>. I*. IIM.
-«••• MIIMlf]. «C«J
1O 0CVTLOPMXM . MAI AS
• lot* trritldS SO.CCI
1M. *0> «•*.
•UTIOr)
Other COM lino
IIKMtriOl HofMtinp.
i.«- a
2.2O-4
0
0
a
0
i.e.- o
0 2.70-)
0 VO.-1
).2e-J }.)e*Z
K OT L1NIT1T10NS 11 OJTA
•W9ICIIIM Or UTUM. CM
•JO CV*XIMTC ISSUCS, SCI
VolotiUzoUon River Sm
0 2.704 Z
«.H-3 t.te.) t
0 2>ei*)
0 *.*e-7
I.M-2 1.0.-1
0 t.1e.)
1.0..4 3. 5.. 4
a ft.i.-s
0 0
Z. 1e*Z Z.le-Z 3
«C MITHODS 11 SdlllUl MCts or
II DISCS IX INIIWCIPHII. ICTUI
niaiiiics. M> ncttio* cwm
rcce
5.-I
W-Z
9t-4
U- )
1*-l
1*-2
4»"?
i»-:
it- >
IX
it
Two plants contribute most to the excess annual incidence
for the air point sources included in our analysis—the NEWPCP
and Refinery B. The risks associated with the NEWPCP are due
primarily to 1,2-dichloropropane coming from the chemical manu-
facturing plant, whereas gasoline vapor is the pollutant of
concern from the refinery. Both of tnese sources/ however,
appear to contribute only minimally to the total human health
risks we could quantify in our analysis.
We have separated the volatilization from the sewer vents
from the NEWPCP because of the distinction made in our modeling
between point and area sources. If •<« add the sewer vent emis-
sions to the NEWPCP and view these points of release as one
interconnected source, the annual excess cancer incidence (upper-
bound) associated with this system is .11 cases. (The sewer vent
emissions account for roughly 19 percent of the .11 cases.)
Inclusion of the exposure associated with the volatiles from the
Delaware River that are discharged by the NEWPCP will raise this
risk calculation still further to .13 annual cancer cases
(volatilization from the Delaware River and the sewer vents
account for 18 percent and 16 percent of the .13 cancer'cases,
respectively), but the baseline cancer incidence estimates for
the NEWPCP system still remain relatively small.
-------
VI-22
Risks to the Most Exposed Individual
Cancer. Our findings on the cumulative lifetime cancer
risks to the MEI from air and drinking water show risks that
range from 1.1 x 10~4 to 3.2 x 10"4 throughout eight industrial
areas of the city (see Table VI-7). With the exception of the
area around the chemical manufacturer, drinking water risks (due
to the presence of chloroform) contribute the majority of the
total cancer risks to the MEI. Drinking water risks are uniform
across the sites; however, this is a function of our assumption
that contaminant levels in the drinking water are uniform
throughout the distribution system. We also assumed that
individuals at the MEI locations are all served by the Baxter
plant.
Table VI-7
PHILADELPHIA IEMP
PHASE II RESULTS INTENDED FOR POLICY DEVELOPMENT1
UPPER-BOUND ESTIMATES OF CUMULATIVE LIFETIME CANCER RISKS
TO THE MOST EXPOSED INDIVIDUAL
(1984 analysis)2
MEI Location
Northeast WPCP*
Refinery B
Chemical Mfr.
Plastic Cabinet Mfr.
Pharmaceutical Mfr.
Garment Manufacturer
Refinery A
Industrial Dry Cleaner
Cumulative Lifetime
Cancer Risk
(Upper-Bound) Inhalation
5.6 x
1.4 x
2.2 x
6.5 x
4.5 x
1.2 x
3.0 x
2.2 x 10~5
10-5
1CT5
10-*
10'6
10"5
10"
Cumulative Lifetime
Cancer Risk
(Upper-Bound) Ingest ion
1.0 x 10-*
1.0 x 10-*
1.0 x 10-*
1.0 x 10-*
1.0 x 10-*
1.0 x 10-*
1.0 x 10-*
1.0 x 10-*
Cumulative Lifetime
Total Cancer Risk
(Upper-Bound) Inhalation
and Inqestion
1.6 x 10-*
1.1 x 10-*
3.2 x 10-*
1.1 x 10-*
1.4 x 10-*
1.1 x 10-*
1.3 x 10-*
1.2 x 10-*
1THE UNIT RISK FACTORS USED IN THIS ANALYSIS ARE BASED ON CONSERVATIVE ASSUMPTIONS THAT GENERALLY
PRODUCE UPPER-BOUND ESTIMATES. BECAUSE OF LIMITATIONS IN DATA AND METHODS IN SEVERAL AREAS OF
THE ANALYSIS, SUCH AS EXPOSURE CALCULATIONS AND POLLUTANT SELECTION, RISK ESTIMATES WERE CALCU-
LATED AS AIDS TO POLICY DEVELOPMENT, NOT AS PREDICTION OF ACTUAL CANCER RISKS IN PHILADELPHIA.
ACTUAL RISKS MAY BE SIGNIFICANTLY LOWER; IN FACT, THEY COULD BE ZERO. THE PROPER FUNCTION OF
THE ESTIMATES IS TO HELP LOCAL OFFICIALS SELECT AM) EVALUATE ISSUES, SET PRIORITIES, AND DEVELOP
CONTROL STRATEGIES FOR THE TOPICS EXAMINED.
2The risk estimates presented in this table were calculated using unit risk factors from 1984.
3Recent reductions in dischargee to the NEWPCP may f»«utt in lower risk numbers than presented in
this table.
Source: IEc,-Inc., Cost Effectiveness Analysis of Strategies to Reduce Human Health Risk in
Philadelphia. U.S. EPA, May 1985.
-------
VI-23
With the exception of two locations—the chemical manufac-
turer and the plastic cabinet fabricator—total air risks are
also relatively constant. However, while chloroform accounts for
roughly 97 percent of water risks throughout the area, the con-
tributions of individual pollutants to the total MET risks from
inhalation can vary from site to site, as demonstrated in
Table VT-8. For example, while 1,2-DCP accounts for the majority
of the lifetime MET risks at the NEWPCP and the chemical plant
location, it is less important in other MEI locations. Sim-
ilarly, gasoline vapors are the major concern around the two
refineries but are not an important contributor in our risk
analysis in areas of the city. Unlike drinking, water, where we
assume a uniform exposure for all areas of the city, the distri-
bution of air requires more consideration from a site-specific
perspective (i.e., the MEI) because of the significant variabil-
ity in the types of pollutants and ambient levels from one area
of the city to the next.
Table VI -8
PHILADELPHIA IEMP
PERCENTAGE CONTRIBUTIONS OF POLLUTANTS TO
CUMULATIVE LIFETIME CANCER RISKS IN AIR BY FACILITY
(1984 analysis)
Chloro- Carbon
MEI Location DCM DCE form Tet. DCP
Northeast WPCP1 0.1 20.8 1.2 0.3 67.9
Refinery B 0.3 4.1 0.5 0.3 5.3
Chemical Mfr. 0.0 2.5 0.2 0.1 94.6
Plastic Cabinet Mfr. 0.9 8.5 1.5 0.7 28.8
Pharmaceutical Mfr. 4.3 1.3 50.2 28.7 4.0
Garment Mfr. 0.6 4.3 3.3 1.8 13.6
Refinery A 0.2 2.1 0.4 Q.2 2.9
Industrial Dry Cleaner 0.3 14.4 1.7 0.7 47.0
DCM: Methylene Chloride TCE: Trichloroethylene
DCE: 1,2-Oichloroethane PCE: Perch loroethylene
DCP: 1,2-Oichloropropane
Note: Rows may not sum to 100% because of rounding.
^Recent reductions in discharges of DCE and DCP to the NEWPCP
button of these pollutants to risks at this site and others.
t
Source: lEc, Inc., Cost-Effectiveness Analysis of Strategies
Philadelphia. U.S. EPA, May 1985.
TCE
1.6
6.1
0.4
13.0
1.9
46.5
2.8
4.0
PCE
4.2
11.2
1.4
23.4
3.5
13.0
5.2
22.4
may lower the
Benzene
0.5
7.9
0.1
1.7
0.5
1.4
9.8
0.9
relative
to Reduce Human Health
Gas
Vapors
3.4
64.1
0.8
21.7
5.4
15.4
76.4
8.4
contri-
Risk in
-------
VI-24
Noncancer Effects. We also explored whether the maximum
pollutant concentrations estimated at each of the MEI locations
presented a hazard for noncarcinogenic health effects. On a
pollutant-by-pollutant basis, the ambient concentrations did not
approach either the original effect threshold levels used in the
analysis or the most recent values. Appendix J provides more
detail on the pollutant concentrations at each MEI location and
the comparison of dose to the no-effect threshold.
We then calculated a hazard index for the five noncarcino-
genic health effects considered at each MEI site—liver, neurobe-
havioral, kidney, reproductive, and fetal developmental—by
considering ambient concentrations in both air and drinking
water. Table VI-9 shows the resulting hazard indices by effect
and MEI location. With the exception of the hazard index for
liver toxicity at the pharmaceutical plant MEI location (.80),
none of the indices approached unity.
Table VI-9
PHILADELPHIA IEMP
THRESHOLD POLLUTANT HAZARD INDEX FOR EACH MEI LOCATION:
AIR AND DRINKING WATER1
(at 1984 control levels)
MEI Location
Northeast MPCP
Refinery 8
Chemical Mfr.
Plastic Cabinet Mfr.
Pharmaceutical Mfr.
Garment Mfr.
Refinery A
Industrial Dry Cleaner
Liver Neuro
Repro Fetal
.58
.57
.61
.57
.80
.58
.57
.58
.00
.00
.00
.00
.10
.01
.00
.00
.23
.22
.26
.22
.28
.22
.22
.23
.01
.00
.01
.00
.47
.01
.00
.01
.08
.11
.08
.07
.13
.08
.17
.08
hazard index is a numerical Indication of the difference
between acceptable and unacceptable ambient exposure levels.
As the index approaches unity, concern for the potential hazard
of the chemical mixture increases. If the index exceeds 1, the
concern is the same as if a no-effect threshold were exceeded.
Source: IEc» Inc., Cost-Effectiveness Strategies to Reduce
Human Health Risk in Philadelphia. U.S. EPA,
May 1985.
-------
VI-25
Cost-Effective Control Strategies
We developed cost-effective pollution control strategies
for reducing aggregrate cancer incidence and risks to the MEI
separately. For each measure of risk, we assessed cost-effective
control strategies for reducing exposures in both media simultan-
eously, as well as in each individual medium. We performed the
latter analyses to generate additional information for the regu-
latory agencies that independently oversee air and water quality
issues in Philadelphia.
We designed our control-options analysis to map out a
continuum of mutually exclusive control strategies for reducing
incremental levels of risk to human health. As a result, as we
move to progressively higher levels of risk reduction, the simu-
lation model reconfigures the mix of control strategies to deter-
mine an optimum set of controls that would provide the desired
level of risk reduction for the least cost. Each time a higher
level of risk reduction is requested, the model returns to the
current control level and reconfigures the optimum mix of con-
trols. We selected this approach because we did not know the
policymaker's preferences ahead of time. Ultimately, the policy-
maker must decide what levels of risk and control expenditures
are acceptable. This analysis provides important information
that will assist policymakers in making this determination.
Aggregate Excess Cancer Incidence
Air and Drinking Water. Table VI-10 shows our least-cost
control strategies for reducing risk from exposure to both air
and drinking water. This table also displays the total cost,
average cost per case reduced, and incremental cost per case
reduced for each control strategy. Figure VI-3 graphs the total
cost for achieving specified levels of risk reduction. It also
shows the percentage of the maximum control cost required to
achieving increasing percentage reductions in annual cancer
cases. We should note that the shapes of the cost curves
presented in this section are a function of the risk reduction
levels selected for the model runs. Further analysis of addi-
tional risk reduction levels would permit a more precise defini-
tion of the curve.
As shown in Table VI-10, it may be possible to achieve a
modest reduction in the baseline cancer incidence—a little over
1 percent—at a net savings by implementing controls that recover
solvents for future |