&EPA
          United States
          Environmental Protection
                Policy, Planning,
                And Evaluation
                (PM-221J
1992
Baltimore Integrated Environmental
Management Project: Phase II
Ambient Air Toxics Report
                         PENNSYLVANIA
                                    Printed on Recycled Paper

-------
                             PREFACE


     This report was prepared under the auspices of the Baltimore
Integrated Environmental Management Project (IEMP).  The
Baltimore IEMP is a collaborative effort of the State of
Maryland, Anne Arundel and Baltimore Counties, the City of
Baltimore and the U.S. Environmental Protection Agency.  The
Environmental Protection Agency (EPA) initiated the project as
part of its pursuit of new approaches to environmental management
and policy.  The purpose of the IEMP is to use an integrated
approach to identify and assess environmental issues that concern
managers, to set priorities for action among these issues, and to
analyze appropriate approaches to manage these problems.

     The Baltimore IEMP represents the second of four geographic
projects that EPA initiated across the country.  The Baltimore
area was chosen, not because it has a significant toxics problem,
but because EPA and local officials wanted to explore better ways
to identify, assess, and manage the human health risks of
environmental pollutants in the area.  Other lEMPs include
Philadelphia, Santa Clara County, and Denver.

     The decision-making structure of the Baltimore IEMP
consisted of two committees, which also served as the means for
State and local participations  the Management Committee and the
Technical Advisory Committee.  The Management Committee, with
members representing Baltimore City, Baltimore County, Anne
Arundel County, and the State, managed the IEMP and set its
overall policy directions.  The Technical Advisory Committee,
composed of technical managers from the City of Baltimore, the
two counties, the State, as well as representatives from the
Regional Planning Council and the academic community, recommended
issues to study, advised the Management Committee on the
technical and scientific aspects of the project, and oversaw and
commented on all EPA and consultant work,  EPA provided
administrative, technical, and analytical support.

     The Baltimore IEMP examined five environmental issues:  air
toxics, .Baltimore Harbor, indoor air pollution, lead paint
abatement, and potential contamination of ground water from
underground tanks.  For further information on these reports or
other IEMP studies contact the Regulatory Integration Division,
the Office of Policy Analysis (PM-220) in the Office of Policy,
Planning and Evaluation, U.S. Environmental Protection Agency,
Washington, D.C. 20460

-------
                        ACKNOWLEDGEMENTS

     Numerous individuals contributed to the preparation of this
report.  We wish to mention the following individuals:

            of the Air ToxicB Workgroup
Tad Aburn, Section Head, Hazardous Pollutant Evaluation Section,
Air Management Administration, Maryland Department of the
Environment

Don Andrew, Administrator, Engineering & Enforcement Program, Air
Management Administration, Maryland Department of the Environment

Anthony S. Bonaccorsi, formerly Director, Environmental Services,
Eastern Stainless Steel Company

Daryl Braithwaite, formerly Program Coordinator, Clean Water
Project

David Filbert, Chief, Bureau of Air Quality Management, Baltimore
County Department of Environmental Protection and Resource
Management

Michael K. Hettleman, President, The Southern Galvanizing Company

Frank Hoot, Assistant Commissioner, Environmental Health,
Baltimore City Health Department (Retired)

Dr. Genevieve Matanoski, Professor, Department of Epidemiology,
Johns Hopkins School of Hygiene and Public Health

Darryl W. Palmer, Environmental Manager, Agricultural Chemical
Group, FMC Corporation

Susan Wierman, Acting Deputy Director, Air Management
Administration, Maryland Department of the Environment

Bruce Windsor, Program Coordinator, American Lung Association of
Maryland.


ptotnhftr-a of the Indoor Air Workgroup who Contributed  to the  Design
of the Joint Air Toxics Study

David Filbert, Baltimore County Department of Environmental
Protection

Joseph Abey, Anne Arundel County Health Department

Dr. Charles Billings, Johns Hopkins University

Elkins W. Dahle, Baltimore City Health Department

-------
Dr. Katharine Parrel I/ Maryland Department of the Environment

Allan Heaver, Heaver Properties

Jack Lodge/ Baltimore Gas and Electric Company

Arthur Nieberding, Mueller Associates/ Inc.

Velma Rector/ American Lung Association of Maryland

Tom King, Mueller Associates, Inc.
      oi! BPArB Bcono111^ c Benefits Brunch who Contributed to the
Chapter on Risk M*T>*o*ffn*nt

Brett Snyder

Joel Scheraga


Itemhqrg of the Balt^^re IBMP Management CTW? ttee

J. James Dieter/ Special Assistant to the Director, Department of
Environmental Protection and Resource Management, Baltimore
County

Max Eisenberg, Assistant Secretary for Toxics, Environmental
Science and Health, Department of the Environment, State of
Maryland

Robert Perciasepe,  (formerly) Assistant Director, Department of
Planning, City of Baltimore

Claude Vannoy, Assistant to the County Executive for Land Use,
Anne Arundel County


MamHqrB of the Bflltj^mftre TUMP Technical Advisory Cotl"B'*t'tee

Jared L. Cohon, Vice Provost for Research, and Professor of
Geography and Environmental Engineering, Johns Hopkins University
(Chairman, Technical Advisory Committee)

Don Andrew, Administrator, Engineering & Enforcement, Air
Management Administration, Department of the Environment, State
of Maryland

Philip Clayton, Manager, Cooperative Clean Water Program,
Regional Planning Council

Emery Cleaves, Principal Geologist, Maryland Geological Survey
(Chairman, Ground-Water Subcommittee)

-------
Ralph Cullison, Baltimore City Department of Public Works, City
of Baltimore
N. Singh Dhillon, Director, Environmental Health, Anne Arundel
County Department of Health
Tom Ervin, Environmental Planner, Anne Arundel County Office of
Planning and Zoning (Chairman, Ecological Subcommittee)
Katherine Farrell, M.D., M.P.H., Chief, Division of Environmental
Disease Control, Department of the Environment, State of Maryland
David Filbert, Chief, Bureau of Air Quality Management, Baltimore
County Department of Environmental Protection and Resource
Management
Frank Hoot, (formerly) Assistant Commissioner, Environmental
Health, Baltimore City Health Department (Chairman, Human Health
Subcommittee)
Sam Martin, Consultant, Vice Chairman of TAG  (represented the
Regional Planning Council during Phase I)
Janice Outen, Director, Division of Environmental Management,
Baltimore County Department of Environmental Protection and
Resource Management
Colin Thacker,  (formerly) Management Assistant, Baltimore County
Department of Environmental Protection and Resource Management
Bill Wolinski, Water Quality Coordinator, Department of Public
Health
Staff of the U.S. Environmental Protection Acrencv
Daniel Beardsley, (formerly) Director, Regulatory  Integration
Division
Arthur Koines,  Chief, Geographic Studies Branch
John Chamberlin,  (formerly) Site Director, Baltimore IEMP
Andrew Manale,  Senior Analyst,  IEMP
Catherine Tunis,  Policy Analyst, IEMP
Ellen Tohn,  (formerly)  Policy Analyst, IEMP
Roberta Grossman,  (formerly) Policy Analyst,  IEMP
Alan Jones,  (formerly)  Policy Analyst, IEMP

-------
SchoolB

Orterio  Villa, Jr.i Director, Central Regional Laboratory, EPA
Region III

John Austin, Central Regional Laboratory, EPA Region III

Gerry Akland, Environmental Monitoring Systems Laboratory, Office
of Research & Development, EPA

Andy Bond, Environmental Monitoring Systems Laboratory, Office of
Research & Development, EPA

Tom Hartledge, Environmental Monitoring Systems Laboratory,
Office of Research & Development, EPA

William Nelson, Environmental Monitoring Systems Laboratory,
Office of Research & Development, EPA

Dr. Lance Wallace, Environmental Monitoring Systems Laboratory,
Office of Research & Development, EPA

Dr. Joelien Leutas, Chief Genetic Bioassay Laboratory, Office of
Research & Development, EPA

Ila Cote, Office of Air Quality Planning and Standards, EPA

Joseph Padgett, Associate Director for Intermedia and
Intergovernmental Programs, Office of Air Quality Planning and
Standards, EPA

Tim Barry, Science Policy Branch, Regulatory Integration
Division, Office of Policy Analysis, EPA

Robert Omelia, Air Management Administration, Department of the
Environment, State of Maryland

Robert Merrey, Bureau of Air Quality Management, Baltimore County
Department of Environmental Protection and Resource Management

Adon Phillips, Baltimore County Office of Central Services

Dr. L. E. Bills and the staff of the Southeastern Community
Mental Health Center, Baltimore County Department of Health

Merreen Kelly, Baltimore County Public Schools

Arthur Himlin, Principal, and staff of Parkville Middle School
(Baltimore County)

George P. Dausch III, Principal, and staff of Dundalk Senior High
School (Baltimore County)

-------
Individual* from Conaultino Firm
Craig Koralek, VERSAR
Dennis Hlinka, VERSAR
Josephina Castellanos, VERSAR
Kevin Jameson, VERSAR
David Sullivan, Sullivan Environmental Associates
Jay Wind, American Management Systems
Elaine Haemisegger, Temple, Barker, & Sloane, Inc.

-------
                        TABLE OF CONTENTS


EXECUTIVE SUMMARY

EPILOGUE

I.   INTRODUCTION	   1-1

     1.  INTEGRATED ENVIRONMENTAL MANAGEMENT PROJECTS   ...   1
     2.  ORGANIZATION OF THE PROJECTS	   2
     3.  THE BALTIMORE I BMP	   2
     4.  THE RESULTS OF THE PHASE I PRIORITY-SETTING
          PROCESS 	   4
     5.  RELATIONSHIP TO THE EPA AND AMA AIR TOXICS
          STRATEGIES  	   4
     6.  OVERVIEW OF THE REPORT	6

II.   OBJECTIVES AND DESIGN OF THE BALTIMORE AMBIENT AIR TOXICS
     STUDY	     II-l

     1.  STUDY OBJECTIVES 	   1
     2.  STUDY DESIGN 	   2
          a.  Pollutant Selection 	   2
          b.  Risk Characterization of the Selected
               Pollutants 	   3
          c.  Risk Management:  A Demonstration 	   4
     3.  HOW THE RESULTS OF THE AIR TOXICS STUDY SHOULD BE
          VIEWED  	   4

III. SUMMARY OF THE PRINCIPLES OF RISK ASSESSMENT AND RISK
     MANAGEMENT AS USED IN THE AMBIENT AIR TOXICS STUDY .    III-l

     1.  RISK ASSESSMENT AND RISK MANAGEMENT	   1
     2.  STEPS IN RISK ASSESSMENT	   2
          a.  Hazard Identification 	   2
          b.  Dose-Response Evaluation  	   3
               i.   Cancer	   4
               ii.  Cancer Caused by Exposures to Complex
                    Mixtures    	   6
               iii. Noncancer Health Effects    	   7
          c.  Human Exposure Evaluation   	   8
          d.  Risk Characterization 	   9
               i.   Cancer	   9
               ii.  Noncancer	11
     3.  INTERPRETING RISK ASSESSMENT RESULTS IN THE
          BALTIMORE AIR TOXICS STUDY  	  13
     4.  RISK MANAGEMENT	14

-------
                   TABLE OF CONTENTS  (CONTD.)


IV.  SELECTING POLLUTANTS.  EVALUATING TOXICITY. AND
     ESTIMATING EXPOSURE  	    IV-1

     1.  POLLUTANT SELECTION METHODOLOGY AND RESULTS  ....   1
          a.  Pollutant Selection for Modelling 	   2
          b.  Pollutant Selection for TEAM Monitoring ....   5
     2.  HAZARD IDENTIFICATION/DOSE-RESPONSE EVALUATION   .  .   7
          a.  Cancer Effects  	   7
               i.   Weight of Evidence	   7
               ii.  Dose-Response Evaluation   	  11
          b.  Noncancer Effects 	  11
               i.   Weight of Evidence	11
               ii.  Dose-Response Evaluation   	  16
     3.  HUMAN EXPOSURE EVALUATION  	  17
          a.  Air Dispersion Modelling	17
               i.   Emissions Assessment	17
               ii.  Emissions Characterization for Key
                    Sources	28
               iii. Dispersion Modelling  	  29
               iv.  Model Performance Evaluation  	  31
          b.  Ambient Air Monitoring	34
               i.   TEAM VOC Sampling	34
               ii.  TEAM Metals Monitoring	39
               iii. Available Area-Wide Sampling Data    ...  42
          c.  Exposed Population  	  42

V.  RISK ASSESSMENT SCREENING RESULTS 	   V-l

     1.  RANKING OF TARGET COMPOUNDS AND SOURCES BY CANCER
          POTENCY-WEIGHTED AMBIENT CONCENTRATIONS 	   3
     2.  RISK ASSESSMENT SCREENING RESULTS  	   4
          a.  Cancer Risks  	   6
               i.   Area-Wide Risks   	   6
               ii.  "Hotspot" Risks	24
          b.  Noncancer Risks	37
               i.   Results Based on Dispersion Modelling  .  .  37
               ii.  Results Based on Monitoring Data  ....  39
          c.  Limitations	41
               i.   Exposure Limitations Based on Dispersion
                     Modelling	41
               ii.  Exposure Limitations Based on
                     Monitoring	44
               iii.  Dose-Response Limitations  	  45
     3.  RISK RESULTS COMPARISONS	46
          a.  Comparison with IEMP and Other Urban-Scale
               Analyses	46
          b.  Comparison of Alternative Approaches to
               Estimating Risks from the Products of
               Incomplete Combustion	•	52
          c.  Comparison of Risks in Air and Drinking
               Water	54

-------
                    TABLE OF CONTENTS (CONTD.)


VI.  RISK MANAGEMENT STRATEGIES FOR REDUCING RISKS;
     A DEMONSTRATION  	    VI-1

     1.  SELECTION OF SOURCES 	   3
     2.  ESTIMATING CONTROL OPTIONS COSTS, EFFECTIVENESS,
          AND BENEFITS	   9
          a.  Feasible Control Options and their Cost-
               Effectiveness  	   9
          b.  Estimating the Benefits Associated with the
               Control Options  	  19
     3.  ANALYSIS OF THE COST-EFFECTIVENESS OF CONTROL
          STRATEGIES	21
          a.  Cost-Effectiveness Analysis Overview  	  21
          b.  Cost-Effectiveness Analysis Results 	  25
               i.   Results:  Annual Excess Cancer
                    Incidence	25
               ii.  Results:  Maximum Increased Lifetime
                    Individual Cancer Risk  	  34
               iii. Limitations	38
     4.   MAXIMIZING THE BENEFITS TO SOCIETY IN CONTROLLING
          AIR TOXICS	39
          a.  Results of the Analysis of Strategies to
               Maximize Benefits  	  39
          b.  Comparison of Benefit and Cost-Effectiveness
               Analyses	39

APPENDIXES

-------
                             TABLES


IV-1      CHEMICALS SELECTED FOR AIR DISPERSION MODELLING IN THE
          BALTIMORE IEMP AIR TOXICS STUDY	IV-6

IV-2      CHEMICALS SELECTED FOR ORD'S TEAM AMBIENT AIR
          MONITORING AS PART OF THE BALTIMORE IEMP AIR TOXICS
          STUDY	IV-8

IV-3      SUMMARY OF POLLUTANTS SELECTED FOR MODELLING AND
          MONITORING IN THE BALTIMORE AMBIENT AIR TOXICS
          STUDY	IV-9

IV-4      UNIT CANCER RISK FACTORS AND THE WEIGHT OF EVIDENCE FOR
          CARCINOGENICITY FOR THE BALTIMORE IEMP TARGET
          COMPOUNDS	IV-12

IV-5      NONCANCER HEALTH EFFECTS AND THRESHOLD VALUES FOR
          TARGET COMPOUNDS IN THE BALTIMORE IEMP AIR
          ANALYSIS	IV-18

IV-6      COMPARISON OF MEASURED VS. PREDICTED CONCENTRATIONS FOR
          BALTIMORE MONITORING SITES DURING THE PERIOD 11/20/83 -
          2/16/84 FOR MODEL RUN URBAN MODE	IV-32

IV-7      COMPARISON OF MEASURED VS. MODELLED DATA FOR THE
          BALTIMORE AND PHILADELPHIA IEMP STUDIES 	 IV-33

IV-8      BALTIMORE IEMP AIR TOXICS STUDY AVERAGE MEASURED
          AMBIENT CONCENTRATIONS AT THE TWO FIXED MONITORING
          SITES: TARGET VOLATILE ORGANICS 	 IV-37

IV-9      COMPARISON OF 1987 AND 1983/84 MONITORING
          RESULTS	IV-40

IV-10     AVERAGE MEASURED AMBIENT CONCENTRATIONS: METALS AT
          FIXED MONITORING SITES	IV-41

IV-11     AVAILABLE AREA-WIDE SUMMARY MONITORING DATA FOR
          BALTIMORE	IV-43

V-l       RELATIVE RANKING OF TARGET COMPOUNDS BY CANCER POTENCY-
          WEIGHTED AMBIENT CONCENTRATIONS 	 V-4

V-2       RELATIVE RANKING OF SOURCES BY CONTRIBUTION TO CANCER
          POTENCY-WEIGHTED AMBIENT CONCENTRATIONS 	 V-5

V-3       AVERAGE INCREASED LIFETIME INDIVIDUAL CANCER RISK USING
          AVAILABLE MONITORING DATA  	V-10

V-4       BALTIMORE IEMP AIR TOXICS STUDY PHASE II RESULTS  . V-ll

-------
                          TABLES (CON'T)

V-5       AVERAGE  INCREASED LIFETIME INDIVIDUAL CANCER RISKS
          BASED ON AVAILABLE MONITORING DATA: SITES WITH THE
          HIGHEST  AND LOWEST MEASURED VALUES  	V-15

V-6       ESTIMATED ANNUAL EXCESS CANCER INCIDENCE FOR SELECTED
          POLLUTANTS MODELLED IN THE BALTIMORE IEMP AIR TOXICS
          STUDY (TOTAL STUDY AREA, 5 KM GRID SYSTEM)  . . .  . V-19

V-7       ESTIMATED ANNUAL EXCESS CANCER INCIDENCE FOR SELECTED
          POLLUTANTS MODELLED IN THE BALTIMORE IEMP AIR TOXICS
          STUDY BY SOURCE	V-21

V-8       BALTIMORE IEMP AIR TOXICS STUDY ESTIMATED ANNUAL  EXCESS
          CANCER INCIDENCE BASED ON MODELLING BY SOURCE AND
          POLLUTANT (TOTAL STUDY AREA, 5 KM GRID SYSTEM) .  . V-23

V-9       AREA-WIDE ANNUAL EXCESS CANCER INCIDENCE USING
          AVAILABLE MONITORING DATA	V-25

V-10      POLLUTANT CONTRIBUTIONS BASED ON MODELLING TO MAXIMUM
          INCREASED LIFETIME INDIVIDUAL CANCER RISK AT THREE
          HOTSPOT  LOCATIONS 	V-28

V-ll      TOTAL SOURCE AND POLLUTANT CONTRIBUTION BASED ON
          MODELLING AT EACH HOTSPOT LOCATION 	 V-30

V-12      BALTIMORE IEMP AIR TOXICS STUDY ESTIMATED ANNUAL  EXCESS
          CANCER INCIDENCE BASED ON MODELLING BY POLLUTANT  IN THE
          GRID CELL OF HIGHEST PREDICTED ANNUAL EXCESS CANCER
          INCIDENCE	V-34

V-13      BALTIMORE IEMP AIR TOXICS STUDY ESTIMATED MAXIMUM
          ANNUAL EXCESS CANCER INCIDENCE BY POLLUTANT AND SOURCE
          IN THE GRID CELL OF HIGHEST ESTIMATED ANNUAL EXCESS
          CANCER INCIDENCE	V-35

V-14      RECEPTOR LOCATIONS WARRANTING FURTHER INVESTIGATION FOR
          NONCANCER EFFECTS: POLLUTANT-SPECIFIC 	V-38

V-15      RECEPTOR LOCATIONS WARRANTING FURTHER INVESTIGATION FOR
          NONCANCER EFFECTS: COMPLEX POLLUTANT MIXTURES . . .V-40

V-16      BALTIMORE IEMP AIR TOXICS STUDY NONCANCER RATIO
          CALCULATIONS FROM EXPOSURE TO TARGET COMPOUNDS AT THE
          TWO FIXED EXPOSURE TO TARGET COMPOUNDS AT THE TWO FIXED
          TEAM MONITORING SITES 	V-42

V-17      A COMPARISON OF PREDICTED AVERAGE LIFETIME INDIVIDUAL
          CANCER RISK IN BALTIMORE WITH OTHER STUDIES . . . .V-47

-------
                             FIGURES


1-1       BALTIMORE IEMP STUDY AREA	1-3

IV-1      AREA SOURCE GRID SYSTEM FOR BALTIMORE STUDY .  .  . IV-27

IV-2      BALTIMORE INTEGRATED ENVIRONMENTAL MANAGEMENT PROJECT
          MAP OF AMBIENT AIR MONITORING IN 1987	IV-35

IV-3      BALTIMORE STUDY AREA MAP OF BOUNDARIES OF THE MODELLING
          DOMAIN FOR THE 5 KM (STANDARD) GRID SYSTEM . .  .  .IV-44

IV-4      BALTIMORE STUDY AREA MAP OF BOUNDARIES OF THE MODELLING
          DOMAIN FOR THE 2.5 KM (REFINE) GRID SYSTEM  .  .  . IV-45

ZV-5      BALTIMORE STUDY AREA MAP OF BOUNDARIES OF THE MODELLING
          DOMAIN	IV-47

V-l       MODELLING DOMAIN BY UTM COORDINATES  	  V-7

V-2       BALTIMORE IEMP AIR TOXICS POINT & AREA SOURCE
          CONTRIBUTION TO AVERAGE LIFETIME INDIVIDUAL CANCER RISK
          BY GRID CELL	V-8

V-3       BALTIMORE IEMP AIR TOXICS ANNUAL EXCESS CANCER
          INCIDENCE, AVERAGE LIFETIME CANCER RISK AND EXPOSED
          POPULATION BY GRID CELL	V-17

V-4       ANNUAL EXCESS CANCER INCIDENCE BY GRID CELL FOR THE
          STUDY AREA	V-18

V-5       BALTIMORE IEMP AIR TOXICS POINT & AREA SOURCE
          CONTRIBUTION TO ANNUAL EXCESS CANCER INCIDENCE BY GRID
          CELL	V-22

V-6       LOCATION OF HOTSPOTS EVALUATED IN THE AIR TOXICS
          STUDY	V-27

VI-1      BALTIMORE IEMP AIR TOXICS STUDY ANALYSIS OF THE COST-
          EFFECTIVENESS OF CONTROL OPTIONS	VI-24

VI-2      TOTAL ANNUALIZED COST VERSUS REDUCTION IN ANNUAL EXCESS
          CANCER INCIDENCE  	 VI-31

VI-3      TOTAL ANNUALIZED COST VERSUS REDUCTION IN MAXIMUM
          LIFETIME INDIVIDUAL CANCER RISK 	   VI-37

VI-4      SUMMARY OF COSTS OF CONTROLLING AIR TOXICS IN BALTIMORE
          COMPARISON OF BENEFIT METHOD AND COST-EFFECTIVE
          SCENARIOS	VI-45

-------
                          TABLES (CON'T)

V-18      A COMPARISON OF PREDICTED MAXIMUM LIFETIME INDIVIDUAL
          CANCER RISK IN BALTIMORE WITH OTHER STUDIES  .  .  .  .V-49

V-19      COMPARISON OF THE RESULTS OF ALTERNATIVE APPROACHES TO
          ESTIMATING THE RISKS FROM THE SEMI-VOLATILE  COMPONENT
          OF THE PRODUCTS OF INCOMPLETE COMBUSTION 	 V-53

VI-1      BALTIMORE IEMP AIR TOXICS STUDY ESTIMATES OF EXCESS
          ANNUAL CANCER INCIDENCE FOR SOURCES SELECTED FOR
          ANALYSIS OF CONTROL OPTIONS 	VI-4

VI-2      SOURCES POSING GREATER THAN 5 X 10'6 INCREASED  LIFETIME
          INDIVIDUAL CANCER RISK 	 VI-7

VI-3      SOURCES & POLLUTANTS INCLUDED IN THE CONTROL
          ANALYSIS	VI-8

VI-4      DESCRIPTION OF CONTROL OPTIONS FOR POINT SOURCES  VI-10

VI-5      DESCRIPTION OF CONTROL OPTIONS FOR AREA SOURCES  . VI-13

VI-6      CONTROL OPTIONS & ANNUALIZED COSTS  	 VI-18

VI-7      HEALTH AND WELFARE EFFECTS QUANTIFIED 	 VI-20

VI-8      CONTROL OPTIONS, ANNUALIZED BENEFITS,  AND COSTS FOR TSP
          AND VOC CONTROLS	VI-22

VI-9      BALTIMORE IEMP AIR TOXICS STUDY SCHEDULE OF  CONTROL
          STRATEGIES FOR REDUCING CANCER INCIDENCE PHASE II VI-27

VI-10     BALTIMORE IEMP AIR TOXICS STUDY SCHEDULE OF  CONTROL
          STRATEGIES FOR REDUCING LIFETIME INDIVIDUAL  CANCER RISK
          AT SITE OF MAXIMUM INDIVIDUAL CANCER RISK .... VI-35

VI-11     CONTROL STRATEGIES BASED ON BENEFITS ANALYSIS  .  . VI-40

-------
                         FIGURE (COMTD.)

VI-5      SUMMARY OF BENEFITS OF CONTROLLING AIR TOXICS IN
          BALTIMORE COMPARISON OF BENEFIT METHOD AND COST-
          EFFECTIVE SCENARIOS	VI-46

VI-6      SUMMARY OF NET ECONOMIC BENEFITS OF CONTROLLING AIR
          TOXICS IN BALTIMORE COMPARISON OF BENEFITS METHOD AND
          COST-EFFECTIVE SCENARIOS 	  VI-47

-------
                         EXECUTIVE SUMMARY

           The Baltimore IEMP Ambient Air Toxics Study


 INTRODUCTION

     This  report describes  the  second phase of  a two-phase  study
 of human health risks  from  ambient air toxics in the greater
 Baltimore  area.1  The results of the first phase of the study are
 presented  in  Baltimore Integrated Environmental Management
 Project;   Phase I Report.   The  air toxics study was conducted as
 part of the Baltimore  Integrated Environmental  Management Project
 (IEMP).  The  Regulatory Integrated Division (RID), Office of
 Policy Analysis, of the U.S. Environmental Protection Agency
 (EPA).  The Baltimore  IEMP  is one of four full-scale geographic
 projects initiated by  RID over  the past six years.  The other
 lEMPs focused on the Philadelphia metropolitan  area, Santa  Clara
 Valley, California, and Denver, Colorado.

     This  executive summary is  divided into five sections:

     •  Integrated Environmental Management, which describes the
        need  for and the basic  concepts of the  IEMP

     •  Baltimore IEMP,  which summarizes the structure of the
        overall project

     •  Ambient Air Toxics  Study Oblectives. which presents the
        goals of the ambient air toxics study

     •  Ambient Air Toxics  Study Design, which  discusses how the
        ambient air toxics  study was completed

     •  Conclusiona. which  summarizes the major findings of the
        ambient air toxics  study and insights on additional
        analysis that  is needed on this topic
     'This study estimated the increase in cancer and noncancer
risks resulting from exposure to ambient  (i.e., outdoor)
concentrations of air toxics.  It did not consider exposures
resulting from (1) indoor air, (2) the workplace, and (3) other
pathways  (e.g., ingestion).

-------
INTEGRATED ENVIRONMENTAL MANAGEMENT

     RID adopted the concept of integrated environmental
management as one solution to the shortcomings of EPA's
traditional focus on individual industries, pollutants, and media
in developing pollution control strategies.  Such a piecemeal
approach may result in environmental programs and regulations
that do not attack the greatest risks to human health first; it
may also lead to a combination of national standards that do not
address the unique environmental issues found at the local level.
The IEMP strives to set priorities by comparing similar risks
across all sources, pollutants, and exposure pathways.  The IEMP,
through its focus on individual geographic areas, is also
designed to address site-specific situations that are often
inadequately addressed by national regulations.

     The IEMP is grounded primarily in the concepts of risk
assessment and risk management.  Rough estimates of risk—that
is, the probability of adverse health effects—are used as a
common measure for comparing and setting priorities among
environmental issues that involve different pollutants, sources,
and exposure pathways.  Risk management is the process by which
decision makers balance programs to reduce human health risks
against the available resources to support those programs.
Chapter III and Appendix A provide more detail on risk assessment
and risk management as used in the IEMP.

     The Baltimore IEMP is also built on the concept that these
projects should be managed at the state and local level, thus
ensuring that issues of greatest local concern or risk are
adequately addressed.  The Baltimore IEMP was particularly
successful in this regard, as detailed below.


BALTIMORE IEMP

     The Baltimore IEMP was a cooperative effort involving the
governments of the State of Maryland, the City of Baltimore,
Baltimore County, Anne Arundel County/and EPA.  A map of the
study area is presented in Figure 1.  The Baltimore area was
chosen, not because it has a significant toxics problem, but
because EPA and local officials wanted to explore better ways to
identify, assess, and manage the potential human health risks of
environmental pollutants in the area.  In addition, Baltimore was
attractive because of the opportunity to establish a management
structure with the potential to continue after completion of the
project.

     The decision-making structure of the IEMP consisted of two
committees, which also represented the vehicles for State and
local participation:  the Management Committee (MC) and the

-------
Technical Advisory Committee (TAG).  The MC, with members
representing Baltimore City, Baltimore County, Anne Arundel
County, and the State, managed and oversaw the IEMP and set its
overall policy direction.  The TAG, composed of technical
managers from the City of Baltimore, the two counties, the State,
as well as representatives from the Maryland Regional Planning
Council and the academic community, advised the MC on the
technical and scientific aspects of the project, oversaw and
commented on all EPA and consultant work, and recommended and
suggested issues to study.  EPA provided administrative,
technical, and analytical support.  In the second phase of the
project, a Risk Assessment Review Panel (RARP), consisting of
scientists from John Hopkins University, advised the committees
on scientific and technical questions related to public health.
Also in Phase II, special workgroups with members from both the
TAG and representatives from industry, public interest groups,
government, and academia were organized around each priority
issue.

     The major task in Phase I of the Baltimore IEMP was to
identify environmental issues of concern in the study area and to
set priorities among them for further study and development of
control strategies in Phase II.  The Baltimore IEMP set
priorities on the basis of available information, supplemented by
data from a brief ambient air monitoring effort conducted by the
EPA.  Five environmental topics were chosen for further
examination in Phase II of the Baltimore IEMP.  In addition to
ambient air toxics, these issues were:

     •  Multi-media metals, which focused on developing cost-
        effective techniques for lead paint removal and dust
        abatement in Baltimore homes

     •  Underground storage tanks, which identified the ground-
        water resources at greatest potential risk if underground
        tanks leak

     •  Indoor air pollution, which sought to develop the
        information necessary to support discussion of possible
        programs to reduce exposure to indoor air pollution and
        to support the expansion of local government capability
        to respond to inquiries concerning indoor air pollution

     •  Baltimore Harbor, which defined current and future uses
        of the Harbor's waters and identified actions, additional
        research, and institutional arrangements necessary to
        help environmental decision makers improve water quality
        and habitat in the Harbor to achieve the desired uses

The results of these studies are summarized in separate reports
available from EPA's RID.

-------
AMBIENT AIR TOXICS STUDY OBJECTIVES

     The air toxics study was designed to test a new approach for
using risk assessment to set research and control priorities
among sources and pollutants contributing to the complex mixture
of ambient air toxics (often referred to as urban soup) in the
Baltimore area.  The Baltimore air toxics workgroup identified
four major project objectives:

     •  To compile the best possible air toxics emissions
        inventory given available resources

     •  To estimate annual average ambient air concentrations of
        selected air toxics resulting from emissions from both
        point (e.g., industrial) and area (e.g., cars and dry
        cleaners) sources; these compounds include volatile
        organic compounds (VOCs), semi-volatile products of
        incomplete combustion (polycyclic organic matter), and
        metals

     •  To estimate increased risks to human health posed by
        these compounds and by individual sources and source
        categories

     •  To demonstrate an approach for establish priorities by
        developing and analyzing control strategies to reduce the
        adverse health and environmental effects from emissions
        of selected air toxics

     In addition, the workgroup saw the Baltimore IEMP ambient
air toxics study as a complement to the air toxics strategies
recently developed by EPA and the Maryland AMA.  EPA's national
air toxics strategy was first announced in June 1985 and was
envisioned as a three-pronged effort for routine releases
consisting of:2

     •  An enhanced and refocused air-toxics control program for
        problems of national scope, building on the existing
        National Emissions Standards for Hazardous Air Pollutants
        (NESHAPs) program

     •  A new federal program to build state capability to deal
        with air toxics within their boundaries

     •  An expanded effort to devise strategies to reduce risks
        from multi-media/multi-source pollutants in specific
        localities where problems may exist
     2Source:  U.S. EPA, "A Strategy to Reduce Risks to Public
Health  from  Air Toxics," June  1985.

-------
The IBMP in its design and implementation was viewed as having
its largest impact on the third component of the strategy.
However, the extensive data collection activities that accompany
an IEMP would also provide State and local officials with the
tools they would need to start building their own programs.

     The air toxics program developed by the AMA comprises three
activities and is consistent with the EPA air toxics strategy:

     •  Continued enforcement of federal programs, such as
        NESHAPS regulations

     •  Recently adopted state regulations for industrial point
        sources

     •  An urban air toxics initiative to address risks from
        exposure to complex chemical mixtures (i.e., urban soup)

The workgroup saw the Baltimore ambient air toxics study
complementing the AMA's activities on its urban air toxics
initiative by providing insights on the relative risks for
different types of sources and pollutants, as well as the
relative costs, effectiveness (e.g., health risk reduction), and
societal benefits of alternative control strategies for reducing
these risks.
AMBIENT AIR TOXICS STUDY DESIGN

     The Baltimore air toxics study consisted of three major
activities:  (1) an initial pollutant selection designed to focus
the air toxics study on the most important pollutants; (2) a
screening investigation of the risks associated with exposure to
the selected air toxics, and (3) a demonstration of several
approaches for identifying and setting priorities among
alternative strategies to reduce the risks.

Pollutant Selection

     The pollutant selection resulted in two sets of compounds:
one for air dispersion modelling and the other for ambient air
monitoring.  The modelling was performed as part of the Baltimore
ambient air toxics study.  Air monitoring was conducted by EPA's
Office of Research and Development (ORD) as part of its Baltimore
TEAM study.  Although the Baltimore IEMP was not part of the
Baltimore TEAM study, we coordinated with ORD to take advantage
of the opportunity to obtain additional site-specific air
     3Code of Maryland Regulations 26.11.15,  Toxic Air
Pollutants.

-------
monitoring data.  In addition to the TEAM monitoring data, we
gathered readily available monitoring data for the Baltimore
area.  These data were drawn from sampling programs conducted by
the AMA over the past three years and the IEMP during Phase I of
the Baltimore IEMP (1983/1984).

     We used the following criteria to select compounds for
modelling:

     •  Likelihood of exposure or presence in the ambient air

     •  Existence of CAG unit cancer risk factors or threshold
        values for noncancer effects*

     •  Feasibility of control

The selection of pollutants for monitoring considered primarily
the goals of the TEAM study.3

A Screening Investigation of Air Toxics Risks

     After completion of the pollutant selection, rough estimates
of risks to human health were developed using the monitoring and
modelled data, standard EPA exposure assumptions, available dose-
response determinations made by EPA, and information on exposed
populations from the U.S. Census.  Several different measures of
risk were calculated in this study to provide insight on  the
relative importance of the pollutants and sources examined.

     First, at the recommendation of the Johns Hopkins Risk
Assessment Review Panel (RARP), we separately ranked the
pollutants and sources examined in the Baltimore IEMP by  their
cancer potency-weighted ambient concentrations.  The cancer
potency-weighted ambient concentrations were developed by
multiplying estimated average ambient air concentrations—either
by pollutant or source—by the appropriate unit risk factor(s).
This approach is advantageous because it  (1) avoids relying on
     4The unit cancer risk factor combines estimates of cancer
potency with EPA standard  exposure assumptions.  The unit  cancer
risk factor  represents  the upper-bound  estimate of  the  increase
in  cancer  risk  associated  with  continuous  lifetime  (i.e.,  70
years) exposure to 1  ug/m3 of a specific compound.

     5Some of the goals of the Baltimore TEAM study, a study
designed to  assess the  significance  of  indoor versus outdoor
exposures  to selected chemicals,  included:   sampling chemicals
examined in  past TEAM studies;  investigating chemicals  that have
not been adequately studied  in  the indoor  environment;  and
contributing new information on indoor  pollutant  levels in light
of  the extensive data already available.

-------
highly uncertain  exposure  assumptions used by EPA in its  risk
assessments  (e.g.,  continuous  exposures for 70 years), and  (2)
does not  incorporate  population weighting factors.  The obvious
downside  to  this  ranking scheme is that it cannot be used to set
priorities in populated versus unpopulated areas.

     The  second approach ranked the pollutants and sources
examined  using rough  quantitative estimates of risk.  This  latter
analysis  considered both cancer and noncancer effects.  For area-
wide cancer  risks,  the most  important risk measures were  average
increased lifetime  individual  cancer risk and annual excess
cancer incidence.   For so-called "hotspot" areas, the two cancer
risk measures of  interest  were maximum increased lifetime
individual cancer risk and annual excess cancer incidence in the
grid cell of highest  predicted annual excess cancer incidence.6'7


     Noncancer risks  were  evaluated in two ways.  First,  we
calculated in the increased  concern for several noncancer effects
(liver toxicity,  kidney toxicity, reproductive, neurological,
fetal, and blood) by  dividing  the predicted or measured ambient
air concentration of  each  pollutant by the no-effect threshold(s)
relevant  to that  pollutant.  If the resulting ratio exceeded one,
we identified these exposures  as a concern for noncancer  effects.
Second, we developed  a "hazard index" that summed individual
pollutant ratios  by effect category.  The latter analysis was
aimed at  examining  the impact  of exposure to complex chemical
mixtures  in the ambient air.   As the index approaches one,
concern for the potential  hazard of the chemical mixture
increases.  If the  index exceeds one, the concern is the  same as
if a no-effect threshold were  exceeded by the same amount by an
individual pollutant.

     We emphasize that the risk estimates presented in this
report are highly uncertain  and were calculated only for
screening purposes.   For example, the preliminary emissions
estimates used in the modelling effort were based on limited
information that was  available in 1985.  Since this time,
emission  estimates  have changed and important new sources have
been introduced.  Because  of these changes, the numbers and the
conclusions presented below  may no longer be valid.  Also,
because of the generally conservative bias in the underlying data
used in the risk  assessment, it is highly unlikely that the true
     6See Chapter III for detail on the different risk measures
and their estimation.

     7The term "hotspot"  refers to specific areas in which people
are exposed to higher than average concentrations of the target
pollutants.

                                8

-------
 risks would be as high as the risk  estimates,  and  they could  be
 considerably lower.

 A Demonstration of Risk Management  Strategies

      Because of the considerable  uncertainties underlying  our
 risk  estimates, as well as the limited resources available for
 developing appropriate control options and  costs,  we decided  that
 the risk management phase of  the  ambient  air toxics study  should
 be conducted as a demonstration.  To  this end,  we  used cost-
 effectiveness and benefit-cost analysis to  evaluate individual
 control  options and control option  strategies  to reduce the risks
 examined.

      A set of risk-based criteria were established to  identify
 the sources included in the risk  management demonstration:

      •  The source must contribute  approximately 1 percent of the
         estimated excess annual cancer incidence

      •  The source is  projected to  pose an  increased lifetime
         individual cancer risk of 5 x 10"6

      •  The source accounts for at  least  1  percent of  the
         estimated annual excess cancer incidence   at the grid
         cell of highest incidence

      For the selected  sources,  we identified feasible  control
 options  and estimated  preliminary costs and pollutant  removal
 efficiencies.   The removal efficiency estimates considered
 individual  toxic compounds, as  well as co-control  of particulates
 and volatile organic compounds  (VOCs).8 Actual costs  and removal
 efficiencies for a control at  a particular  facility may differ
 from  the engineering estimates  we generated.  Thus, a  detailed
 site-specific analysis  is  required  to confirm whether  an
 identified  option is practically  feasible.

      In  the cost-effectiveness  analysis,  pollutant-specific
 removal  efficiencies were  used  to quantify  the reduction in risk
 afforded by each control option.  For the benefit-cost  analysis,
 we used  available EPA  studies  to  quantify the economic  value  of
 reducing each ton of particulate  and VOC  associated with the
 control  options  evaluated.  A  computer (mixed integer  program)
 model was used to identify the  least-cost mix of control options
 that would  achieve either  a specified level of risk reduction or
monetary benefits.
     8Particulates and VOCs are called criteria pollutants
because they serve to define ambient air quality under the Clean
Air Act.

-------
CONCLUSIONS
Limitations

     To properly interpret the results presented below, the
following major limitations must be considered:

     •  This study was based on rough emissions data developed at
        the start of the project; budget and time constraints
        prevented us from updating the risk estimates when new or
        better emissions data became available.

     •  Because of conservative assumptions regarding the potency
        of compounds studied and the length of exposure (i.e., an
        assumed continuous lifetime (70 year) exposure), it is
        highly unlikely that the true risks would be as high as
        the estimates, and they could be significantly lower.

     •  We identified control options only for the purpose of
        developing rough ranges of expected costs associated with
        controls and the possible reductions in emissions that
        they realize; additional analysis will be needed to
        confirm whether these options are economically and
        technically feasible.

Because of these uncertainties, the results of the Baltimore
ambient air toxics study are best used to indicate which
pollutants and sources deserve further investigation.


Emissions Inventory

     •  We were successful in compiling a rough emissions
        inventory for the purpose of developing screening risk
        estimates. The rough emissions inventory was built using
        available data from the Maryland AMA and engineering
        estimates generated by EPA's technical contractor.  The
        risk estimates based on this inventory allowed us to
        develop a preliminary ranking of pollutants and sources
        deserving closer scrutiny.

     •  further work is needed to develop an emissions inventory
        that could support regulatory decisions.  To this end,
        several modifications are needed.  New sources need to be
        added to the inventory, and the emissions estimates at
        existing sources must be changed to reflect current
        controls and operating practices, especially at Point
        Source A.
                                10

-------
Screening Risk Assessment Results;  Cancer

     Pollutants

     We ranked the risk-potential of the pollutants examined in
the Baltimore IEMP using (1) the RARP approach and (2)
preliminary estimates of human health risk (both incidence and
individual risk) based on monitoring and modelling.  The results
of this effort are shown in Table 1 and support the following
conclusions:

     •  POM, chromium VI. benzene, and perchloroethvlene rank
        most consistently near the too.  Looking across the
        different ranking schemes and risk measures, POM,
        chromium VI, and benzene are generally the top three
        pollutants of concern.

     •  POM clearly ranks highest based on the modelled results,
        while perchloroethvlene ranks highest based on the
        monitoring data.  It is interesting to note that
        perchloroethylene ranks significantly lower (i.e., in the
        bottom half) based on modelled data.  POM was not
        considered categorically in the monitoring study;
        however, monitoring data were available for one of POM's
        constituents, B(a)P.

     •  Based on the RARP ranking. POM ranks five times higher
        than the next closest pollutant, chromium VI.  The
        modelled results also show the same trend.

     •  The available area-wide monitoring indicates that
        potential perchloroethvlene risks are twice as high as
        the next highest ranked pollutant, chromium VI.
        Continuing down the rank, perchloroethylene risks are
        about three times higher than benzene, about seven times
        higher than chloroform, and over ten times higher than
        all other pollutants included in the area-wide
        monitoring.  However, it should be noted that the
        perchloroethylene monitoring data are highly uncertain.

     •  The monitoring data show that all pollutants individually
        pose average increased lifetime individual cancer risks
        greater than 1 x 10'".  The risk results based on source
        modelling indicate that at the "hotspots," only two-
        thirds of the pollutants examined pose maximum increased
        lifetime individual cancer risks greater than 1 x 10" .
                                11

-------
     •  Based on monitoring, three pollutants—perchloroethvlene,
        hexavalent chromium, and benzene—account for
        approximately 87 percent of the estimated area-wide
        average increased lifetime individual cancer risk.  The
        relative contribution of each of these three compounds
        is:  perchloroethylene (48 percent), hexavalent chromium
        (24 percent), and benzene (15 percent).

     •  The modelling results indicated that POM alone accounts
        for about two-thirds of the area-wide average increased
        lifetime individual cancer risks.   Chromium VI
        contributes an additional 15 percent, and a variety of
        other pollutants account for 10 percent or less.  It is
        interesting to note that when looking at the maximum
        increased lifetime individual cancer risks, chromium VI
        poses the highest concern, followed by POM.

     Sources

     Similar to the risk ranking performed for individual
pollutants, we evaluated and ranked the potential risks posed by
selected sources and source categories in the Baltimore study
area.  Two ranking schemes were established using (1) the RARP
risk-ranking approach and (2) preliminary estimates of
quantitative risk (both incidence and individual risk).  Because
of the extreme difficulty in relating ambient air concentrations
from monitoring back to individual release points, the ranking of
sources was based solely on modelled data.

     •  Based on the RARP ranking approach, which indicates the
        potential for a source category to pose a risk to human
        health. Point Source A ranked approximately seven times
        higher than the next highest source, road vehicles.
        Point Source A also ranked over ten times higher than all
        other point sources combined.

     •  The preliminary risk estimates show that point and area
        sources account for similar percentages of the estimated
        annual excess cancer incidence.  Point and area sources
        contribute approximately 45 percent and 55 percent of the
        total, as shown in Figure 2.  Most of the area source
        fraction is attributable to road vehicles (33 percent).

     •  The rouohlv even contribution of point and area sources
        to the estimated annual excess cancer incidence does not
        hold for individual pollutants.  This is easily seen when
        the two key pollutants in the modelling analysis of
        cancer incidence, POM and chromium VI, are examined more
        closely.  Whereas both point and area sources account for
        similar portions of the estimated annual excess cancer
        incidence associated with POM (approximately 44 percent
        versus 56 percent, respectively),  point source emissions

                                13

-------
                         FIGURE 2
     SOURCE CATEGORIES CONTRIBUTING TO ANNUAL
      EXCESS CANCER INCIDENCE IN THE BALTIMORE
                         AREA
100%
 80%
 60%
 40%
 20%
  0%
               5%
7%
                                                  1%
     Point Sources Road Vehicles  Heating  Solvent Usage   Other
                       Source Category

-------
        account for almost all (99.6 percent) of the estimated
        annual excess cancer incidence associated with hexavalent
        chromium.

     •  Approximately 80 percent of the estimated annual excess
        cancer incidence is attributable to road vehicles,
        heating, and Point Source A.  The other point sources
        evaluated (Point Sources B, C, and D) account for an
        additional 10 percent; solvent usage and other sources,
        primarily minor point sources, account for the final 10
        percent.

     •  As shown in Figure 3, point sources were the malor
        contributors to the average increased lifetime individual
        cancer risks.  Point sources account for almost all of
        the risk at those locations with the highest estimated
        lifetime individual cancer risks.

     •  As shown in Figure 4. point sources account for roughly
        92 percent of the maximum increased lifetime individual
        cancer risks at the three "hotspots" evaluated in the
        Baltimore IEMP.  Road vehicles accounts for an additional
        5 percent.


        the estimated maximum increased lifetime individual
        cancer risk at two of the three "hotsoots."  At the third
        hotspot, Point Sources A, B, and C together account for
        roughly 83 percent of the increased lifetime individual
        cancer risk.  Road vehicles and heating account for the
        remainder.

     Preliminary Cancer Risk Estimates

     Before summarizing our preliminary cancer risk results,
several limitations are noted:

     •  This study estimated the increase in cancer resulting
        from exposure to ambient (i.e.. outdoor) concentrations
        of air toxics.  It did not consider exposures resulting
        from (1) indoor air, (2) the workplace, and (3) other
        pathways (e.g., ingestion).

     •  This study used conservative estimates of increased
        cancer risk to establish priorities among pollutants and
        sources.  The risk estimates are calculated using
        modelled and monitored concentrations and EPA unit cancer
        risk factors.  There is considerable uncertainty in the
        estimated concentrations, which could either overstate or
        understate the true concentrations  (see Chapter IV).
        Unit cancer risk factors combine CAG potency estimates
        with EPA exposure assumptions.  The CAG potency estimates

                                15

-------
                                                FIGURE  3

                                          Baltimore IEMP Air Toxics

                                  Point & Area Source Contribution  to  Average
                                   Lifetime Individual Cancer Risk  by Grid Cell
                                      Average Lifetime Individual Cancer Risk
                                  Point Source Contribution to Average Lifetime

                                              individual Cancer Risk
                                      Area Source Contribution  to Increased

                                                Individual Cancer Risk
                         0.000
                                                                           UTU COORDINATES
  u: The risk scale for the area source contribution
  i^ntficanui larger than that used for the avenge
ifeame individual cancer risk and point source contribution
  jphs.
16

-------
                         FIGURE 4
         SOURCE CATEGORIES CONTRIBUTING TO
       MAXIMUM INCREASED LIFETIME INDIVIDUAL
         CANCER RISKS IN THE BALTIMORE AREA
100%
 80%
 60%
 40%
 20%
  0%



-




/







/






/
92%




5% 1% 1% 1%
/ A , ,
1 . V / . / / . / / . /
     Point Sources Road Vehicles  Heating  Solvent Usage
                       Source Category
Other

-------
        provide a plausible upper limit to the cancer risk of a
        compound (see Appendix A); however,  the true value of the
        risk is unknown and may be as low as zero.   The exposure
        assumptions are extremely conservative in that they
        assume continuous exposure to outdoor air for 70 years.
        Because of the generally conservative bias in the
        information/ it is highly unlikely that the true risks
        would be as high as the estimates/ and they could be
        considerably lower.


     •  Based on the model performance evaluation,  the dispersion
        model consistently underpredicted ambient air
        concentrations for all pollutants considered, except
        trichloroethvlene.  Predicted ambient air concentrations
        of benzene and benzene-ring compounds (e.g., xylene,
        toluene, and ethyl benzene) are likely to be
        underestimated by at least a factor of two.  Estimated
        concentrations of trichloroethylene may have overstated
        by a factor of two.  The predicted values for the other
        compounds evaluated (e.g., chloroform, carbon
        tetrachloride, lf2-dichloroethane, 1,2-dichloropropane)
        are best seen as lower estimates of actual ambient air
        concentrations.  Consequently, our exposure estimates
        based on modelling are understated;  however, our
        estimates of risk using these modelled values are still
        more likely to overstate than understate true risk.

     Bearing these important limitations in mind, the following
risk results are presented:

     •  Based on modelling, the estimated average increased
        lifetime individual cancer risks in Baltimore are
        generally in the same range as those predicted for other
        urban study areas.  The comparison was made using results
        from:  the Philadelphia IBMP; the Kanawha Valley, West
        Virginia, IEMP; the Santa Clara Valley, California, IEMP;
        and the South Coast Air Quality Management study of the
        Los Angeles basin.  The estimated maximum increased
        lifetime individual cancer risks tended to be lower in
        Baltimore than in these other areas.

     •  Based on modelling, the area-wide average increased
        lifetime individual cancer risk was fairly uniform across
        all sections of the study area, approximately 1.5 x 10' .
        The heavily industrialized southeast section of the study
        area had the highest average lifetime individual cancer
        risk.
                                18

-------
0  The available area-wide ambient air monitoring indicated
   an eatimated average increased lifetime individual cancer
   risk of 3-Q x 10.This result is approximately a
   factor of three higher than that predicted by the
   modelling.

•  The average increased lifetime individual cancer risks at
   the two TEAM monitoring sites were eatimated to be
   approximately 1.5 x 10'" and 1.2 x IP"1, respectively.
   These results are very close to the estimated area-wide
   averages based on modelling.

•  The maximum increased lifetime individual cancer risks at
   the three "hotsoots" evaluated were estimated to range
   from 5.4 x IP"1 to 1.3 x 10"*.  These results  are
   approximately four to nine times higher than the
   estimated area-wide average individual risks.

•  Based on modelling and area-wide monitoring, four and
   twelve annual excess cancer cases were predicted,
   respectively.  The difference between the estimates can
   be explained largely by higher monitoring-based exposure
   estimates for perchloroethylene, hexavalent chromium, and
   benzene.

•  The modelling results indicate an excess of 0.15 annual
   cancer cases in the arid cell of highest predicted cancer
   incidence.  This grid cell accounts for less than 1
   percent of the study area and approximately 4 percent of
   the area-wide estimated excess annual cancer incidence.

•  The average increased individual cancer risk in the grid
   cell of highest incidence, while twice the area-wide
   average, is less than the average increased individual
   risk estimates in other arid cells.  Figure 5 shows that
   the grids with the highest average increased lifetime
   individual cancer are not necessarily the grids with the
   highest annual excess cancer incidence.  This information
   indicates that population density can be a more important
   factor than exposure levels in identifying high-incidence
   areas.

Air Toxics Risk Perspective

•  The increased cancer risks from ingestion of chloroform,
   a trihalomethane. in drinking water are comparable to
   those from air toxics.  The annual excess cancer
   incidence from ingestion of chloroform in drinking water
   was estimated at approximately three cases; the average
   increased lifetime individual cancer risk was computed
   roughly at 1 x 10"*.  More detail on this  issue can be
   found in Chapter V.

                           19

-------
                 FIGURE  5

           Baltimore IEMP Air Toxics

Annual Excess Cancer Incidence, Average Lifetime
 Cancer  Risk and Exposed Population by Grid Ceil
         Annual Excess Cancer Incidence
   Average Lifetime Individual Cancer Risk
           Exposed  Population
                       20

-------
     •  In our comparison of the approach used to quantify POM
        risks in the Baltimore IBMP and the BfalP surrogate
        approach used in the Sixth Month Study, we found that the
        B(alP surrogate approach resulted in greater estimates of
        annual excess cancer incidence (6.8 versus 2.1\, but
        lower estimates of the increaaed lifetime individual
        cancer risks for the three hotaoot sites (Hotspot #1;
        7.91 x 10"J va. 3.7 x 10'"; Hotsoot *2t  1.4 x  10"*  va.  6.3  x
        10'"-. and Hotspot »3t 1.3 x 1Q'J vs.  6.3 x IQ'M.   The
        reason for this seeming contradiction is that the B(a)P
        surrogate approach uses only one unit cancer risk factor
        for all emissions of particulates that result from
        incomplete combustion of organic matter.  In contrast,
        the unit cancer risk factors we used for the semi-
        volatile fraction of the product of incomplete combustion
        in the Baltimore IBMP are specific to types of sources of
        combustion products.


Screening Risk Assessment Results»   Noncancer

     As with our cancer risk assessment, our findings for
noncancer effects are preliminary and subject to several
important caveatss

     •  This study used preliminary no-effect threshold values to
        estimate noncancer risks.  No-effect threshold values for
        six categories of non-cancer effects were developed by
        EPA toxicologists by pollutant.   These values are
        extremely uncertain and have not all undergone extensive
        peer review.

     •  The no-effect threshold value for blood effects is
        especially uncertain.   This threshold value had not yet
        undergone peer review and is, thus, subject to change.
        If this no-effect threshold value is found to be accurate
        (2.45 ug/m3  or approximately  0.8 ppb),  monitoring data
        across the country indicate that the benzene no-effect
        threshold value for blood effects is exceeded in almost
        every urban area in the country.

     Our major findings on the need to further explore noncancer
effects are summarized below:

     •  The modelled data indicate that there is a need to
        further examine benzene and xvlene exposures in the study
        area.  The refined (2.5 km) grid analysis shows benzene
        ambient concentrations exceed the no-effect threshold for
        blood effects in the southeast quadrant of the study
        area.  The discrete receptor analysis supports the need
        to investigate benzene.  It also identifies a section of

                                21

-------
        the study area in which xylene exposures may need to be
        looked into more carefully because they exceed the no-
        effect thresholds for liver, kidney,  reproductive,
        neurological, fetal, and blood effects.

     •  The available monitoring data and the TEAM data for the
        two fixed sites both show that benzene exposures exceed
        the threshold for blood effects.  Further work is needed
        to confirm our measured concentrations of benzene.

     •  The noncancer risk results were driven primarily bv
        exposure to a single pollutant, and not mixtures.
        Further work is needed to verify the noncancer threshold
        values to confirm this finding.


Risk Management;  A Demonstration

     •  we identified control options only for the purpose of
        developing rough ranges of expected costs associated with
        controls and the possible reductions in emissions that
        thev realize.  Additional analysis will be needed to
        confirm whether these options are economically and
        technically feasible.

     •  Preliminary annualized costs of individual controls
        ranged from a savings of S220.000 for the recovery of
        trichloroethvlene from decreasing operations to S150
        million for converting residences currently burning oil
        for heat to natural gas.  Control options and their costs
        were identified using available information and
        engineering judgment.  Additional site-specific analysis
        would be required to confirm the feasibility of these
        options.

     •  Eight control options analyzed in this project could
        result, if implemented, in  savings while reducing risks.
        These control options are designed to recover solvents.

     •  Of the control options that did not result in a net
        savings, control potion 4 at Point Source C achieved the
        most incidence reduction per dollar spent.  However, this
        option would reduce only 0.2 excess cancer cases every  70
        years.

     •  Implementing either control option 2 or control option  3
        at Point source B would result in approximately 10 cancer
        cases avoided over a 70-vear period.  Of the control
        options that will not result in a net  savings, these
        options are  some of the most cost-effective.  Option 3
        for miscellaneous industrial use of methylene chloride  is


                                22

-------
also fairly cost-effective, but will only reduce three
cancer cases over a 70-year period.

Relatively small reductions in the area-wide annual
excess cancer incidence were identified using coat-
effectiveness analysis.  The control strategies
identified by the model to lower the estimated annual
excess cancer incidence ranged from a reduction of
approximately 7 percent at a cost savings (attributable
to solvent recovery at area sources) to a reduction of 24
percent at an annual cost of cost about $211 million.
Roughly one case every two years could be avoided at a
cost of approximately $1.4 million.

The control strategies identified would reduce the annual
excess cancer incidence at the location of the highest
estimated annual excess cancer incidence 10.15 excess
cases per year) between 39 and 43 percent. This area
accounts for less than 1 percent of the study area
evaluated.

We can achieve significant reductions in the maximum
increased lifetime individual cancer risk (approximately
46 percent^ at an annualized cost of S46 million.  The
most important costs would be borne by a single facility.
Also, after control the resulting maximum increased
lifetime individual cancer risk, about 7 x 10**, would
still be significantly above the average area-wide
lifetime individual cancer risk of about 1 to 2 x 10"*.

The strategies for reducing annual excess cancer
incidence and the maximum increased lifetime individual
cancer risk would reduce benzene ambient levels below the
no-effect threshold for noncancer effects.  However, the
strategies considered in this study would not reduce
xylene ambient air concentrations, which also caused
concern for noncancer health effects.

Controlling emissions of air toxics, and considering the
co-control for ozone precursors and particulates. can
lead to net health and welfare benefits of approximately
SB.5 million per year, while reducing annual excess
cancer incidence bv approximately 16 percent.  Selection
of control strategies to achieve a risk reduction level
above 19 percent will result in benefits failing to keep
pace with the control cost increases.
                        23

-------
I. INTRODUCTION
1. INTEGRATED ENVIRONMENTAL MANAGEMENT PROJECTS

     This report describes the second phase of a two-phase study
of human health risks from air toxics in the greater Baltimore
area.  The results of the first phase of the study are presented
in Baltimore Integrated Environmental Management Project; Phase I
Report.  The study was conducted as part of the Baltimore
Integrated Environmental Management Project (IEMP).  The
Regulatory Integration Division (RID) of the Environmental
Protection Agency (EPA) initiated the project as part of its
pursuit of new approaches to environmental management and policy.
The purpose of the IEMP was to identify and assess the
significance of selected environmental issues, to set priorities
for action among these issues, and to analyze alternative
approaches to manage the risks that are assessed.

     EPA adopted the concept of integrated environmental
management as a potential solution to the shortcomings of the
traditional approach for pollution control.  The traditional
approach of focusing on one pollutant or class of pollutants
within each medium at a time may result in environmental programs
and regulations that do not use resources as efficiently as
possible.  Grounded in the concepts of risk assessment and risk
management, the IEMP uses estimates of risk—that is, the
probability of adverse effects—as a common measure for comparing
and setting priorities among environmental issues that involve
different pollutants, sources, and exposure pathways and that may
affect human health, ecosystems, and resources.  The need to set
priorities is prompted by the realization over the past ten years
that hundreds of chemicals present in our environment can pose
some risk of causing cancer or other adverse health effects.

     Comparing risks helps to establish priorities.  It allows
environmental managers to focus limited resources in a manner
that will achieve the greatest public benefit.  The structure of
the lEMPs is also built on the concept that these projects should
be managed at the state and local level, thus ensuring that
issues of greatest local concern or risk are adequately
addressed.  The Baltimore IEMP was particularly successful in
this regard.
                               I-l

-------
2. ORGANIZATION OF THE PROJECTS

     Each IEMP is divided into two phases.  In the first, project
managers establish the decision-making structure of the project,
identify key environmental issues, and set priorities among them.
Risk is but one of the criteria used in ranking issues; the
others include analytical feasibility, relevance to EPA, state
and local program objectives, and the potential for effective
mitigation.  In the second, the IEMP studies the priority issues
in greater detail and develops and evaluates potential strategies
for their control or resolution.
3. THE BALTIMORE IEMP

     The Baltimore IEMP was a cooperative effort involving the
governments of the State of Maryland, the City of Baltimore,
Baltimore County, Anne Arundel County, and EPA.  A map of the
study area is presented in Figure 1-1.  The Baltimore area was
chosen, not because it has a significant toxics problem, but
because EPA and local officials wanted to explore better ways to
identify, assess, and manage the potential human health risks of
environmental pollutants in the area.  In addition, Baltimore was
attractive because of the opportunity to establish a management
structure with the potential to continue after completion of the
project.  This project represents the second of four, full-scale
geographic projects initiated to date.  The others are the
Philadelphia IEMP, the Santa Clara, California IEMP, and the
Denver IEMP.

     The decision-making structure of the IEMP consisted of two
committees that also served as vehicles for State and local
participation!  the  Management Committee (MC) and the Technical
Advisory Committee (TAG).  The MC, with members representing
Baltimore City, Baltimore County, Anne Arundel County, and the
State, managed- and oversaw the IEMP and set its overall policy
direction.  The TAG, composed of technical managers from the City
of Baltimore, the two counties, the State, as well as
representatives from the Maryland Regional Planning Council and
the academic community, advised the MC on the technical and
scientific aspects of the project, oversaw and commented on all
EPA and consultant work, and recommended and suggested issues to
study.  EPA provided administrative, technical, and analytical
support.  In the second phase of the project, a Risk Assessment
Review Panel, consisting of scientists from Johns Hopkins
University, advised the committees on scientific and technical
questions related to public health.  Also in Phase II, special
workgroups with members from both the TAG and representatives
from industry, public interest groups, government, and academia
were organized around each priority issue.
                                1-2

-------
       FIGURE 1-1

BALTIMORE IEMP STUDY AREA
   PRINCE
  GEORGE'S
    CO.
             1-3

-------
4.  THE RESULTS QF THE PHASE I PRIORITY-SETTING PROCESS

     The major task in Phase I was to identify environmental
issues of concern in the study area and to set priorities among
them for further study and development of control strategies in
Phase II.  The Baltimore IEMP set priorities on the basis of
available information, supplemented by data from a brief ambient
monitoring effort conducted by EPA.1

      Five environmental topics were chosen for further
examination in Phase II of the Baltimore IEMP:

     1)   Ambient air toxics.  The goal was to investigate and
          demonstrate alternative approaches for reducing risks
          from ambient air toxics in urban areas like Baltimore
          at the least cost to society.

     2)   Multi-media metals.  The goal was to develop cost-
          effective techniques for lead paint removal and dust
          abatement in Baltimore homes.

     3)   Underground storage tanks.  The goal was to develop a
          strategy for identifying those ground-water resources
          at greatest potential risk if underground tanks leak.

     4)   Indoor air pollution.  The goal was to develop the
          information necessary to support (1) discussion of
          possible programs to reduce exposures to indoor air
          pollution and (2) the expansion of local government
          capability to respond to inquiries concerning indoor
          air pollution.

     5)   Baltimore Harbor.  The goals were to define current and
          future uses of the Harbor's waters and identify
          actions, additional research, and institutional
          arrangements necessary to help environmental decision
          makers improve water quality and habitat in the Harbor
          to achieve the desired uses.


5.  RELATIONSHIP TO THE EPA AND AMA AIR TOXICS STRATEGIES

     While this report summarizes primarily the results of the
Baltimore IEMP air toxics study, it is important to note the ways
in which the Baltimore IEMP—especially the Baltimore air toxics
     'please see Chapter IV of Baltimore integrated Environmental
Management Prolect: Phase I Report (hereafter referred to as the
Phase I report) for a detailed account of the priority-setting
process in the first phase of the IEMP.

                               1-4

-------
study—has complemented the air toxics strategies recently
developed by  EPA and the Maryland Air Management Administration
(AMA).

     EPA's national air toxics strategy was first announced in
June 1985 and was developed considering both routine and
accidental releases.2  For routine releases of hazardous air
pollutants (the focus of the IEMP, as well as the AMA's program),
EPA's strategy was envisioned as a three-pronged effort
consisting of:

     (1)  An  enhanced and refocused air toxics control program
          for problems of national scope, building on the
          existing National Emissions Standards for Hazardous Air
          Pollutants (NESHAPS) program

     (2)  A new federal program to build state capability to deal
          with air toxics within their boundaries

     (3)  An  expanded effort to devise strategies to reduce risks
          from multi-media/multi-source pollutants in specific
          localities where problems may exist

The IEMP in its design and implementation can have its largest
impact on the third component of the strategy.  However, the
extensive data collection activities that accompany an IEMP can
also provide  state and local officials with the tools they need
to start building their own programs.

      The air toxics program developed by Maryland's Air
Management Administration (AMA) consists of three activities:
(1) continued enforcement of federal programs, such as NESHAPS
regulations,  (2) recently adopted state regulations for
industrial point sources, and (3) an urban air toxics initiative
to address risks from exposures to complex chemical mixtures
(sometimes referred to as "urban soup") in metropolitan areas
like Baltimore.3   This  program is very consistent with the  EPA
air toxics strategy.  The Baltimore IEMP air toxics study has its
greatest influence on the AMA urban air toxics initiative.  It
complements the State's activities in this area by providing
insights on the relative risks for different types of sources and
pollutants, as well as the relative cost-effectiveness of
alternative control strategies for reducing these risks.
     2For further information,  see:  U.S.  EPA,  "A Strategy to
Reduce Risks to Public Health from Air Toxics," June 1985.

     3The recently adopted industrial point source regulations
can be found in:  Code of Maryland Regulations 26. 11. 15, Toxic
Air Pollutants.

                               1-5

-------
6.  OVERVIEW OF THE REPORT

     This report is organized into six chapters including this
introduction.  These chapters are:

     •  Chapter lit  Ob1actives and Design of the Baltimore
        Ambient Air Toxics Study

     •  Chapter III;  Summary of the Principles of Risk
        Assessment and Risk Management as Used in the Ambient Air
        Toxics Study

     •  Chapter IV:  Selecting Pollutants. Evaluating Toxicitv.
        and Estimating Exposure

     •  Chapter V;  Risk Assessment Screening Results

     •  Chapter VI;  Risk Management Strategies for Reducing
        Risks;  A Demonstration

     In addition, this report contains an Executive Summary that
summarizes the objectives and findings of the Baltimore IEMF
Ambient Air Toxics Study.  An Epilogue follows the Executive
Summary that presents -the Ambient Air Toxics Workgroup's insights
on the EPA/State relationship/ the overall IBMP process/ and
technical issues.  Appendixes are included to provide more detail
on some of the supporting data.  Additional background
information is referenced throughout the report and is available
from EPA's RID.
                               1-6

-------
II.  OBJECTIVES AMD DESIGN OF THE BALTIMORE AMBIENT AIR TOXICS
     STUDY

     This chapter presents an overview of the Baltimore air
toxics study, focusing on the study objectives, the study design,
and how the results of this study should be viewed.


1.  STUDY OBJECTIVES

     The Baltimore IEMP air toxics study was designed to test a
new approach for using risk assessment to set control and
research priorities among sources and pollutants contributing to
the complex mixture of ambient air toxics (often referred to as
"urban soup") in the Baltimore area.

     The Ambient Air Toxics Workgroup identified four major
project objectives of the Baltimore air toxics study:

     •  To compile the best possible air toxics emissions
        inventory given available resources

     •  To estimate annual average ambient air concentrations of
        selected air toxics resulting from emissions from both
        point (industrial) and area (e.g., cars and dry cleaners)
        sources; these compounds included volatile organic
        compounds (VOCs), semi-volatile products of incomplete
        combustion (polycyclic organic matter), and metals

     •  To estimate increased risks to human health posed by
        these compounds and by individual sources and source
        categories

     •  To demonstrate alternative approaches for establishing
        priorities by developing and analyzing control strategies
        to reduce the adverse health and environmental effects
        from emissions of selected air toxics

     These objectives would allow us to answer the following
analytical questions:

     •  Can we develop a screening methodology to identify and
        characterize health risks posed by selected air toxics
        from exposures occurring in the "urban soup?"
                               II-l

-------
        What is the relative contribution of point sources versus
        area sources to urban air toxics in the Baltimore area?

        What is the relative importance of specific point or area
        sources to urban air toxics risk?

        What is the relative importance of specific air
        contaminants, by source, to urban air toxics risk?

        What are the relative costs and effectiveness of
        alternative control options to reduce human health risk
        from air toxics in the Baltimore area?

        Will the control of carcinogenic emissions lead to
        reduced .concern for noncarcinogenic health effects from
        air toxics?

        Will the control of air toxic emissions also result in
        lower emissions of criteria air pollutants, in particular
        precursors to ozone, such as volatile organic compounds
        (VOCs) and particulates?

        What are the health and welfare benefits that accrue from
        reduced emissions of VOCs and particulates which result
        from controlling air toxics emissions at specific
        sources?
2.  STUDY DESIGN

     The Baltimore air toxics study consisted of three major
activities:  (1) an initial pollutant selection designed to focus
the air toxics study on the most important pollutants; (2) a
screening investigation of the risks associated with exposure to
the selected air toxics, and (3) a demonstration of alternative
analytical approaches for identifying strategies to reduce the
risks (i.e., cost-effectiveness and benefit-cost analysis).  In
this section, we describe briefly the general steps that we
performed as part of each of these activities.  The pollutant
selection process and results are discussed in more detail in
Chapter IV.  The results of the exposure and risk
characterizations are detailed in Chapters IV and V,
respectively.  Finally, Chapter VI summarizes the approach and
results of our demonstration project to explore alternative risk
management strategies.

     a.  Pollutant Selection

     Because of time and resource constraints, the first task in
the air toxics study was to identify the most important
pollutants that could be examined in detail.  Two sets of
compounds were identified for the purpose of air dispersion

                               II-2

-------
modelling and for ambient air monitoring.  These lists were
compiled using data we collected on emissions, exposures,
potencies, the potential for control, and criteria developed by
the Workgroup.

     b.  Risk Characterization of the Selected Pollutants

     After completing the pollutant selection, we focused our
efforts on collecting the additional data that would allow us to
perform a quantitative risk assessment.  As discussed in Chapter
III, there are four steps in risk assessment:  hazard
identification, dose-response evaluation, human exposure
evaluation, and risk characterization.  The Baltimore IEMP did
not perform the first two steps.  We relied on EPA determinations
regarding hazard identification and dose-response for our
studies.  To quantify human exposure, information was needed on
ambient air levels, duration of exposure, and the number of
exposed individuals.

     We generated data on ambient levels using results from
predictive (i.e., fate and transport) modelling and ambient air
monitoring.  To use dispersion models, one must have information
on pollutant emissions by source and meteorological conditions.
We generated the emissions data by compiling an emissions
inventory of point and area sources using available information
from the AMA and engineering estimates prepared by our technical
contractor.  These data were quality controlled and entered into
our data management system, PIPQUIC.

     The monitoring data used in this study were collected by
(1) EPA's Office of Research and Development as part of a special
monitoring study called TEAM (Total Exposure Assessment
Methodology)1 that they were conducting in the Baltimore area,
(2) EPA's Regulatory Integration Division during Phase I of the
Baltimore IEMP, and (3) AMA as part of their routine and special
project toxics monitoring.

     To calculate exposure to the estimated ambient air
concentrations, we relied on standard EPA exposure assumptions,
which  are highly conservative in that they assume continuous
exposure to predicted concentrations for a lifetime of 70 years.
Chapter III and Appendix A provide more information on the EPA
exposure assumptions.   Information on the population exposed by
      'The focus of the TEAM study is to develop methods for
quantifying  exposures  to  pollutants present  in the  immediate
vicinity of  an individual as  he  or she  goes  about his  or her
daily activities.  Readers interested in  this study should
contact  U.S.  EPA,  Office  of Research and  Development,
Environmental Monitoring  Systems Laboratory, for more
information.
                               II-3

-------
geographic area was obtained from the 1980 U.S. Census at the
block grid/enumeration district level.

     The risk characterization combined all of the information
generated as part of the above steps.  For carcinogenic effects/
we estimated (1) maximum increased lifetime individual cancer
risk/ (2) average increased lifetime individual cancer risk/ and
(3) annual excess cancer incidence.

     For noncancer effects/ we calculated the ratio of the
estimated ambient air concentration to the no-effect threshold
for individual pollutants.  An exceedance of the threshold
indicates that pollutant-specific exposures require additional
investigation.  Because individuals in urban environments are
exposed to complex chemical mixtures, we also explored whether
exposure to multiple pollutants with the same noncancer effects
required additional study using a "Hazard Index."  The hazard
index is calculated simply by summing the ratios of estimated
ambient air concentration to the no-effect threshold for all
pollutants with the same noncancer effect.  If the index exceeds
one, there is a need to further examine these exposure levels.

     c.  Risk Management:  A Demonstration

     Because of the considerable uncertainties underlying our
risk estimates, as well as the limited resources available for
developing appropriate control options and costs, the Workgroup
decided that the risk management phase of this study should be
conducted as a demonstration.  To this end/ we analyzed different
approaches for identifying strategies to reduce the risks
examined.

     For the sources selected as the focus of our demonstration,
we identified feasible control options and estimated their
associated costs, efficiencies, and benefits by employing
engineering estimates and EPA technical documents developed to
support various regulatory activities.  We then packaged these
options into strategies for consideration by local risk managers
using two approaches:  cost-effectiveness and benefit-cost
analysis.


3.  HOW THE RESULTS OF THE AIR TOXICS STUDY SHOULD BE VIEWED

     The Phase II study of air toxics in the Baltimore area was
intended as a demonstration of an analytical approach for ranking
strategies to reduce the increased risk from toxics in the
                               II-4

-------
ambient air.2  We made no determinations  regarding  the  absolute
magnitude of risk nor the need for regulatory action to reduce
emissions of air toxics, in general.  The risk estimates are
intended only for ranking risks and establishing pollutant and
source category priorities for more detailed evaluation.  Because
of conservative assumptions (that is, protective of public
health) regarding the potency of compounds studied and the length
of exposure  (i.e., an assumed continuous lifetime (70 year)
exposure), it is highly unlikely that the true risks would be as
high as the estimates, and they could be significantly lower.

     We identified control options only for the purpose of
demonstrating a risk management approach.  Rough ranges of
expected costs associated with controls and the possible
reductions in emissions that they realize were used in the
demonstration.  Economic and engineering considerations may
preclude their use for the specific sources identified.

     Finally, we emphasize that this study was based on rough
emissions data developed at the start of the project.  Budget and
time constraints prevented us from updating the risk estimates
when new or better emissions data became available.  Thus, the
results summarized in this report may not reflect current
conditions and are best used to indicate where additional
analysis is needed to verify our estimated exposures and risks.
     2This study estimated the increase in cancer and noncancer
risk resulting from exposure to ambient (i.e., outdoor)
concentrations of air toxics.  It did not consider exposures
resulting from (1) indoor air, (2) the workplace, and (3) other
pathways  (e.g., ingestion).
                               II-5

-------
III.  SUMMARY OF THE PRINCIPLES OF RISK ASSESSMENT AND RISK
      MANAGEMENT AS USED IN THE AMBIENT AIR TOXICS STUDY

     The Baltimore IBMP air toxics study was a demonstration of
techniques for making decisions on policies that address
environmental and human health risks.  These techniques fall into
the two categories of environmental decision-making:  risk
assessment and risk management.  Below, we provide a general
overview of each of these concepts.  We then describe the key
tasks associated with risk assessment and risk management as
applied to the Baltimore IEMF air toxics study.

1.  RISK ASSESSMENT AND RISK MANAGEMENT

     As recently defined by the National Academy of Sciences
(NAS), risk assessment is the scientific activity of evaluating
the toxic properties of a chemical and the conditions of human
exposure to it both to ascertain the likelihood that exposed
humans will be adversely affected, and to characterize the nature
of the effects they may experience.1   A risk assessment can be
either quantitative, emphasizing a reliance on numerical results,
or qualitative.  For cancer, the Baltimore IEMP uses quantitative
risk assessment to establish priorities among pollutants and
sources.  For noncancer health effects, we can only identify
whether exposures—either modelled or measured—exceeding an
estimated threshold are generally more appropriate subjects for
further investigation because of data limitations.

     It is important to emphasize that our risk estimates were
intended to help decision makers understand the relative
importance of different problems and set priorities among them.
They serve as aids in developing public policies for controlling
environmental risks.  The risk assessments do not examine disease
incidence in the local population and then attempt to link
incidence with environmental exposures.  This was not an
epidemiological study; the risk assessments were not intended to
and do not answer questions such as what caused a statistically
higher rate of cancer in one neighborhood or part of the
community.  Nor do they make a definitive statement concerning
     *Risk Assessment in the Federal Government;   Managing the
Process.  (Washington, D.C.:  National Academy Press, 1983).
Additional information in this section is drawn from Principles
of Risk Assessment;  A Nontechnical Review. (U.S. EPA, 1985).

                              III-l

-------
the absolute risk posed by a particular substance, pollutant, or
exposure pathway.  Rather the data provide a rough, but
conservative indication of potential human health risks.  The
actual risk is not likely to be much higher than the estimated
risk, but it could be considerably lower.  The reader is referred
to Appendix A for a more detailed discussion of the issues
surrounding EPA's method of quantifying cancer risks.

     In contrast to risk assessment, the NAS describes risk
management as a process of evaluating (1) whether an assessed
risk is sufficiently high to present a public health concern and
(2) the appropriate response to the potential chronic health
hazard.  The risk management process entails consideration of
political, social, economic, and engineering information during
the evaluation process.  It also involves value judgments,
especially when viewing the acceptability of risk and the
reasonableness of the costs of control.

     The sections that follow describe in more detail the steps
involved in completing a quantitative risk ^assessment for the
selected air toxics studied in Baltimore.  This chapter concludes
with a discussion of the demonstration project that was conducted
to assist in the risk management phase of the air toxics study.


2.  STEPS IN RISK ASSESSMENT

     Risk assessment can be divided into four major steps:

     •  Hazard identification

     •  Dose-response evaluation

     •  Human exposure evaluation

     •  Risk characterization

Each of these steps is discussed below in the context of the
Baltimore air toxics study.

     a.  Hazard  Identification

     Hazard identification involves reviewing relevant human
epidemiological  studies, animal experiments, and  toxicological
tests bearing on whether or not an agent can cause an increase in
the incidence of a health condition (e.g., cancer, birth
defects).  It is important to note that in the Baltimore IEMP, we
relied solely on the hazard determinations already made by EPA.

     In deciding the potential hazard of a compound, EPA has
evaluated the available studies using sound biological and
statistical considerations and procedures.  In addition, it has

                              III-2

-------
characterized the nature and strength of the evidence of
causation, which is an important component of the hazard
identification process. • For compounds associated with noncancer
effects, EPA has included uncertainty factors into the
calculation of an acceptable exposure level or threshold.  These
uncertainty factors account for differences in animal and human
sensitivity, differences in human sensitivities, and differences
in experimental versus actual  (or expected) exposure levels.  For
agents identified as potential human carcinogens, EPA has
developed a weight-of-evidence stratification scheme that
reflects a synthesis of the conclusions drawn from the different
pieces of information on the carcinogenicity of the agent.  There
are five levels of the EPA stratification scheme:

     •  Group A:  Human Carcinogen

     •  Group Bt  Probable Human Carcinogen

     •  Group C:  Possible Carcinogen

     •  Group D:  Not Classified

     •  Group E:  No Evidence of Carcinogenicity for Humans

Appendix A discusses these groups in more detail.

     Agents that are judged to be in the EPA weight-of-evidence
stratification groups A and B are regarded as suitable for
quantitative risk assessments.  Agents judged to be in group C
are generally regarded as suitable, but judgments in this regard
are made on a case-by-case basis.  For the agents assigned to
groups A, B, and, to a lesser extent, C, a quantitative
dose-response relationship has been developed.

     b.  Dose-Response Evaluation

     A dose-response assessment is the process of characterizing
the relationship between the dose of an agent administered or
received and the extent of toxic injury or disease.  The
information used in this type of assessment is derived from
animal studies or, less frequently, from studies in exposed human
populations.  Thus, in most cases, the dose-response evaluation
involves extrapolation between species and exposure levels (i.e.,
from high to low doses).  In completing our risk assessment for
selected air toxics in the Baltimore area, we relied primarily on
the "dose-response relationships" or potency estimates already
developed by EPA for three effect categories:  cancer, cancer
caused by exposures to complex mixtures, and noncancer.  Each of
these is described below.
                              III-3

-------
        i.  Cancer

     EPA'3 Carcinogen Assessment Group (CAG) relies heavily on
data from animal tests in determining the toxicity (especially
carcinogenicity) of chemicals and in estimating their potency in
producing adverse health effects.2   The reasons  are as  follows:

     •  Human testing is not possible for ethical reasons.

     •  Use of animals as proxies for humans is consistent with
        current scientific knowledge of the mechanism of
        carcinogenicity in humans,  although certain species of
        animals may not be suitable models for the action of
        certain chemicals.  Known human carcinogens, with the
        single possible exception of arsenic, have been found to
        be carcinogenic in some animal system.  Although the
        reverse is not necessarily true,  it does indicate the
        potential for carcinogenicity in a mammalian species.
        Consequently, "in the absence of adequate data on humans,
        it is reasonable, for practical purposes, to regard
        chemicals for which there is sufficient evidence of
        carcinogenicity in animals as if they presented a
        carcinogenic risk to humans."3

     •  Epidemiological studies of the chronic health effects
         (such as cancer) of long-term environmental exposures to
        chemicals of concern are problematic; they require
        extensive high-quality data on historical exposures and
        causes of disease and death and can generally only be
        applied to situations where exposures have occurred for
        twenty or more years.

     •  A high-quality negative epidemiological study, while
        useful, cannot prove the absence of an association
        between chemical exposure and human cancer.  For the
        population and levels of exposure studied, it can
        determine an upper-limit or range for the estimates of
        risk and the statistical likelihood of the study to
        detect an effect.

     •  Understanding the mechanism of action of a chemical in
        producing an adverse health effect  through  studies on
         animals is important in determining its toxicity and
      2For more information on the methodology used by CAG in its
does-response  evaluation,  please  see:   "Guidelines  for
Carcinogenic Risk Assessment, as  published  in the Federal
Register.  September 24,  1986.

      3 Preamble, International Agency for Research on Cancer
 (IARC) monograph.
                               III-4

-------
        potency at environmental exposures, as opposed to
        occupational exposures for which most epidemiological
        data exist.

     The use of animal data in assessing the carcinogenicity of
chemicals and developing potency estimates is not without
problems.  There are major uncertainties in extrapolating from
animals to humans.  There are important species differences in
uptake, metabolism, and organ distribution of carcinogens, as
well as species and strain differences in target-site
susceptibility.  And human populations are variable with respect
to genetic constitution, diet, occupational and home environment,
activity patterns, and other cultural factors.

     Another major area of uncertainty involves estimating the
potency of chemicals at the low environmental exposures found in
IBMP studies from the high doses or exposure levels observed in
occupational or experimental settings—a process called "low-dose
extrapolation."  For practical reasons, responses at low doses
cannot be measured directly either by animal experiments or by
epidemiological studies.  Low-dose extrapolation must, therefore,
be based on current understanding of the mechanisms of
carcinogenesis.

     In estimating the potency of compounds at low environmental
exposure levels, CAG uses mathematical models that are designed
to produce conservative estimates.  The potency estimates made on
the basis of these models (where actual exposures are consistent
with exposure assumptions that underlie the model) should be
regarded as representing the plausible upper-bound for the risks;
the true value of the risk is unknown and may be as low as zero.
They are not best estimates of cancer potency.4   We emphasize
that use of the upper-bound potency estimates does not
necessarily give a realistic prediction of actual risks to
humans.  In addition, because we present risk estimates as sums
of the separate risks from multiple compounds, the uncertainty
regarding the final estimate is commensurately increased.

     The potency estimates are generally expressed as "unit
cancer risk factors."  The unit cancer risk factor combines
estimates of cancer potency with EPA standard exposure
assumptions.  The unit cancer risk factor represents the upper-
bound estimate of the increase in cancer risk associated with
continuous lifetime (i.e., 70 years) exposure to 1 ug/m3  of  a
specific compound.
     *  Alternative procedures  for cancer risk assessment  exist
that account for conflicting cancer risk estimates.  However, the
suitability of the procedure depends on the uses to which the
risk estimates will be used.

                              III-5

-------
        ii. Cancer Caused by Exposures to Complex Mixtures

     The complex mixtures of chemicals produced by the incomplete
combustion of fossil fuels (e.g., automobile exhaust, coke oven
emissions, smoke from woodburning stoves) have long been known to
contain compounds, such as polycyclic organic matter (POM) which
includes benzo(a)pyrene, a human carcinogen.  However, these
mixtures present special problems in estimating carcinogenic
potency because combustion of fuels can produce thousands of
organic gaseous and particulate pollutants, as well as inorganic
materials, such as metals.  Also, the relative composition of
these pollutants within these mixtures, and hence potency of the
overall mixture, varies with fuel source.  Testing and developing
potency estimates for every constituent of the products of
incomplete combustion from a source represents a formidable task.

     Previous studies9 used one constituent of a carcinogenic
class of compounds within the complex mixture, such as
benzo(a)pyrene  (BaP), as a surrogate for all carcinogenic species
and applied its potency estimate to the class, such as POMs.  An
alternative approach assumed that the carcinogenic potency of all
combustion products was the same, regardless of the source.6
Neither of these approaches has been found to be consistent with
current scientific understanding of the relationship of
mutagenicity and human carcinogenicity or with the relative
composition of complex mixtures.

     In the Baltimore ambient air toxics study, we used the
comparative potency method.  Scientists estimate the risk of a
suspect carcinogen (e.g., diesel emissions), for which there are
no epidemiological cancer data, by comparing the potency of the
agent in mutagenicity and carcinogenicity bioassays to those of
known human carcinogens.  The known human carcinogens are coke
oven, roofing tar, and cigarette smoke tar emissions.  The
following equation is used in the example for diesel emissions:

   Estimated Hunan Risk* • Human Risk,  x  B

   Where:
   d = diesel
   c = coke ovens
   B = the relative bioassay potency of the bioassay potency
       (diesel) to the bioassay potency (coke ovens)
     3 One example of such a study is U.S.  EPA,  Santa Clara
Vallev Integrated Environmental Management Protect; Revised Stage
One Report. Office of Policy, Planning, and Evaluation, May 30,
1986.

     6 U.S. EPA, Office of Policy Analysis, The Air Toxics
Problem in the  United States: an Analysis of Cancer Risks  for
Selected Air Toxics. U.S. EPA, May 1985.

                              III-6

-------
The relative bioassay potency is obtained by calculating the
ratio of the observed mutagenic activity in tests of diesel
emissions to the observed mutagenic activity in tests of coke
oven emissions (the slopes of the dose response functions from
the same in vitro or in vivo bioassays).

        iii.  Noncancer Health Effects

     The noncancer health effects we examined included systemic
effects, such as liver and kidney toxicity, fetal developmental
(teratological), neurobehavioral, and blood effects.  Only
effects associated with long-term exposures were considered.
They cover a very broad range of severities with regard to human
health—from life-threatening to only mildly debilitating.
Agents that give rise to these effects are often referred to by
EPA as "systemic toxicants" because they affect the functioning
of the various organs.  Based on scientists' understanding of
physiological mechanisms, these systemic toxicants are assumed to
have identifiable exposure thresholds (both for the individual
and the population) below which the adverse effects are not
observable.

     EPA uses animal data to calculate these thresholds.
Threshold values, expressed in terms of mg/kg/day are called
"reference doses" (RfOs).  As discussed above, the estimates
include uncertainty factors to account for differences in animal
and human sensitivity, differences in human sensitivities, and
differences in experimental versus actual (or expected) exposure
levels.

     EPA has established a method for calculating RfDs for
exposures that occur only through ingestion—oral RfDs.  Thus,
there are no Agency-approved inhalation RfDs for target
pollutants in this air toxics study.  We have calculated
inhalation thresholds using either Agency-approved oral RfDs, if
they existed for our target pollutants, or threshold values
estimated by RID toxicologists using the same method employed by
EPA in developing oral RfDs.  The procedures we followed have
considerable precedence at the Agency.

     Because the effects for which the thresholds were developed
cover a wide gamut of impacts on human health, exceedance of a
threshold generally does not imply a life-threatening situation.
Furthermore, it is assumed that exposures occur over a lifetime
for most effects (excluding fetal developmental) that could be
caused by environmental exposures.  Our exposure assessment
represents only a snapshot in time and not the kind of analysis
necessary for an epidemiological study.  The exceedance of a
threshold for an individual pollutant indicates the need to
further investigate this compound.


                              III-7

-------
     In Baltimore, as in any urban environment, environmental
exposures generally involve numerous compounds.  To account for
exposures to complex chemical mixtures, one can also sum the
threshold ratios for pollutants that contribute to the same
noncancer health effect.  By summing these ratios, one creates a
so-called "Hazard Index," which is discussed in more detail later
in this chapter.  If the estimated Hazard Index exceeds "1," it
identifies exposures that need further exploration.

     We refer the reader to Appendix A for a more comprehensive
discussion of the analytical approach for developing thresholds.

     c.  Human Exposure Evaluation

     This step in the risk assessment process involves measuring
or estimating the intensity, frequency, and duration of human
exposures to a compound in the environment.  It can also require
determining the size of the exposed population.  There are two
major ways the Baltimore IEMP estimated exposures to chemical
compounds in the urban air: (1) dispersion modelling of ambient
concentrations, and  (2) monitoring of ambient concentrations.
These approaches are traditionally used by air quality programs
and have been adopted by the IEMP.

     The IEMP approach requires knowing the concentrations of a
chemical in the ambient air at a residential location.  The
information on concentrations at receptor sites was coupled with
assumptions about how ambient concentrations relate to actual
human exposures, or dose, to arrive at an estimate of exposure
for an individual or population.  We followed EPA standard
assumptions, called exposure constants, that account for the
amount of air or water a typical person takes in each day, and
the weight of an average person.  Specifically, EPA assumes  that
an average person weighs 70 kilograms, breathes 20 cubic meters
of air and drinks two liters of water each day.

     We estimated ambient air exposures at fixed points
throughout the study area using dispersion (i.e., predictive
transport and fate) modelling and monitoring at fixed outdoor
sites.  Ambient monitoring data often provide the most direct
measure of environmental conditions.  However, there are two
important limitations to their use:

     •  Monitoring data  cannot generally be used to predict
        maximum levels as data are  often taken from a few
        specific  points  and must be extrapolated over the  area  of
        concern.  In contrast, modelling can take  into account
        geographic variability and  can pinpoint the sites  of
      7 Versar,  Inc.,  A Screening  Methodology for Air  Quality
Analysis.   U.S.  EPA,  April  1985 draft.
                               III-8

-------
        maximum ambient concentrations.  Modelling can also be
        used to evaluate source contribution.

     •  Monitoring data most often cannot link observed ambient
        concentrations to sources, which is essential for making
        risk management decisions.  Phase II of the Baltimore
        IEMP focused on gathering the data needed to identify
        sources.

Consequently, for the purpose of estimating risks at the highest
exposure levels, we relied primarily on dispersion modelling.  To
estimate risks at average exposure levels, we used both modelled
and monitoring data.  We used monitoring data to evaluate how
well our models predict actual concentrations.  It is difficult,
to use modelling alone to predict total ambient exposures.
Because background levels of pollutants are rarely known, they
cannot be incorporated into the dispersion model when it is run
to estimate ambient air concentrations.  The result is a range of
likely concentrations, bordered by our modelled and monitored
values.

     d.  Risk Characterization

     Risk characterization is the process of estimating the
incidence of a health effect under various exposure scenarios.
Generally, it involves the integration of the data and analysis
stemming from the three steps of risk assessment described above.
The type of risk characterization that we performed for
carcinogens versus systemic toxicants was somewhat different, as
detailed below.

        i.  Cancer

     Cancer risk is defined as either 1) the increased
probability that an individual continuously exposed for a
lifetime to one or more chemicals will contract cancer, referred
to in this report as either the maximum or average increased
lifetime individual cancer risk, or 2} where a number of people
are likely to be exposed, the excess number of cancer cases per
year in the population exposed to one or more chemicals, referred
to as the annual excess cancer incidence.  Where simultaneous
exposures to multiple compounds is assumed, these risk estimates
for individual compounds are then summed to produce a cumulative
estimate of risk.  The methods used to calculate each risk
measure are discussed separately.

     Maximum Increased Lifetime Individual Cancer Risk.  We
define the maximum increased lifetime individual cancer risk as
the increased probability that an individual exposed to the
greatest amount of one or more chemicals will contract cancer
over a lifetime (70 years).  Generally, this calculation is
computed by multiplying the maximum ambient concentration at a

                              III-9

-------
discrete receptor location (generated through our air dispersion
modelling exercise) by the unit cancer risk factor.   In  the
Baltimore IEMP, a discrete receptor location was defined as a
residential area.  Thus, high concentrations in areas not defined
as residential by U.S. Geological Survey maps or by visual siting
were not considered possible maximum individual risk sites.
             increased        KMIM™            Unit Cancer
     Lifetime Individual -    Concentration^ x  Risk Factor!
     Cancer Risk

     Where :
     i = Target pollutant i
     k = Receptor location k

We then summed the maximum increased lifetime individual cancer
risks for each target pollutant to estimate the total maximum
increased lifetime individual cancer risk from all study
pollutants .

     Average Increased Lifetime Individual Cancer Risk.  We
define the average increased lifetime individual cancer risk as
the increased probability that an individual exposed to the
average amount of a chemical will contract cancer over a lifetime
(i.e., 70 years).  We calculated the average increased lifetime
individual cancer risk for each pollutant by multiplying the
average ambient pollutant concentration in a defined area by the
exposure constant and the appropriate unit cancer risk factor.
These values were summed to arrive at the aggregate average
increased lifetime individual cancer risk posed by all target
pollutants .

     The estimates of average ambient air concentration were
based on both measured and modelled data.  The results of the
dispersion modelling were used to estimate average increased
lifetime individual cancer risks for the study area, as a whole.
The results of the available ambient air monitoring were used to
estimate average increased lifetime individual cancer risks at
each of the fixed monitoring site locations, as well as for the
study area.

     Annual Excess Cancer Incidence.  We define annual excess
cancer incidence as the excess number of cancer cases per year in
a defined population resulting from specified environmental
exposures.  Annual cancer incidence is calculated by dividing the
      8The unit cancer risk factor combines estimates of cancer
 potency with EPA standard exposure  assumptions.  The unit  cancer
 risk  factor represents  the upper-bound  estimate of  the  increase
 in  cancer risk  associated with  continuous  lifetime  (i.e.,  70
 years)  exposure to 1 ug/m3 of a specific compound.
                              111-10

-------
lifetime excess cancer incidence by 70, the assumed human
lifetime.  The lifetime excess cancer incidence is calculated by
multiplying the average increased individual cancer risk for each
pollutant, as discussed above, by the population and summing
across all pollutants.  These calculations are represented by the
following equations:

Annual Excess      - V^ Lifetime Excess       '      70 years
 Cancer Incidence    Z-* Cancer Incidence!
                      1
                               -or-

Annual Excess  • V^ V^ Average Increased     Estimated    •    70
 Cancer          L^ Z-* Individual Cancer  *   Popula-
 Incldence        1   j
     Where t
     i =  Target pollutant i
     j =  Either grid cell j (dispersion modelling) or the study
          area (available monitoring data)

     The annual excess cancer incidence was calculated for the
entire study area and the grid cell of highest predicted
incidence.  Dispersion modelling was used to generate area-wide
annual excess cancer incidence estimates, as well as to identify
the grid cell of maximum annual excess cancer incidence.  As
discussed in more detail in Chapter IV, the dispersion model was
programmed to generate estimates of ambient concentration for a
series of small boxes or grid cells that are overlayed on the
study area.

     As a policy check on the incidence results generated using
the dispersion modelling results, we developed rough estimates of
annual excess cancer incidence for the study area using the
available area-wide monitoring data.  The incidence calculations
were performed for each pollutant by multiplying the measured
concentration (averaged across all monitoring locations) by the
appropriate unit cancer risk factor (which combines the cancer
potency scores and the exposure constants) and an estimate of the
total exposed population in the study area.  Although there are
numerous limitations with this approach, the results can provide
the decision maker with additional insight on the range of
potential human health risks posed by selected air toxics using
both monitoring and modelled data.

        ii.  Noncancer

     There are two risk measures for noncancer health effects.
For single pollutants, the measure of noncancer health effects is
the ratio of the estimated pollutant ambient air concentration —
based on either modelling or monitoring — to its no-effect
threshold.  Exposures that are less than the no-effect threshold

                              Ill-ll

-------
are not likely to be associated with noncancer health effects and
are, therefore, less likely to be of regulatory concern.
Conversely, as the frequency of exposures exceeding the threshold
increases and as the size of the excess increases, the
probability also increases that adverse effects may be observed
in a human population.  Such exceedances indicate the need to
examine the exposures more carefully.  Because a single substance
may have many different thresholds for the different health
effects associated with it, we examine each health effect for
each substance separately.  We can also explore the need to
further investigate exposures to multiple compounds that have the
same noncancer effect, as discussed below.

     Because 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 can calculate a "hazard index"  to
assess the potential hazard from cumulative exposures to the
target compounds.  The hazard index is used to indicate when
simultaneous exposures to multiple pollutants with the same
systemic effects could require additional investigation, although
exposures to each individual pollutant did not exceed the no-
effect threshold.  The index is based on the assumption of dose
additivity and is defined by the following equation:
                 HI - Ei/AIi! + B2/ALj +  ...+

     Where :
        HI  =  Hazard index for a particular category of health
               effect
        Et  =  Ambient concentration of pollutant i
        ALj. =  Threshold value for pollutant i  for a particular
               health effect category

After first calculating the ratio of the ambient concentration
for the pollutant to the threshold for the health effect
(Et/ALt), we summed these ratios across all pollutants with  the
same effect.

     In this study, the hazard index is intended to be a
numerical indication of whether exposures to complex mixtures of
pollutants in the environment deserve additional study.  As the
index approaches 1, 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 pollutant, thus indicating the need to
investigate further the observed exposure levels.
     9 U.S. EPA, Environmental Criteria and Assessment Office,
Guidelines for  the Health Risk Assessment of Chemical Mixtures.
Final Report, September 1985, pp. 12-14.

                              111-12

-------
3.  INTERPRETING RISK ASSESSMENT RESULTS IN THE BALTIMORE AIR
    TOXICS STUDY

     We have already highlighted some of the uses and limitations
of risk assessment.  We restate and elaborate these points below
to ensure that the reader properly interprets the  estimates of
risk presented in Chapters V and VI of this report:

     •  The estimates of maximum and average increased lifetime
        individual cancer risk and annual excess cancer incidence
        from exposure to toxics should not be interpreted as
        precise or absolute estimates of future health effects.
        The simplifying assumptions and uncertainties in both the
        toxicology and exposure components are simply too great
        to justify a high level of confidence in the precision of
        the results.  They are approximations of the potential
        upper-bound risks to human health and are designed to
        permit relative comparison across different sources,
        pollutants, and exposure pathways.

     •  The potency and threshold estimates used in this study
        are consistently conservative (i.e., upper-bound) in the
        direction of overestimating risk for the particular
        pollutants and exposure pathways we have assessed.  There
        is the possibility that the cancer risk from any or all
        of the pollutants could be zero.

     •  We may understate increased risks from ambient air toxics
        exposures to the extent that we have not considered all
        pollutants, sources, and exposure scenarios (such as
        short-term exposures to very high levels).   Current
        technology, for example, does not allow scientists to
        identify all compounds in air or water.  This is because,
        first of all, monitoring techniques are selective in the
        kinds of compounds they can effectively and reliably
        capture.  Second, laboratory analysis to determine the
        identity and quantity of monitored samples must be
        targeted at specific chemical or types of chemicals of
        concern.  Modelling, on the other hand, is limited by the
        quality of the data on emissions of pollutants.  And
        third, even if one were able to identify all compounds to
        which one is being exposed through a particular medium,
        one would have only a rough notion of the toxicity of
        only a very small fraction of the total number of
        compounds.
     l°This study estimated only the increase in cancer and
noncancer risk resulting from exposure to ambient (i.e., outdoor)
concentrations of air toxics.  It did not consider exposures
resulting from (1) indoor air, (2) the work place, and (3) other
pathways (e.g., ingestion).
                              111-13

-------
4. RISK MANAGEMENT

     A major focus of the Phase II IEMP work was in demonstrating
an approach to identify the best (namely, the most cost-
effective) strategy for managing the estimated risks.   The
National Academy of Sciences defines risk management as the
process of evaluating whether an assessed risk is sufficiently
high to present a public health concern and the appropriate
response to the potential human health hazard.  Risk management
considers not only the level of risk posed by a particular
pollutant or source of pollution but also the feasibility and
cost of control, public preferences, and institutional
capabilities.  Setting priorities for research into the risk
posed by pollutants is also a key aspect of risk management.  Key
activities are investigating control alternatives, estimating
their efficiency in controlling pollution, developing estimates
of their cost, and then assessing their cost-effectiveness.

     In the Baltimore air toxics study, the criteria of
cost-effectiveness (i.e., the cost of control per unit of human
health risk reduced) and benefit-cost (i.e., the cost of control
per dollar of benefits accruing to society) were used to evaluate
alternative control strategies.  Benefits estimates were included
in this analysis to capture additional positive outcomes from
controlling air toxics (other than risk reduction) that may
offset the costs of control.  For example, controlling benzene
emissions from cars can also lead to the reduction of volatile
organic compounds (VOCs) that cause smog.  These analyses can
consider risks that result from exposure to primary as well as
secondary (intermedia) releases.  However, the Baltimore air
toxics study considered only primary ambient air releases.

     We developed rough information on the costs and efficiencies
for a set of potentially feasible control options.  We then
ranked these control options by their cost-effectiveness, as well
as their benefits.  This information was used in a mathematical
model, referred to as a mixed integer program, to obtain the
lowest cost combination of control options to achieve a given
level of benefits or risk reduction.  The control options and the
computer model are discussed in more detail in Chapter VI and
Appendix C.

     The estimates of the costs of control are, like the risk
estimates, highly uncertain.  They are best used in developing a
relative ranking of control strategies, not in generating
definitive estimates of the resources necessary to solve the
selected environmental problems.
                              111-14

-------
IV.  SELECTING POLLUTANTS. EVALUATING TOXICITY. AND ESTIMATING
     EXPOSURE

     In this chapter, we describe the activities undertaken to
select the study pollutants and characterize the risks and
exposures associated with these pollutants.  First, we discuss
the approach we used to identify the study pollutants, including
a description of the data gathering activities that supported the
pollutant screening.  Second, we present what is known about the
toxicity of each compound evaluated.  This discussion combines
the risk assessment steps of hazard identification and dose-
response evaluation.  It also discusses the weight-of-evidence
issue, which is important to consider when viewing the potential
toxicity of each pollutant examined.  We then summarize how we
assessed the extent to which Baltimore residents could be exposed
to the target pollutants in the ambient air.  The results of the
risk assessment for the study pollutants are presented in
Chapter V.


1.  POLLUTANT SELECTION METHODOLOGY AND RESULTS

     We selected two sets of compounds to analyze:  one for air
dispersion modelling and the other for ambient air monitoring.
The air dispersion modelling was performed as part of the
Baltimore IBMP ambient air toxics study.  Air monitoring was
conducted by EPA's Office of Research and Development (ORD) as
part of its Baltimore TEAM study.1   Although the  Baltimore IEMP
was not part of the Baltimore TEAM study, we coordinated with ORO
to take advantage of the opportunity to obtain additional site-
specific air monitoring data.  In addition to the TEAM monitoring
data, we gathered all readily available monitoring data for the
Baltimore area.  These data were drawn from sampling programs
conducted by the AMA over the past three years and the IEMP
during 1983/1984.
     'The TEAM study is a study designed by EPA's Office of
Research and Development to assess the significance of indoor
versus outdoor exposures to selected chemicals.  Readers
interested in this study should contact U.S. EPA, Office of
Research and Development, Environmental Monitoring Systems
Laboratory for more information.

                               iv-1

-------
      Different factors influenced our choice of study compounds
for dispersion modelling and ambient air monitoring.  For the air
dispersion modelling/ we were constrained by the set of chemicals
for which we had estimates of emissions from sources.  Even for
these chemicals, we could not always model a compound with any
reasonable degree of accuracy (e.g., a compound, such as
formaldehyde, may be expected to be produced in significant
amounts from other compounds through atmospheric processes).   For
the ambient air monitoring conducted by ORD, we were limited by
sampling technology and resources.  Furthermore, we were
constrained to select compounds for monitoring which could be
expected to be found in both the outdoor and indoor environments
to complement the Baltimore TEAM study mentioned earlier.

     In what follows, we describe the criteria used to select
compounds for air dispersion modelling, the procedure followed to
choose compounds included in the TEAM ambient air monitoring
program, and the final list of pollutants analyzed in the
Baltimore air toxics study.

     a.  Pollutant Selection for Modelling

     We used the following three criteria in selecting compounds
for modelling:

     1. Likelihood of exposure or presence in the ambient air.
        To make this determination, we developed a rough
        emissions inventory of air toxics from both point and
        area sources.  We also used available monitoring data.
        The following sources were used:

        i.     1983 Survey for the Maryland Toxic Substance
               Registry fTSRl.  This survey (referred to in this
               document as the 1983 TSR survey) compiled
               information on toxic substances used, produced or
               handled by industry.  The State's 1985 TSR survey
               had not yet been completed when the pollutant
               selection was completed.

        ii.    1983 IEMP monitoring data.  As discussed in the
               Phase I report, IEMP sampled for ten VOCs for a
               period of three months starting in November of
               1983.

        The rough inventory—which was entered into a computer
        database that could be managed through the IEMP data
        management system  (PIPQUIC)—includes process and boiler
        emissions from approximately 250 point sources, as well
        as emissions from area sources (e.g., dry cleaners and
        degreasers) and nontraditional sources  (e.g., wastewater
                               IV-2

-------
        treatment plants).2  The inventory was  a compilation of
        data generated by the AMA and EPA.  More detail is
        provided on the emissions inventory later in this chapter
        under Section 3 entitled "Human Exposure Evaluation."

        Existence of GAG unit cancer risk factors or threshold
        values for noncancer effects.3The Agency has  approved
        numerous RfDs for the oral route of exposure, but none
        for inhalation.  Toxicologists from EPA's Regulatory
        Integration Division (RID) and EPA's Air Office have
        adjusted these oral RfDs, where possible, for the
        inhalation route of exposure.  We acknowledge that these
        are not official EPA values and are useful only for
        identifying sources and pollutants that warrant further
        investigation.  The unit cancer risk factors are
        developed by EPA's Carcinogen Assessment Group.

        Sources t

        i.     EPA's Cancer Assessment Group fCAGU  CAG unit
               cancer risk factors represent the Agency's
               consensus on estimates for carcinogens and are
               commonly used by all EPA's program offices.4

        ii.    EPA's Office of Air Quality Programs and Standards
               (OAQPS1.  OAQPS is charged with developing
               regulations mandated by the Federal Clean Air Act
               for controlling air toxics.

        iii.   Regulatory Integration Division toxicolooists.
               RID toxicologists and health scientists estimated
               thresholds for noncancer health effects based on
               reviews of toxicological literature.
     2An overview of the PIPQUIC system is  contained in:   PIPQUIC
User's Guide. American Management Systems, Inc. for U.S. EPA,
Office of Policy Analysis, Regulatory Integration Division,
November 1985.

     3The unit cancer risk factor combines  estimates of cancer
potency with EPA standard exposure assumptions.  The unit cancer
risk factor represents the upper-bound estimate of the increase
in cancer risk associated with continuous lifetime  (i.e., 70
years) exposure to  1 ug/m3 of a specific compound.

     *The unit cancer risk factors can be obtained  directly from
CAG or through ORD's Integrated Risk Information System  (IRIS).
Information on IRIS may be obtained by contacting EPA's Office of
Health and Environmental Assessment.

                               IV-3

-------
     3. Feaaibilitv of control.  Pollutants with limited
        potential for reducing exposure to the chemical through
        engineering measures or changes in production processes
        were excluded.  For example, some toxics are emitted from
        natural sources.  Control of such emissions cannot be
        achieved readily through an air toxics regulatory
        program.

        Sources:

        i.     Maryland's Air Management Administration engineers

        ii.    EPA contractors to the Baltimore IEMP air toxics
               study

        Source: EPA's Office of Research and Development.

     To account for known differences in the quality of the data,
we used professional judgment in applying the above criteria.  In
general, any individual compound met the first two criteria if 1)
over a metric ton per year was being emitted into the air and it
was carcinogenic or it had an RfD that could conceivably be
exceeded by environmental levels, or 2) it was emitted in
substantial quantities from a single source (over a metric ton
per year) and was toxic or carcinogenic.

     In addition to individual compounds, we also chose to
examine emissions of the larger molecular weight, semi-volatile
class of compounds called polycyclic organic matter (POM) that is
produced by incomplete combustion of organic matter from a
variety of combustion processes.  A recent nationwide study of
air toxics conducted by EPA had shown that the products of
incomplete combustion could be significant contributors to
airborne risk (U.S. EPA, Office of Policy Analysis, The Air
Toxics Problem in the United Statest An Analysis of Cancer Risks
for Selected Air Toxics, May 1985).  Unfortunately, the unit
cancer risk factor that we have for POM from different categories
of sources are generally not of the same quality as for many of
the volatile organic compounds included in the study, as detailed
in Section 2 of this chapter.  [The exception is the unit cancer
risk factors for POM from coke oven emissions which is based on
human epidemiological data.]  Nevertheless, we decided that it
was preferable to provide rough estimates of potential risk from
particulate organics than to ignore this important source of risk
altogether.3
     3Joellen Lewtas,  "Combustion Emissions:   Characterization
and Comparison of Their Mutagenic and Carcinogenic Activity," in
Hans Stich, ed., Carcinogens and Mutacrens in the Environment,
Volume V (CRC Press, Inc., Baton Rouge, 1985), pp. 59-72.  Also,
Roy Albert, J. Lewtas, S. Nesnow, T. Thorslund, and E. Anderson,

                               IV-4

-------
         The  compounds  that were  selected  for modelling are  shown
 in  Table IV-1,  ranked  by rough estimates  of total annual
 emissions in the  study area.  Since the emissions data were drawn
 from different  EPA and State databases generated for different
 purposes,  they  vary greatly in quality.   The toxicities of  the
 compounds are discussed under Section 2 of this chapter.

      b.   Pollutant Selection for TEAM Monitoring

      The monitoring data used in the Phase II Baltimore air
 toxics study came from several sources; however, the only
 monitoring that took place while the air  toxics study was being
 conducted came  from ORD as part  of their  Baltimore TEAM study.
 We  coordinated  with the ORD monitoring effort because we saw the
 potential to use  the TEAM monitoring data in our dispersion model
 performance  evaluation.6  To ensure success of this project, we
 worked with  ORD to develop a list of pollutants for sampling that
 included many of  the pollutants  modelled  in the IEMP ambient air
 toxics study.

      Building on  previous indoor air toxics work, the Workgroup
 and ORD  identified the pollutants for monitoring starting with
 four lists of compounds.   The first had been compiled by Dr.
 Joyce McCann at the University of California's Lawrence Berkeley
 Laboratory.   Under a contract with the U.S.  Department of
 Energy,  Dr.  McCann and her coworkers had  conducted a
 comprehensive review of  data on  indoor air pollutants.  The list
 she provided us contained those  chemicals that were suspected to
 be  present in the indoor environment and were known to be toxic
 at  environmental  levels,  but whose concentrations had not been
 adequately determined  for the purpose of estimating risks to the
 general  population.  The second  was a list of compounds that the
 Baltimore  IEMP  Indoor  Air Workgroup had identified as candidates
 for inclusion in  the TEAM study.  This list in turn was developed
 on  the basis  of knowledge of chemicals that are used in common
 consumer products  and  that are toxic to some degree.  The third
 list  included a set of priority  compounds developed by the Air
 Toxics Workgroup  using professional judgment.  Finally, the
 fourth was a  list  of compounds that TEAM had studied in previous
"Comparative Potency Method for Cancer Risk Assessment:
Application to Diesel Particulate Emissions," Risk Analysis. Vol.
3, No. 2, 1983.

     6The model performance evaluation is discussed later in this
chapter  (see Section 3(a)(iv)).

     7Personal  communication between Andrew Manale, EPA and Dr.
Joyce McCann, Lawrence Berkeley Laboratory and the University of
California, Berkeley, March 1986.

                               IV-5

-------
                           TABLE IV-1
        CHEMICALS SELECTED FOR AIR DISPERSION MODELLING
             IN THE BALIMORE IEMP AIR TOXICS STUDY

     (Data Drawn from the 1983 Survey for the Maryland Toxic
  Substances Registry and 1983 National Emissions Data System)
POLLUTANT
TOTAL
EMISSIONS
(kkg/yr)1
SINGLE SOURCE
EMISSIONS
(kkg/yr)
USE BY MAJOR
POINT SOURCES
(kkg/yr)
Oraanic oases
     Toluene
     Benzene
     Tetrachloroethylene
     Xylene
     Methyl chloroform
     Methylene chloride
     Trichloroethylene
     Formaldehyde
     Glycol ethers
     Ethyl benzene
     Ethylene oxide
     Chloroform
     Phenol
     Carbon tetrachloride
     Ethylene dichloride
     Ethylene dibromide

  Oraanic Particulates

     POM2
     Benzo(a)pyrene

  Metals

     Chromium
     Cadmium
     Arsenic
5923.00
3038.00
2341.00
1762.00
1696.00
1097.00
 769.00
 709.00
 638.00
 232.00
  45.00
  30.00
  19.00
   3.80
   3.35
   1.71
N.A.
N.A.
6105.00
   1.60
   1.00
         2056.44
         2585.52
            N.A.
           83.40
           16.00
           77.36
          197.70
           82.60
           15.64
            4.46
           66.16
           51.80
            0.01
            6.40
            N.A.
            N.A.
                N.A.
                N.A.
2.30
1.40
0.40
             22618.49
             68839.70
               974.33
             20265.98
              1817.85
               938.68
               457.23
              1188.43
                 N.A.
                 N.A.
              2530.63
                15.88
               897.22
              1652.92
                26.62
                 N.A.
                 N.A.
                 N.A.
                         46108.76
                          2430.84
                            83.92
Note: The data in this chart are of varying quality.

N.A. = not available

1  Metric tons/year
2  POM = polycyclic organic matter.   POM includes  carbon
containing molecules having more than one ring structure.
Polyaromatic compounds, such as benzo(a)pyrene, are included in
this class of compounds.

                               IV-6

-------
studies.  Budget considerations caused EPA to restrict the number
of target compounds for TEAM monitoring to 27.  These compounds
are listed in Table IV-2.


2.  HAZARD IDENTIFICATION/DOSE-RESPONSE EVALUATION

       We present below the hazard assessments for the substances
included in the Baltimore ambient air toxics study.  Table IV-3
lists all compounds included in the study, as well as the effects
(i.e., cancer versus noncancer) that could result from exposure
to these substances.8  As discussed earlier in this report,  we
relied primarily on hazard determinations and dose-response
evaluations already made by EPA.  The one exception is for
polycyclic organic matter (POM), which results from the
incomplete combustion of organic material.  This is because POM
represents a large class of compounds and because the
constituents of POM and their portion of the total composition of
emissions vary with source category.  EPA recently identified
products of incomplete combustion as posing potentially
significant risks from exposure.9

     a.  Cancer Effects

          i.  Weight of Evidence

     Before we present the dose-response information for the
carcinogenic compounds examined in this study, it is important to
emphasize that not all carcinogens are equal in terms of potency
or weight of evidence.  As discussed in Chapter III, EPA has
developed a stratification scheme that classifies compounds into
one of five groups (i.e., A, B, C, D, E) based on weight of
evidence.  In general, compounds that are judged as either Group
A or B carcinogens are regarded as suitable for quantitative risk
assessment.  Group C carcinogens are also suitable for
quantitative risk assessment, but a decision to include these
pollutants must be made on a case-by-case basis.  Because the
weight-of-evidence determination is critical information for the
decision maker, we have included this designation on all tables
summarizing risk results by pollutant.  It was impossible to
assign an overall weight-of-evidence score in instances where we
     8Several compounds included in the TEAM sampling were
excluded from the Baltimore ambient air toxics study because of
limited health effects information.  These compounds are
limonene, decane, dodecane, nonane, octane, and undecane.

     9For more information, see U.S.  EPA,  Office of Policy
Analysis, The Air Toxics Problem in the United States: An
Analysis of Cancer Risks for Selected Pollutants, May 1985.  This
study is also referred to as the "Six Month Study."

                               IV-7

-------
                               TABLE IV-2

        CHEMICALS SELECTED FOR ORD'S TEAM AMBIENT AIR MONITORING
             AS PART OF THE BALTIMORE IEMP AIR TOXICS STUDY
                                               STRAIGHT CHAIN
   VOLATILE                                       ALIPHATIC
   ORGANICS                 METALS              HYDROCARBONS

Benzene1                   Arsenic1               Decane3
Carbon                     Cadmium1               Dodecane
tetrachloride2             Chromium1              Nonane3
Chlorobenzene3             Lead3                  Octane3
Chloroform1                                       Undecane3
Dichlorobenzene3
Ethyl benzene1
Ethylene dibromide
(EDB)1
Formaldehyde1
Limonene3
Methyl chloroform
(TCA)1
Methylene chloride
Styrene3
Tetrachloroethane
Tetrachloroethylene1
(Perchloroethylene)
Trichloroethylene
(TCE)1
Vinyl chloride3
Vinylidene chloride3
Xylene1
   1     The ambient concentrations of these compounds were modelled
        as well as monitored.

   2     Although we had originally chosen carbon tetrachloride for
        both air dispersion modelling and ambient air monitoring
        based on 1983 emissions data/ more recent data from 1985
        indicated that the compound was little used by industry in
        the Baltimore area.  Consequently, we dropped it from the
        modelling study but included it in the monitoring effort.

   3     This compound was chosen primarily because of its interest
        to the TEAM monitoring study.  Modelling was not possible
        due to lack of data on either emissions from point or area
        sources or the likelihood that it is produced through
        atmospheric transformations of other air contaminants.


                                  IV-8

-------
1-1
VO
                                          TABLE IV-3

                      SUMMARY OF POLLUTANTS SELECTED FOR  MODELLING AND MONITORING
                               IN THE BALTIMORE AMBIENT AIR  TOXICS STUDY

                                                                AVAILABLE
POLLUTANT MODELLED (1)
ORGANIC GASES:
Benzene
Carbon tetrachloride
Chlorobenzene
Chloroform
Dichlorobenzene
Ethyl benzene
Ethylene dibromide
Ethylene dichloride
Formaldehyde
Glycol ethers
Methyl chloroform
Methylene chloride
Methyl isobutyl ketone
Perch loroethylene
Phenol
Propylene dichloride
Styrene
Toluene
Trichloroethylene
Vinyl chloride
Vinyl idene chloride
Xylene
ORGANIC P ARTICULATES:
Benzo(a)pyrene
POM

X
X

X

X
X
X
X
X
X
X
X
X


X
X


X

X
X
TEAM
MONITORING (I

X
X
X
X
X
X
X
X

X
X
X


X

X
X
X
X



MONITORING
) DATA (1)

X
X

X

X
X


X
X
X
X

X

X
X
X



X

CANCER
EFFECT

X
X

X


X
X
X


X
X

X


X
X
X


X

NONCANCER
EFFECT

X
X
X
X
X
X
X
X
X
' X
X
X
X
X
X
X

X
X
X
X




                                               (continued)

-------
                                      TABLE IV-3

                  SUMMARY OF POLLUTANTS SELECTED FOR MODELLING AND MONITORING
                           IN THE BALTIMORE AMBIENT AIR TOXICS STUDY

                                                           AVAILABLE
                                            TEAM          MONITORING   CANCER   NONCANCER
POLLUTANT                   MODELLED (1)  MONITORING (1)   DATA (1)    EFFECT    EFFECT


METALS:

  Arsenic                       XX                XX
  Cadmium                       XX                XX
  Chromium                      XX                XX
  Lead                                        XX                    X
(1) Please see Tables IV-4 and IV-5 for more information on the health effects associated
    with these compounds.  Several compounds in the TEAM monitoring were not included  in
    the Baltimore Ambient Air Toxics Study because of limited health effects data.
    These compounds include:  tetrachloroethane, limonene, decane, dodecane, nonane,
    octane, and undecane.

-------
summed risks across pollutants and/or sources.  However, the
uncertainty surrounding these summed risk estimates is
proportionately increased.

          ii.  Dose-Response Evaluation

     Table IV-4 presents data on the carcinogenicity of the
target compounds, including POM.  The data are grouped according
to whether or not the target compounds were modelled, monitored,
or both and by general chemical class, i.e., volatile organic
(organic gas) or related compound, metal, or POM.  Two types of
information are presented!  1) a qualitative evaluation of the
strength of evidence that a substance is carcinogenic and 2) a
quantitative estimate of the unit cancer risk developed by EPA's
Carcinogen Assessment Group (GAG), except where noted.  We must
again remind the reader that POM, unlike the other compounds,
represents a large group of chemicals that differs in composition
according to source category.

     With the exception of coke oven emissions, the hazard
assessment for POM was conducted by the Genetic Bioassay Branch,
Genetic Toxicology Division, Health Effects Research Laboratory
within EPA's Office of Research and Development.  A comparative
potency approach was used in developing the unit cancer risk
estimates.   The approach involves comparing the tumorigenic and
mutagenic potencies of particulates from different source
categories with combustion and pyrolysis products (coke ovens,
roofing tar, and cigarette smoke) known to cause lung cancer in
humans.

     b.  Noncancer Effects

          i.  Weight of Evidence

          Similar to the situation for pollutants identified as
carcinogens, there can be considerable uncertainty surrounding
whether an agent poses noncancer effects.  As a result, EPA
includes uncertainty factors into the calculation of acceptable
exposure levels or thresholds for noncarcinogens based on a
synthesis of all available information.  These uncertainty
factors account for differences in animal and human sensitivity,
differences in human sensitivities, and differences in
     10Joellen Lewtas,  "Combustion Emissions! Characterization
and Comparison of Their Mutagenic and Carcinogenic Activity," in
Hans Stich, ed.. Carcinogens and Mutagena in the Environment,
Volume V (CRC Press, Inc., Baton Rouge, 1985), pp. 59-72.  Also,
Roy Albert, J. Lewtas, S. Nesnow, T. Thorslund, and E. Anderson,
"Comparative Potency Method for Cancer Risk Assessment:
Application to Diesel Particulate Emissions," Risk Analysis. Vol.
3, no. 2, 1983.

                              IV-11

-------
                          TABLE  IV-4

     UNIT CANCER RISK FACTORS AND THE WEIGHT OF EVIDENCE
 FOR CARCINOGENICITY FOR THE BALTIMORE IEMP TARGET COMPOUNDS
Target Compounds
Unit Cancer
Risk Factor1
Weight of Evidence
  Classification2
Organic gases
Benzene
Benzo ( a ) pyrene
Carbon tetrachloride
Chloroform
Chlorobenzene
Dichlorobenzene
Ethyl benzene
Ethylene dibromide
Ethylene dichloride
Ethylene oxide
Formaldehyde
Glycol ethers
Methyl chloroform
Methylene chloride
Perchloroethylene
Phenol
Propylene dichloride
Styrene
Tetrachloroethylene
Toluene
Trichloroethylene
Vinyl chloride
Vinyl idene chloride
Xylene
Metals
Arsenic
Cadmium
Chromium VI
Lead

8.0 x 10'6
3.3 x 10°
3.7 x 10-6
2.3 x 10'5
N/A See Note 3
N/A See Note 4
N/A See Note 3
2.2 x 10'*
2.6 x 10'3
1.0 x 10'4
1.3 x 10'3
N/A See Note 6
N/A See Note 7
4.1 x 10'6
5.8 x 10'3
N/A See Note 3
1.8 x 10'3
N/A See Note 9
4.8 x 10'J
N/A See Note 3
1.3 x 10'6
2.6 x lO'6
5.0 x 10'3
N/A See Note 3

4.3 x 10°
1.8 x 10°
1.2 x 10'2
N/A See Note 9

A
B2
B2
B2



B2
B2
Bl
Bl (see note 7)


B2
C

C

B2

B2
A
C


A
Bl
A

                         (continued)
                            IV-12

-------
                        TABLE  IV-4  (continued)
          UNIT CANCER RISK FACTORS AND THE WEIGHT  OF  EVIDENCE
      FOR CARCINOGENICITY FOR THE BALTIMORE  IEMP TARGET COMPOUNDS
     Target Compounds
     Unit
    Cancer
 Risk Factor1
                                                  Weight
                                               of Evidence
                                              Classification2
Emission source

Automobiles:

     Gasoline-catalyst
     Gasoline-non-catalyst
     Diesel

Trucks

     Heavy duty diesel
     Light duty gasoline
      (non-catalyst—van)

Residential heating

     Oil
     Coal
     Wood

Industrial & Utility
 Power Plants
     Oil
     Coal

Municipal Waste
 Incineration

Coke ovens
                                 See Note  11
5.1 x 1(
1.6 x 10
3.0 x 1(T5
                                          5
                                         '5
                                 See Note 12
                                 2.0 x 10'
                                 3.0 x 10
        -*
                                 See Note 12

                                 9.0 x 10-6
                                 1.0 x 10'5
                                 1.0 x 10'5
                                 See Note 12
                                 See Note 12
                                 8.0 x 10-B

                                 See Note 13
                                 6.5 x 10-s
                        N.A.
                        N.A.
                        N.A.
                        N.A.

                        3.0 x 10'7
                        8.0 x 10'8

                        N.A.
Notes:
N/A  - Not applicable; compound is not currently considered  to be a
       human carcinogen through inhalation.  See pollutant-specific
       notes for more detail.
N.A. = Not available

                               (continued)

                                 IV-13

-------
                TABLE  IV-4  (continued)

                         NOTES

The unit cancer risk factors are developed by EPA's GAG,
unless otherwise indicated.   The unit cancer risk factor
combines estimates of cancer potency with EPA standard
exposure assumptions.  The unit cancer risk factor
represents the upper-bound estimate of the increase in
cancer risk associated with continuous lifetime (i.e.,
70 years) exposure to 1 ug/m3 of  a  specific compound.

EPA classification: A • carcinogenic in humans,
Bl = probably carcinogenic in humans (limited human data),
B2 = probably carcinogenic in humans (insufficient human
evidence but sufficient animal evidence), C = possibly
carcinogenic in humans.

At this time, EPA's Cancer Assessment Group (CAG) has not
found this compound to be carcinogenic from inhalation
exposures and, therefore, not suitable for quantitative risk
assessment.

Within this group of compounds, p-dichlorobenzene is a
CERCLA (Superfund) hazardous substance that has been
identified as a potential carcinogen by CAG.  The weight of
evidence for p-dichlorobenzene is B2/C.

The unit cancer risk factor and evaluation of weight of
evidence are from EPA's Office of Toxic Substances.

We examined methyl and butyl cellosolve as the
representative glycol ethers in this study.  Insufficient
data exist for an adequate evaluation of the potential
carcinogenicity of these compounds.  Therefore, they are not
analyzed as carcinogens in this report.

The carcinogenicity of methyl chloroform is under review by
EPA's CAG.  Until a final determination is made by CAG, we
cannot include this compound in our risk assessment.

Styrene is one of the CERCLA  (Superfund) hazardous
substances that has been identified as a potential
carcinogen by CAG.

EPA's Office of Toxic Substances is  "moderately concerned11
about lead and is evaluating the potential  carcinogenicity
of lead.  Memo from Jim Cogliano, CAG to Jack  Kooyoomjian,
Emergency Response Division, Office of Emergency and
Remedial Response, USEPA, "Potential carcinogens for
designation or updating," June 2,  1987.


                             IV-14

-------
                      TABLE IV-4  (continued)

                              NOTES

10    The kinds and amounts of compounds within this chemical
     class vary depending on the combustion source.
     Consequently, the unit cancer risk factor varies with source
     type.  The unit risk values are to be used with estimates of
     total particle exposures (i.e., the combination of POM and
     elemental carbon).

11    These numbers were generated by the comparative potency
     method for estimating unit cancer risk.  Generally, there is
     more uncertainty surrounding these numbers than for CAG unit
     cancer risk factors.  (Source: Joelien Lewtas, "Combustion
     Emissions: Characterization and Comparison of Their
     Mutagenic and Carcinogenic Activity," in Hans Stich, ed.,
     Carcinogens and Mutaaens in the Environment. Volume V (CRC
     Press, Inc., Baton Rouge, 1985), pp. 59-72.  Also, Roy
     Albert, J. Lewtas, S. Nesnow, T. Thorslund, and E. Anderson,
     "Comparative Potency Method for Cancer Risk Assessment:
     Application to Diesel Particulate Emissions, "Risk Analysis.
     Vol. 3, No. 2, 1983.

12    There is greater uncertainty surrounding these comparative
     potency estimates than for other compounds because of
     limited tumorigenicity data.

13    The potency estimate for POM from coke oven emissions is
     based on CAG's potency estimate which is derived from human
     epidemiological data.  Because the CAG unit cancer risk
     factor is only for POM, the value was adjusted to reflect
     total particle emissions (i.e., POM and elemental carbon).
                                 IV-15

-------
experimental versus actual exposure levels.  EPA also includes an
additional modifying factor—ranging from greater than zero to
ten—to reflect qualitative professional judgments regarding
scientific uncertainties not covered by uncertainty factors, such
as the completeness of the overall database and the number of
animals in the study.11  Thus, the reader should keep in mind
that the quality of evidence underlying the threshold values used
in this report vary widely by pollutant.

          ii.  Dose-Response Evaluation

     To assess the noncancer health risks for the selected
pollutants, we compared measured and modelled concentrations with
estimated human threshold values for noncancer health effects.
The threshold values were derived from EPA-developed reference
doses (RfD) or estimates from EPA's Regulatory Integration
Division (RID) and Office of Air Quality Planning and Standards
(OAQPS) toxicologists.  In general, the RfD is an estimate (with
the uncertainty spanning potentially an order of magnitude or
greater) of a continuous lifetime human exposure (considering
sensitive subpopulations) that is not likely to present an
appreciable risk of adverse health effects.  Doses below the RfD
are not likely to be associated with any health risks.  However,
as the frequency and size of exposures exceeding the RfD
increase, the probability that adverse health effects may be
observed in the exposed population increases, thus indicating a
need for further investigation.  The RID threshold values were
developed using the same procedures used in deriving RfDs.
However, these threshold values have not undergone EPA review.
In converting to dose from concentration in the ambient air,
standard EPA exposure assumptions were used, which assume that
the typical person weighs 70 kg and breathes 20 cubic meters of
air each day for 70 years.

     The threshold values are expressed mostly in terms of six
broad health effects categories!  renal (kidney), hepatic
(liver), reproductive (male and female), fetal developmental
(teratological), neurobehavioral, and blood effects.  Additional
broad health effects categories—mutagenicity, nonspecific
cellular effects, gastrointestinal, respiratory and
cardiovascular—were examined but were not included in the final
analysis because lower thresholds for the target pollutants were
available for the other health endpoints.  Few of the target
pollutants are known or have been shown to produce these other
effects at environmental levels.  If a compound was shown to be
generally toxic, but was not necessarily linked with the
     "U.S. EPA, Office of Solid Waste Workshop on Risk and
Decision Making, "Principles of Risk Assessments  A Nontechnical
Review," p. IV-3.

                              IV-16

-------
malfunctioning of a specific organ, we grouped its threshold
value under the rubric of "non-specific cellular" effects.

     Existence of a threshold does not imply a dose-response
relationship nor is a threshold for a health effect for one
pollutant necessarily comparable in severity to that of another
pollutant.  Thus, for example, the health effect for pollutant A
which falls within the category of liver effects may be life-
threatening.  On the other hand, the health effect for pollutant
B, which does or does not fall within the same effects category,
may not be.

     We present possible noncancer health effects, estimated
thresholds, and sources used in deriving the thresholds in
Table IV-5.  The compounds are grouped according to their
chemical structures into four subgroups since some are potential
metabolic precursors for others.  The reader should keep in mind
that the quality of evidence underlying these threshold values
varies widely.


3.  HUMAN EXPOSURE EVALUATION

     To perform an exposure evaluation, it is necessary to have
information on ambient concentrations and the number of
individuals exposed to these levels.  We estimated ambient air
levels of the target pollutants using air dispersion modelling
and ambient air monitoring.  Ambient monitoring that took place
during the study was conducted by EPA's TEAM with the assistance
of EPA's Region III Central Regional Laboratory.  In addition, we
gathered existing monitoring data for the Baltimore area to use
in our exposure determinations.

     a.  Air Dispersion Modelling

     Air dispersion modelling requires (1) emissions estimates
for the selected pollutants by source category and (2) computer
models to estimate ambient concentrations resulting from source
emissions.  This section describes the preliminary emissions
inventory that we compiled as part of this study and the steps
taken to refine our emission estimates for key sources.  It also
describes the dispersion models used in this analysis and the
results of our model performance evaluation.  The model
performance evaluation was designed to ascertain our model's
precision in predicting actual ambient levels by comparing the
modelled results with measured values.

        i.  Emissions Assessment

     The rough preliminary emissions inventory compiled for the
risk assessment phase of the air toxics study—which was updated
from the one used in the pollutant selection process—includes

                              IV-17

-------
                              TABLE IV-5

             NONCANCER HEALTH EFFECTS  AND THRESHOLD VALUES
        FOR TARGET COMPOUNDS IN THE BALTIMORE IEMP AIR ANALYSIS
Substance
Health Effect   Threshold Value1
  Category        (ug/nr)         Source1
Modelled compounds

   Organic gasesi

   Benzene and derivatives

Benzene
Ethyl benzene


Phenol


Toluene
Xylene
  blood
  fetal/dvlopnntl

  liver
  kidney

  liver
  kidney

  liver
  kidney
  reproductive
  blood
  neurobehavioral
                  respiratory
  fetal/dvlpmntl

  liver
  neurobehavioral

  respiratory
  cardiovascular
  blood

  kidney
  reproductive
  fetal/dvlpmntl
   2.451   Snyder et al, 1980
  41.3    1986 RID analysis

 350.0    IRIS RfD*
 350.0    IRIS RfD*

  40.0    IRIS RfD*
  40.0    IRIS RfD*

1050.0    IRIS RfD*
1050.0    IRIS RfD*
 500.0    Matsumoto et al., 1971
 1050.0   CITT, 1980
 580.0    Hanninen et al, 1976;
          Seppalainen et al.,
          1978
1010.0    Von Dettingen et al.,
          1942; Bruckner &
          Peterson, 1981
 476.0    Hudak & Ungvary, 1978

 215.0    Bowers et al, 1982
 215.0    EPA, 1984; Savolainen
          et al, 1979
 215.0    Hipolito, 1980
 215.0    Hipolito, 1980
 215.0    Hipolito, 1980;
          Browning, 1965
 215.0    EPA, 1984
  52.8    Ungvary et al., 1980
  52.8    Ungvary et al., 1980
                                 IV-18

-------
                        TABLE IV- 5 (continued)
             NONCANCER  HEALTH EFFECTS AND THRESHOLD VALUES
        FOR TARGET  COMPOUNDS  IN  THE  BALTIMORE  I BMP AIR ANALYSIS
Substance
Health Effect
   Category
Threshold value1
    (ug/ms)        Source
  Halogenated Aliphatics
Carbon
  tetrachloride
Chloroform
Ethylene
  dichloride
Ethylene
  dibromide
Methyl
  chloroform

Methylene
  chloride
 liver

neurobehavioral
kidney
reproductive
fetal/dvlpmntl

fetal/dvlpmntl

liver
neurobehavioral

kidney
reproductive

liver
neurobehavioral
kidney

gastrointestinal

reproductive
           (male)

reproductive
        (female)
liver
kidney

nonspecific
   cellular

 liver
 fetal/dvlpmntl
kidney
neurobehavioral
 Perchloroethylene liver
                  kidney
                  fetal/dvlpmntl
      2.45   IRIS RfD; EPA, 1980 &
             1984; Smyth,1935 & 36
      2.45   Holler, 1973
    108.0    EPA, 1980
    430,0    Adams et al., 1952
     24.2    Schwetz et al., 1974

      2.4    Murray et al., 1979
             Schwetz et al., 1974
     35.0    IRIS RfD4
     11.7    EPA 1985, Challen et
             al., 1858
     22.5    Heywood et al., 1979
      2.4    Schwetz et al., 1974

     26.0    Kozik, 1957
     26.0    Kozik, 1957
     26.0    Heppel et al., 1946
             Hoffman et al., 1971
     26.0    EPA, I982a
      1.75   Hurtt & Zenick, 1985
             NTP, 1982

      11.9    Short et al.,  1977
      12.8    NTP, 1982
      12.8    NTP, 1982

     315.0    IRIS RfD*
     210.0
     210.0
     699.0
    8600.0

      69.9
      69.9
     909.0
IRIS RfD*
Schwetz et al, 1975
NTP 1985 draft
EPA, 1985a
                              IRIS  RfD*
                              IRIS  RfD*
                              Nelson  et  al.,
               1979
                                 IV-19

-------
                        TABLE  IV-5  (continued)
             NONCANCER HEALTH.EFFECTS AND THRESHOLD VALUES
        FOR TARGET COMPOUNDS IN THE BALTIMORE IEMP AIR ANALYSIS
Substance
Health Effect
 Category
Threshold Value1
(ug/m3)
Source
Propylene
  dichloride
 liver

 kidney
Trichloroethylene kidney
   (TCE)           liver
                  neurobehaviora1
   Other Qrqanicflt

Benzo(a)pyrene
     308.0    Basu,  et al.  1984
              NTP 1983
     308.0    NTP 1983

    3770.0    Tucker,  1982
      25.9    EPA,  1984
      25.9    Grandjean et  al,  1955
Ethylene oxide3

Glycol ethers6




Formaldehyde7

   Metale;

Arsenic


Cadmium
Chromium VI
fetal/dvlpmntl       93.3
reproductive
blood, kidney
liver

nonspec. cell.       12.37
nonspecific          13.3
cellular effects

fetal/dvlpmntl       21.0

liver                 0.4
respiratory           2.0
kidney                0.24
reproductive        119.0

fetal/dvlpmntl       17.0
liver                17.0
reproductive         17.0
nonspecific cell.    17.5
              Hanley et  a1.,1984
              OTS,  1986
              Heywood  & Sortel,'79
              Schroeder  &
              Kitchener, 1971
              Friberg, 1950
              Comm. Eur. Com.,1978
              Kjellstrom et al,'77
              Scharpf et al., 1972

              Gale, 1978
              NIOSH, 1975
              Gale, 1978
              IRIS RfD4
                                 IV-20

-------
                       TABLE  IV-5  (continued)

            NONCANCER HEALTH EFFECTS AND THRESHOLD VALUES
       FOR TARGET COMPOUNDS IN THE BALTIMORE IEMP AIR ANALYSIS
Substance
   Health Effect
     Category
Threshold Value
    (ug/m3)
Source
Compounds onlv monitored
 Organic aaaeai
 Benzene derivatives
Chlorobenzene8  nonspec. cell.      104.9     Inne et al, 1969
Dichlorobenzene liver
(para & ortho   blood
 forms)         kidney
                neurobehavioral
Styrene
cardiovascular
respiratory

liver
blood
 Halogenated aliphatics

Vinylidene      liver
 Chloride

Vinyl chloride  liver
 Metalat

Lead
nonspecific
    cellular
  315.0     NTP, 1982
  315.0     Varashavskaya,1967
  315.0     NTP, 1982
  315.0     Hollingsworth et
            al. 1956
  315.0     NTP, 1982
  315.0     NTP, 1982

  699.3     IRIS RfD*
  699.3     IRIS RfD4
                     31.5     IRIS RfD*


                      4.5     EPA Office of
                              Drinking Water
     1.5     Health Effects
            Assessment Document
            prepared by  the
            Environmental Criteria
            and Assessment Office,
            USEPA, Cincinnati, Ohio
            (May  1986)
                                IV-21

-------
                        TABLE IV-5 (continued)

                                NOTES


 1 Quality of the data for the thresholds varies greatly.  Those
  derived from RfDs are generally more reliable.  Noncancer health
  risks were assessed in two ways:  (1) for individual pollutants/ the
  modelled and measured ambient air concentrations were compared with
  the estimated threshold values, and (2) for mixtures of pollutants
  with the same systemic effect, a hazard index was created (see
  Chapter III).  If the index exceeds the value "1", this indicates
  the need to  further investigate these exposure levels.  The hazard
  index is created by summing the ratio of the predicted or measured
  ambient air  concentration to the threshold  value for all pollutants
  associated with the same health effect category.

 2 All individual studies referenced can be obtained by contacting the
  Regulatory Integration Division, U.S. EPA,  Washington, D.C.

 3 This threshold value is being reviewed by EPA's Environmental
  Criteria and Assessment Office.

 4 IRIS = Integrated Risk Information System.  IRIS is an EPA database
  that maintains the most recent information  of RfDs and CA6 values
  for use in risk assessment/risk management.  Information on IRIS
  can be obtained by contacting EPA's Office  of Health and
  Environmental Analysis, Office of Research  and Development.

 3 The Risk Analysis Branch of EPA's Office of Toxic Substances has
  identified tentative threshold values for ethylene glycol
  monomethyl and monoethyl ether, not for the monobutyl ether form
  which is the glycol ether of primary concern in this study on the
  basis of industrial use of glycol ethers in the Baltimore study
  area.  Toxicological data on ethylene glycol butyl ether suggest
  that its toxicity is similar to that of ethylene glycol monomethyl
  ether, producing primarily blood, liver, and kidney effects.  The
  threshold values given are for ethylene glycol monomethyl ether and
  should serve, for the purpose of screening noncancer risks,  as
  conservative estimates of threshold values  for the glycol ether
  group.

6 RID scientists estimated the threshold for health effects based
  upon the results of an evaluation of toxicological data on health
  risks that was conducted by the Risk Analysis Branch of EPA's
  Office of Toxic Substances (Assessment of Health risks to Garment
  Workers and  Certain Home Residents from Exposure to Formaldehyde,
  December 1986).   The report has yet to be formally peer or
  administratively reviewed.

7 The threshold value should be-considered tentative because it has
  been subject to very limited EPA review.

                               IV-2 2

-------
manufacturing process and boiler emissions from approximately 250
point sources.  It also includes estimates of emissions from area
sources and nontraditional sources (such as cooling towers and
sewage treatment plants).  The inventory covered more than 200
pollutants.  Although this inventory was relatively
comprehensive/ it is important to note that the quality of the
emissions estimates varies considerably by source.

     It should also be noted that the Workgroup made a conscious
decision in 1986 to stop making revisions to the emissions
database because of time and resource constraints.  New and more
refined emissions information for some of the major sources
included in this study became available after 1986, but were not
included in this report.

           (1)  Point Sources

     We estimated emissions from point (industrial) sources
using data from the 1985 TSR survey.12  This survey generated
data on hazardous materials that were being used, produced or
handled at selected industrial facilities.  Based on this
information, AHA estimated plant-wide daily emissions using
registration files on plant operations, EPA emission factors for
certain kinds of manufacturing facilities, and engineering
judgment.  The daily emission rates were then converted to annual
rates using plant operating schedules.

     There were three major limitations in using the Maryland
1985 TSR survey to estimate emissions.  First, the survey did not
consider all sources and pollutants.  Second, it was not possible
to differentiate between fugitive emissions and stack emissions
at specific point sources.  Third, the survey did not cover
emissions from fuel combustion (i.e., metals, aldehydes, and
polycyclic organic matter—POMs).  Finally, only total plant-wide
use information was available to estimate emissions (i.e.,
emissions were- not broken down within a plant by process or
stack).  For large facilities with emissions coming from many
locations at the plant, our inability to identify the location of
these emissions could limit the accuracy of results from the
dispersion model.

     We supplemented the point source emissions inventory with
emissions estimates for source categories that were not
originally included, such as boilers.  For large industrial
facilities, we assigned emissions to processes or areas within
the facility where the emissions were likely to occur.  We
     12After pollutant selection, source emissions were
reestimated using the more recent TSR survey from the fall of
1985.  This more recent survey was not yet available when
emissions estimates were generated for the pollutant selection.

                              IV-23

-------
 assumed that all emissions were released from stacks in our
 modelling.

      We calculated volatilization of  organics during wastewater
 treatment at the sewage treatment plants using data  from
 influent, effluent/  and sludge  monitoring.  We used  mass balances
 and  estimates of destruction rates during treatment  to  estimate
 emissions of the volatiles.   We also  calculated emissions
 resulting from sewage sludge incineration on  the  basis  of our
 estimates on the type of incinerator, air pollution  controls,
 sludge  composition and throughput, and  literature estimates of
 destruction  and removal efficiencies.13

      To complement the existing process data  we developed an
 algorithm and a set  of emission factors to estimate  boiler
 emissions for the point sources in the  toxics inventory.14  These
 unique  emission factors were based on fuel type,  boiler type and
 size, and assumed in-place controls.  We then multiplied the
 factors by fuel consumption  reported  in the AMA registration
 files to estimate metals,  formaldehyde,  and*1 total suspended
 particulate  (TSP)  emissions  from point  source boilers.   We
 included TSPs so that we could  later estimate risk from exposure
 to POMs, as  described below.

     We estimated emissions  of  POMs using a technique developed
 in conjunction with  EPA's  Office of Air Quality Programs and
 Standards (OAQPS).   We calculated POM emissions, defined as  the
 benzene-extractable  portion  of  combustion particulate matter, by
 first estimating TSP emissions  from point and area combustion
 sources.  EPA's Genetic Bioassay Laboratory provided estimates  of
 the percentage of  extractable organics  by source category.   The
 resulting POM emission rates were later modelled and multiplied
 by source-specific potency scores  also  provided by the  Genetic
 Bioassay Laboratory.   For  point  sources,  emissions of POMs were
 estimated for industrial coal and oil combustion, coke  ovens, and
municipal waste incinerators.

     Because  hexavalent chromium is highly potent, we made an
attempt  to estimate  hexavalent  chromium,  as opposed to  total
chromium, emissions  as  accurately as the data would allow.  This
      The emission calculations are documented in:  U.S. EPA,
Office of Policy Analysis, Regulatory Integration Division,
Baltimore Phase II Air Toxics Emissions Assessment,  Versar Inc.
for EPA, EPA Contract No. 68-01-7053, November 6, 1987.

     "ibid.

                              IV-24

-------
determination was made using best engineering judgment and EPA
emission factors.15

           (2)  Area Sources

     We designated all sources not included in the category of
major point sources as area sources.  Examples of area sources
that were included in the analysis are:  motor vehicles; gasoline
service stations; solvent usage by commercial and small
industrial facilities (dry cleaning establishments; businesses
that do degreasing, surface coating, printing and publishing; and
manufacturers and users of rubber and plastics); industrial,
commercial, institutional, and residential heating (including
wood stoves and fireplaces); cooling towers; the drinking water
distribution system from which trihalomethanes volatilize; waste
oil combustion; agricultural burning, and minor point sources not
modelled individually.  We calculated emissions from these
sources, except the minor point sources, using an assortment of
emission factors and algorithms developed by RID and EPA's Office
of Air and Radiation.

     Emissions from cars and trucks were estimated using a
computer model called MOBIL3 that calculates total hydrocarbon
emissions on the basis of vehicle miles traveled in rural and
urban sections of the study area.  These estimates of total
hydrocarbon emissions were subsequently disaggregated into
estimates for specific pollutant emissions on the basis of
emission factors developed by OAQPS and published in the
scientific literature.

     Area sources of POMs included gasoline- and diesel-powered
road vehicles; small industrial boilers not included in the point
source inventory; residential, institutional, and commercial
heating; and wood stoves and fireplaces.  We used techniques
similar to those described earlier for point sources to estimate
POM emissions as source category-specific percentages of TSP
emissions.17  All of the emissions data were entered into the
PIPQUIC database.
     l5The  emission  calculations  are documented in:  U.S.  EPA,
Office of  Policy Analysis, Regulatory Integration Division,
Baltimore  Phase II Air Toxics Emissions Assessment.  Versar Inc.
for EPA, EPA Contract No. 68-01-7053, November 6, 1987.

     "ibid.

     "The  emission  calculations  are documented in:  U.S.  EPA,
Office of  Policy Analysis, Regulatory Integration Division,
Baltimore  Phase II Air Toxics Emissions Assessment.  Versar Inc.
for EPA, EPA Contract No. 68-01-7053, November 6, 1987.

                              IV-25

-------
     Area sources are not uniformly distributed within a
geographic area,  Thus, we distributed the area source emissions,
except road vehicles, on the basis of USGS land use data.  The
entire study area was subdivided into quadrangles of irregular
shape and size composing the area source grid shown in
Figure IV-1.  Land use within each quadrangle was then broken
down into various use categories, including residential,
commercial and services, industrial, and transportation
categories.  Emissions associated with different area source
activities were then apportioned into the grids depending on use.
For example, dry-cleaning emissions were distributed
proportionately to commercial and service land use categories.
For motor vehicles, emissions were distributed over the area
source grid cells on the basis of the computer model MOBIL3,
previously described.

           (3)  Quality Control

     Quality assurance/quality control was an integral part of
our data gathering efforts.  The number of emission estimates and
factors, algorithms, combined with multiple applications and
manipulations of data necessitated that we continuously review
the data and algorithms to assure that all data entered into
PIPQUIC were as accurate as possible and that PIPQUIC tools and
algorithms were operating as designed.  Thus, the massive size
and interrelational nature of the PIPQUIC database meant that
quality control became a routine part of all data gathering and
analysis.

     Some of the specific measures taken were the following:

     •  EPA's technical contractors reviewed AMA's plant-wide
        point source emissions estimates, which were based on the
        1985 TSR survey, to identify any apparent errors.

     •  Technical staff at EPA's RID reviewed all emission
        factors and algorithms developed by the technical
        contractor for area sources, nontraditional sources, and
        point source boilers, to assure that values were correct
        and applied properly.  Many of these factors underwent
        further review by other EPA Offices, such as the Office
        of Air Quality Planning and Standards.

     •  As soon as data were entered into PIPQUIC, EPA's
        technical contractors checked printouts again to assure
        that all algorithms and factors were correct.

     •  EPA's technical contractors periodically generated new
        printouts to compare with previous printouts to guarantee
        that any internal modifications to PIPQUIC tools did not
        incorrectly change PIPQUIC output values.


                              IV-26

-------
                             FIGURE IV-1
       Area Source Grid System for Baltimore Study
4401
                 331  331  341  344  391 3»  311 3M  371 371  3*1 3M  391
                                      Note:
                                      	County Line
                                      	Area Source Grid
                                      UTM  - Universal Transverse Mercator
                               IV-2 7

-------
        ii.  Emissions Characterization for Kev Sources

     Because of the large number of point sources that make up
the emissions inventory, we ranked facilities to identify which
sources to emphasize in our analyses.  We estimated the portion
of the plant-wide emissions for these facilities to be assigned
to a particular process or stack.  This step was necessary to
improve the accuracy of the modelling because (as mentioned in
the previous section) the rough emissions inventory provided only
a single figure for the emissions of a pollutant from a plant.

     We ranked facilities using three approaches: (1) total
emissions of study pollutants, (2) emissions of study pollutants
weighted by unit cancer risk factors, and (3) emissions of study
pollutants weighted by both exposed population and the unit
cancer risk factor.  In the first ranking based on emissions, one
point source accounts for nearly 50 percent of the total point
source emissions of study pollutants in the study area; the top
10 facilities contribute 80 percent of the point source
emissions.

     While the product of emissions and unit cancer risk factors
does not reflect actual risk, it does consider two important
factors that contribute to human health risk:  the volume of
emissions and the toxicity of the pollutant(s).  The same point
source that accounted for most of the emissions also dominated
the second ranking in terms of toxicity-weighted emissions.
Several smaller sources of emissions ranked high in the second
approach because of the estimated high toxicity-weighted
emissions of hexavalent chromium.

     Multiplying the risk-weighted value by the population in the
vicinity of each facility gave an indication of which sources are
likely to be exposing the largest populations to the greatest
volumes of the most potent carcinogens.  The relative rankings of
point sources by this measure differed very little from the
rankings based on toxicity-weighted emissions.  This indicates
that variation in potencies of the pollutant is more significant
than any variation in population among receptor grid cells.

     On the basis of these three rankings, we selected 15
facilities as major point sources.

     For complex integrated major point sources, we broke down
emissions by process, plant area, or stack.  We distinguished
between process and boiler emissions and assigned representative
or unique stack parameters for each release, basing our selection
on each facility's largest VOC or TSP stack as listed in the
AMA's registration files.  We visited some major point sources to
update and verify emissions, gather modelling input information,
such as stack parameters and building dimensions inside and


                              IV-28

-------
 adjacent to the facility,  and gather data  on feasible control
 options  (see Chapter VI).

      We  classified approximately 40  additional  facilities,  for
 modelling purposes,  as  "intermediate point sources."   For these
 plants,  emissions  from  boilers and process emissions  were
 modelled separately.  We chose a single  process stack and one
 boiler stack based on registry data  to be  representative of
 process  and boiler emissions,  respectively.   We made  no attempt
 to  further break down emissions for  these  operations.  As a
 general  rule,  we chose  stacks  emitting the largest amount of VOC
 as  being the representative  process  stack,  while the  boiler stack
 emitting the greatest quantity of particulates  was selected as
 the representative boiler  stack (since the pollutant  of concern
 would be POM).

      Because of  the  small  volume of  emissions from the remaining
 200 very small  point  sources,  we modelled  these "minor point
 sources"  as  area sources,  with area  source release specifications
 (height,  temperature, and  exit velocity).   Each of these minor
 point sources  is still  listed  in PIPQUIC as  a point source.

      The decision  rules used in our  modelling of large,
 intermediate, and  area  sources may have resulted in
 underestimates of  maximum  increased  lifetime individual cancer
 risk  and the need  to  further investigate noncancer effects.  By
 assuming that emissions from all large and  intermediate sources
 were  released from the  stack with the greatest  VOC or  particulate
 emissions, we often modelled the tallest stack.  Emissions from a
 taller stack will  disperse better, thus resulting in  lower
 maximum  concentrations.  Similarly,  modelling all small volume
 emitters  as  area sources—although practical—eliminated our
 ability  to consider peak concentrations.

         iii.  Dispersion Modelling

            (1)   Choosing the Model

      In  choosing an air guality model for evaluating urban
 exposures in Baltimore, we selected  a model that is (1) cost-
 effective in estimating annual  exposures, and (2) capable of
 addressing urban area sources  as well as point  sources.  While a
 number of readily  available  models fulfill these desired
 characteristics, we chose two  off-the-shelf EPA Gaussian
dispersion models:  the Climatological Dispersion Model (COM)  and
 the Industrial Source Complex  Long-Term (ISCLT) model.18
     18U.S. EPA, Guideline on Air Quality Models {Revised!. U.S.
Environmental Protection Agency, Office of Planning and
Standards, Research Triangle Park,  N.C., EPA-450/2-78-027R, 1986.
                              IV-29

-------
     The CDM is an EPA preferred model that is appropriate for
estimating the annual average concentrations for both point and
area sources located in flat, urban regions.19  We used CDM to
estimate the larger/ urban area source impact.  The integration
technique used in the CDH model for area sources theoretically
produces more accurate results than are obtained through the more
simplified treatments found in most other models.

     The ISCLT model is also an EPA recommended model for
evaluating point source releases (including small industrial area
sources) located in regions of flat or gently rolling terrain.20
ISCLT is generally the model of choice when evaluating major
industrial sources, especially facilities for which detailed data
are available on release specifications such as building downwash
and vent releases.  We chose this model to estimate the annual
concentrations in the neighborhoods surrounding each major
facility.

           (2)  Running the Model

     We modelled 55 release points at 52 major and intermediate-
size facilities (one of the large facilities had three release
points modelled).  The remaining minor facilities were grouped
into 23 area sources and modelled.  We used detailed release
specifications (when available) to assess the microscale
concentrations (that is, concentrations at discrete receptor
sites).  With the exception of a few large facilities, most
facilities were modelled using one representative stack.

     We used hourly meteorological data that had been collected
at a downtown monitoring site during Phase I of the Baltimore
IEMP study in 1983/1984.  These data consist of wind direction,
wind speed, and horizontal and vertical turbulence information.
We derived the temperature and stability data by using the
National Weather Service stations at Baltimore-Washington
International (BWI) Airport, and obtained mixing height data from
Washington (Dulles) Airport.  When data were missing from the
downtown station, we used surface data from BWI Airport.   We then
reformatted the hourly meteorological data into day/night joint
frequency distributions (frequency of wind directions, wind
speeds, and stabilities), as required by both our air dispersion
     "irwin, J.S., T. Chico, and J. Catalano, CDM 2.0
Climatoloqical Dispersion Model—User's Guide. U.S. Environmental
Protection Agency.

     ZOU.S. EPA,  Industrial Source Complex  (ISC^ Dispersion Model
User's Guide. Second Edition, Volume 1.  U.S. Environmental
Protection Agency, Research Triangle Park, N.C., Publication No.
EPA-450/4-86-005a, 1986.

                              IV-30

-------
models, Industrial Source Complex Long-term (ISCLT) Model and
Climatological Dispersion Model (COM).

        iv.  Model Performance Evaluation

     We conducted a limited evaluation of the performance of the
dispersion model using monitoring data that had been collected
for three months at ten stations in 1983-84 during Phase I of the
Baltimore IEMP.21  The evaluation allowed us to assess the
adequacy of the modelling conducted for the exposure assessment
for the Baltimore IEMP project.

     We ran COM (for urban area sources) and ISCLT (for point
sources) to estimate the pollutant concentrations at locations
corresponding to the monitoring sites and monitoring periods.  We
then compared the modelled and measured concentrations.

     Table IV-6 displays the modelled and measured ambient air
concentrations for the ten compounds and ten monitoring sites.
For all pollutants considered, except trichloroethylene, the
models consistently underpredicted ambient air concentrations.

     We statistically compared measured and modelled data for
bias and correlation.  Good correlation indicates that our
modelling reliably predicts ambient levels.  Bias tells us the
extent to which the model under- or overpredicts actual levels.

     Table IV-7 displays the bias and correlation coefficients
for the Baltimore IEMP study (with and without the contribution
of point source emissions) and the Philadelphia IEMP study.  When
comparing model performance in the Baltimore and Philadelphia
IEMP studies, the bias for the Baltimore model performance is
substantially greater for most compounds and the correlation
substantially lower.  The higher bias could be attributed to the
following factors:   (1) a potential underestimation of the
Baltimore point or area source emissions,  (2) failure to account
for background concentrations imported  from outside the study
area or by chemical  transformations,  or  (3) that the models are
systematically underpredicting ambient  levels due to the
treatment of atmospheric and meteorological conditions.
      "The monitoring data collected in 1987 as part of the TEAM
 study,  previously described,  were not  available in  time for this
 evaluation.

      2ZA more detailed model performance check would also include
 partitioning of the results as a function of meteorological
 conditions,  and performing a follow-up verification of point
 source  emissions data.   However, we did not conduct a detailed
 performance  check because the 1983-84  data are outdated.

                               IV-31

-------
M
Ul
Isj
                                                            Table IV-6
                  Comparison  of Measured  vs.  Predicted Concentration*  for Baltimore  Monitoring Sites
                                 During  the Period 11/20/83  -  2/16/84 for Model  Hun llrbun Mode I
      Compound
Measured Cone. (mg/m3)
Benzene
Cartxm Telrachtoiide
ChkMDfbim
Eltiyl Beniene
Xytone
Toluene
1.2CMct)kxoelhane
1 .2 OfcNw opropane
Tilchloroelhylene
Peichtoroelhylene
Predicted Cone. (mo,/m3)
Benzene
Carbon TeUachtorUe
Chloroform
Ethyl Benzene
Xytene
Toluene
1.2DJcWoroelhane
1.2Olchloropropene
Trkhloraethylene
Peicrttoroeinytene
Ratio ol Pred./Meaa.
Benzene
Caibon Tetrechlortde
Chloroform
Eltiyl Benzene
Xytone
Toluene
l.20fchtoroelhane
1 .2 OlcMofopr opane
Trfchtoroethylene
Perchloroetliytene
t
1200
1 10
210
930
3040
1640
020
030
090
4 BO
1
421
000
005
062
283
479
001
000
144
434
1
035
000
002
007
009
029
005
000
160
090
2
950
090
040
640
21 10
900
020
020
050
540
2
223
000
007
080
205
312
001
000
160
578
2
023
000
017
012
010
035
006
000
320
107
3
1020
060
020
800
2080
990
030
020
050
700
3
261
000
004
049
181
320
001
000
140
3 10
3
026
000
018
006
009
032
002
000
279
044
4
1290
130
070
740
2010
1600
050
070
390
600
4
397
000
003
038
214
404
001
000
092
244
4
031
000
004
005
Oil
025
001
000
024
041
5
1260
090
too
470
1370
930
040
030
1 40
930
5
928
000
003
036
1 98
267
001
000
104
240
5
074
000
003
008
014
029
002
000
074
026
6
550
090
320
290
570
750
260
010
040
1 SO
6
992
000
002
025
1 45
206
001
000
064
163
e
1 80
000
001
009
025
027
000
000
161
109
7
1210
120
260
620
1720
7.60
070
200
1 10
320
7
930
002
006
038
917
738
001
000
076
186
7
077
001
002
006
053
097
001
000
069
058
a
1060
1 40
470
570
1640
480
020
040
100
390
8
669
000
003
041
303
438
001
000
090
225
8
063
000
001
00?
018
091
004
000
090
058
9
780
070
060
520
1360
550
100
040
020
290
9
660
000
002
023
185
399
001
000
061
ISO
9
085
000
003
004
014
072
001
000
306
052
10
1030
060
1 10
630
1920
740
020
020
030
240
10
S96
000
002
020
1 19
268
000
000
056
1 35
10
058
000
002
003
006
036
002
000
185
056
       Note: BalUmoto monitoring data used In tins analysis wen collected m 1983/84
Sites
  1 Gulltord
  2NEPO
  3SWPD
  4 HoteWrd
  9 Uumtaft
6 Chesapeake Terrace*
7 Sun A Chesapeake
6 Fl Henry
9 Coasl Guutti
10 Rivet a Beach

-------
                                      Table IV- 7
              Comparison  of Measured  vs.  Modelled  Data
         for  the  Baltimore and Philadelphia  IEMP  Studies  [1]
                                      Bias [2]
                    Baltimore IEMP   Baltimore IEMP
Compound

Benzene
Carbon  Tetrachlonde
Chloroform
Ethyl Benzene
Xylene
Toluene
1,2 Dichloroethane
1,2 Oichloropropane
Trichloroethylene
Perchloroethylene
                                      Philadelphia IEMP
                                       Paint and Area
Compound

Benzene
Carbon Tetrachlonde
Chloroform
Ethyl Benzene
Xylene
Toluene
1,2 Dichloroethane
1,2 Dichloropropane
Trichloroethylene
Perchloroethylene
m mm wuiv
9.74
0.96
1.63
5.82
16.93
7.89
0.62
0.48
0.04
2.03
r wi ••% «ip»« ^^ • •• ••
4.27
0.96
1.62
5.80
15.07
5.51
0.62
0.48
0.03
1.97
3.70
2.00
2.70
6.00
16.60
8.60
0.20
0.40
1.30
1.70
                                     Correlation
                    Baltimore IEMP    Baltimore IEMP
                                      Philadelphia IEMP
                                        Paint and Area
Mrvai wniv
0.36
0.00
-0.27
0.56
0.49
0.19
-0.45
0.00
0.01
0.39
-0.20
0.36
-0.04
0.55
0.08
0.08
-0.45
0.00
0.01
0.39
0.62
0.28
•0.49
0.38
0.51
0.53
0.92
0.99
0.02
0.80
 [1] For the compoeae period 11/83 • 3/84
 [2] Average measured concentration - Average modelled concentrauon
 Note: Baton
tormg data ueed in ttiia analysis were collected in 1983/84.

                           IV-3 3

-------
     Of the ten compounds considered, we found that ethyl
benzene, perchloroethylene, xylene, and benzene had the best
correlation in the model runs.  Of these four, ethyl benzene
correlated highest in the Baltimore area with a value of 0.56.
The remaining six compounds, chloroform, carbon tetrachloride,
toluene, 1,2-dichloroethane, 1,2-dichloropropane, and
trichloroethylene (TCE) had negligible lower correlations.

     This analysis indicates that our air dispersion modelling
underestimates actual ambient concentrations of the selected
pollutants, with the exception of TCE.  Predicted ambient air
concentrations of benzene and benzene-ring compounds, like
xylene, toluene, and ethyl benzene, are likely to be
underestimated by at least a factor of two.  TCE may have been
overestimated by a factor of two.  The values for the other
compounds are at best seen as lower estimates of ambient
concentrations.  Consequently, our exposure estimates based on
modelling are understated; however, our estimates of risk using
these modelled values are still more likely to overstate than
understate true risk.

     b.  Ambient Air Monitoring

     EPA's Office of Research and Development—in coordination
with the AHA and the Baltimore County Department of Health—
conducted monitoring of the ambient air at two fixed sites
(Dundalk and Parkville) in the Baltimore study area as part of
its TEAM study (see Figure IV-2).  The sampling focused on 27
compounds and their isomeric forms, of which 18 were VOCs
(volatile organic compounds, i.e., organic gases), 5 were
straight-chain aliphatic hydrocarbons, and 4 were metals.  The
modelling analysis addressed 13 of these 27 pollutants.

     In addition, we compiled readily available monitoring data
on organics and metals for the Baltimore area.  These data came
from various AMA studies and the 1983/1984 IEMP monitoring study.
There is considerable uncertainty surrounding these data and not
all of the results have undergone extensive peer review.
Nonetheless, we felt that it was important to include this
information to demonstrate a different approach for estimating
exposure.

        i.  TEAM VOC Sampling

     The sampling of all VOCs and straight-chain aliphatic
hydrocarbons occurred concurrently at the two fixed sites over a
six-week period between March 23 and May 1, 1987.  Table IV-8
presents a summary of the maximum and average measured pollutant
concentrations of the target VOCs and hydrocarbons included in
the Baltimore ambient air toxics study.
                              17-34

-------
                        FIGURE IV-2
Baltimore Integrated Environmental Management Project
         Map of Ambient Air Monitoring in 1987
                                      t = Monitoring Site
                                   Parkvilte Fixed
                                   Monitoring Site
                                             Dundalk Fixed
                                             Monitoring Site
                          IV-3 5

-------
     There are several limitations to the data that should be
noted  in properly interpreting the monitoring results. In
Table  IV-8, we reported the level of confidence in each pollutant
exposure level.  In particular, a'confidence level of 3 means
that the exposure estimate is probably invalid.  A low confidence
level  was assigned in situations where:

     (1)   The quality control techniques could not confirm a
           pollutant's presence (vinyl chloride)

     (2)   There were analytical problems in quantifying ambient
           concentrations of methylene chloride at Parkville, and
           vinylidene chloride at both Dundalk and Parlcville, as
           well as specific isomers of a compound (e.g., xylene)

     (3)   Pollutant emissions from sources near the monitoring
           site were suspected of interfering with the ambient
           sampling (dichlorobenzene)

     (4)   Measured concentrations may have been overstated
           because of coelution with another substance (styrene
           with xylene)

     The concentrations of pollutants at the Dundalk site,
downwind of major industrial sources, were generally lower than
anticipated.  Vinylidene chloride was the clear exception;
however, there are no known sources of vinylidene chloride in the
area and, most importantly, the measured concentrations are
highly uncertain.  Additional exploration is needed to verify
whether vinylidene chloride are, in fact, present in the ambient
air.

     The dichlorobenzenes were also unexpectedly detected in the
ambient air.  The known sources of emissions of the
dichlorobenzenes in Baltimore are only consumer products.
Statistical analysis of the relationship of their concentrations
with wind direction suggested that at least some portion of the
emissions may have come from vents from the building upon which
the monitoring station was located.  We were not able to identify
products used in the building which contained the compounds.

     The Parkville monitoring results for pollutants that have
been sampled in the past were generally at anticipated levels,
based  on the results of previous lEMPs.  The surprises were for
methylene chloride, vinylidene chloride, and the
dichlorobenzenes.  However, the measured levels of methylene
chloride and vinylidene chloride are suspect because of the
analytical problems in quantifying their concentrations.
Methylene chloride, like the dichlorobenzenes,  is heavily used in
consumer products.
                              IV-36

-------
                              Table IV-8
                   BALTIMORE IEMP AIR TOXICS STUDY

               AVERAGE MEASURED AMBIENT CONCENTRATIONS           .
     AT  THE TWO  FIXED MONITORING SITES: TARGET VOLATILE ORGANICS1

                                (ug/m3)

                                Dundalk
Compound
   Average   ,
Concentration
   Maximum
Measured Value
                                      Confidence
                                        in
                                      Exposure.
                                      Estimate
Benzene
Carbon tetrachloride
Chlorobenzene
Chloroform       .
m-Dichlorobenzene
o-Dichlorobenzene
p-Dichlorobenzene

Ethylenendibromide {EDB)
Formaldehyde      .
Methylene Chloride5
Percnloroethylene
Styrene
Trichloroethane (TCA)
Trichloroethylene (TCE)
Vinyl chlorioV    ,
Vinylidene chloride7
m-Xylene
                          3S:

                           3.8

                          21.9
                          75.0
                         22!

                          s:
                            .0
                               Parkville
Compound
   Average   ,
Concentration
                                      Confidence
                                        in
                         Maximum      Exposure.
                      Measured Value  Estimate
Benzene
Carbon
Chlorol
Chloro:
m-Dich
o-Dich
p-DichJ
tetrachloride
Benzene
:orm
.orobenzene
.orobenzene
[orobenzene
Ethyl benzene
Ethylene dibromide (EDB)
Formaldehyde ,
Methylene Chloride
Percnlorpethylene
Styrene
Trichloroethane (TCA)
Trichloroethylene (TCE)
Vinyl chloride8^ , ;
Vinylidene chloride
m-Xylener
5.2
1.1
1.5
N.D.
7.2
4.8
5.7
5.9
0.1
25610
ii:l
5.9
0.01
N.D.
68.4
15.4
19

16
48
39
27
55
J
27
109
0
ili
.2
.0
.6
. 3
. 8
.3

'.7
:jj
!4
:8
2
1
2
3
3
3

1
3
2
1
1
   N.D.:    Not  detected
                              (continued)
                                 IV-3 7

-------
                        Table IV-8 (continued)

                                 NOTES


   Monitoring occurred between March  26,  1987  and May 1,  1987 at
   Parkville Middle School in the Parkyille section.of Baltimqrr
   County and Merritt Point Mental Health Facility in the Dundalk
   section of Baltimore County.   The air monitoring was conducted as
   part  of the 1987 Baltimore Total Exposure Assessment Methodology
   fTEAM)  Study.

   Monitoring samples for which  no concentration of the target L
   compounds  was  detected were given the value zero.  On the other
    Sana,  samples  for which trace amounts of the target compounds  were
    etected but the concentrations were below quantification were
   assigned half  the value for the level of detection for that sample
   and target compound.

   There are  three levels of confidence in exposure estimates. Level
   1 means that the exposure estimate is probably good.  Level 2  means
   that  the exposure estimate may be good.   Finally, level 3 means
   that  the exposure estimate is probably not good.  The evaluation is
   based on EPA audit reports and statistical analysis by RID staff.

   It was  not possible to quantify the concentration of m-
   dichlorobenzene at Dundalk.

   There is some  evidence that methylene chloride was being emitted
   from  sources within or on top of the building where the monitoring
   station was located.  Consequently, the measured values may not be
   representative of concentrations to which people may be exposed in
   the ambient, as opposed to indoor,  environment.

   Quality control procedures could not confirm the presence of vinyl
   chloride in monitoring samples.

   There were analytical problems in quantifying ambient air
                                         at both Parkville ar__
                                         are highly uncertain and
niBiw wwtw anaj.ytiuaj. UCUUJ.«HIIS J.IL uueijii.ii,		
concentrations of vinylidene chloride at both Parkville and
Dundalk.  Thus, these measured values arc
  deserve  further investigation.

8  The  laboratory was  only able to quantify the concentration of the
  meta isomer of xylene.

9  The  laboratory analytical level of  detection (0.37  micrograms per
  cubic meter)  was not adequate to measure chloroform in tne ambient
  air  of Parkville.

10 The  value  given for styrene is  probably high because of coelution
  with the xylene isomer  o-xylene.
                                 IV-38

-------
     Because the 1983 IEHP monitoring effort (discussed in the
Phase I report) did occur under similar meteorological
conditions, we compared the averages for compounds present in the
two sets of samples.  The results shown in Table IV-9 are
generally similar, differing by less than a factor of two.  The
earlier data appear to be the higher of the two.  The exceptions
are chloroform/ trichloroethylene, and, to a lesser extent,
perchloroethylene.

        ii.  TEAM Metala Monitoring

     EPA's ORD collected 26 24-hour samples of both total
suspended particulates (TSP) and small respirable particulates of
10 microns and smaller (PM10) over a six week period from March
26, 1987 to May 1, 1987 using high volume evacuated canisters.
The sampling occurred concurrently with the VOC monitoring at the
two fixed monitoring sites in Parkville and Dundalk.  The samples
were analyzed for only four metals: arsenic, cadmium, chromium,
and lead.

     Table IV-10 summarizes the monitoring results for the
metals.  The concentrations represent simple arithmetic averages.
The highest ambient air concentrations are reported for lead.
For all the metals, the maximum value for any sample was
generally twice the average measured value.  The averages were
roughly the same for each site.

     The limitations of the results are again emphasized.  At
both monitoring sites, cadmium and chromium were detected in only
a small number of the samples.  Furthermore, the analysis could
not discern between trivalent and hexavalent chromium.
Consequently, the values are for total chromium.  At Dundalk,
arsenic was not found in any of the samples.  The reported
concentration for arsenic at this site represents half of the
analytical detection limit.

     ORD's treatment of samples with no detectable levels of
metals differed from their treatment of samples with no
detectable levels of VOCs.  There is no standard method for
calculating average levels of pollutants where pollutants are
present at levels below the levels of detection.  EPA felt that
if a particular VOC was regularly not detected, there was a
reasonable likelihood that it was indeed not present in the
ambient air or was present in negligible quantities.  On the
other hand, it was known from data on sources and emissions that
all of the metals were likely to be present in the ambient air on
the sampling days, but the sensitivity of the analytical
procedures was simply not adequate to detect and quantify them.
ORO also collected samples for metals over a much smaller period
of time than for VOCs (roughly 40 days versus about 14 days for
metals) and hence the confidence in the estimated ambient levels
were likely to be lower.  Finally, the targeted metals are

                              IV-3 9

-------
                             Table  IV-9

        COMPARISON OF 1987 AND 1983/84 MONITORING RESULTS1

                              (ug/m3)
                                1983-1984        1987
     Compound                 Monitoring2    Monitoring

Benzene                         10.4              4.6
Carbon tetrachloride             1.0              0.7
Chloroform                       1.7              O.I3
Ethyl benzene                    6.2              3.7
Xylene                          17.8             10.54
Trichloroethylene                1.0              0.2
Perchloroethylene                4.6              1.4
  Values represent averages for values from all monitoring sites.
  Although more compounds were monitored, this table presents
  only those compounds that overlapped in the two monitoring .-.
  programs.

2 Source: Versar,  Inc.,  Draft Data Report for Monitoring and
  Analytical Activities to Determine Ambient Air Concentrations
  of Selected Toxic Pollutants in Baltimore. Maryland. April 18.
  1984.

3 Because chloroform was  detected at only one site,  only this
  average value for this one site is given.

* Value is for ra-xylene  only.
                              IV-40

-------
                           Table  IV-10

    AVERAGE MEASURED AMBIENT CONCENTRATIONS: METALS AT FIXED
                        MONITORING SITES
                                                   .3,1
            (nanograms per cubic meter of air—ng/m )
                                                    Avg.  Across
Compound1

Arsenic
maximum
Cadmium6
maximum
Chromium7
maximum
Lead
maximum
Dundalk
TSP
1.83
-
1.6
7.6
3.1
5.8
43.0
88.0
PM10
1.83
-
1.1
3.6
2.0
4.2
36.7
73.0
Parkville Sites
TSP
1.8*
4.1
0.9
3.2
1.7
3.6
37.0
87.0
PM10
1.93
3.6
0.9
3.0
2.0
—
33.3
57.0
TSP PM10
1.8 1.8

1.2 1.0

2.4 2.0

40.0 35.0

1 One  nanogram is  equivalent  to one one-thousandth  of  a micro-
  gram, which, in turn, is  equal to one one-millionth of a gram.

2 The  value for one-half the  detection limit was  given for those
  samples for which the target compound was not detected.  The
  figures in the table are simple arithmetic averages.   Eighteen
  samples were collected in Dundalk and seventeen in Parkville.

3 in the samples collected in Dundalk no arsenic  was detected
  within the laboratory's analytical detection limit.   Conse-
  quently, the value for one half the detection limit for the
  samples was used in estimating exposure levels.

* Arsenic was detected in only one TSP sample in  Parkville.

3 Arsenic was detected in only one PM10 sample in Parkville.

6 Cadmium was detected in eight TSP and five PM10 samples in
  Dundalk and two TSP and 2 PM10 samples in  Parkville.

7 Chromium was detected in eight TSP and two PM10 samples in
  Dundalk and in only two TSP  samples in Parkville.
                               IV-41

-------
 generally much more potent carcinogens than the VOCs such that
 even relatively low levels can pose significant risks.  Thus,
 assigning the value of half the detection limits to samples with
 no detected metal would likely provide more reasonable estimates
 of exposure.  Clearly, if actual ambient concentrations are even
 less than half the detection limit, risks may be overstated.

         iii.  Available Area-Wide Sampling Data

      To supplement the TEAM monitoring data,  we compiled readily
 available air toxics monitoring data for several sites throughout
 the Baltimore metropolitan area.   This summary consists of
 monitoring data generated by several sampling programs conducted
 by the AMA and the 1983/1984 IEMP VOC monitoring.   Table IV-11
 presents the lowest and highest measured pollutant value at a
 given monitoring site and the average measured pollutant
 concentrations across all monitoring locations (the supporting
 data can be found in Appendix B).   We emphasize that these data
 are highly uncertain; however,  we feel that it is  important to
 include them in this analysis to  demonstrate  the importance of
 evaluating area-wide exposures  for screening  purposes  using both
 modelled and measured data.

      c.   Exposed Population

      We subdivided the study area  into 5 km squares (grid cells)
 composing what we call the standard arid and  centered  a refined
 grid consisting of 2.5 km grid  cells over the most densely
 populated portion of the study  area.   Figures IV-3 and
 IV-4 show the siting of the  standard and refined grids,
 respectively.   The total area covered by the  standard  grid is
 1600 km (40  km by 40 km)  with 64 nodes in all and  a total area of
 400  km (20 km by 20 km)  covered by the refined grid, again with
 64 nodes.  The refined grid  is  offset from the standard grid by
 1.25 km because we placed the refined grid over the most densely
 populated  portion of the study  area.

      We  used the 1980 Census  Bureau Block Grid/Enumeration  .
 District data  to determine the  population within the squares of
 each grid.   The total population in the  study area  is  about  1.6
 million.   We used these  population  estimates  to estimate the
 annual excess  cancer incidence, above background,  that  would
 occur as a consequence of exposure  to the modelled  concentrations
 within each  square.   These population estimates were also  used to
 determine  the  number of  people  at risk of  noncancer effects  as a
 result of  average  pollutant levels  within the  squares.

      Each grid  cell  is identified by  the  Universal Transverse
Mercator  (UTM)  coordinates for  its  southwest node.  We henceforth
 refer to a specific  cell or location  within the study area by  its
 UTM coordinates.
                              IV-4 2

-------
                                            TABLE IV-11

                    AVAILABLE AREA-HIDE SUMMARY MONITORING DATA FOR BALITMORE  (1)
M
   POLLUTANT
LOWEST AND HIGHEST
 AVGERAGE MEASURED
   CONCENTRATION
      (ug/m3)
  AVERAGE MEASURED
CONCENTRATION ACROSS
ALL MONITORING SITES
       (ug/m3)
                                                                                   NUMBER OF
                                                                                MONITORING  SITES
Arsenic
Benzene
Benzo ( a ) pyrene
Cadmium
Carbon tetrachloride
Chloroform
Chromium
Ethylene dichloride
Propylene dichloride
Ethyl benzene
Lead
Methylene chloride
Methyl isobutyl ketone
Perchloroethylene
Toluene
Methyl chloroform
T r i ch 1 or oet hy 1 ene
VinyL chloride
Xylene
<0.01
3.90
<0.01
<0.01
0.20
0.10

-------
                FIGURE IV-3
          Baltimore Study Area
Map of Boundaries of the Modelling Domain
   for the 5 km (Standard) Grid System
                 IV-4 4

-------
               FIGURE IV-4
          Baltimore Study Area
Map of Boundaries of the Modelling Domain
   for the 2.5 km (Refined) Grid System
                  IV-4 5

-------
     We also identified 118 locations, called discrete receptors.
for use in estimating the highest increased lifetime individual
risks.  The 118 discrete receptor locations were chosen because
of their location relative to ten major emission sources.  They
represent discrete geographic locations or points.  The locations
for these discrete receptors are also identified by their UTM
coordinates.  We show the location of these discrete receptors in
Figure IV-5.
                               IV-4 6

-------
                  FIGURE IV-5
           Baltimore Study Area
Map of Boundaries of the Modeling Domain
                          Leaend:

                          + Location of discrete modelling receptor.
                     iv-4 7

-------
V. RISK ASSESSMENT SCREENING RESULTS

     This chapter presents the results of the risk assessment
screening analyses used to address the key questions raised in
Chapter II:

     •    Can we develop a screening methodology to identify and
          characterize health risks posed by selected air toxics
          from exposures occurring in the "urban soup"? l|2

     •    What is the relative importance of point sources versus
          area sources to urban air toxics in the Baltimore area?

     •    What is the relative contribution of specific point and
          area source categories to air toxics risk?

     In the Baltimore air toxics study, a relative ranking of
pollutants and sources was - developed using two approaches.
First, at the recommendation of the Johns Hopkins Risk Assessment
Review Panel (RARP), a ranking of pollutants and sources by
cancer potency-weighted ambient concentrations was developed.
This approach is advantageous because it (1) avoids relying on
highly uncertain exposure assumptions used by EPA in its risk
assessments (e.g./ continuous exposures for 70 years), and (2)
does not incorporate population weighing factors.  The second
approach consisted of conducting a quantitative risk assessment
for the pollutants and sources examined.  This latter analysis
considered both cancer and noncancer effects.

     We emphasize that the risk estimates presented in this
chapter are highly uncertain and were calculated only for
screening purposes.  For example, the preliminary emissions
estimates used in the modelling effort are based on limited
     *This study estimated the increase in cancer and noncancer
risks resulting from exposure to ambient (i.e., outdoor)
concentrations of air toxics.  It did not consider exposures
resulting from (1) indoor air, (2) the workplace, and (3) other
pathways  (e.g., ingestion).

     2As  discussed in Chapter II,  Summary of the Principles  of
Risk Assessment and Management, we define "urban soup" as the
combination of toxic emissions from both point and area sources
to form elevated concentrations of pollutants in urban areas.

                               v-l

-------
information that was available in 1985.  Since this time,
emission estimates at some sources have changed significantly
(e.g., Point Source A), while important new sources have been
introduced.  Because of these changes, the numbers and the
conclusions presented below may no longer be valid.  Also,
because of the generally conservative bias in the underlying data
used in the risk assessment, it is highly unlikely that the true
risks would be as high as the risk estimates, and they could be
considerably lower.

     This chapter is organized into four sections.  The first
describes the methodology and results for the RARP cancer
potency-weighted ranking scheme.  The second summarizes the
results of the quantitative risk assessment for cancer and
noncancer effects.  The third section provides perspective on the
risk assessment screening results by comparing (1) the Baltimore
risk screening estimates with the results of other IBMP and
urban-scale studies of air toxics, (2) the approach used in the
Baltimore study to estimate risks from the products of incomplete
combustion with other approaches used by EPA, and  (3) the
Baltimore risk screening results for drinking water from Phase I
with the results of this more recent study of air toxics.  The
final section of this chapter summarizes the major findings from
the risk assessment screening analyses.


1.   RANKING OF TARGET COMPOUNDS AND SOURCES BY CANCER POTENCY-
     WEIGHTED AMBIENT CONCENTRATIONS

     As stated in Chapter III, quantitative risk assessment
incorporates numerous assumptions that are not readily
verifiable.  Some of the more important ones are  (1) the
percentage of time that people spend at given locations in the
study area, (2) the amount of air inhaled, (3) body-weight, and
(4) lifetime.  Where these assumptions are inappropriate, the
cancer risk estimates and the conclusions drawn from these
estimates may be misleading.

     in this section, at the suggestion of the Johns Hopkins
RARP, we describe how we ranked pollutants and sources by a
method that is simpler than the IEMP risk assessment approach
described in Chapter III.  This alternative method avoids the
reliance on assumptions that relate to population exposure by
leaving out the population weighting factor from  the quantitative
risk equation.  We calculated the average area-wide concentration
of each target pollutant using the 5 km modelling grid system and
then multiplied this value by the pollutant's corresponding unit
cancer risk factor.  The product is a cancer potency-weighted
average ambient concentration for each target pollutant in the
study area.  The cancer potency-weighted average  concentration is
simply a measure of each pollutant's potential human health risk
in the Baltimore area.  Pollutants were compared  and ranked by
                               V-2

-------
normalizing against the cancer potency-weighted concentration for
one of the more common pollutants in the ambient air.

     We chose the potency-weighted average concentration of
chloroform as the base value and then divided the values for the
other pollutants by this number.  The result, as shown in
Table V-l, is a ranking of pollutants in descending order of
potency-weighted concentration ratios.  POM ranked highest with a
score approximately four times higher than the next highest
pollutant, chromium, and roughly twice as high as the summed
scores for all other pollutants.  Carbon tetrachloride produced
the lowest score.

     To rank sources, we first summed the potency-weighted
average concentrations resulting from each source or source
category and then normalized against the value for gasoline
marketing.  Again, the potency-weighted concentration is simply a
measure of the potential risk to human health posed by a source
or source category in the Baltimore area.  We included only those
sources that ranked high on the basis of quantitative risk
assessment (see Table v-8) to allow comparison of the two
approaches.  We present the results of this''-ranking in Table V-2.
Point Source A ranks highest with a score approximately seven
times higher than the next highest source, road vehicles, and
roughly twice as high as the summed scores for all other sources.
Gasoline marketing produced the lowest score.


2. QUANTITATIVE RISK ASSESSMENT SCREENING RESULTS

     In the Baltimore ambient air toxics study, risks were
quantified considering two health endpointst   cancer and
noncancer effects.  As discussed in Chapter IV, the calculations
for estimating the risks associated with ambient air exposures
used data on (1) ambient air concentrations based on either
dispersion modelling or monitoring, (2) population data (1980
U.S. Census), and (3) EPA unit cancer risk factors (see
Table IV-2) and threshold values for noncancer effects
(Table IV-3).3   We summarize the results  of the risk  screen for
cancer and noncancer effects below.  We also list the limitations
that the reader must consider to interpret and use the results
presented in this chapter.
     3The unit cancer risk factor combines  estimates  of  cancer
potency with EPA standard exposure assumptions.  The unit cancer
risk factor represents the upper-bound estimates of the increase
in cancer risk associated with continuous lifetime (i.e., 70
years) exposure to 1 ug/m3 of a specific compound.

                               v-3

-------
                                    TABLE  V-l
                        RELATIVE RANKING  OF TARGET COMPOUNDS
                             BY CANCER POTENCY-WEIGHTED
                               AMBIENT CONCENTRATIONS
POLLUTANT
POM
CHROMIUM- 6
BENZENE
FORMALDEHYDE
METHYLENE
CHLORIDE
ARSENIC
CADMIUM
PERCHLOROETHYLENE
TRICHLOROETHYLENE
CHLOROFORM
ETHYLENE DICHLORIDE
ETHYLENE DIBROMIDE
ETHYLENE OXIDE
CARBON
TETRACHLORIDE
UNIT
CANCER RISK
FACTOR-
6.50E-05
1.20E-02
8.00E-06
1.30E-05
4.10E-06

4.30E-03
1.80E-Q3
4.80E-07
1.30E-06
2.30E-05
2.60E-05
2.20E-04
l.OOE-04
3.70E-06

AVERAGE
CONCENTRATION

-------
                                  TABLE V-2
                  RELATIVE RANKING OF SOURCES  BY CONTRIBUTION
                          TO CANCER POTENCY-WEIGHTED
                            AMBIENT CONCENTRATIONS


SOURCE
UNIT
CANCER
RISK
POLLUTANT FACTOR
AVERAGE
CONCEN-
TRATION
(ug/m3) (1)

POTENCY-
WEIGHTED
CONC.
SUM OF RATIO OF
POTENCY- SOURCES
WEIGHTED (gas
CONC. mrktng-1)
   Point  sources:

 Point  Source A   POM       6.50E-05
                 Chrom.-6  1.20E-02
                 Benzene   8.00E-06

 Point  Source B   Chrom.-6  1.20E-02

 Point  Source C   Chrom.-6  1.20E-02

•Point  Source D   Chrom.-6  1.20E-02
                                          0.86406   0.0000562   0.0000905       489
                                          0.00245   0.0000294
                                          0.60625   0.0000049

                                          0.00027   0.0000033   0.0000033        18

                                          0.00026   0.0000031   0.0000031        17

                                          0.00011   0.0000013   0.0000013        7
                 POM        0.00005
                 Fnnldhyde 1.30E-05
                 Benzene   8.00E-06
   Area sources:

Road vehicles



Solvent usage



Heating



Gas marketing    Benzene   8.00E-06
Met. CL
PERC
TCE
POM
Arsenic
Cadmium
4.10E-06
4.80E-07
1.30E-06
0.00001
4.30E-03
1.80E-03
0.11494  0.0000057
0.27422  0.0000036
0.41250  0.0000033

1.34609  0.0000055
0.68156  0.0000003
0.68000  0.0000009

0.45625  0.0000046
0.00088  0.0000038
0.00100  0.0000018
0.0000126
                                                             0.0000067
                                                             0.0000101
68
                36
                55
                                         0.02313  0.0000002  0.0000002
(1)  Modelled concentration.
                                      V-5

-------
     a.  Cancer Risks

     We used different measures of cancer risk to characterize
urban soup in the Baltimore area.  For area-wide risks, the most
important risk measures were:  average increased lifetime
individual cancer risk and annual excess cancer incidence.  For
so-called "hotspot" areas, the two risk measures of interest
were:  the highest increased lifetime individual cancer risk at
three non-contiguous locations, and annual excess cancer
incidence in the grid cell of highest predicted cancer
incidence.*'5

          i. Area-Wide Risks

             (1)  Average Increased Lifetime Individual Cancer
                  Risk

     Results Based on Dispersion Modelling.  We estimated the
average increased lifetime individual cancer risk posed by the
pollutants examined in the study area by (1) multiplying the
estimated average ambient air concentration by pollutant for each
5 km grid cell by the appropriate unit cancer risk factor, (2)
calculating an arithmetic average across all grid cells
evaluated, and (3) summing across all pollutants examined.  The
resulting value indicated an average increased lifetime
individual cancer risk for the study area of approximately
1.5 x  10"*.  The study area is  shown with its UTM  (XY)
coordinates in Figure V-l.

     In general, we found that the average increased lifetime
individual cancer risk was fairly uniform across all grid cells.
The one exception was in the heavily industrialized southeast
section of the study area.  These findings are graphically
presented in Figure v-2.

     Figure V-2 also shows the relative contribution of point and
area sources to the estimated average increased lifetime
individual cancer risk by grid cell.  As shown, point  source
emissions are the major contributor to the average increased
lifetime individual cancer risk  in the study area.  Area source
emissions pose more or less uniform average increased  lifetime
individual cancer risks in the area of downtown Baltimore; lower
average increased lifetime individual cancer risks were estimated
at the fringes of the study area.  The higher average  increased
     *See Chapter II, Summary of the Principles of Risk
Assessment and Risk  Management,  for detail on  the different  risk
measures and  their estimation.

     5The term "hotspot" refers to specific areas in which people
are exposed to higher  than  average concentrations of the  target
Dollutants.
pollutants

                               v-6

-------
                       FIGURE V-1

                   MODELLING DOMAIN
                  BY UTM COORDINATES
U
T
M
        BALTIMORE COUNTY
   4356
   4346
                                PARKMUE
                                   OUMMLK
BALTIMORE
  CITY
                 ANNE ARUNDEL  > .
                   COUNTY

                  UTM X  	
                      SPARROWS PONT
                          V-7

-------
                                               FIGURE  V-2

                                         Baltimore  IEMP  Air Toxics

                                  Point &  Area Source Contribution to Average
                                   Lifetime Individual Cancer Risk by  Grid Cell
                                      Average Lifetime Individual  Cancer Risk
                                  Point Source Contribution to Average Lifetime

                                              Individual Cancer Risk
                                      Area Source Contribution to Increased

                                                Individual Cancer Risk
                         0.000
                                                                            UTM COORDINATES
.Vote: The nsk scale for the area source contribution
is significant? larger than that used for the average
lifetime individual cancer risk and point source contribution
graphs.
V-8

-------
lifetime individual cancer risks in the heavily industrialized
southeast section of the study area (see the right-hand corner of
the second graph in Figure V-2) are attributable primarily to
point source emissions.  Point sources also contribute more to
the lower average increased lifetime individual cancer risks
estimated in the northwest section of the study area, but to a
lesser extent that in the southeast section.

     Results Based on Monitoring Data.  Similar to the risk
analysis based on the modelled data, we used available monitoring
data from the past four years for the Baltimore area to
characterize area-wide risks.  These data came from sampling
programs conducted by the AMA in 1985 through 1987 and from the
IEMF air toxics monitoring performed in 1983/1984.  We also
estimate the average increased lifetime individual cancer risks
at two specific locations within the study area based on the
monitoring data available at two sites used in the TEAM study—
Dundalk and Parkvilie.

     We estimated the average increased lifetime individual
cancer risk for the Baltimore study area by (1) calculating the
average concentration by pollutant across all of the AHA and IEMP
monitoring sites (see Appendix B for more detail), (2)
multiplying by the appropriate unit cancer risk factor, and (3)
summing across pollutants.  The same steps were also taken to
estimate the average increased lifetime individual cancer risk at
the Dundalk .and Parkville sites.  In determining the average
increased lifetime individual cancer risk, we assumed that all of
the detected chromium was the hexavalent or highly carcinogenic
form.  In actuality, this is unlikely to be the case.  Thus, the
average increased lifetime individual cancer risk from chromium
exposure must be seen as worse-case.

     Table V-3 summarizes the total and pollutant-specific
average increased lifetime individual cancer risks using the
available monitoring data for the Baltimore area.  All pollutants
individually pose increased lifetime individual cancer risks
equal to or greater than 1 x 10"£ (i.e., greater than one chance
in one million).  Combined, the total average increased lifetime
individual cancer risk is approximately 5.0 x 10** area-wide.
Three pollutants account for most (approximately 87 percent) of
the estimated average increased lifetime individual cancer risk.
Perchloroethylene accounts for almost half (48 percent) of the
total, followed by hexavalent chromium (24 percent) and benzene
(15 percent).

     Table V-4 summarizes the total and pollutant-specific
average increased lifetime individual cancer risks using the TEAM
monitoring data gathered at the Dundalk and Parkville sites.
The reader should note that the average increased lifetime
individual cancer risk is reported individually for vinylidene
chloride, but is not included in the summed risks because of the
                               V-9

-------
                        TABLE V-3

      AVERAGE INCREASED LIFETIME INDIVIDUAL CANCER RISK
           USING AVAILABLE MONITORING DATA (1/2)

        RESULTS INTENDED FOR POLICY DEVELOPMENT ONLY
                                        AVERAGE INCREASED
    POLLUTANT                          LIFETIME INDIVIDUAL
[Weight of Evidence]                      CANCER RISK
Arsenic [Bl]
Benzene [A]
Benzo(a)pyrene [B2]
Cadmium [Bl]
Carbon tetrachloride [B2]
Chloroform [B2]
Chromium VI [A]
Ethylene dichloride [B2]
Propylene dichloride [C]
Ethyl benzene [N/A]
Lead [N/A]
Methylene chloride [B2]
Methyl isobutyl ketone [N/A]
Perchloroethylene [B2]
Toluene [N/A]
Methyl chloroform [N/A]
Trichloroethylene [B2]
Vinyl chloride [A]
Xylene [N/A]
7.0E-06
7.5E-05
2.5E-06
2.2E-06
3.4E-06
3.2E-05
1.2E-04
1.6E-05
8.6E-06
N/A
N/A
9.8E-06
N/A
2.4E-04
N/A
N/A
1.1E-06
N/A
N/A
     TOTAL                                   5.2E-04
N/A:  Not Applicable

(1) Source of monitoring data:  See Appendix B.

(2) This study uses conservative estimates of increased cancer
    risk from ambient (i.e., outdoor) exposure to establish
    priorities among pollutants and sources.  The risk estimates
    are calculated using modelled or monitored concentrations
    and EPA unit cancer risk factors.  There is considerable
    uncertainty in the estimated concentrations, which could
    either overstate or understate the true concentrations (see
    Chapter IV).  Unit cancer risk factors combine CAG potency
    estimates with EPA exposure assumptions.  The CAG potency
    estimates provide a plausible upper limit to the cancer risk
    of a compound (see Appendix A); however, the true value of
    the risk is unknown and may be as low as zero.  The exposure
    assumptions are extremely conservative in that they assume
    continuous exposure to outdoor air for 70 years.  Because of
    the generally conservative bias in the information, it is
    highly unlikely that the true risks would be as high as the
    estimates and they could be considerably lower.
                              V-10

-------
                               TABLE V-4

                    BALTIMORE IEMP AIR TOXICS STUDY         .
            PHASE  II  RESULTS  INTENDED  FOR  POLICY DEVELOPMENT1


          AVERAGE INCREASED LIFETIME INDIVIDUAL CANCER RISKS
                   FROM EXPOSURE TO TARGET COMPOUNDS,
                   AT THE TWO FIXED MONITORING  SITES2
                            Average"
   Target Compound
                        Increased Lifetime      CAG
                           Individual.      Classificat
   Cancer Risk
                   Level'
I           Confidence
      ion  in Exposure
            Estimate
     Volatile OrganicBi

   Benzene       ,   .,
   Carbon tetrachloride
   Chlorobenzene
   Chloroform      6
   mDichlorobenzene
   oDichlorobenzene
   pDichlorobenzene
   Ethylene dibromide
   Formaldehyde
   Methyl chloroform
   Methylene Chloride
   Percnloroethylene
   Styrene          ,
   Tetrachloroethane
   Trichloroethylene  „
   Vinylidene chloride

        Metalst
3.1 x IQ'l
7.4 x 10'7

2.3~x 10'*
   N.Q.


4.4~x 10^
2.2 x 10'3
     1.7 x 10
     2.9 x 10
        -5
        -1
             -7
   N.D.
5.2 x 10 ,
1.7 x 10°
                         A
                         B2

                         B2
IZ/2

B2
B2
C
B2
B2
C
Cumulative Lifetime Individual Cancer Risk*10
                             1.5 x 10'
                               Par]
                            Average
           •kville
                                   1
                           1


                           2
                           1
                                        1
                                        3
Arsenic!
Cadmium ,
Chromium- VI
7.7 x 10'*
2.0 x 10'*
2.4 x ID'5
A
Bl
A
2
   Target Compound
Increased Lifetime       CAG
   Individual      classification
   Cancer Risk          Level
                                Confidence
                               in Exposure
                                 Estimate
        Volatile organicai

   Benzene
   Carbon tetrachloride
   Chlorobenzene
   Chloroform
   mDichlorobenzene
   oDichlorobenzene
    §Dichlorobenzene
    thylene dibromide
   Formaldehyde
   Methyl chloroform ,,
   Methylene chloride"
   Percnloroethylene
   Styrene          .,
   Tetrachloroethane
   Trichloroethylene  rt
   Vinylidene chloride

        Metalsi
      .2 x
      .1 x
        N.D.
     2.2 x
     2.1 x
     1.0
     1.0
   N.D.
1.3 x 1
3.4 x I
-5

-6
-S

-5


-3

-6
             .
             '!
             '3
                    t;

                    B2
B2
Bl/2

B2
B2
C
B2
B2
C
               2
               1
                                   1
                                   1

                                   3
                           1
Arsenic!
Cadmium ,
Chromium-vi
8.2
1.6
2.4
: x 10'*
x 10't
x 10
A
Bl
A
3
2
2
Cumulative Lifetime Individual Cancer Risk
                                          114
                             1.2 x 10'
                                  V-ll

-------
1
                         TABLE V-4 (CONTINUED)



                                 NOTES
N.Q. = Not quantified
N.D. = Not detected
  This study uses conservative estimates of increased cancer risk from
  ambient  (i.e., outdoor) exposure to establish priorities among
  pollutants and sources.  The risk estimates are calculated using
  modelled or monitored concentrations and EPA unit cancer risk
  factors.  There is considerable uncertainty in the estimated
  concentrations, which could either overstate or understate the true
  concentrations (see Chapter IV).  Unit cancer risk factors combine
  CAG potency estimates with EPA exposure assumptions.  The CAG
  potency estimates provide a plausible upper limit to the cancer risl
  of a compound  (see Appendix A); however, the true value of the risk
  is unknown and may be as low as zero.  The exposure assumptions are
  extremely conservative  in that they assume continuous exposure to
  outdoor air for 70 years.  Because of the generally conservative
  bias in the information, it is highly unlikely that the true risks
  would be as high as the estimates, and they could be considerably
  lower.

2 The two  fixed ambient monitoring stations were located at the
  Parkville Middle School in the Parkville section of Baltimore County
  and the Merritt Point Mental Health Facility in the Dundalk section
  of Baltimore County respectively.

3 Average monitored values are from the sampling period March 26 to
  June 1,  1987.

* Unit risk  (potency) estimates are current as of May 1987.

5 There are three levels  of confidence in exposure estimates.  Level 1
  means that the exposure estimate is probably good.  Level 2 means
  that the exposure estimate may be good.  Finally, level 3 means that
  the exposure estimate  is probably no good.  The evaluations are
  based on quality control audits.

6 m-dichlorobenzene levels could not be quantified at Dundalk.

7 NO tetrachloroethane was detected at a detection limit of 0.37
  micrograms per cubic meter.

8 Average  concentrations for the PM10 samples  (i.e., particle sizes
  equal to or  smaller  than  10 microns) were used in estimating risks.

9 Average  chromium concentrations  for the PM10  samples were used in
  estimating exposure  levels.  We  assumed that  all of the chromium
  that was collected was chromium-VI.  The risk from exposure to
  chromium is  therefore  overestimated for the  purpose of this
  screening  exercise  since  the actual percentage of chromium-VI  in  the
  sample  is  likely to  be considerably lower.
                               (continued)

                                 V-12

-------
                         TABLE V-4  (CONTINUED)

                                 NOTES

10 The cumulative  lifetime individual  cancer risk estimate does not
  include vinylidene chloride because of low confidence in the
  exposure  estimate.

11 For the monitoring samples  in  Parkville, no chloroform was detected
  at the analytical level of  detection of the laboratory/ which was
  0.3 micrograms  per cubic meter.

12 The monitoring  results for  methylene chloride in the ambient air of
  Parkville are highly questionable due to analytical problems in
  quantifying the amounts in  the monitoring samples.  Consequently,
  for the purposes of  comparison,  the risk result is given, but it is
  not included in the  cumulative risk estimate.

13 No tetrachloroethane was detected at a detection limit of 0.37
  micrograms per  cubic meter.

  The cumulative  lifetime individual  cancer risk estimate does not
  include the estimate for methylene  chloride and vinyl chloride
  because of low  confidence  in the exposure estimates.

  Vinylidene chloride  was detected in the sampling, but there were
  analytical problems  in quantifying  ambient concentrations.  Because
  vinylidene chloride  could  be an  important compound, the  individual
  risk  estimate for vinylidene chloride  is reported in the  risk
  screening results; however, it is not  included in the hazard  index.
  Additional analysis  is needed to confirm the  presence of  vinylidene
  chloride  in the ambient air.
14
                                  V-13

-------
extreme uncertainty surrounding its actual measured value.6   The
average increased lifetime individual cancer risk summed across
all pollutants at the Dundalk site is 1.5 x 10~* or two chances
in ten thousand and 1.2 x 10"* or roughly one chance in ten
thousand at Parkville.  At the Dundalk site, ethylene dibromide
and benzene exposures account for approximately half of the
predicted average increased lifetime individual cancer risk.  An
additional 42 percent of the risk is accounted for by hexavalent
chromium (16 percent), formaldehyde (15 percent), and methylene
chloride (11 percent).  At Parkville, roughly 55 percent of the
estimated average increased lifetime individual cancer risk is
attributable to benzene (35 percent) and hexavalent chromium  (20
percent); ethylene dibromide and formaldehyde each contribute an
additional 18 percent.

     For the pollutants common to the TEAM sampling and the
available area-wide monitoring, there is close agreement in the
estimated risks for most compounds.  With the exception of
perchloroethylene, which was found at a lower ambient level in
the TEAK study, the average increased lifetime individual cancer
risks by pollutant at the Dundalk and Parkville sites fall within
the range of the estimated average increased lifetime individual
cancer risks using the lowest and highest average measured values
from the available Baltimore area monitoring data (see
Table V-5).

     We cannot easily compare the estimated ambient levels and,
hence, the average increased lifetime individual cancer risks for
our target pollutants from the monitoring and modelling studies.
The monitoring efforts were short-term (for example, the TEAM
sampling covered roughly six weeks during the transitional period
when winter ends and spring begins), whereas the modelling
estimates are annual averages that take into account the
different meteorological and industry operating conditions that
occur throughout the year.  Nevertheless, the average increased
lifetime individual cancer risks from just the compounds that are
common to both the modelling and the monitoring studies are
close.  From modelling, we would expect the average increased
lifetime individual cancer risks to be upwards of 1 x 10" .  The
average increased lifetime individual cancer risks based on
monitoring data for this joint set of pollutants within the study
area are approximately 5 x 10~*.  The average increased lifetime
     6As discussed in Chapter IV,  there were analytical problems
in quantifying the concentrations of vinylidene chloride.  Based
on the advice of EPA's Office of Research and Development, we
show the estimated risk for vinylidene chloride in Table V-4 to
indicate its potential hazard as part of the risk screen;
however, we have not included this in the summed risks.  ORD has
recently completed additional monitoring in Baltimore that may
help to confirm the ambient air concentrations of vinylidene
chloride.  The results will be available in early 1989.

                               V-14

-------
                               TABLE  V-5
        AVERAGE  INCREASED LIFETIME  INDIVIDUAL  CANCER  RISKS
              BASED ON AVAILABLE MONITORING DATA:
     SITES  WITH  THE HIGHEST AND LOWEST MEASURED VALUES  (1),  (2)

                   RESULTS INTENDED FOR POLICY DEVELOPMENT ONLY
    POLLUTANT
 [Weight of Evidence]
Arsenic  [Bl]
Benzene  [A]
Benzo(a)pyrene  [B2]
Cadmium  [Bl]
Carbon tetrachloride  [B2]
Chloroform [B2]
Chromium VI [A]
Ethylene dichloride [B2]
Propylene dichloride  [CJ
Ethyl benzene [N/A]
Lead [N/A]
Methylene chloride [B2]
Methyl isobutyl. ketone [N
Perchloroethylene [B2]
Toluene  [N/A]
Methyl chloroform [N/A]
Trichloroethylene [B2]
Vinyl chloride  [A]
Xylene [N/A]
	AVERAGE LIFETIME INDIVIDUAL	
               CANCER RISK
LOWEST MEASURED      HIGHEST MEASURED
     VALUE                VALUE
   1,
   7,
6.9E-06
3.1E-05
6.6E-09
  .8E-06
  .4E-07
2.3E-06
2.4E-05
5.2E-06
1.8E-06
    N/A
    N/A
2.9E-06
    N/A
3.5E-05
    N/A
    N/A
1.3E-08
    N/A
    N/A
                      7.3E-06
1,
3
1,
5,
1,
  •OE-04
  2E-06
  8E-06
  2E-06
  1E-04
2.2E-04
6.8E-05
3.6E-05
    N/A
    N/A
1.7E-05
    N/A
5.4E-04
    N/A
    N/A
5.1E-06
    N/A
    N/A
N/A:  Not Applicable

(1) Source of monitoring data:  See Appendix B.

(2) This study uses conservative estimates of increased cancer
    risk from ambient (i.e., outdoor) exposure to establish
    priorities among pollutants and sources.  The risk estimates
    are calculated using modelled or monitored concentrations
    and EPA unit cancer risk factors.  There is considerable
    uncertainty in the estimated concentrations, which could
    either overstate or understate the true concentrations (see
    Chapter IV).  Unit cancer risk factors combine CAG potency
    estimates with EPA exposure assumptions.  The CAG potency
    estimates provide a plausible upper limit to the cancer risk
    of a compound (see Appendix A);  however, the true value of the
    risk is unknown and may be as low as zero.   The exposure
    assumptions are extremely conservative in that they assume
    continuous exposure to outdoor air for 70 years.  Because of
    the generally conservative bias  in the information, it is
    highly unlikely that the true risks would be as high as the
    estimates, and they could be considerably lower.
                                v-15

-------
individual cancer risks for Dundalk and Parkville are 1.3 x 10"*
and 1.0 x 10"*, respectively.

             (2)  Annual Excess Cancer Incidence

     Results Based on Dispersion Modelling.  We determined the
estimated annual excess cancer incidence by pollutant for the
study area using the average increased lifetime individual cancer
risk estimates for each 5 km grid cell/ multiplying this value by
the number of people estimated to live in the grid cell, summing
across all cells, and then dividing this number by 70 years (an
assumed lifetime).  We then summed across pollutants to estimate
area-wide annual excess cancer incidence.

     The results of each step in the calculation of annual excess
cancer incidence are illustrated in Figure V-3.  Each block in
the graph at the bottom corresponds to the number of people who
live in the area represented by the grid cell.  As shown, the
greatest number of people live in downtown Baltimore.  The blocks
in the middle graph show the average increased lifetime
individual cancer risk for each grid cell.  Finally, in the top
graph, we show the estimated annual excess cancer incidence due
to the combination of the average increased lifetime individual
cancer risk and population living in the area represented by the
grid cell.  The highest blocks correspond to the downtown area of
Baltimore.

       Although the average increased lifetime individual cancer
risk was highest in the southeast section of the study area, the
highest predicted annual excess cancer incidence was for the
downtown area because of the high concentration of people living
in this area.  Clearly, the differences in population among cells
have the greatest impact on our estimated annual excess cancer
incidence for a given pollutant.  Blocks representing the
relative importance of the estimated annual excess cancer
incidence are superimposed upon a map of the study area in
Figure V-4.  The weighing of our average increased lifetime
individual cancer risk estimates by population allows us to
identify the particular sources and pollutants that contribute
more to exposures where most individuals currently reside.

     Table V-6 details the estimated annual excess cancer
incidence associated with ambient air exposures to each of the
pollutants  (from all sources)  included in the study.  As shown,
the modelling analysis indicates that exposure to the selected
sources and pollutants analyzed in Baltimore will result in a
conservative estimate of approximately four excess cancer cases
each year.  This estimate is highly uncertain because of the
limitations in  the emissions calculations and the dispersion
modelling, as discussed in Chapter IV.
                               V-16

-------
               FIGURE  V-3
           Baltimore IEMP Air Toxics
Annual Excess Cancer Incidence, Average Lifetime
 Cancer Risk and Exposed Population by Grid Cell
        Annual  Excess Cancer Incidence
  Average Lifetime  Individual Cancer Risk
          Exposed Population
                      V-17

-------
 Annual Excess
Cancer Incidence
          0.44
                                              FIGURE V-4



                                    Annual Excess Cancer Incidence  by

                                         Grid Cell for the Study Area

-------
                             TABLE V-6

      ESTIMATED ANNUAL EXCESS CANCER INCIDENCE FOR SELECTED
    POLLUTANTS  MODELLED  IN  THE BALTIMORE IEMP AIR  TOXICS  STUDY
               (TOTAL  STUDY AREA, 5 KM GRID SYSTEM)1

           RESULTS  INTENDED FOR POLICY DEVELOPMENT ONLY
                         ANNUAL EXCESS
                        CANCER INCIDENCE     PERCENTAGE
POLLUTANT             (Weight of Evidence)    OP TOTAL
Polycyclic organic
 matter (POM)r               2.05 (N.A.)        57.4
Chromium (VI)                0.53 (A)           14.8
Methylene chloride           0.20 (62)           5.6
Benzene                      0.20 (A)            5.6
Arsenic                      0.20 (A)            5.6
Formaldehyde                 0.19 (Bl)           5.3
Cadmium                      0.09 (Bl)           2.5
Perchloroethylene (PCE)      0.04 (B2)           1.1
Trichloroethylene (TCE)      0.03 (B2)           0.8
Chloroform                   0.02 (B2)           0.6
Ethylene dichloride (EDC)    0.01 (B2)           0.3
Ethylene dibromide (EDB)    <0.01 (B2)           0.1
Benzo(a)pyrene (B(a)P)      <0.01 (B2)           <0.1
Carbon tetrachloride        <0.01 (B2)          <0.1
           TOTAL1            3.56              100.0

N.A. = Not Available~~~

1   This study uses conservative estimates of increased cancer
   risk from ambient (i.e., outdoor) exposure to establish
   priorities among pollutants and sources.   The risk estimates
   are calculated using modelled or monitored concentrations and
   EPA unit cancer risk factors.  There is considerable
   uncertainty in the estimated concentrations, which could
   either overstate or understate the true concentrations (see
   Chapter IV).  Unit cancer risk factors combine CAG potency
   estimates with EPA exposure assumptions.   The CAG potency
   estimates provide a plausible upper limit to the cancer risk
   of a compound (see Appendix A); however,  the true value of the
   risk is unknown and may be as low as zero.  The exposure
   assumptions are extremely conservative in that they assume
   continuous exposure to outdoor air for 70 years.  Because of
   the generally conservative bias in the information, it is
   highly unlikely that the true risks would be as high as the
   estimates, and they could be considerably lower.

2   The POM risk estimates are, in part, based on source-specific
   unit cancer potency factors that have not undergone extensive
   peer review.  Thus, these numbers are subject to change.

3   Totals may not sum because of rounding.

                               V-19

-------
     Almost all of the estimated annual excess cancer incidence
 (approximately 94 percent) can be attributed to six pollutants:
 POMs, chromium (VI), methylene chloride, benzene, arsenic, and
 formaldehyde.  Most of the estimated annual excess cancer
 incidence  (approximately 57 percent) results solely from POM
 exposures.  However, the POM estimates are based on unit cancer
 risk factors that have not been extensively reviewed and are thus
 more uncertain.  An additional 15 percent of the estimated annual
 excess cancer incidence can be attributed to hexavalent chromium.
 Because there was no practical way for us to measure the actual
 levels of hexavalent chromium in the ambient air, we could not
 ascertain the accuracy of our exposure estimates (which assumed
 that all chromium concentrations were hexavalent, its most potent
 form) and hence the estimates of risk from this chemical.
 Methylene chloride, benzene, arsenic, and formaldehyde each
 account for between 5 and 6 percent of the total.

     Table V-7 expands on Table V-6 by showing the portions of
 the estimated annual excess cancer incidence attributable to
 point and area sources.  In the Baltimore study area, point and
 area sources account for fairly similar percentages of the
 estimated annual excess cancer incidence (45 percent and 55
 percent, respectively).  Figure V-5 shows this graphically.

     The roughly even contribution of point and area sources to
 the estimated annual excess cancer incidence does not hold for
 individual pollutants.  This is easily seen when the two key
 pollutants in the analysis of cancer incidence, POM and
 hexavalent chromium, are examined more closely.  Whereas both
 point and area sources account for the estimated annual excess
 cancer incidence associated with POM (approximately 44 percent
 versus 56 percent, respectively), point source emissions account
 for almost all (99.6 percent) of the estimated annual excess
 cancer incidence associated with hexavalent chromium.

     Table V-8 adds an important dimension to the analysis by
 identifying the specific source categories that account for the
estimated annual excess cancer incidence in the Baltimore area.
As shown, approximately 81 percent of the estimated annual excess
cancer incidence can be attributed to three sources:  road
vehicles, Point Source A, and heating.   These three sources also
account for all of the estimated annual excess cancer incidence
 associated with POM, benzene, formaldehyde, ethylene dibromide,
and ethylene dichloride.  Approximately 89 percent of the
estimated animal excess cancer incidence resulting from exposure
to POM ambient air concentrations comes from road vehicles and
Point Source A.  The majority of the estimated annual excess
cancer incidence associated with hexavalent chromium
 (approximately 87 percent) can be attributed to Point Sources A,
B, and C.
                               V-20

-------
                                                         TABLE V-7
NI
                                     ESTIMATED ANNUAL EXCESS CANCER INCIDENCE FOR SELECTED
                            POLLUTANTS MODELLED IN THE BALTJMORE, IEMP AIR TOXICS  STUDY BY SOURCE (1)

                                           (TOTAL STUDY AREA. 5KM GRID SYSTEM)

                                     RESULTS INTENDED FOR POLICY DEVELOPMEHT ONLY
                POLLUTANT
           [Weight of Evidence]
                                  A.J
POLYC ORG MATTER (POM)(2) [N
CHROMIUM HEXAVALENT [A]
METHYLENE CHLORIDE [B2]
BENZENE [A)
ARSENIC [A]
FORMALDEHYDE [Bl]
CADMIUM [Bl)
PERCHLOROETHYLENE (PCE)
TRICHLOROETHYLENE (TCE)
CHLOROFORM  [B2]
ETHYLENE DICHLORIDE   (EDC) [B2]
ETHYLENE BROMIDE  (EDB)  [B2]
CARBON TETRACHLORIDE  [B2]
ETHYLENE OXIDE [Bl]
     TOTAL   (3)
                              1C]
                              [B2]
                                     ANNUAL
                                     EXCESS
                                   INCIDENCE:
                                     POINT
                                    SOURCES
 0.90
 0.53
 0.01
 0.09
 O.OS
<0.01
 0.02
<0.01
<0.01
<0.01
<0.01
<0.01
          PERCENTAGE
           OF TOTAL:
         POINT SOURCES
25.2
14.8
 0.2
 2.5
 1.5
 0.1
 0.6
<0.1
 0.1
                                                              0.1
        ANNUAL
        EXCESS
        INCIDENCE:
        AREA
        SOURCES
 1.15
<0.01
 0.20
 0.11
 0.15
 0.18
 0.07
 0.04
 0.03
 0.02
<0.01
<0.01
          PERCENTAGE
           OF TOTAL:
         AREA SOURCES
32.4
 0.1
 5.5
 3.2
 4.1
 5.2
 1.9
 1.0
 0.9
 0.6
 0.1
 0.1
                                             1.60
                                                       44.99
                                                                          1.95
                                                                                    54.93
        ANNUAL
        EXCESS
        CANCER
        INCIDENCE
 2.05
 0.53
 0.20
 0.20
 0.20
 0.19
 0.09
 0.04
 0.03
 0.02
<0.01
<0.01
<0.01
<0.01
                                                                                                      3.56
      N.A.  =  Not  Available

      (1)   This study  uses  conservative  estimates of  increased cancer  risk  from ambient  (i.e., outdoor)
           exposure  to establish priorities among pollutants and sources.   The risk estimates are calculated
           using  modelled or  monitored concentrations and  EPA unit cancer risk factors.  There is
           considerable uncertainty in the estimated  concentrations, which  could either  overstate or
           understate  the true concentrations  (see  Chapter IV).  Unit  cancer  risk  factors  combine CAG
           potency estimates  with EPA exposure assumptions.  The CAG potency  estimates provide a plausible
           upper  limit to the cancer risk of a compound  (see Appendix  A); however, the true  value of the
           risk  is unknown  and may be as low as zero.  The exposure assumptions are extremely conservative
           in that they assume continuous exposure  to outdoor air for  70 years.  Because of  the generally
           conservative bias  in the information, it is highly unlikely that the true  risks would be as  high
           as the estimates,  and they could be considerably lower.

      (2)  The POM risk estimates are based on source specific unit  risk  factors that have not undergone
           extensive peer  review.  Thus, these numbers are subject  to  change.

      (3)  The numbers shown  may not sum to  the totals because of  rounding.

-------
              FIGURE V-5

         Baltimore IEMP Air Toxics
    Point & Area Source Contribution to Annual
      Excess  Cancer Incidence by Grid Call
          Annual  Excess  Cancer Incidence
         Point Source Contribution to Annual

              Excess Cancer Incidence
          Area Source Contribution to Annual

               Excess Cancer Incidence
0.26
0.23
0.20
0.17
0.14
0.11
0.09
0.06
0.03
                                             UTM COORDINATES
                         V-22

-------
f
N)
OJ
               POLLUTANT
         (Weight of Evidence]
                                                              TABLE V-B

                                                   BALTIMORE 1EMP AIR TOXICS  STUDY

                                              ESTIMATED ANNUAL EXCESS CANCER  INCIDENCE
                                            BASED ON MODELLING BY SOURCE AND  POLLUTANT (1)
                                               (TOTAL STUDY AREA. 5 KM GRID SYSTEM)

                                             RESULTS INTENDED FOR POLICY DEVELOPMENT ONLY
                                 ROAD     POINT            SOLVENT    POINT    POINT     POINT     ALL
                               VEHICLES SOURCE A  HEATING    USAGE  SOURCE B SOURCE C   SOURCE D  OTHER
                                                          TOTAL ANNUAL
                                                          EXCESS CANCER
                                                            INCIDENCE
ARSENIC  [A]
BENZENE  [A]
CADMIUM  (Bl]
CARBON TETRACHLORIDE  [B2]
CHLOROFORM  [B2]
CHROMIUM HEXAVALENT  [A]
ETHYLENE OXIDE  [Bl]
FORMALDEHYDE  [Bl]
METHYLENE CHLORIDE  [B2]
POLYC ORG MATTER (POM)[N.A.] (2)
ETHYLENE DIBROMIDE 
-------
      In  reviewing  these  results,  it  should be noted  that  there  is
considerable uncertainty underlying  the modelled estimates of
ambient  concentration.   The model performance evaluation  (see
Chapter  IV) suggests  that we have underpredicted actual ambient
levels for some pollutants.  There is also information, according
to the Maryland AHA,  showing that modelled ambient air
concentrations at  Point  Source A  are generally much  greater than
those that have been  found at the facility using a detailed
monitoring network.   Because Point Source A contributes so
heavily  to the estimated human health risks, any revision to its
estimated emissions will have a significant impact on our risk
results.  Finally, our emissions  estimates were based on the
rough information  available at the start of this study.
Completion of a more  detailed analysis considering the changes
over  recent years  could  lead to different findings.

      Results Based on Monitoring.  Using available monitoring
data  for the Baltimore area, we estimated the annual excess
cancer incidence by simply multiplying the area-wide average
increased lifetime individual cancer risks by the total exposed
population used in the modelling  analysis, approximately 1.6
million.  Table V-9 presents the  results of these calculations.
As shown/ we predict  a larger annual excess cancer incidence—
roughly  12 annual  cases—using monitoring versus modelled data.
For comparison, Table V-9 shows the modelled annual excess cancer
incidence for those pollutants common to the monitoring and
modelling studies.  As shown, the difference in the predicted
annual excess cancer  incidence is explained by significantly
higher incidence estimates based on  the monitoring data for
perchloroethylene, chromium VI, and  benzene.

      There is considerable uncertainty surrounding the incidence
calculations based on monitoring data.  Although these monitoring
data  represent the best  readily available information for the
Baltimore area (other than the TEAM  results), the results used in
this  analysis are drawn,  by pollutant, from different studies,
and not  all results have  undergone extensive peer review.  Also,
exposures were estimated by assuming that all individuals are
exposed  to the same ambient pollutant levels represented by the
average  of the measured  concentrations across the monitoring
sites.   Despite these limitations, the annual excess cancer
incidence estimates based on the available monitoring data
suggest  that some exposures could be dramatically higher than
those characterized by dispersion modelling.  Accordingly, the
actual ambient air concentrations and sources of
perchloroethylene, chromium VI, and benzene deserve further
attention.

          ii.  "Hotspot" Risks

     A hotspot is defined as an area in which people are exposed
to higher than average concentrations.  Given available data and
resources, dispersion modelling is the only practical way to


                              V-24

-------
                           TABLE V-9
              AREA-WIDE ANNUAL EXCESS CANCER INCIDENCE
               USING AVAILABLE MONITORING DATA (1),(2)

          RESULTS INTENDED FOR POLICY DEVELOPMENT ONLY
    POLLUTANT
[Weight of Evidence]
  ANNUAL EXCESS
CANCER INCIDENCE
  ANNUAL EXCESS
CANCER INCIDENCE
 (MODELLED FROM
   TABLE V-6)
Arsenic [Bl]
Benzene [A]
Benzo(a)pyrene [B2]
Cadmium [Bl]
Carbon tetrachloride [B2]
Chloroform [B2]
Chromium VI [A]
Ethylene dichloride [B2]
Propylene dichloride [C]
Ethyl benzene [N/A]
Lead [N/A]
Methylene chloride [B2]
Methyl isobutyl ketone [N/A]
Perchloroethylene [ B2 ]
Toluene [N/A]
Methyl chloroform [N/A]
Trichloroethylene [B2]
Vinyl chloride [A]
Xylene [N/A]
0.16
1.72
0.06
0.05
0.08
0.73
2.74
0.37
0.20
N/A
N/A
0.22
N/A
5.44
N/A
N/A
0.03
N/A
N/A
0.20
0.20
<0.01
0.09
<0.01
0.02
0.53
0.01
N.A.
N/A
N/A
0.20
N/A
0.04
N/A
N/A
0.03
N.A.
N/A
     TOTAL
        11.79
                                                            1.32
N/A:   Not applicable; not a proven human carcinogen
N.A.:  Not available

(1) Source of monitoring data:  See Appendix B.

(2) This study uses conservative estimates of increased cancer risk
    from ambient (i.e., outdoor) exposure to establish priorities among
    pollutants and sources.  The risk estimates are calculated using
    modelled or monitored concentrations and EPA unit cancer risk
    factors.  There is considerable uncertainty in the estimated
    concentrations, which could either overstate or understate the true
    concentrations (see Chapter IV).  Unit cancer risk factors combine
    CAG potency estimates with EPA exposure assumptions.  The CAG
    estimates provide a plausible upper limit to the cancer risk of a
    compound (see Appendix A); however, the true value of the risk is
    (see Appendix A); however, the true value of the risk is unknown
    and may be as low as zero.  The exposure assumption are extremely
    conservative in that they assume continuous exposure to outdoor
    air for 70 years.  Because of the generally conservative bias
    in the information, it is highly unlikely that the true risks
    would be as high as the estimates, and they could be
    considerably lower.
                                  V-25

-------
pinpoint those areas with above-average exposure levels.  We
relied on the results of our dispersion modelling to locate (1)
discrete receptor locations with the highest estimated increased
lifetime individual cancer risk, and (2) the grid cell (using the
2.5 km refined grid system) of highest predicted annual excess
cancer incidence.

               (1)  The Highest Estimated Increased Lifetime
                    Individual Cancer Risks at Three Non-
                    Contiguous Sites

     As Figure V-2 suggests, emissions from point sources can
lead to relatively high localized average increased lifetime
individual cancer risks.  We examined these average increased
lifetime individual cancer risks in greater detail by focusing
our attention on the three non-contiguous hotspot locations
(separated by more than 2 kilometers) with the highest increased
lifetime individual cancer risks.  Figure V-6 shows the location
of the three hotspots.  The purpose of choosing non-contiguous
sites was to analyze hotspots resulting from the emissions from
different combinations of sources.  These hotspot locations were
modelled as discrete receptor locations rather than as a grid
cell.

     Table V-10 shows the increased lifetime individual cancer
risks by pollutant at the three hotspots.  Their exact location
is identified by UTM coordinates.  There is approximately a
factor of two difference between the highest and the lowest of
the three individual cancer risks (1.3 x 10° versus 5.4 x  10' ).
The hotspot with the highest increased lifetime individual cancer
risk is approximately a factor of nine higher than the estimated
average area-wide increased lifetime individual cancer risk
(1.5 x 10"*);  the  hotspot with the lowest increased lifetime
individual cancer risk is approximately four times higher than
the area-wide average.

     At the hotspot exhibiting the highest increased lifetime
individual cancer risk/ POM and hexavalent chromium account for
roughly 96 percent of the increased lifetime individual cancer
risk.  In addition, exposure to benzene in the ambient air
results in an increased lifetime individual cancer risk of
2.2 x 10'3.   These pollutants  are also  the  key compounds of
concern at the hotspot with the second highest increased lifetime
individual cancer risk.  At the third hotspot site, POM and
hexavalent chromium continue to account for most  (approximately
79 percent)  of the estimated increased  lifetime individual cancer
risks.  However,  arsenic  (6.1 x  10'5) and cadmium  (2.6  x 10' )
also contribute  significantly to these  increased  lifetime
individual cancer risks.

     Table V-ll  summarizes  the  increased lifetime individual
cancer risks  by  pollutant and individual source for each of the
three hotspot locations in  order of highest to lowest  risk.   As


                               V-26

-------
              FIGURE V-6

        LOCATION OF HOTSPOTS
  EVALUATED IN THE  AIR TOXICS STUDY
BALTIMORE COUNTY
                      \
     BALTIMORE
     \ CITY
       /ANNE ARUNDEL
         COUNTY/^
                                HOTSPOT 3
                              UTM: 4350.90 364.55
  HOTSPOT 1
UTM: 4343.00 374.12
                                 HOTSPOT 2
                               UTM: 4339.85 374.88
                  V-27

-------
                    TABLE V-10

   POLLUTANT CONTRIBUTIONS BASED ON MODELLING TO
 MAXIMUM INCREASED LIFETIME INDIVIDUAL CANCER RISK
           AT THREE HOTSPOT LOCATIONS (1)

RISK RESULTS INTENDED FOR POLICY DEVELOPMENT ONLY
MAXIMUM LIFETIME
POLLUTANT INDIVIDUAL CANCER
[Weight of Evidence]
Hotspot 1: 4343.00 374.12
CHROMIUM [A]
POLYCLYCLIC ORGANIC MATTER (POM) [N.A.] (3)
BENZENE [A]
ARSENIC JA]
FORMALDEHYDE fBll
METHYLENE CHLORIDE [B2]
CADMIUM [Bl]
TRICHLOROETHYLENE [B2]
PERCHLOROETHYLENE [C]
CARBON TETRACHLORIDE [82)
ETHYLENE DICHLORIDE [B2]
ETHYLENE OXIDE [B1J
ETHYLENE DIBROMIDE [B2]
CHLOROFORM [B2]
TOTAL (4)
Hotspot 2: 4339.85 374.88
POLYCLYCLIC ORGANIC MATTER 
-------
                      TABLE V-10 (CONTINUED)

                              NOTES

1  Only the sites with the highest increased lifetime individual
  cancer risk that were separated by at least two kilometers were
  chosen.  The purpose was to examine the sites of maximum
  lifetime individual cancer risk affected by the emissions from
  different combinations of sources.

2  This study uses conservative estimates of increased cancer risk
  from ambient  (i.e., outdoor) exposure to establish priorities
  among pollutants and sources.  The risk estimates are
  calculated using modelled or monitored concentrations and EPA
  unit cancer risk factors.  There is considerable uncertainty in
  the estimated concentrations, which could either overstate or
  understate the true concentrations (see Chapter IV).  Unit
  cancer risk factors combine CAG potency estimates with EPA
  exposure assumptions.  The CAG potency estimates provide a
  plausible upper limit to the cancer risk of a compound (see
  Appendix A); however, the true value of the risk is unknown and
  may be as low as zero.  The exposure assumptions are extremely
  conservative in that they assume continuous exposure to outdoor
  air for 70 years.  Because of the generally conservative bias
  in the information, it is highly unlikely that the true risks
  would be as high as the estimates, and they could be
  considerably lower.

3  The POM risk estimates  are based  on  source-specific unit  risk
  factors of which only POM from coke oven emissions has
  undergone extensive peer review.  Thus, these numbers are
  subject to change.  There is no overall weight-of-evidence
  determination for POM.

*  Totals  may not sum because of  rounding.
                              V-29

-------
                                                     TABLE V-ll
                                      TOTAL SOURCE AND POLLUTANT CONTRIBUTION
                                  BASED ON MODELLING AT EACH HOTSPOT LOCATION (1,2)

                                      (HOTSPOT LOCATION 1:  4343.00   374.12)
                                         (URBAN SCALE, 2.5 KM GRID SYSTEM)
                                     RESULTS  INTENDED FOR POLICY DEVELOPMENT ONLY


POLLUTANT


ROAD
[Weight of Evidence] VEHICLES
ARSENIC (A]
BENZENE [A] 3
CADMIUM [Bl] <1
CARBON TETRACHLORIDE [B2]
CHLOROFORM [ B2 ]
CHROMIUM HEXAVALENT [A]
ETHYLENE OXIDE [B1J
FORMALDEHYDE [ Bl ] 3
METHYLENE CHLORIDE [B2]
POLY ORG MATTER (POM) [NA](3) 2
ETHYLENE DI BROMIDE (EDB)[B2] <1
ETHYLENE DICHLORIDE (EDC)fB2]
-------
                                                 TABLE  V-ll  (CONTINUED)
                                       TOTAL SOURCE AND POLLUTANT CONTRIBUTION
                                   BASED ON  MODELLING  AT EACH  HOTSPOT  LOCATION(1,2)
                                         (HOTSPOT LOCATION 2:  4339.85   374.88)
                                           (URBAN SCALE.  2.5 KM GRID SYSTEM)
                                       RESULTS INTENDED FOR POLICY  DEVELOPMENT  ONLY
POLLUTANT
ROAD POINT SOLVENT POINT POINT
[Weight of Evidence] VEHICLES SOURCE A HEATING USAGE SOURCE B SOURCE C
ARSENIC [A]
BENZENE [A] 2
CADMIUM [Bl] <1
< CARBON TETRACHLORIDE [B2]
' CHLOROFORM [ B2 ]
£ CHROMIUM HEXAVALENT [A]
ETHYLENE OXIDE (Bl]
FORMALDEHYDE [ Bl ] 3
METHYLENE CHLORIDE [B2]
POLY ORG MATTER (POM) [NA](3) 2
ETHYLENE DIBROMIDE (EDB)[B2] <1
ETHYLENE DICHLORIDE (EDC)[B2]<1
TRICHLOROETHYLENE [B2]
PERCHLOROETHYLENE 1 C 1
<1. OOE-06 2. OOE-06 <1. OOE-06 <1. OOE-06
.OOE-06 3.00E-05
.OOE-06 <1. OOE-06 1. OOE-06 <1. OOE-06


8.20E-05 3. OOE-06 4. OOE-06

.OOE-06 1. OOE-06
4. OOE-06
. 70E-05 6.00E-04 2. OOE-06
.OOE-06
.OOE-06
<1. OOE-06 <1. OOE-06
<1. OOE-06
TOTAL
INCREASED
LIFETIME
INDIVIDUAL
POINT ALL CANCER
SOURCE D OTHER
2.50E-06
2. OOE-06
1.20E-06
4. OOE-09
3.90E-07
2. OOE-06 2. OOE-06
6.20E-OB
6. OOE-07
4. OOE-07
<1. OOE-06
9.60E-08
1.20E-07
7.60E-07
6.90E-07
RISK
4.50E-06
3.40E-OS
2.20E-06
4. OOE-09
3.90E-07
9.30E-05
6.20E-08
4.60E-06
4.40E-06
6.29E-04
9.60E-08
1.20E-07
7.60E-07
6.90E-07
TOTAL INCREASED LIFETIME
  INDIVIDUAL CANCER RISK (4)
3.20E-05  7.12E-04  6.OOE-06  4.OOE-06  3.OOE-06  4.OOE-06   2.OOE-06 1.08E-05  7.74E-04
                                                           (continued)

-------
to
                                                   TABLE V-ll (CONTINUED)

                                           TOTAL SOURCE AND POLLUTANT CONTRIBUTION
                                     BASED ON MODELLING AT EACH HOTSPOT LOCATION  (1,2)

                                           (HOTSPOT LOCATION 3:  4350.90  364.55)
                                             (URBAN SCALE, 2.5 KM GRID SYSTEM)
              POLLUTANT
         [Weight of Evidence]
  ROAD     POINT
VEHICLES SOURCE A
HEATING
    ARSENIC  (A]
    BENZENE  [A]                    4.OOE-06
    CADMIUM  [Bl]                  <1.00E-06
    CARBON  TETRACHLORIDE  [B2]
    CHLOROFORM  [B2]
    CHROMIUM HEXAVALENT  [A]
    ETHYLENE OXIDE  [Bl]
    FORMALDEHYDE   [Bl]              4 . DOE-06
    METHYLENE CHLORIDE   [B2]
    POLYC ORG MATTER  (POM)[NA](3)   3.70E-05
    ETHYLENE DIBROMIDE  (EDB)[B2]   <1.00E-06
    ETHYLENE DICHLORIDE (EDC)[B2]  <1.00E-06
    TRICHLOROETHYLENE  [B2]
    PERCHLOROETHYLENE  [C]
           i.OOE-06
          9.OOE-06
 .OOE-06
 .OOE-06
 .OOE-06
                    2.OOE-06
          5.00E-05  4.OOE-06

DEVELOPMENT ONLY


SOLVENT POINT POINT
USAGE SOURCE B SOURCE C
5.40E-05 <1.00E-06

2.20E-05


3.01E-04 l.OOE-05


6.00E-06



l.OOE-06
l.OOE-06




POINT ALL
SOURCE 0 OTHER
3.00E-06
L.60E-06
2.00E-06
l.OOE-08
6.90E-07
3.00E-06 <1.00E-06
9.00E-08
9.00E-07
1.50E-06
4. OOE-06
1.40E-07
1.80E-07
2.00E-07
3.00E-07
TOTAL
INCREASED
LIFETIME
INDIVIDUAL
CANCER
RISK
6.10E-OS
9.60E-06
2.60E-05
l.OOE-08
6.90E-07
3.23E-04
9.00E-08
6.90E-06
7.SOE-06
9.50E-05
1.40E-07
1.80E-07
1.20E-06
1.30E-06
    TOTAL INCREASED LIFETIME
      INDIVIDUAL CANCER RISK (4)

    NA = Not Available
4.50E-05  6.30E-05  1.20E-OS  8.OOE-06  3.77E-04  l.OOE-05  3.OOE-06  1.46E-05  S.33E-04
    (1)   This study uses conservative  estimates of  increased cancer risk  from ambient  (i.e., outdoor)  exposure  to
         establish priorities  among  pollutants and  sources.  The risk estimates  are calculated using modelled or
         monitored concentrations  and  EPA unit cancer risk factors.  There  is considerable  uncertainty in  the estimated
         concentrations, which could either  overstate or understate the true concentrations  (see Chapter IV).   Unit
         cancer risk factors combine CAG potency estimates with EPA exposure assumptions.   The CAG  potency estimates
         provide a plausible upper limit to  the cancer  risk of a compound (see Appendix A);  however, the true value of
         the risk is unknown and may be as low as zero.  The exposure assumptions are  extremely conservative in that
         they assume conintuous exposure to  outdoor air for 70 years.  Because of the  generally conservative bias  in  the
         information, it is highly unlikely  that the true risks would be  as high as the estimates,  and they could  be
         considerably lower.

    (<2)   Risk numbers are in scientific notation.   Thus 1E-06 is the same as 1 x 10 to the  sixth power or  one in a million.

    (3)   The POM risk estimates are  based on source specific unit  risk factors that have not undergone
          extensive peer review.   Thus, these numbers are subject  to change.
    (4)  The numbers shown may not sum to  the  totals  because of  rounding

-------
shown, the lifetime individual cancer risks at each of these
sites are dominated by emissions from Point Source A.  At the
first two receptor locations, Point Source A accounts for
approximately 91 percent of the increased lifetime individual
cancer risks.  At the third hotspot location, Point Source A is
still a major player but its contribution to the increased
lifetime individual cancer risk is lower, approximately 69
percent.  Point Sources A, B, and C together account for roughly
83 percent of the increased lifetime individual cancer risks at
the third site.  Road vehicles and heating account for an
additional 11 percent.  The pollutants of concern at each of the
hotspot locations are also of important contributors to the
estimated annual excess cancer incidence.

               (2)  Estimated Annual Excess Cancer Incidence in
                    the Grid Cell of Highest Predicted Annual
                    Excess Cancer Incidence

     In Table V-12, we provide a detailed examination of the 2.5
km grid with the highest estimated area-wide annual excess cancer
incidence.  For this analysis, we used the more accurate
estimates associated with the 2.5 kilometer (refined) grid system
(see Figure IV-4).

     The grid with the highest estimated annual excess cancer
incidence can be found in the center of the study area.  The
average increased lifetime individual cancer risk at this
location is 3 x 10"*, and the predicted annual excess cancer
incidence is approximately 0.15.  The grid, which covers less
than 1 percent of the study area, accounts for about 4 percent of
the estimated area-wide annual excess cancer incidence.  Almost
99 percent of the annual excess cancer cases are attributable to
seven compounds:  hexavalent chromium, POMs, arsenic, cadmium,
benzene, methylene chloride, and formaldehyde.  Hexavalent
chromium and POMs are the primary contributors to the annual
excess cancer incidence in the grid cell of highest incidence, as
well as the estimated annual excess cancer incidence for the
study area.  However, hexavalent chromium, and not POM, explains
most of the estimated annual excess cancer incidence in the grid
cell of highest incidence.

     Table V-13 arrays the estimated annual excess cancer
incidence in the grid cell of highest incidence by facility and
pollutant.  Almost 91 percent is attributable to four sources:
Point Source A, Point Source B, road vehicles, and heating.
Point Source B accounts for almost half of the estimated annual
excess cancer incidence at the location of maximum incidence.
Approximately 85 percent of the estimated annual excess cancer
incidence associated with hexavalent chromium is also
attributable to Point Source B.  Almost all of the estimated
annual excess cancer incidence associated with POM can be
explained by environmental releases from Point Source A, road
vehicles, and heating.


                               V-33

-------
                             TABLE V-12

                  BALTIMORE  IEMP AIR TOXICS STUDY

               ESTIMATED ANNUAL  EXCESS CANCER INCIDENCE
          BASED ON MODELLING  BY  POLLUTANT IN THE GRID CELL OF
          HIGHEST PREDICTED ANNUAL EXCESS CANCER INCIDENCE (1)

                 RESULTS  FOR POLICY DEVELOPMENT ONLY
POLLUTANT
[Weight of Evidence]
Chromium (VI) [AJ
Polycyclic organic
matter (POM) [N.A.] (2)
Arsenic [A]
Cadmium [Bl]
Benzene [A]
Methylene chloride [B2]
Formaldehyde [Bl]
Perchloroethyiene (Perc) [C]
Trichloroethylene (TCE) [B2]
Chloroform [B2]
Ethylene dichloride (EDC) [B2]
Ethylene dibromide (EDB) [B2]
Ethylene oxide [Bl]
Carbon tetrachloride [B2]
ANNUAL EXCESS
CANCER
INCIDENCE
0.10
0.04
0.01
Oft 1
. Ul
<0.01
<0.01
<0.01
<0.01
<0.01
f\ f\ 1
<0 . 01
<0.01
<0.01
<0.01
<0.01
PERCENTAGE
OF TOTAL
47.2
29.5
9.6
a i
*t « x
3n
. u
2.7
2.6
0.5
0.4
01
• i
0.1
<0.1
       TOTAL (3)
0.15                  100.0
N.A. = Not available.
    This study uses conservative estimates of increased cancer risk
    from ambient (i.e., outdoor) exposure to establish priorities
    among pollutants aAd sources.  The risk estimates are calculated
    using modelled or monitored concentrations and EPA unit cancer
    risk factors.  There is considerable uncertainty in the estimated
    concentrations, which could either overstate or understate the
    t?Se concentrations (see Chapter IV).  Unit Cancer risk factors
    combine CAG potency estimates with EPA exposure assumptions.  The
    CAG potency estimates provide a plausible upper limit to the
    cancer risfc of a compound (see Appendix A); however, the true risk
    is unknown and may be as low as zero.  The exposure.assumptions
    are eSremely conservative in that they assume continuous exposure
    II outdoor air for 70 years.. Because °f the generally conservative
    bias in the information, it is highly unlikely that the true risks
    would be as high as the estimates and they could be considerably
    lower.

 i7\ The POM risk estimates are based, in part, on source-specific
 (  } unit risk factors that have not undergone extensive peer review.
    Thus, these numbers are subject to change.

 (3) Totals may not sum because of rounding.
                                    V-34

-------
OJ
Ul
                                                      TABLE V-13

                                          BALTIMORE IEMP AIR TOXICS STUDY

                                   ESTIMATED MAXIMUM ANNUAL EXCESS CANCER INCIDENCE
                                 BY POLLUTANT AND SOURCE IN THE GRID CELL OF HIGHEST
                                     ESTIMATED ANNUAL EXCESS CANCER INCIDENCE (1)
                                          (URBAN SCALE, 2.5 KM GRID SYSTEM)

                                       RESULTS INTENDED FOR POLICY DEVELOPMENT ONLY
             POLLUTANT
         [Weight of  Evidence]
                              POINT    POINT    ROAD
                            SOURCE B SOURCE A VEHICLES
                HEATING
                SOLVENT    POINT     POINT
                 USAGE  SOURCE C SOURCE D
ARSENIC  [A]
BENZENE  [A]
CADMIUM  [Bl]
CARBON TETRACHLORIDE  [B2]
CHLOROFORM   [B2]
CHROMIUM HEXAVALENT  [A]
ETHYLENE OXIDE  [Bl]
FORMALDEHYDE  [Bl]
METHYLENE CHLORIDE  [62]
POLYC ORG MATTER  (POM)[N.A.]
ETHYLENE DIBROMIDE (EDB)[B2]
ETHYLENE DICHLORIDE (EDC)[B2]
TRICHLOROETHYLENE  [B2]
PERCHLOROETHYLENE  [C]

TOTAL ANNUAL EXCESS
   CANCER INCIDENCE (3).(4)

PERCENTAGE OF TOTAL (3).(4)
                                     0.06
                                   (2)
                                     0.07
                                     47.6Z
<0.01
 0.02
 0.02
 13.8Z
<0.01

 0.02
<0.01
<0.01
 0.02
 13.52
<0.01

<0.01

<0.01
 0.00


  O.OZ
                                                                TOTAL
                                                            ANNUAL EXCESS
                                                      ALL      CANCER
                                                     OTHER    INCIDENCE
0.01

<0.01
<0.01
<0.01
<0.01

<0.01
<0.01
<0.01
<0.01
<0.01
<0.01



<0.01
<0.01

 0.00


  o.oz
<0.01




<0.01








<0.01
<0.01
<0.01
<0.01
<0.01
<0.01 0.01
<0.01
<0.01
<0.01
<0.01
<0.01
<0.01
<0.01
<0.01
0.01
<0.01
<0.01
<0.01
<0.01
0.07
<0.01
<0.01
<0.01
0.04
<0.01
<0.01
<0.01
<0.01
                                                                                 0.00
                                                                                  O.OZ
                                               0.00
                                                O.OZ
                                                                                                    0.04
                                               26.71
                                                            0.15
100.OZ
     N.A.  = Not Available

     (1)  This study uses conservative estimates  of  increased  cancer  risk  from ambient (i.e.. outdoor) exposure to
         establish priorities among pollutants and  sources.   The  risk estimates are calculated using modelled or
         monitored concentrations and EPA unit cancer  risk  factors.  There  is considerable uncertainty in the estimated
         concentrations, which could either overstate  or  understate  the true concentrations (see Chapter IV).  Unit
         cancer risk factors combine CAG potency estimates  with EPA  exposure assumptions.  The CAG potency estimates
         provide a plausible upper limit to the  cancer risk of  a  compound (see Appendix A); however, the true value of
         the risk is unknown and may be as low as zero.   The  exposure assumptions are extremely conservative in that
         they assume continuous exposure to outdoor air  for 70  years.   Because of the generally conservative bias in the
         information, it is highly unlikely that the true risks would be  as high as the estimates, and they could be
         considerably lower.
                                                   (continued)

-------
                                     TABLE V-13  (CONTINUED)
(2)  The POM risk estimates are based,  in part,  on source-specific  unit  risk  factors  that  have
    not undergone extensive peer review.  Thus,  these  numbers  are  subject  to change.

(3)  The POM risk estimates are based on source  specific  unit  risk  factors  that  have  not undergone
    extensive peer review.  Thus, these number  are subject  to change.

(A)  The numbers shown may not sum to the totals because  of  rounding.

-------
     b.  Noncancer Risks

          i.  Results Based on Dispersion Modelling

     The analysis of noncancer risks considered two exposure
scenarios:  (1) modelled concentrations that were generated for a
subset of the study area using the refined (i.e., 2.5 km) grid
system, and (2) modelled concentrations resulting from an in-
depth analysis of ambient air concentrations at 118 discrete
locations near roughly 40 sources.7  The analysis of  noncancer
risks focused on 18 pollutants, of which 12 are also
carcinogenic.

     As discussed earlier in Chapter II, we analyzed noncancer
effects in two ways.  First, we divided the predicted ambient air
concentration of each pollutant evaluated in the modelling
exercise by the no-effect threshold(s) for those noncancer
effects (e.g., liver toxicity, kidney toxicity, reproductive,
neurological, fetal or blood) relevant to the pollutant,  If the
resulting ratio exceeded one, we identified these pollutant-
specific exposures as requiring additional investigation.
Second, we developed a "hazard index" that .summed individual
pollutant ratios by effect category.  The latter analysis was
aimed at examining the impact of exposure to complex chemical
mixtures in the ambient air.  If the index exceeded one, we
identified these exposures as deserving further analysis.

               (1)  Pollutant Specific

     In both the refined grid and discrete receptor analysis, we
found locations at which the estimated benzene and xylene ambient
air concentrations exceed several no-effect thresholds.  The
refined (2.5 km) grid analysis indicates that there is a need to
further examine benzene exposures in the southeast quadrant of
the study area (see Table V-14) where benzene ambient concentra-
tions exceed the no-effect threshold for blood effects.  The
results from the discrete receptor analysis support these
findings and show additional sites at which benzene exposures
should also be studied in greater detail.  The discrete receptor
analysis also identifies a section of the study area in which
xylene exposures may need to be more carefully examined because
they exceed the threshold levels for the following noncancer
     7See Chapter IV for an overview of the three levels at which
modelling and population exposures were considered:  5 km grid
cell, 2.5 km grid cell, and discrete receptor locations.  Figures
IV-2 through IV-4 illustrate these grid locations graphically.

     8For the purpose of this analysis, we assumed that a ratio
greater than or equal to 0.95 was equivalent to 1.0.

     'ibid.

                               V-37

-------
                                        TABLE V-14

                  RECEPTOR LOCATIONS WARRANTING FURTHER INVESTIGATION
                    FOR NONCANCER EFFECTS:  POLLUTANT-SPECIFIC (1)

                    RESULTS INTENDED FOR POLICY DEVELOPMENT ONLY
    RECEPTOR
    LOCATION
   REFINED GRID
                CONCENTRATION
                TO THRESHOLD
                  RATIO (2)
                 POLLUTANT
                 OP CONCERN
                   NONCANCER
                   EFFECT (3)
4342.25
4342.25
4342.25
4342.75
4342.75
4342.75
364.75
367.25
369.75
364.75
367.25
369.75
1.1
1.6
5.0
1.3
1.1
1.4
BENZENE
BENZENE
BENZENE
BENZENE
BENZENE
BENZENE
BLOOD
BLOOD
BLOOD
BLOOD
BLOOD
BLOOD
  DISCRETE SITE
4339.85
4340.22
4340.65
4340.85
4341.15
4341.41
4342.00
4342.52
4343.00
4343.15
4350.92
4350.92
4350.92
4350.92
4350.92
4350.92
374.86
374.95
375.10
375.31
375.30
375.30
374.25
374.00
374.12
373.90
370.55
370.55
370.55
370.55
370.55
370.55
                                1.7
                                1.6
                                1.3
                                1.2
                                1.2
                                1.1
                                1.3
                                1.2
                                1.1
                                1.1
                                1.5
                                1.5
                                6.2
                                1.5
                                6.2
                                1.5
                                       BENZENE
                                       BENZENE
                                       BENZENE
                                       BENZENE
                                       BENZENE
                                       BENZENE
                                       BENZENE
                                       BENZENE
                                       BENZENE
                                       BENZENE
                                        XYLENE
                                        XYLENE
                                        XYLENE
                                        XYLENE
                                        XYLENE
                                        XYLENE
                                       BLOOD
                                       BLOOD
                                       BLOOD
                                       BLOOD
                                       BLOOD
                                       BLOOD
                                       BLOOD
                                       BLOOD
                                       BLOOD
                                       BLOOD
                                       LIVER
                                       KIDNEY
                                    REPRODUCTIVE
                                    NEUROLOGICAL
                                    FETAL/DEVELOPMENT
                                       BLOOD
(1)  The noncancer effects analysis did not consider personal exposures
      (e.g., indoor and work).

(2)  The threshold values underlying these ratio calculations are based
     on data of uneven quality.   Furthermore,  ezceedance of a threshold
     value of greater than one  does not necessarily indicate severity
     of effect; it simply indicates the need to further explore these
     exposure levels.  For the  purpose of this study,  it was assumed
     that a ratio greater than  or equal to 0.95 was equivalent to "1"
     through rounding.

(3)  The blood effect thresholds for benzene and xylene have not yet
     undergone peer review, thus the results are subject to change.
                                        V-38

-------
effects:  liver, kidney, reproductive, neurological, fetal, and
blood.

     We explored source contribution for the refined grid cell
and the discrete site with the highest concentration-to-threshold
ratios.  For the refined grid cell, benzene emissions from Point
Source A account for approximately 94 percent of the ambient
concentrations that appear to pose a concern for blood effects.
This finding is subject to the limitation stated above regarding
the uncertainty (potentially overstated) of the modelled results
for Point Source A.  At the discrete receptor site with the
highest concentration-to-threshold ratios, almost all of the
ambient air concentrations of xylene can be attributed to Point
Source E.

               (2)  Complex Pollutant Mixtures

     Using the hazard index we identified additional discrete
receptor locations where exposures to multiple pollutants may
deserve a closer examination because of concern for blood
effects.  As shown in Table V-15, there sites are primarily in
the southeast quadrant of the study area, Similar to the
pollutant specific results.   The compound principal concern is
benzene.

          ii.  Results Based on Monitoring Data

     We investigated the increased concern for noncancer effects
looking at (1) average area-wide exposures based on the available
monitoring data for the Baltimore area and (2) exposures in the
vicinity of individual monitoring sites from the TEAM study
(Dundalk and Parkville) and the available area-wide Baltimore
monitoring data.  We explored whether the measured ambient air
concentration of each pollutant examined exceeded any relevant
no-effect thresholds.  We also calculated a hazard index for each
noncancer effect.  We should note again that the existence of a
threshold value does not suggest anything about the severity of
effect.  Also, the data underlying the threshold values are of
varying quality.

               (1)  Area-Wide

     Using the average area-wide ambient pollutant levels
suggested by the available Baltimore monitoring data (see
Table IV-11), it appears that benzene exposures warrant further
investigation for blood effects in the Baltimore study area.
Because of the limitations in using data from a limited number of
fixed monitoring sites to typify an entire region, further
sampling and analysis are needed to validate these findings.  The
hazard index calculation did not indicate the need to further
study additional pollutant exposures.
                              V-39

-------
                                     TABLE V-15

                   RECEPTOR LOCATIONS WARRANTING FURTHER INVESTIGATION
                  FOR NONCAHCER EFFECTS:   COMPLEX POLLUTANT MIXTURES (1)

                    RESULTS INTENDED FOR POLICY DEVELOPMENT ONLY
    DISCRETE
    RECEPTOR
    LOCATION
4342.93
4343.20
4343.45
4343.58
4343.70
4346.27
4346.68
362.98
362.95
362.78
362.55
362.28
367.88
367.77
                 HAZARD
                INDEX (2}
1.0
1.0
0.96
1.0
0.95
            PRIMARY POLLUTANT
               OF CONCERN
             (RATIO VALUE)
BENZENE
BENZENE
BENZENE
BENZENE
BENZENE
BENZENE
BENZENE
(0.92)
(0.92)
(0.94)
(0.93)
(0.88)
(0.93)
(0.87)
                      NONCANCER
                        EFFECT
BLOOD
BLOOD
BLOOD
BLOOD
BLOOD
BLOOD
BLOOD
(1)  The noncancer effects analysis did not consider personal exposures
     (e.g.,  indoor and work).

(2}  The threshold values underlying these hazard indexes are based
     on data of uneven quality.   Furthermore,  exceedance of a threshold
     value or a hazard index of  greater than one does not necessarily
     indicate severity of effect; it simply indicates the need to further
     investigate these exposure  levels.  For the purpose of this study.
     it was  assumed that a ratio greater than or equal to 0.95 was
     equivalent to "1" through rounding.  The hazard index represents the
     sum of  all pollutant-specific ratios with the same systemic effect
     at a particular location.

(3)  The blood effect threshold  for benzene has not yet undergone peer
     review, thus the results  are subject to change.
                                         V-40

-------
                (2)   Individual Monitoring Sites

      TEAM Data (Dundalk and Parkville).   Table V-16  presents  the
 concentration-to-threshold ratios  calculated by  pollutant and
 noncancer effect, as well as the hazard  indexes.   These
 calculations were made using the average measured  pollutant
 concentrations from the two fixed  TEAM monitoring  sites.   Note
 that we did not include the concentration-to-threshold ratios for
 methylene chloride  and vinylidene  chloride in calculating the
 hazard indexes because of our concern regarding  the  validity  of
 the  monitoring data (see Chapter IV).

      Our analysis of individual pollutants indicates that benzene
 exposures warrant further investigation  at the Dundalk and
 Parkville sites; the measured benzene concentrations in these
 areas exceed the threshold for blood effects.  In  addition, if
 verified,  the vinylidene chloride  ambient levels at  each  site and
 methylene chloride  at Parkville may require additional analysis
 because of an increased concern for liver effects.   The hazard
 index calculations  did not identify additional pollutants
 warranting further  study.

      Available Monitoring Data  for the Baltimore Area.    Based on
 the  available monitoring data for  the Baltimore area (see
 Appendix B for a summary of  the average measured concentrations
 by monitoring site),  benzene exposures—even at the  monitoring
 site with  lowest average measured  level—need to be  examined  more
 closely.   This finding is supported by the TEAM data and  the
 results  from the dispersion  modelling.

      c.  Limitations

      The risk calculations presented in this section build
 directly on  the estimates  of  exposure (either measured or
 modelled)  and dose-response.  The  underlying uncertainties
 associated with the  exposure  and dose-response calculations are
 briefly  discussed below.

           i.   Exposure  Limitations Baaed on Dispersion Modelling

     Although  time constraints  have not permitted a  thorough
 evaluation of  model  performance using the  1987 TEAM  ambient data,
 a similar exercise has  been completed using the ambient air data
 gathered during 1983  to 1984  in Baltimore.  The results of this
 effort suggest  that  the model may  be systematically  underpredict-
 ing ambient air risks  for  some  pollutants.  However,   facility-
 specific monitoring  performed by the AMA suggests that ambient
 levels associated with  Point  Source A are most likely overstated.

     There are  four major  factors  introducing uncertainty into
our estimates of exposure:
                              V-41

-------
                                    TABLE V-16
                         BALTIMORE IEHP AIR TOXICS STUDY

          NONCANCER RATIO CALCULATIONS FROM EXPOSURE TO TARGET COMPOUNDS

                       AT THE TWO FIXED TEAM MONITORING SITES (1), (2)

                                 Dundalk
POLLUTANT (3)
                   REPRO-     NEURO-
   LIVER   KIDNEY  DUCTIVE  BEHAVIORAL
                            FETAL BLOOD (4)
Arsenic
Benzene
Cadmium
Carbon tetrachloride
Chlorobenzene
Chloroform
Chromium-VI (5)
mDichlorobenzene
oDichlorobenzene
pDichlorobenzene
Ethyl benzene
Ethylene dibronu.de
Formaldehyde (6)
Lead
Methyl chloroform
Methylene chloride
Perchloroe thy lane
Styrene
Trichloroethylene
Vinylidene chloride (7)
m-Xylene


0.003
0.061
0.000
0.002
0.000
N.Q.
0.009
0.018
0.004
0.017
0.024
0.005
0.162
0.009
0.006
0.015
1.061
0.026


0.005
0.001

0.004

N.Q.
0.009
0.018
0.004
0.017
0.024
0.005
0.049
0.009

0.000

0.026


0.000
0.000

0.042
0.000




0.123
0.024
0,005





0.105



0.061

0.009

N.Q.
0.009
0.018


0.024
0.005
0.004


0.015

0.026

0.093
0.000
0.006

0.042
0.000

0.009
0.018


0.024
0.005
0.162
0.001



0.105
0.000
1.572




0.000
N.Q.




0.138
0.024
0.005


0.006


0.026
HAZARD INDEXES (8)
     0.3
0.2
0.3
0.2
0.4
1.8
POLLUTANT (3)
             Parkville

                   REPRO-     NEURO-
LIVER     KIDNEY   DUCTIVE  BEHAVIORAL
                           FETAL  BLOOD £4)
Arsenic
Benzene
Cadmium
Carbon tetrachloride
Chlorobenzene
Chloroform
Chromium-VI (5)
mDichlorobenzene
oDichlorobenzene
pDichlorobenzene
Ethyl benzene
Ethylene dibromide
Formaldehyde (6)
Lead
Methyl chloroform
Methylene chloride
Perchloroethylene
Styrene
Trichloroethylene
Vinylidene chloride (7)
m-Xylene
HAZARD INDEXES (9)


0.002
0.408
0.014
N.D.
0.000
0.023
0.015
0.018
0.017
0.006
0.022
0.019
1.219
0.031
0.017
0.000
2.171
0.071
0.6


0.004
0.009
Q.014
N.D.

0.023
0.015
0.018
0.017
0.006
0.022
0.019
0.366
0.031

0.000
0.071
0.2


0.000
0.002
0.014
N.D.
0.000




0.043
0.022
0.019




0.291
0.4



0.408
0.014
N.D.

0.023
0.015
0.018


0.022
0.019



0.000
0.071
0.6

0.127
0.000
0.041
0.014
N.D.
0.000





0.022
0.019
1.219
0.002


0.291
0.5
0.000
2.135


0.014

0.000
0.023
0.015
0.018


0.130
0.022
0.019


0.017

0.071
2.5
N.Q. - Not Quantified in sampling
N.D. - Not detected in sampling
                                         (continued)

                                       V-42

-------
                      TABLE V-16  (CONTINUED)

                               NOTES

 1.    The two fixed ambient monitoring  stations  were located at
      the Parkville Middle School  in the  Parkville section of
      Baltimore County and the Merritt  Point Mental Health
      Facility at the Dundalk  section of  Baltimore County.

 2.    The noncancer ratios are defined  as (1)  the  ratio  of  the
      measured ambient concentration of the  pollutant and  the
      threshold for a specific noncancer  health  effect category,
      and (2)  a "hazard index" which is calculated by summing the
      pollutant-specific ratios for  each  noncancer health  effect.

 3.    Risk assessments for metals  used  the average concentrations
      in  the  PM10 samples.

 4.    For the purpose of this  screening analysis,  the health
      category of blood  effects includes  non-specific cellular
      effects.

 5.    We  assumed  that all  of the chromium that was collected was
      chromium VI.   The  risk from  exposure to chromium is
      therefore overestimated  for  the purpose of this screening
      exercise since the actual percentage of chromium VI in the
      sample  is likely to  be considerably lower.

 6.    The noncancer  health  effects of formaldehyde were  included
      in  this  category even though the  known effects  are of  a less
      specific  nature.

 7.    Vinylidene  chloride was  detected  in the sampling, but  there
      were analytical  problems  in quantifying ambient
      concentrations.  Because  vinylidene chloride could be  an
      important compound, the  ratio of  the ambient concentrations
      to  the threshold value is  given for the purposes of
      comparison; however,  it  is not included in the  hazard  index.
     Additional analysis is needed to  confirm the presence  of
     vinylidene chloride in the ambient air.

8.   The monitoring results for vinylidene chloride  at Dundalk
     are highly questionable due to analytical problems in
     quantifying ambient concentrations.   Consequently,  it  is not
     included in the cumulative hazard index.  Additional
     analysis is needed to confirm the presence of vinylidene
     chloride in the ambient air.

9.   The monitoring results for methylene chloride and vinylidene
     chloride in the ambient air of Parkville are highly
     questionable due to analytical problems in quantifying the
     amounts in the monitoring samples.  Consequently, these
     compounds are not included in the cumulative hazard index.

                               V-43

-------
     •    First, exposure modelling can be limited by the quality
          and coverage of the emissions database.  Although the
          air toxics inventory that we compiled for this study
          represents a significant data collection effort, it is
          impossible to account for all sources.  For example,
          the inventory does not address fugitive emissions which
          can be a major contributor to ambient levels.  Based on
          the model performance evaluation, our modelling
          analysis significantly underestimated exposures to all
          pollutants, except for trichloroethylene (a more
          thorough discussion on this topic can be found in
          Chapter IV).

     •    Second, modelling can be hampered by the lack of
          adequate detail on releases and meteorological
          conditions at a facility.  Simplifying assumptions
          concerning release parameters and meteorological
          conditions could result in either underestimates or
          overestimates (as in the case of Point Source A) of
          exposure.

     •    Third, dispersion modelling cannot adequately address
          the products of atmospheric chemical transformation,
          which may be the driving force behind the problem of
          "urban soup."  As scientific knowledge expands/
          dispersion modelling may be able to accommodate trans-
          formation products more easily.  For the moment,
          however, existing models are best suited for quantify-
          ing ambient levels of relatively stable compounds.

     •    Fourth, as we begin to add the pollutant-specific
          risks, the uncertainties in the risk calculations are
          compounded.

          ii.  Exposure Limitations Based on Monitoring

     Many of the limitations of the TEAM monitoring program and
the available monitoring data for the Baltimore area have already
been discussed, but a few of them are worth repeating for the
reader:

     •    The quality control analysis in the TEAM study
          indicated that actual concentrations may be overstated
          by 30 percent for the samples collected at Parkville.
          However, it does not appear that this overestimate
          changes our findings on individual risks and noncancer
          effects dramatically: average increased lifetime
          individual cancer risks at Parkville could fall from
          3.5 x 10~3 to 2.5 x 10"3,  and  the  hazard  indexes  for
          liver and blood (the two noncancer effect categories of
          concern) will still remain above one.
                               v-44

-------
      •     The available monitoring  data  for the  Baltimore area
           are highly uncertain.   These data stem from different
           monitoring programs  conducted  at  different  times.
           Also, not  all results have undergone extensive  peer
           review.

      •     Unlike dispersion modelling, ambient data alone cannot
           be  used  to identify  sources and their  relative
           contributions.

      •     Additional ambient air monitoring should be performed
           to  obtain  a better baseline estimate of methylene
           chloride ambient levels.

      •     Vinylidene chloride, a possible human  carcinogen
           (Class C),  could be  an important  factor in  the  risk
           calculations  and hence its measured concentrations in
           the ambient air need to be confirmed.  As a  result,
           additional analysis  is needed to  confirm its presence
           in the ambient air.

      •    As we begin to add the pollutant-specific risks, the
          uncertainties in the risk calculations are compounded.

          iii.  Dose-Response Limitations

     Chapter. Ill and Appendix A describe in detail the principals
and limitations of quantitative risk assessment.   The major
caveats relating to  the unit cancer risk factors and no-effect
thresholds for noncancer effects are briefly restated below:

               (1)  Cancer

     •    As discussed throughout the report, EPA relies on
          models that yield a plausible,  upper-bound estimate of
          potency rather than a "best guess" estimate.  Thus, the
          true risks are unlikely to be as high as our risk
          estimates suggest,  and could be significantly lower.

     •    The POM unit cancer risk factors have not undergone
          peer review and are thus subject to change.

               (2)   Noncancer

     •    The no-effect threshold  for blood effects (the
          threshold most often  exceeded)  has not  undergone peer
          review and is, therefore,  subject to change.

     •    Numerous  uncertainty  factors  have been  incorporated
          into the  setting of a no-effect threshold.   Thus,  the
          quality of  evidence underlying  the threshold values
          used in  this report vary widely by pollutant.
                              V-45

-------
      •     Exceedance of  a threshold does  not  indicate  severity of
           an effect; it  merely indicates  the  need to further
           examine an observed exposure  level.


 3.   RISK RESULT COMPARISONS

      To  provide perspective  on the  modelled risk  estimates  for
 the  Baltimore study area,  we compared our findings on  modelled
 risk with the results from other  lEMPs  and studies of  urban
 areas.   Because we used  a new approach  to estimate the lifetime
 individual cancer risks  from exposure to  the  semi-volatile
 fraction of the products of  incomplete  combustion (POMs), we also
 compared our POM risk results with  the  estimated  increased
 lifetime individual cancer risks  from the products of  incomplete
 combustion summarized in a 1985 EPA study. °   Finally,  we
 compared modelled risks  attributable to air toxics in  the
 Baltimore area with those associated with drinking water.   Each
 of these comparisons is  detailed  below.

      a.   Comparison with IBMP and Other Urban-Scale Analyses

      Table V-17 shows a  comparison  of the average increased
 lifetime individual cancer risks  and Table V-18 the maximum
 increased lifetime individual cancer risks in Baltimore with  the
 results  of seven other urban air  toxic  studies.   The air toxics
 studies  developed estimates  of average  and maximum increased
 lifetime individual cancer risks  for individual toxic  pollutants.
 These studies  are briefly described below.

      The EPA "Six Month  Study."   The first attempt by  EPA,  using
 a variety of  techniques  based largely on  existing data and  on
 extrapolations from existing data,  to estimate the scope and
 character of  the air toxics  problem nationally.   It comprises  a
 series of  studies,  including the  following: "35 County Study,"
 which produced estimates of  air toxics  risk in 35  U.S. counties;
 the  "Ambient Air Quality Study,"  which  used available  ambient  air
 data  to  estimate cancer  risk;  and the "NESHAPs" study, which
 analyzed air  toxics risk based on available national emissions
 data.
     IOU.S. EPA, The Air Toxics Problem in the United States;  An
Analysis of Cancer Risks for Selected Pollutants, June 1985 (the
so-called "Six Month Study)."

                               V-46

-------
                                                   TABLE V-17

                       A COMPARISON OF PREDICTED AVERAGE LIFETIME INDIVIDUAL CANCER RISK
                                        IN BALTIMORE WITH OTHER STUDIES

                                 RESULTS INTENDED FOR POLICY DEVELOPMENT ONLY1-2

POLLUTANTS
[WEIGHT OP
EVIDENCE 1
Arsenic [A]
Benzene [A]
Benzo-a-
pyrene [B2]
Cadmium [Bl]
Carbon Tet. [B2]
Chloroform [B2J
Chromium VI [A]
1.2Dichloro-
ethane [B2]
Ethylene
Dibromide [B2]
Ethylene
Oxide [Bl]
Forma Idehyde(Bl)
Methylene
Chloride [B2]
Perchloro-
ethylene (C)
POMs [N.A.]
Trichloro-


NESHAPS
STUDY
1.4E-06
9.8E-06


4.9E-06
4.2E-06
<7.00E-07
7.7E-06
1.3E-05
8.4-06
1.5E-05
<7.00E-07

<7.00E-07

2.8E-06


35
COUNTY
STUDY
1.4E-05
2.7E-OS

1.4E-06
1.4E-06
2.8E-07
2.1E-07
2.0E-05
2.8E-06
1.4E-06

1.5E-05


9.8E-06

1.1E-05
AMBIENT
AIR
QUALITY
STUDY
1.8E-05
7.6E-OS

1.4E-06
4.2E-06
2.6E-OS
3.2E-OS
7.4E-05



5.8E-05


7.7E-06

7.7E-06

SOUTH
COAST
l.OE-09
3.9E-04


6.7E-06
l.OE-08

7.1E-04

5.0E-08


2.4E-OS

3.0E-06



SANTA
CLARA
IEMP
1.5E-05
2.0E-05


4.0E-06
l.OE-OS
6.0E-08
2.0E-05

2.0E-07
2.0E-06

6.0E-07

2.0E-06

l.OE-07


KANAWHA
IEMP
1.3E-05
2.9E-05

4.3E-06
2.0E-06
1.5E-05
1.4E-04



5.9E-04

2.6E-05

1.5E-06

1.8E-06


PHILA BALT
IEMP STUDY
8.6E-06
1.9E-OS 8.7E-06

9.9E-09
3.8E-06
1.5E-06 3.7E-09
4.6E-06 9.0E-07
2.2E-05
2.6E-06 2.1E-07
1.7E-07
8.4E-08
8.1E-06
8.7E-06

1.7E-06 1.7E-0
8.9E-05
1.3E-06 1.5E-06
ethylene [B2]
                                                   (continued)

-------
                                                  TABLE V-17  (CONTD.)

                                                         NOTES

   N.A.  - Not available.

   1  This study uses conservative estimates of increased cancer risk from ambient (i.e.,  outdoor) exposure to
      establish priorities among  pollutants and sources.  The  risk estimates are calculated using modelled
      concentrations and  EPA unit cancer  risk  factors.   There  is considerable uncertainty in the  estimated
      concentrations, vhich could either  overstate or understate the true concentrations (see Chapter IV).  Unit
      cancer risk factors combine CAG potency  estimates  with EPA exposure assumptions.  The CAG patency estimate
      provide a plausible upper limit to  the cancer  risk of a  ocmpound  (see Appendix A); however, the true value  of
      the risk is unknown and may be ae low as zero.  The exposure assumptions are extremely conservative in that
      they assume continuous exposure to  outdoor  air for 70 years.  Because of the generally conservative bias in the
      information, it is  highly unlikely  that  the true  risks would be as high as the estimates, and they could be
      considerably lower.

   *  The unit cancer risk factors used to estimate risk vary among the studies.   The reader is directed to each
      report for details.  Differences in modelling  assumptions, which  vary across studies, can slightly affect
      predictions of risk.
i
•I*
CO

-------
                                                   TABLE V-1B

                        A COMPARISON OF PREDICTED MAXIMUM LIFETIME INDIVIDUAL CANCER RISK
                                         IN BALTIMORE WITH OTHER STUDIES

                                  RESULTS INTENDED FOR POLICY DEVELOPMENT ONLY1'2

POLLUTANTS
[WEIGHT OF
EVIDENCE 1
Arsenic [A]
Benzene [A]
Benzo-a-
pyrene [B2]
Cadmium (Bl]
Carbon Tet. [B2]
Chloroform [B2]
Chromium VI [A]
*. 1,2 Dichloro-
«• ethane [B2]
Ethylene
Dibromide [B2J
Ethylene
Oxide [Bl]
Formaldehyde [Bl]
Methylene
Chloride [B2]
Perchloro-
ethylene [C]
POMs [N.A.]

35
NESHAPS COUNTY
STUDY STUDY
6.5E-03
8.0E-03

7.5E-04
5.8E-04 6.0E-06
3.0E-03
1.6E-01 3.6E-03
2.9E-04
1.6E-04
6.8E-03
6.1E-04
l.OE-05
4.6E-04 l.OE-05

AMBIENT
AIR SANTA
QUALITY CLARA KANAWHA
STUDY IEMP IEMP
4.0E-03
1.5E-04 2.0E-04 9.0E-09
2.5E-05
1.5E-03
1.5E-04 4.0E-04
7.7E-05 2.0E-06 2.3E-03
1.4E-03

6.0E-06
2.0E-04 4.6E-03
4.9E-05
<1.0E-05 5.0E-04
1.9C-05 l.OE-05



BALT
STUDY
6.1E-05
3.4E-OS
3.2E-08
2.6E-06
4.8E-08
4.5E-06
7.5E-04
2.3E-07
1.8E-07
6.8E-07
8.7E-06
4.2E-05
3.8E-06
6.3E-04
Trichloro-
 ethylene [B2]
l.OE-04
2.1E-05
2.6E-05
2.0E-05
6.0E-06
                                                                                  1.8E-06
                                                    (continued)

-------
                                                  TABLE V-18 (CONTD.)

                                                         NOTES

   N.A. - Not available.

   1  This study uses conservative estimates of increased  cancer risk from ambient  (i.e.,  outdoor)  exposure  to
      establish priorities among pollutants and sources.   The risk estimates  are  calculated using modelled
      concentrations and EPA unit cancer risk factors.  There is considerable uncertainty  in the estimated
      concentrations, which could either overstate or understate the  true  concentrations  (see Chapter  IV).   Unit
      cancer risk factors combine CAG potency estimates with EPA exposure  assumptions.  The CAG potency estimate
      provide a plausible upper limit to the cancer risk of a ocmpound (see Appendix A); however, the  true value of
      the risk is unknown and may be  as  low as zero.  The  exposure assumptions are  extremely conservative in that
      they assume continuous exposure to outdoor air for 70 years.  Because of the  generally conservative bias in the
      information, it is highly unlikely that the true risks would be as high as  the estimates, and they could be
      considerably lower.

   2  The  unit  cancer risk factors used  to estimate risk vary among the studies.  The reader is directed to each
      report for details.  Differences in modelling assumptions,  which vary across  studies,  can slightly affect
      predictions of risk.
m
o

-------
     Philadelphia Study  flEMPl.  Like the Baltimore IEMP project,
the study used dispersion modelling and ambient air and drinking
water monitoring to examine the extent of toxic pollutant
releases in the Philadelphia metropolitan area.

     Kanawha Vallev Study HEMP1.  A Joint EPA-IEMP/West Virginia
study of chronic exposures to toxic air pollution and other
toxics-related issues in this heavily-industrialized area.'2

     Santa Clara Study flEMPK  A multimedia toxics study
performed by EPA-IBMP in conjunction with the California Bay Area
Air Quality Management District.13

     South Coast Study.  A study of the Los Angeles basin
performed by the California South Coast Air Quality Management
District.  It involved both ambient air monitoring and dispersion
modelling.14

     To compare the pollutant-specific lifetime individual cancer
risks from these preceding studies with the Baltimore IEMP, we
developed estimates of maximum and average increased lifetime
individual cancer risks by pollutant.  We identified the average
increased lifetime individual cancer risk by multiplying
population-weighted average exposure data for each pollutant by
its CAG unit cancer risk factors.  We determined the maximum
increased lifetime individual cancer risk by identifying the
location with the highest concentration of the particular
pollutant and multiplying this number by its unit cancer risk
factor.  This is different from the hotspot exposure analysis
discussed earlier in this chapter, which considered only the
highest lifetime individual cancer risks summed across all
pollutants.
     "Source:  U.S.  EPA, Office of Policy Analysis, Final Report
of the Philadelphia  Integrated Environmental Management Project,
December 1986.

     "Source:  U.S.  EPA, Kanawha Vallev. West Virginia Toxics
Screening Study Report. Environmental Services Division (U.S. EPA
Region 3) and the Regulatory Integration Division, Philadelphia,
Pennsylvania, 1987.

     "Source:  U.S.  EPA, Office of Policy Analysis, Santa Clara
Vallev Integrated Environmental Management Protect;  Revised
Stage One Report. May 30, 1986.

     "Shikiya, D.L.  Chung, E. Nelson, R. Rapport, The Magnitude
of Ambient Air Toxics Impacts from Existing Sources in the South
Coast Air Basin:  1987 Air Quality Management Plan Working Paper.
Revision #31, South Coast Air Quality Management District, El
Monte/ California; 1987.

                               V-51

-------
     For those compounds for which comparison data were
available, average increased lifetime individual cancer risks
were generally in the same range as those predicted in the other
urban studies.  Estimates of maximum increased lifetime
individual cancer risks for individual pollutants, however,
tended to be generally somewhat lower.

     Caution must be employed when comparing estimates of maximum
and average increased lifetime individual cancer risk across the
various studies.  Differences in modelling assumptions regarding
release specifications and receptor placement can significantly
affect predictions of lifetime individual cancer risk.

     b.   Comparison of Alternative Approaches to Estimating
          Risks from the Products of Incomplete Combustion

     In the EPA "Six Month Study," ambient concentrations of PICa
were calculated using the ambient levels of benzo(a)pyrene
(B(a)P) as surrogates for the levels of the semi-volatile or
polycyclic organic (POM) component of the products of incomplete
combustion (PIC).  The study quantified these risks using a unit
cancer risk factor of 0.42 (ug/m3)'1 derived from a review of
existing epidemiological data.  Although there were numerous
limitations with this approach, which are discussed in the  "Six
Month Study," it was the only available tool for exploring PIC
risks at the time the study was completed.  In Table V-19 we
present the estimated annual excess cancer incidence for the
study area and the estimated increased lifetime individual cancer
risks at each of the three hotspot locations using the two
approaches.

     The B(a)P surrogate approach results in over three times
greater estimates of annual excess cancer incidence than the POM
approach used in this study.  On the other hand, the surrogate
approach gives much lower estimates of estimated increased
lifetime individual cancer risks for the three hotspot sites.

     The reason for this seeming contradiction is that the  B(a)P
surrogate approach uses only one unit cancer risk factor for all
emissions of particulates that result from incomplete combustion
of organic matter.  In contrast, the unit risk factors we used
for the semi-volatile fraction of the product of incomplete
combustion in the Baltimore IEMP are specific to types of sources
of combustion products.  As shown in Table IV-5, EPA's ORD  has
found  that the relative carcinogenic potency of POM from
different  source types can differ by over five orders of
magnitude  (x  100,000).  Thus, the risk estimates depend heavily
on the type of combustion sources located in the study area.
Additional analytical work and peer review of both approaches are
required before the results of one approach over the other  can be
supported.
                               V-52

-------
                           TABLE V-19

       COMPARISON OF THE RESULTS OF ALTERNATIVE APPROACHES
         TO ESTIMATING THE RISKS FROM THE SEMI-VOLATILE
        COMPONENT OF  THE  PRODUCTS OF  INCOMPLETE COMBUSTION1

           RESULTS  INTENDED FOR  POLICY DEVELOPMENT ONLY

                              POM             B(a)P SURROGATE
                           APPROACH           APPROACH
  Annual Excess Cancer
    Incidence                     2.05                 6.8

  Estimated Lifetime
    Individual Cancer
    Risk at Hotspot          3.7 x 10'4            7.9 x 10°
    Site 1

  Estimated Lifetime
    Individual Cancer
    Risk at Hotspot          6.3 x 10"4            1.4 x 10"
    Site 2

  Estimated Lifetime
    Individual Cancer
    Risk at Hotspot          6.3 x 10"4            1.3 x 10
    Site 3
1   This study uses conservative estimates of increased cancer
   risk from ambient (i.e., outdoor) exposure to establish
   priorities among pollutants and sources.   The risk estimates
   are calculated using modelled or monitored concentrations and
   EPA unit cancer risk factors.  There is considerable
   uncertainty in the estimated concentrations, which could
   either overstate or understate the true concentrations (see
   Chapter IV).   Unit cancer risk factors combine CAG potency
   estimates with EPA exposure assumptions.   The CAG potency
   estimates provide a plausible upper limit to the cancer risk
   of a compound (see Appendix A); however,  the true value of the
   risk is unknown and may be as low as zero.  The exposure
   assumptions are extremely conservative in that they assume
   continuous exposure to outdoor air for 70 years.  Because of
   the generally conservative bias in the information, it is
   highly unlikely that the true risks would be as high as the
   estimates/ and they could be considerably lower.
                               V-53

-------
     c.  Comparison of Risks in Air and Drinking Water

     In Phase I of the Baltimore IBMP, the potential cancer risks
from ingestion of drinking water were examined.  In particular,
the analysis looked at trihalomethanes (THMs).  The Phase I risk
assessment was performed using average ambient data for finished
drinking water at the Ashburton and Montebello drinking water
treatment plants.  Only exposures from ingestion of drinking
water were modelled; there was no consideration of alternative
exposure pathways, such as inhalation of volatilized compounds
from showering.

     The Phase I results suggested that the cancer risks from
chloroform, a THM, are roughly comparable to those from air
toxics»  the annual excess cancer incidence was estimated at
approximately three cases; average increased lifetime individual
cancer risks were estimated at approximately 1 x 10" .  The Phase
II air toxics results indicated that the estimated annual excess
cancer incidence is somewhat higher, but the average increased
lifetime individual cancer risks are generally in the same order
of magnitude.
                               V-54

-------
VI.   RISK MANAGEMENT STRATEGIES  FOR REDUCING RISKS!
      A DEMONSTRATION

      This chapter demonstrates how two analytical techniques—
cost-effectiveness and benefit-cost analysis—can be used to
identify  control priorities  in the risk management phase of the
IBMP.   In conducting these analyses, we identified (1) control
options,  (2) rough estimates of  control costs,  (3) the
effectiveness of the control options in reducing risks to human
health, and  (4) the  dollar benefits afforded by each control
option in terms of reduced mortality, morbidity, and adverse
welfare effects (e.g., materials damage, visibility, and
agricultural effects).

      In the cost-effectiveness analysis, we ranked control
options and identified the set of control options that would
reduce risk to a specified level at the least cost.  In the
benefit-cost analysis, we identified the control strategy—that
is, the set of control options—that would lead to the greatest
monetary  benefit to  society at a given level of cost.  We
examined  these two approaches because the risk manager may decide
to implement a specified control strategy on the basis of various
factors,  including baseline risks, cost-effectiveness of the
remedial  program, and the economic benefits to society of the
control plan.

     The  results of  these two studies can be used to shed light
on policy questions  relating to risk management that were raised
in Chapter II of this document, namely:

     What are the relative coats and effectiveness of alternative
     control options and control option strategies to reduce
     human health risk from air toxics in the Baltimore area?

     Will the control of carcinogenic emissions lead to reduced
     concern for noncarcinogenic health effects from air toxics?

     Will the control of air toxic emissions also result in lower
     emissions of  criteria air pollutants,  in particular ozone
     precursors, such as volatile organic compounds (VOCs),  and
     particulatea?
                              VI-l

-------
     What are the health and welfare benefits that will accrue
     from reducing emissions of VOCa and participates as part of
     controls to lower air toxics emissions at specific sources?

We emphasize that the results presented in this chapter are
highly uncertain and are included primarily to illustrate
different approaches for establishing risk management priorities.

     This chapter is divided into four sections.  The first
describes the process by which sources in the Baltimore area were
selected for inclusion in our risk management demonstration
project.  The second section describes the control options
identified for the screened sources based on a preliminary
engineering analysis.  It also presents rough estimates of
control costs and control effectiveness from the standpoint of
reducing human health risk and increasing dollar benefits to
society.  The third and fourth sections describe the methodology
and results for the cost-effectiveness and benefit-cost analyses,
respectively.


1.  SELECTION OF SOURCES

     The first step in conducting either a cost-effectiveness or
benefit-cost analysis is to select the sources for which control
options will be evaluated.  An analysis of feasible control
options can be very time consuming/ even at a preliminary level.
Because of the large number of sources and pollutants in the
study area, we focused the analysis on the sources and pollutants
that were likely to contribute most to cancer risk.

     The Ambient Air Toxics Workgroup established criteria to
identify the point and area sources that were included in the
analysis.  A point or area source needed to meet only one of the
screening criteria to be included in the analysis:

     (1)  Contributes at least 1 percent (through rounding) of
          the estimated excess cancer incidence in the study area

     (2)  Projected to pose an average increased lifetime
          individual cancer risk of 5 x 10  or  greater  (through
          rounding) for the target pollutants considered

     (3)  Contributes at least 1 percent (through rounding) to
          the incidence at the refined grid cell of highest
          incidence.

The value for the increased lifetime individual cancer risk
criterion was chosen to ensure—given the uncertainty of the risk
estimates—that point sources with emissions of carcinogens that
result in risks in excess of 1 x 10"5  (a common  de minimis  risk

                               VI-2

-------
level used by various states and EPA programs) would be included
in the analysis.   We also decided to use the 1 percent criterion
to focus the control analysis on those sources that contribute
most to cancer risk.  Although we did not explicitly consider
noncancer effects in selecting sources, we did evaluate the
impact of controls to reduce cancer risks on exposures that lead
to noncancer effects.

     We did not arbitrarily choose these criteria.  Our review of
the data in Chapter V suggested that these criteria represented a
clear break in the importance of sources and pollutants in terms
of overall cancer risk from ambient air toxics.

     we show a comparison of the selected sources and pollutants
with regard to their estimated annual excess cancer incidence in
Table VI-1.  The area source categories discussed in Chapter V
are broken out in greater detail here for the purpose of
identifying control options.  Each of these subcategories also
had to meet at least one of the selection criteria to be
included.  The category of road vehicles comprises four different
types of road vehicles,  Solvent usage is represented by the
major subcategories of solvent use:  degreasing, dry cleaning,
and the catchall category for methylene chloride use,
miscellaneous industrial methylene chloride use.  Heating is
characterized by residential woodstoves, commercial oil.
combustion, residential oil combustion, and residential coal
combustion.  We also chose to include gasoline marketing (that
is, gasoline stations, truck stops, and other commercial
operations involving the sale of gasoline).  This was done
because the AMA had already estimated control costs and VOC
removal efficiencies.

     The sources chosen on the basis of maximum increased
lifetime individual cancer risk are listed in Table VI-2.  Four
of the six point sources, were also selected because they
contribute approximately 1 percent or more to the total estimated
excess cancer incidence.

     The final list of 15 sources and their pollutants for which
controls were identified is presented in Table vi-3.  Constraints
on time and resources prevented us from analyzing control options
for heavy and light duty gasoline road vehicles, Point Source E,
and Point Source F.  However, because this was a demonstration
project, it was unnecessary to include all sources.

     Although the control-options analysis focused on the removal
of toxics, we estimated co-control of other study pollutants as
well.  For example, we assumed that controls to reduce benzene
emissions at Point Source A, would also reduce ambient air
releases of toluene and xylene.  [Only a detailed engineering
examination of the facility could ensure that the control options
for one air toxic would indeed control other discharges at the

                               VI-3

-------
           Table  VI-1
BALTIMORE IBMP AIR TOXICS STUDY
ESTIMATES OF EXCESS ANNUAL CANCER INCIDENCE ,
FOR SOURCES SELECTED FOR ANALYSIS OF CONTROL OPTIONS1

Sources
POINT SOURCES
Point Source A
POM2--coke ovens
Chromium- 6
Benzene
Arsenic
Cadmium
Point Source B
Chromiura-6
Arsenic
Cadmium
Point Source C
Chromium- 6
Arsenic
Point Source D
Chromium- 6
AREA SOURCES
Road vehicles:
Light duty gasoline
POM
Formaldehyde
Benzene
EDB
EDC
Cadmium
Heavy duty gasoline
POM
Formaldehyde
Benzene
EDB
EDC
Current
Annual
Emissions
t kkor/vr 1


863 63

0.20
0.10
0.10

0.28
0.07

0.19
TOTAL3:

vehicles
34.63
!3I:?8
3:18
0.08
vehicles
10!60
3?:SS
1.32
Light duty diesel vehicles
POM— light duty diesel 24.44
Formaldehyde 32.50
Control
Estimated Annual
Excess Cancer
Incidence
rweiaht of Evidence 1

1.18
0.90
8:i§
8.00
.00
N.A.]
'A] J
AJ
'AJ
•Bl]
0.21
0.17
8:8?
A]
A]
si]
0.12
0.11
0.01
A]
A]
0.05
0.05 [A]
1.56

0.72
0.52
Q»o|
0.10
o!oo
N.A.]
Bl] J
A!
B21
B2
Bl
0.35
8:33
0.01
0.00
0.00
[N.A.I
Bl]
Al
B2]
B2
0.12
8:4!
N.A.]
Bl]

Percentage
of Total
Cases
32.59
24.82
4.98
2.77
0.01
0.01
5.77
4.58
0.83
0.36
3.44
3.16
0.28
1.30
1.30
43.10

19.85
14.47
2-S2
0.06
0.05
9.67
l-Q
0.09
o!o2
3.39
3.04
0.35
           (continued)
              VI-4

-------
                      Table VI-1 (continued)

                 BALTIMORE  IEMP AIR TOXICS STUDY

           ESTIMATES OF EXCESS ANNUAL CANCER  INCIDENCE
      FOR SOURCES SELECTED  FOR ANALYSIS OF CONTROL OPTIONS
Annual
Emissions
Sources t kka/vr >
Heavy duty diesel vehicles
POM— heavy duty diesel 59.31
Forma Idehyde 9.60
Solvent usage
" Methylene chloride 172.00
Percnloroethylene 315.00
Trichloroethylene 660.00
-Drycleaning
Perchloroethylene 1 , 825 . 00
-Miscellaneous industrial
Methylene chloride use 753.00
Heating
-Woodstoves
POM 112.00
-Commercial oil combustion
POM— distilled oil 12.60
POM-com. residual oil 14.00
Forma Idehyde 0.60
Chromium- o 0 . 0 0 1
Cadmium 0.51
Arsenic 0.34
-Residential oil combustion
POM— distilled oil 83.60
Formaldehyde 7.80
Chromium-5 0.003
Cadmium 0.33
Arsenic 0.05
-Residential coal
POM 133.30
Cadmium 0.01
Arsenic 0.24
-Gasoline marketing
Benzene 25.20
EDC 2.10
EDB 0.21
Estimated Annual
Excess Cancer
Incidence
fWeioht of Evidence!
0.02
0.01
0.00

Ul
0.01
0.03
O.*03

[N.A. ]
1B1]

ci]
B2]
[C]
0.12
0.12 [B2]


0!03 [N.A.]
J:JJ
oioo
o'.os
8. 10
.07
0.00
ul
0.09
0.05
0.00
0.04
8:81
8:88
N.A. ]
N.A.]
sa]
1],
[N.A.]
|j j
[N.A.]
I"1
I®}
;B2J
Percentage
of Total
Cases
0.42
0.32
0.10

1.76
0.75
3.27
3.27

0.91
0.91
CL36
0.01
0.01
0.83
1.42
2.92
2.06
0.10
0.03
2.40
1.38
0.01
1.01
8:8
8:81
TOTAL RISK SUBJECT TO
   CONTROL ANALYSIS:
3.3 annual cases
1.76             48.57%

     91.7%  of total risk
                              VI-5

-------
                       TABLE VI-1 (CONTD.)

                              NOTES

N.A. = Not Available

1  This study uses conservative estimates of increased cancer
   risk from ambient (i.e., outdoor) exposure to establish
   priorities among pollutants and sources.  The risk estimates
   are calculated using modelled or monitored concentrations and
   EPA unit cancer risk factors.  There is considerable
   uncertainty in the estimated concentrations, which could
   either overstate or understate the true concentrations (see
   Chapter IV).  Unit cancer risk factors combine GAG potency
   estimates with EPA exposure assumptions.  The CAG potency
   estimates provide a plausible upper limit to the cancer risk
   of a compound (see Appendix A); however/ the true value of the
   risk is unknown and may be as low as zero.  The exposure
   assumptions are extremely conservative in that they assume
   continuous exposure to outdoor air for 70 years.  Because of
   the generally conservative bias in the information, it is
   highly unlikely that the true risks would be as high as the
   estimates, and they could be considerably lower.

2  POM - Polycyclic organic matter produced through incomplete
   combustion of fossil fuels.  With the exception of coke oven
   emissions, POM unit risk factors used in the assessment of
   risks have not undergone extensive EPA review and thus are
   subject to change.

3  The numbers shown may not sum to the totals because of
   rounding.
                               VI-6

-------
                         TABLE VI-2

           SOURCES POSING GREATER THAN 5 X 10~6
         INCREASED LIFETIME INDIVIDUAL CANCER RISK1
       Risk Estimates  Are  for Policy  Development Only
                                INCREASED LIFETIME INDIVIDUAL
       SOURCE                          CANCER RISK
  Point Source A                      1.1 x 10~3
  Point Source B                      3.8 x 10"J
  Point Source C                      9.9 x 10~^
  Road vehicles                       7.1 x 10"^
  Point Source E                      3.8 x 10";?
  Heating                             2.4 x 10"|
  Solvent usage                       1.4 x 10~£
  Point Source D                      1.2 x 10~-|
  Point Source F                      6.3 x 10~6
This study uses conservative estimates of increased cancer
risk from ambient (i.e., outdoor) exposure to establish
priorities among pollutants and sources.  The risk estimates
are calculated using modelled or monitored concentrations and
EPA unit cancer risk factors.  There is considerable
uncertainty in the estimated concentrations, which could
either overstate or understate the true concentrations (see
Chapter IV).  Unit cancer risk factors combine CAG potency
estimates with EPA exposure assumptions.  The CAG potency
estimates provide a plausible upper limit to the cancer risk
of a compound (see Appendix A); however, the true value of the
risk is unknown and may be as low as zero.  The exposure
assumptions are extremely conservative in that they assume
continuous exposure to outdoor air for 70 years.  Because of
the generally conservative bias in the information, it is
highly unlikely that the true risks would be as high as the
estimates, and they could be considerably lower.
                           VI-7

-------
         TABLE  VI-3  SOURCES & POLLUTANTS INCLUDED  IN
                        THE  CONTROL ANALYSIS
         Source
Point Sources
Point Source A


Point Source B

Point Source C

Point Source 0

Area Sources
Degreasing (solvent usage)


Dry Cleaning (solvent usage)

Other (misc.) Solvent Usage

Light Duty Diesel Road Vehicles

Heavy Duty Diesel Road Vehicles

Commercial Oil Heating


Residential Oil Heating


Residential Coal Heating



Residential Wood Stoves



Gasoline Marketing*
Pollutants


poiycyctic  organic matter, hexavalent chromium,
benzene, arsenic, cadmium,  toluene, xylerte

hexavalent chromium, nickel, cadmium, arsenic

hexavalent chromium, nickel, arsenic

hexavalent chromium


tnchloroethylene. methylene chloride.
perchloroethylene

perchloroethylene

methylene chloride

polycyclic  organic matter, cadmium,  formaldehyde

polycyclic  organic matter, cadmium, formaldehyde

hexavalent chromium, nickel, formaldehyde
toluene, benzene, polycychc organic matter

hexavalent chromium, nickel, formaldehyde
toluene, benzene, poiycyctic organic matter

arsenic, cadmium,  hexavalent chromium, nickel.
formaldehyde,  toluene, benzene, polycydie organic
matter

polycyclic organic  matter, arsenic, cadmium,
hexavalent chromium, formaldehyde,
benzene,  toluene, nickel, phenol

benzene,  EDC, EDa xylene. toluene, ethyl benzene
  Gasoline Marketing was included in the analysis despite the fact that the source
   contributed less than 1% of the modeled incidence in the study ares.
                                         VI-8

-------
same level of effectiveness.]  In addition, we generated data on
the co-removal of VOCs and TSFs which were used in the benefit-
cost analysis, as described in Section 2 below.


2.   ESTIMATING CONTROL OPTIONS COSTS. EFFECTIVENESS. AND
     BENEFITS

     a.   Feasible Control Options and their Cost-Effectiveness

     We analyzed one or more potentially feasible air pollution
controls for each of the key point and area sources using
engineering estimates.  We then approximated capital and annual
operating costs and the effectiveness in pollutant removal for
each control.  For the cost-effectiveness analysis, the pollutant
removal efficiencies were specific to individual toxic
constituents to permit calculation of the reductions in risk
after control.  For the benefit-cost analysis, the removal
efficiencies were identified for the pollutant categories of VOCs
and particulates to permit calculation of benefits.

     Actual costs and removal efficiencies for a control at a
particular site may differ from these engineering estimates.
Thus, a detailed site-specific analysis is required to confirm
whether an identified opti-   is feasible.  We do not claim that
all chosen control options   3 practical in consideration of the
cost of reducing emissions,   -it only that each option is probably
technically feasible.

     We based the pollution  sntrol analysis on literature
review, engineering judgmer:   discussions with AHA staff, and
telephone conversations wit:, pollution control equipment vendors.
Alternative control technologies were grouped into one or more
appropriate control "options."  In some cases, a control option
is a single removal technology, while in other cases an option
represents a series of controls on a single emission point or
separate controls on multiple release points within a given
facility.  Controls for point sources were identified by first
consulting with AMA staff responsible for monitoring permit
compliance of the respective facilities.  Each State inspector
familiar with a facility was asked to identify control options
that could be applied to reduce toxic emissions.

     Tables VI-4 and VI-5 summarize the options for point and
area sources respectively.  Costs and effectiveness of controls
in reducing emissions were calculated on the basis of the
literature and estimates provided by vendors.  We computed
capital and annual (operating and maintenance) costs in May 1987
dollars.  For Point Source A, we grouped controls by the major
location at the facility where the emissions occur.  We present a
more detailed description of baseline emissions, the
effectiveness of each control option in reducing emissions from

                               VI-9

-------
                           TABLE VI-4 DESCRIPTION of CONTROL OPTIONS for POINT SOURCES
Source
Point Source A








Plant Area
A&B Plants



Benzene/Lit ol
Plant


Storage Tanks
Mills
Release Pt.
Tar decanter, tar intercepting
sump, flushing liquor
circulation tank
Tar storage tanks, tar
dewatering tanks
Excess ammonia liquor
storage tanks
Light oil condensor
Light oil sump
Benzene condensor
Light oil benzene mixture
storage tanks
Benzene storage tanks
Chrome plating line
Option/Alternative
Current Control
Alternative 1
Current Control
Alternative 2
Alternative 3
Current Control
Alternative 4
Alternative 5
Current Control
Alternative 6
Current Control
Alternative 7
Current Control
Alternative 8
Current Control
Alternative 9
Alternative 10
Current Control
Alternative 1 1
Alternative 12
Current control
Alternative 13
Option Description
Uncontrolled
Coke oven gas blanketing system
Uncontrolled
Wash Oil Scrubber
Coke oven gas blanketing system
Uncontrolled
Wash Oil Scrubber
Coke oven gas blanketing system
Uncontrolled
Coke oven gas blanketing system
Uncontrolled
Cover
Uncontrolled
Coke oven gas blanketing system
Uncontrolled
Wash Oil Scrubber
Coke oven gas blanketing system
Uncontrolled
Wash Oil Scrubber
Coke oven gas blanketing system
Packed bed wet scrubber
Packed bed wet scrubber
Efficiency
97
90
97.9
90
97.7
97.9
98.2
97.9
89.9
98.2
90
97.9
50
I
»-•
O

-------
                     TABLE VI-4 DESCRIPTION of CONTROL OPTIONS lor POINT SOURCES (continued)
Source
Point Source A
[continued)



Plant Area
Coke Ovens



Release Pt.
11 and 12 Batteries


A Battery
	 1 	
Option/Alternative
Current Control
Option 1
Option 2
Current Control
Option Description
Meets proposed NESHAP limitations for
coke oven doors, i.e.. 10% leaking
doors, lor 75% control. Charges are
currently every 32 seconds lor 97%
control. Four percent of lids leak.
Construct two new batteries to replace
the existing batteries. The new
batteries wouild be designed to meet
the propuosed NESHAP.
Modify charging to 16 second intervals.
lor 98% control. Unit leaking lids to 3%
ol topside ports through maintenance
and inspection
Meets proposed NESHAP
Efficiency

1.3
13
—
M
I

-------
                            TABLE VI-4 DESCRIPTION of CONTROL OPTIONS for POINT SOURCES (continued)
Source
Point Source B
Point Source C


Point Source D

Plant Area






Release Pt.
Electric arc furnace (E AF)
Argon-oxygen decarbunzation
vessel (ADD)
Electric arc furnace (EAF)
Argon-oxygen decarbunzation
vessel (AOO)
EAF and AOO
Process Line

Option/Alternative
Current Control
Option 1
Current Control
Option 2
Current Control
Option 1
Current Control
Option 2
Option 3
Current Control
Option 1
Option Description
Direct evacuation control, segmented
canopy hood
Scavenger ducts, cross draft partitions,
closed roof
Close Filling Hoods
Segmented canopy hood, scavenger
ducts, cross draft partitions, and
closed roof
Segmented canopy hood, closed roof
Direct evacuation control, scavenger
ducts, and cross draft partitions
Close Fitting Hoods
Segmented canopy hood, scavenger
ducts, and cross draft partitions
Vinyl strip doors to block wind
circulation
Horizontal packed bed wet scrubber,
venturi scrubber
Second packed bed wet scrubber
Efficiency
99%
99%
99%
99%
50%

40%
I
!-•
N>
       * Control alternatives for the A&B Plant, Benzene/Lrtol Plant, and Storage Tanks were grouped into a series of Options as shown in Table VI-6

       The selection of control options and the estimation ol capital and annual costs and removal efficiencies are projections and estimations based
        on discussions with AMA staff, literature review and engineering judgement. Actual costs as well as feasibility and effectiveness could differ
        considerably from our estimates.

-------
                                            TABLE VI-S   DESCRIPTION of CONTROL OPTIONS lor AREA  SOURCES
       Area  Source
Current
 Control
             Option  1
                                                                                            Option 2
                                                Option 3
       DRY CLEANING
  None
Quickly identify and repair leaks

Regenerative filter wastes not to
exceed 25kg of solvent per 100 kg
of wet waste material

Distillation wastes not to exceed
 60 kg of solvent per 100 kg of wet
waste materials

Better housekeeping

Overall control efficiency = 20%
                                                                                Quickly identify and repair leaks

                                                                                Regenerative filler wastes not to
                                                                                exceed 25kg of solvent per 100 kg
                                                                                of wet waste material

                                                                                Distillation wastes not lo exceed
                                                                                60 kg of solvent per 100 kg of wet
                                                                                waste materials

                                                                                Belter housekeeping

                                                                                Carbon adsorption systems for all
                                                                                commercial dry cleaners

                                                                                Overall  control efficiency = 58%
 I
t-«
Ul
                                                                                    Quickly identify and repair leaks

                                                                                    Regenerative filter wastes not lo
                                                                                    exceed 25kg of solvent per  100 kg
                                                                                    of wet waste material

                                                                                    Distillation wastes not to exceed
                                                                                    60 kg of solvent per 10O kg of wet
                                                                                    waste materials

                                                                                    Belter housekeeping

                                                                                    Carbon adsorption systems tor all
                                                                                    coin-op and commercial dry cleaners

                                                                                    Overall control  efficiency  = 65%
       DECREASING
  Nona
Cold Cleaners
Cover during idle time (control
efficiency of 90%)

Dram racks with 30 second dram
and control efficiecy of 50%
during operation of degreaser

Open Top Vapor Degieasers
Cover during ide time (control
efficiency of 90%)

Increase freeboard ratio lo 0 75
dunng operation (control
efficiency of 27%)
Cold Cleaners
Cover dunng idle time (control
efficiency of 90%)

Dram racks with 30 second drain
and control efficiecy of 50%
ddurmg operation of degreaser

Open Top Vapor Degreasers
Cover during ide time (control
efficiency of 90%)

Use refrigerated freeboard device
(control efficiency of 60% for
vaporization losses and 29% of
earn/out losses)
                                                                                                                   Cold Cleaners
                                                                                                                   Cover during idle lime (control
                                                                                                                   efficiency of 90%)

                                                                                                                   Drain racks with 30 second dram
                                                                                                                   and control efficiecy of 50%
                                                                                                                   ddurmg operation of degreaser

                                                                                                                   Open Top Vapor Degreasers
                                                                                                                   Cover during ide time (control
                                                                                                                   efficiency of 90%)

                                                                                                                   Use carbon adsorption unit
                                                                                                                   (control efficiency of 70% for
                                                                                                                   vaporization losses and 3O% of
                                                                                                                   carryoul losses)

-------
                               TABLE VI-S  DESCRIPTION of CONTROL OPTIONS for AREA SOURCES  (continued)
Area Source
OTHER INDUSTRIAL



GAS MARKETING
ROAD VEHICLES
Light Duty Gas
Road Vehicles
Light Duty Diesel
Road Vehicles
Heavy Duty Diesel
Road Vehicles
Current
Control
None



Stage 1
Annual Inspection
None
None
Option 1
Enlarged condensation zone
Waste recovery facility
Manual enclosure
Overall control efficiency 21%
Stage II (vapor balance system at pump
Overall control efficiency 70 5%
Cease Annual Inspection
Catalytic Convenor
Catalytic Convenor
Option 2
Enlarged condensation zone
Waste recovery facility
Automatic enclosure
Overall control efficiency 32 8%
)



Option 3
Enlarged condensation zone
Waste recovery facility
Refrigerated condenser
Overall control efficiency 36 6%




M
I

-------
                                     TABLE VI-S  DESCRIPTION of CONTROL OPTIONS for AREA SOURCES   (continued)
Area Source
HEATING
Residential
Wood Stoves
Commercial Oil
Heating
Residential Oil
Heating
Residential Coal
Heating
Current
Control
None
None
None
None*
Option 1
Stop using wood stoves Instead
use mam fossil fuel heating system
(gas.coal or oil)
Replace oil-lired furnaces with
natural gas systems
Replace oil-firedfurnaces with
natural gas sytems
Replace coal-fired units with
natural gas systems
Option 2

Replace existing burner with dual
burner and bum natural gas


Option 3




U»
^    • The burning of coal is illegal, but does occur
The selection of control options and the estimation oi capital and annual costs and removal efficiencies are projections and estimations based
on discussions with AMA staff, literature review and engineering judgement  Actual costs as well as feasibility and effectiveness could differ
considerably from our estimates

-------
the baseline, and capital and annual costs for each option for
each of the four point sources in Appendix C.1

     We calculated the costs and removal efficiencies of controls
for area sources for the total study area, rather than the
individual facility.  Because of the large number of small
facilities making up each area source category, we assumed that
the frequency of facilities within a source category per unit
population (e.g., dry cleaners) was the same for Baltimore as for
the nation as a whole.  We then identified existing and potential
air pollution controls and estimated removal efficiencies and
associated capital and annual costs on the basis of the
following:  EPA publications, such publications as background
documents for proposed and promulgated standards and control
technology assessment documents, and national data on the
distributions of sources of different sizes.  A summary of the
estimated costs and emission reduction associated with the
selected control options for each area source can be found in
Appendix C.

     The control option costs for both point and area sources do
not take into account the age or usefulness of current equipment
that must be replaced under the option.  For example, a
forty-year old coke oven at Point Source A that has already
outlived its useful life would/ under Control Option #1, be
replaced by a new oven that meets current federal emission
standards.  To the extent that Point Source A would make the
investment in new equipment regardless of the control option or
the control only fractionally reduces the useful life of existing
equipment, we overestimate the actual costs.  Furthermore, we do
not take into account the resale or income from replaced
equipment or scrap.

     Similarly, both commercial facilities and private residences
would, under the respective control options, be expected to
switch from coal and oil combustion to natural gas for heating
purposes.  Yet a certain, probably high, percentage of both
commercial facilities and private residences—given the age of
the property in the study area—have old equipment that must be
replaced regardless of the needs of air quality.  The costs to
these residences of purchasing natural gas-burning equipment
rather than oil or coal is substantially less than the costs we
estimate.

     Thus, our estimates of costs, particularly for the more
expensive options, may be greatly exaggerated.  We could not
     *For more information,  see:  Versar Inc.,  Baltimore  Phase  II
Air Toxics Control Options Analysis, prepared for U.S. EPA,
Office of Policy Analysis, EPA Contract No. 68-01-7053,  November
6, 1987.

                              VI-16

-------
generate more realistic estimates without a much more thorough
analysis.  Yet these cost estimates should suffice for a general
comparison of options and strategies for control and the setting
of priorities in this demonstration project.  They are not
adequate for an accurate and realistic determination of the
financial burden of specific controls.

     Table VI-6 presents a summary of the total annualized costs,
the annual reduction in the estimated annual excess cancer
incidence for each control option, the annualized cost per case
reduced, and the percent reduction in cases expected.
Annualization is the process of spreading the present value of an
expenditure over time considering an appropriate discount rate.
In this analysis, the capital costs of the control equipment were
annualized over the estimated life of the equipment using a 10
percent discount rate.  Total annualized costs were estimated by
adding the annualized capital costs to the annual operating
costs.

     The greatest annual reduction in excess cancer incidence
comes with control option 3 on Point Source B.  The lowest
annualized cost per cancer case reduced—a cost savings—is
roughly $200,000 for control option 2 for reclaiming
trichloroethylene (TCE) from degreasing, which would lead to an
estimated reduction in the estimated annual excess cancer
incidence of 0.014 (an annualized cost per case of approximately
$13 million).  At the opposite end of the spectrum, controls on
the use of oil for residential heating would cost about $150
million annually and would lead to only about 0.045 fewer cases
per year—an annualized cost per case of $3.3 billion.  What
should be clear from Table VI-6 is that no one control option
alone is likely to suffice in reducing total cancer incidence,
should the risk manager deem current predictions of excess cancer
incidence to be unnecessarily high.

     Table VI-6 is an extremely important table for the risk
manager.  By showing the annual cost per case of each option
evaluated, the risk manager can assess the cost-effectiveness of
implementing controls at individual facilities, as well as at a
combination of facilities.  By ordering the cost-effectiveness
ratios for each option from highest to lowest, the risk manager
can develop a strategy for achieving increasing levels of risk
reduction at least cost.  In a more complicated analysis that,
for example, also considered intermedia transfers resulting from
pollution control, it would not be as easy to identify the
optimal control strategy.  Thus, computer modelling is often
utilized to assist with such analyses.  In this demonstration
project, the latter approach was utilized and is discussed in
Section 3 below.
                              VI-17

-------
SOURCE

HD DIESEL VEHICLES
LO DIESEL VEHICLES
POINT SOURCE A-COKE OVENS

POINT SOURCE A
POINT SOURCE A-BENZENE
POINT SOURCE B


POINT SOURCE C
POINT SOURCE D
GASOLINE MARKETING
OEGREASING-PERCHLOROETHYLENE
DEGREAS1NG-TRICHLORQETHYLEME
OEGREASING-HETHYL. CHLORIDE
ORYCLEANING
MISC. IND.-METHYL. CHLORIDE
HOOOSTOVES
COMMERCIAL OIL COMBUSTION

RESIDENTIAL COAL COMBUSTION
RESIDENTIAL OIL COMBUSITON
CONTROL
 OPTION
   1
   1
   1
   2
   1
   1
   2
   3
   4
   5
   1
   2
   3
   1
   2
   3
   4
   1
   1
   I
   2
   3
   1
   2
   3
   1
   1
   2
   3
   1
   2
   3
   1
   1
   2
   1
   1
   CAPITAL
    COST

  $5.941.000
    S956.320
1238,000.000
    $433.000
    {460.000
    S770.000
    WO. 000
    $200.000
    1150,000
    $210,000
     $70,000
    1240,000
    $310.000
    $120,000
    $160.000
    $280,000
      $5,000
     $10,000
  $8,300,000
     $60.000
    $140.000
    $390.000
    $120.000
    $330,000
    $900.000
      $6,000
          $0
  $2,750.000
  $4.470,000
    $390.000
    $830,000.
  $1,210,000
          $0
$163.110.000
 $20.040,000
 $18.100.000
$699,270,000
                                                            TABLE VI  - 6

                                                 CONTROL OPTIONS & ANNUALIZED COSTS  (1.2)

                                                 .CAPITAL     ANNUAL     ANNUALIZED  ANN. CASE   ANNUAL COST  \ REDCTN
                                                   LIFE    OPERATING COST     COST      REDUCTN     PER CASE    IN CASES
10
10
25
25
15
15
15
15
15
15
10
10
10
10
10
10
10
15
10
15
15
15
15
15
15
15

10
10
15
15
15

35
35
35
35
$0
$0
$243,000
$243.000
$200.000
$158.000
$64.000
$4,000
$54,000
$60.000
$20.000
$60.000
$80.000
$30.000
$40.000
$70,000
$200
$60,000
$970,000
($50.000)
($80.000)
$70,000
($130.000)
($220.000)
($220.000)
($50.000)
$110,000
$170.000
$110,000
($120.000)
($130,000)
($130,000)
$220,000
$8,920.000
$8,920,000
$1.300.000
$77.700.000
$966,600
$155,593
$26,423,000
$290.630
$259.800
$259.255
$117.904
$30.290
$73.721
$87,609
$31.392
$99,059
$130,450
$49.529
$66,039
$115,569
$1.014
$61,015
$2,320,0*00
($42.112)
($61.594)
$121.274
($114,223)
($176.614)
($101,674)
($49.211)
$110.000
$277.549
$837,449
($68,726)
($20.877)
$29,082
$220,000
$25.830,000
$10,998,000
$3,180.000
$150.000,000
0.004
0.021
0.080
0.007
0.090
0.073
0.059
0.036
0.009
0.023
0.040
0.154
0.194
0.001
0.004
0.005
0.003
0.019
0.006
0.001
0.002
0.003
0.007
0.014
0.016
0.010
0.006
0.017
0.020
0.025
0.039
0.044
0.033
0.104
0.104
0.096
0.045
$226.370,023
$7,251,007
$332,066,427
$42,207,676
$2.886.667
$3,572.411
$2.001,572
$839.058
$8.230,415
$3,832,894
$792.042
$642,704
$671.434.
$69,340.600
$15,409,100
$23.654.474
$362.143
$3.251,553
$375.057.737
($31.697,204)
($27.567.647)
$47.267.149
($15.677.667)
($12.745.340)
($6,354.625)
($5.065.838)
$18.246.445
$15.924,943
$42,789,365
($2,764.839)
($535,308)
$665,275
$6.579.228
$248.878.183
$105,968,341
$33.095.450
$3,322.784.810
0.1181
0.5929
2.2016
0.1905
2.4901
2.0079
1.6298
0.9988
0.2478
0.6324
1.0966
4.2644
5.3755
0.0198
0.1186
0.1352
0.0775
0.5217
0.1711
0.0368
0.0618
0.0710
0.2016
0.3834
' 0.4427
0.2688
0.1668
0.4822
0.5415
0.6877
1.0791
1.2095
0.9252
2.6715
2.8715
2.6585
1.2490
1  The selection of control options and the estimation of capital and annual costs and removal efficiencies
     are projections and estimations based on discussions with AHA staff, literature review and engineering
     judgement.  Actual costs as well as feasibility and effectiveness could differ considerably from
     our estimates.

2  This study uses conservative estimates of Increased cancer risk from ambient (I.e.. outdoor) exposure to establish
     priorities among pollutants and sources.  The risk estimates are calculated using modelled or monitored
     concentrations and EPA unit cancer risk factors.  There is considerable uncertainty in the estimated
     concentrations, which could either overstate or understate the true concentrations (see Chapter IV).  Unit cancer
     risk factors combine CAG potency estimates with EPA exposure assumptions.  The CAG potency estimates provide
     a plausible upper  limit to the cancer risk of a compound (see Appendix A); however, the true value of the risk
     is unknown and may be as low as zero.  The exposure assumptions are extremely conservative in that they assume
     continuous exposure to outdoor air for 70 years.  Because of the generally conservative bias In the information.
     it is highly unlikely that the true risks would be as high as the estimates, and they could be considerably lower.
                                                           VI-18

-------
      The results in Table VI-6 are presented only as  part of  our
 demonstration project.   There is considerable uncertainty
 surrounding our estimates of emissions,  risks,  and cost.   A more
 detailed analysis would be necessary to  validate  these  results.

      b.    Estimating the Benefits Associated with the Control
           Options

      An  important economic benefit that  can  result from control
 of  toxic air pollutants is the concomitant reduction  in emissions
 of  pollutants that impair visibility and cause smog,  including
 the adverse respiratory health effects associated with  smog.
 These pollutants—photochemically reactive compounds, such as
 VOCs in  general,  that contribute to ozone formation;  total
 suspended particulate (TSP)  matter often emitted  by combustion
 sources;  oxides of sulfur and nitrogen;  and  carbon monoxide (also
 called "criteria pollutants"2)—are frequently controlled by the
 same control technologies designed to reduce health risks
 attributed to toxic air pollutants.   Information  on the changes
 in  health and environmental  effects resulting from reductions in
 emissions and exposure  levels to criteria air pollutants can
 change the relative effectiveness and outcomes  of  control
 strategies directed at  toxic air pollutants.

      To  conduct a benefit-cost  analysis,  we  developed benefits
 estimates for each control option selected to reduce air toxics
 emissions.   The benefits  estimates  required  three  pieces of
 information:   (1)  the effectiveness  of each  control option in
 reducing emissions of particulates  and VOCs,  (2) the benefits
 categories related to particulates  and VOCs  that are of interest,
 and  (3)  an estimate of  the economic  value for each  ton  of
 particulate and VOC reduced  by  benefit category.  We used  the
 data  presented in Appendix C on the  effectiveness of control
 options  for air toxics  emissions  in  reducing  emissions  of
 particulates  and  VOCs.  We based  our  analysis of economic
 benefits  on the quantified health and welfare effects listed  in
 Table  VI-7.   Appendix D provides  more detail  on the benefits
 calculations  and  the sources of  the  benefits  numbers by category.
 The  procedure we  followed to estimate benefits required several
 simplifying assumptions:

      •    For the control options analyzed, a reduction of
           1  percent in  total particulate  emissions translates
           into a  1  percent reduction  in ambient concentrations of
           particulate matter.
     2 They are called criteria pollutants because they serve to
define ambient air quality under the Federal Clean Air Act.

                              VI-19

-------
                            TABLE VI-7

              Health and Welfare Effects  Quantified
Pollutant

Total suspended
 particulate matter
Effects

Mortality
 - Carcinogenic
 - Noncarcinogenic
   (Adverse Respiratory
   Effects)

Morbidity
 -  Acute Effects
    (Work Loss Days,
    Restricted Activity
    Days, Medical Costs)

Welfare
 -  Material Effects
 -  Visibility Effects
Volatile organic compounds
Mortality
    Carcinogenic

Morbidity
 - Acute Effects
   (Work Loss Days,
   Restricted Activity
   Days, Medical Costs,
   Eye Irritation, Coughing,
   Chest Discomfort, Shortness
   of Breath, Headaches)

 - Chronic Effects
   (Respiratory Disease)

Welfare
 -  Agricultural Effects
 -  Material Effects
 -  Ornamental Plant Effects
                              VI-20

-------
     •     For the control options analyzed,  a reduction of 1
           percent in VOC emissions  results  in a  0.6  percent
           reduction in ambient air concentrations of  ozone.

      •    Reductions in the ambient air concentrations  of
           particulate matter and ozone are evenly distributed
           over the Baltimore study area.

      We  estimated the expected economic benefits  that would
 result from reducing one ton of emissions for both particulate
 matter and volatile organic compounds  using  the results of
 readily  available statistical analyses.  These statistical
 studies  provided a quantitative estimate of  the relationship
 between  changes in health and welfare  effects and predicted
 changes  in the levels of ambient concentrations of particulates
 and  ozone.

      To  arrive at an estimate of the economic benefits  of  each
 control  option,  we multiplied the total  reductions in emissions
 of particulates and VOCs that each control option would provide
 by our estimates of economic benefit per ton.   The results of
 this  exercise are summarized in Table  VI-8.   The  expected
 economic benefits from reducing emissions of  particulate matter
 are about  $14,000 per ton reduced.   The  expected  benefits  of
 reducing ozone levels resulting from controls on  the  emissions of
 VOCs  are $345 per ton reduced.   Both of  these estimates of
 benefits fall within the ranges of  the estimated  economic
 benefits of  reducing TSP and ozone  concentrations  performed for
 other cities  under conditions similar  to those modelled in
 Baltimore.
3.  ANALYSIS OF THE COST-EFFECTIVENESS OF CONTROL STRATEGIES

     In this section, we present the results of our analysis of
cost-effective-strategies for reducing risks from ambient air
toxics and the results.  Figure VI-1 illustrates the key pieces
of information required to conduct a cost-effectiveness analysis.
The information necessary for the activities represented in the
boxes in the left-hand side of the diagram was generated in the
risk assessment phase of the study.  The activities represented
on the right-hand side were conducted once key sources and
pollutants posing risk had been identified, as described above.
We used a computer (mixed integer programming) model to select
those options that would provide a given percent reduction in
risk at the lowest cost for the sources and pollutants examined.

     a.  Cost-Effectiveness Analysis Overview

     For the purposes of our study, we defined cost-effectiveness
as the cost of control per unit of health risk reduced.  We
defined effectiveness as the percent reduction in either the

                              VI-21

-------
                                                                  TABLE VI-B

                                         CONTROL OPTIONS, ANNUALIZEO BENEFITS.  AND COSTS FOR TSP AND VOC CONTROLS   U.2)
ANNUAL ANNUAL
VOC TSP ANNUAL
CONTROL REDUCED REDUCED REDUCED
SOURCE Tin OPTION EMISSIONS EMISSIONS CAHCERS
(KKC/TR) (KKC/YR)


















M
1
to
to

















HEAVY-DUTY DIESEL VEHICLES
LIGHT-DUTY DIESEL VEHICLES
POINT SOURCE A - COKE OVENS

POINT SOURCE A - METALS
POINT SOURCE A - BENZENE




POINT SOURCE B - METALS


POINT SOURCE C - METALS



POINT SOURCE C
GASOLINE MARKETING
DECREASINC-PERCBLOROETHYLENE


DECREAS1NG-TRICHLOROETHYLENE


DEGREASING-METHYLEHE. CHLORIDE
DRYCLEAN1NC-PERCBLOROBTHYLENE


MISC. INDUSTRIAL-METHYL. CHLORIDE


WOODSTOVES-RESIDENTIAL HEATING
COMMERCIAL OIL COMBUSTION

RESIDENTIAL COAL COMBUSTION
RESIDENTIAL OIL COMBUSTION
1119
309
0
0
0
648
522
299
81
223
0
0
0
0
0
0
0
0
2955
80
135
155
155
290
335
62
368
1061
1188
158
248
278
93.6
-3.5
-5.3
146.12
0.4



20
12.5
77
7
0
0
0
0
0
0
39
34
72
a
7
15
17
0.94
0
0
0
0
0
0
0
0
0
0
0
0
0
0
148
71.09
71.09
534.94
213.9



0.004
0.021
0.080
0.007
0.090
0.073
0.059
0.036
0.009
0.023
0.040
0.154
0.194
0.001
0.004
0.005
0.003
0.019
0.006
O.OOI
0.002
0.003
0.007
0.014
0.016
0.010
0.006
0.017
0.020
0.025
0.039
0.044
0.033
0.104
0.104
0.096
0.045



ANNUAL
CANCER CONTROL
BENEFITS
(1987 $)
$8,540
$42,857
$159,143
$13,771
$180,000
$145,143
$117,811
$72,200
$17,914
$45,714
$79,269
$308,257
$388,571
$1,429
$8,571
$9.771
$5,600
$37,714
$12.371
$2,657
$4,469
$5.131
$14,571
$27.714
$32,000
$19,429
$12,057
$34,857
$39,143
$49,714
$78.000
$87.429
$66,877
$207,571
$207,571
$192,171
$90,286
VALUE
PER REDUCED
CAHCERS
TOTAL
ANNUAL TSP
BENEFITS
(1987 $)
$274,000
$171,250
$1.054,900
$95,900
$0
$0
$0
$0
$0
$0
$534,300
$465,800
$986,400
$109,600
$95,900
$205,500
$232.900
$12.878
$0
$0
$0
$0
$0
$0
$0
$0
$0
$0
$0
$0
$0
$0
$2.027,600
$973,933
$973,933
$7.328,678
$2,930,430
TSP
BENEFIT
PER TON
TOTAL
ANNUAL VOC
BENEFITS
(1987 $>
$386,055
$106,605
$0
$0
$0
$223,560
$180,090
$103,155
$27,945
$76,935
$0
$0
$0
$0
$0
$0
$0
SO
$1,019,475
$27,600
$46,575
$53,475
$53,475
$100,050
$115,575
$21.390
$126,960
$366,045
$409,860
$54,510
$85,560
$95,910
$12,292
($1,898)
($1,898)
$50,411
$138
VOC
BENEFIT
PER TON
TOTAL
ANNUAL
BENEFITS
(1987 $)
$668,595
$320,712
$1,214,043
$109.671
$180,000
$368.703
$297,901
$175,355
$45,859
$122,649
$613,569
$774,057
$1,374.971
$111.029
$104,471
$215,271
$238,500
$50,592
$1,031,846
$30,257
$51,044
$58,606
$68,046
$127,764
$147,575
$40,819
$139,017
$400,902
$449,003
$104,224
$163,560
$183,339
$2,126.769
$1,179,607
$1,179,607
$7,571,261
$3,020,834



TOTAL
ANNUAL
CONTROL
COSTS
(1987 $)
$966,600
$155.593
$26,423,000
$290,630
$259,800
$259,255
$117,904
$30,290
$73,721
$87,609
$31,392
$99,059
$130,450
$49,529
$66.039
$115,569
$1,014
$61,315
$2.320,000
($42,112)
($61,594)
$121.274
($114,223)
($176.614)
($101,674)
($49,211)
$110.000
$277,549
$837,449
($68,726)
($20,877)
$29;082
$220,000
$25,830.000
$10,998,000
$3,180,000
$150,000,000



TOTAL
ANNUAL
NET BENEFITS
(1987 $)
($298,003)
$165.119
($25.208.957)
($180,959)
($79.800)
$109.448
$179,997
$145,065
($27,862)
$35,040
$582,177
$674.998
$1,244,521
$61 , 500
$38,432
$99.702
$237,486
($10,723)
($1,288,154)
$72,369
$112.638
($62,668)
$182,269
$304,378
$249.249
$90,030
$29,017
$123.351
($388,446)
$172,950
$184,437
$134,257
$1,906,769
($24,650,193)
($9,818,393)
$4.391,261
($146.979.146)



                                                                           $2,000,000    $13700
$345
1.   The selection of control options and the  estimation of capital  and annual coses and removal efficiencies are projections
      and estimations based on discussions with AHA staff, literature review and engineering  Judgment.,  Actual costs as well as

-------
                                                                   TABLE Vl-B (continued)


    2.  This study usea conservative  estimates of increased cancer risk fron ambient (i.e., outdoor) exposure to establish
       priorities among poilutanta and  sources.  The risk estimates are calculated using nodelled or monitored concentrations
       and EPA unit cancer risk factors.  There is considerable uncertainty in the estimated concent radons, which could either
       overstate or understate the true concentrations, (see Chapter IV).  Unit cancer risk factors combine CAG potency estimates
       with EPA exposure assumptions.   The  CAC potency estimates provide a plausible upper limit to the cancer risk of a compound
        (see Appendix A); however, the true  value of the risk is unknown and may be as low aa gero.  The exposure assumptions are
       extremely conservative ta chat they  assume continuous exposure to outdoor air for 10 years.  Because of the generally
       conservative bias in the Information, it is highly unlikely that the true risks would be as high as the estimates, and
       they could be considerably lower.
<
M
(O

-------
                           Figure VI -1
          Baltimore IEMP  Air Toxics  Study
        Analysis  Of The  Cost-Effectiveness
                    Of  Control Options 1

                     Information  Flow
                              Sources
                          (1987 control levels)
                           • Point   • Area
  Pollutant Release
                       Feasible
                     Control Option
  Exposure Pathway

 Dispersion Modeling
                     Control Costs
 Human Health Effects
(Baseline Cancer Risks)
  • annual excess cancer
  • maximum lifetime
    individual cancer nsk
Computer
 Model
                  Control Effectiveness
                    (Percent removal
                       efficiency)
                     Cost-Effective Control Strategies
                  •Cost of the reduction in annual excess
                   cancer incidence
                  •Cost of the reduction in maximum
                   lifetime individual cancer risk
      Cost-effecfivness is defined as the cost of control per unit of health nsk reduced.
                             VI-24

-------
annual excess cancer incidence or the maximum increased lifetime
individual cancer risk.  It is important to note that this
analysis 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 provide
information useful to local risk managers in setting control
priorities.  In particular this analysis took an approach that
allows the risk manager to determine (1) which individual control
options are most cost-effective, and (2) the cheapest set (or
combination) of control options from for all sources included in
the analysis that will achieve any predetermined percentage risk
reduction level.  We used a computerized optimization technique,
known as a mixed integer program/ to identify the least-cost
control strategy to meet several specified risk reduction levels,

     We chose the different percent reductions in risk to human
health to identify a series of control strategies for different
levels of risk reduction.  As we requested progressively higher
levels of risk reduction, the computer model reconfigured the set
of control options (above current controls) to determine the
optimum strategy to provide the desired level of risk reduction
at the least cost.  We employed this strategy because we did not
know the risk manager's preferences ahead of time.  Comparing
costs to the percent reduction in cancer risk (above current
controls) aids the risk manager in deciding how much risk
reduction to target.

     The results of the cost-effectiveness analysis are provided
below for two measures of human health risk:  annual excess
cancer incidence (both area-wide and for the grid cell of maximum
annual excess cancer incidence), and maximum increased lifetime
individual cancer risk.  In some situations, the risk manager may
be faced with a need to implement two different control
strategies to reduce both measures of risk.  In our discussion
that follows/ we explore whether this situation occurs in the
Baltimore area.  We also assess qualitatively how the different
control strategies will affect noncancer effects.  This section
concludes by summarizing the major limitations of the analysis.

     b.  Coat-Effectiveness Analysis Results

          i. Results;  Annual Excess Cancer Incidence

                (1)  Area-Wide Control Options

     Individual Options.  Table VI-6 lists the cost-effectiveness
(i.e./ the annual cost per cancer case avoided) for all control
options evaluated in this analysis.  As shown/ the options range
from a net annual saving of about $32 million per case avoided
from perchloroethylene degreasing (control option 1) to a maximum

                              vi-25

-------
of over $3 billion from residential oil combustion (control
option 1).  By inspecting the cost-effectiveness of each option,
in conjunction with a policy determination of risk reduction
goals, the decision maker can explore where additional control
may be warranted.  For example, of the individual options to
reduce at least annual excess cancer incidence over ten years,
control option 2 at Point Source B ranks best at annualized cost
of approximately $643,000.  It is also interesting to note the
numerous control options that could result in a savings while
reducing cancer risk.

     Control Strategies.  Using our automated model, we were able
to ranking and compile options that would achieve various
predetermined risk reduction levels.  Table VI-9 shows the least-
cost control strategies for reducing different levels of the
estimated upper-bound annual excess cancer incidence posed by the
target pollutants.  This table also displays the total cost,
average cost per case reduced, and incremental cost per case
reduced for each control strategy.  As can be seen from this
table as well as from the graph in Figure VI-2, the total
annualized cost of implementing the control strategies associated
with each of the points A through K ranges from a low of negative
$200,000 (that is, a cost savings) to a high of almost $211
million for the controls considered.

     We estimate that we can achieve a modest reduction in the
estimated annual excess cancer incidence (by roughly 7 percent)
and save money by implementing solvent recovery programs for area
sources emitting perchloroethylene, TCE, and methylene chloride
(again at cost savings), and controls at Point Sources B and C at
modest cost.  The cost savings for solvent usage are the result
of controls that lead to the recovery of solvent and reduced
solvent consumption.

     Roughly one excess cancer case every two years (see point 0)
could be avoided at a cost of $1.4 million per year (that is,
about $2.8 million per case avoided).  This is within the range
of values often given by economists for the value of a
statistical life saved ($400,000 to $7,000,000).3

     The cost per excess cancer case avoided for a given set of
control strategies increases with increasing requirements on risk
reduction.  The costs begin to rise rapidly with the reduction of
0.51 cancer cases per year at point D (a 14 percent reduction
from baseline risk) because the incremental (or marginal) cost
for each additional case reduced begins to rise sharply.
Controls are required on all of the point sources and most of the
area sources.  The jump in cost is explained by the selection of
     3 U.S.  EPA,  "Regulatory Impact  Guidelines,"  December 1983,
EPA 230-01-84003.

                              VI-26

-------
                         Table Vl-9
              BALTIMORE  IEMP AIR TOXICS STUDY
SCHEDULE OF CONTROL STRATEGIES FOR  REDUCING CANCER INCIDENCE
    PHASE II  RESULTS  INTENDED FOR POLICY DEVELOPMENT1-2

Point
on Cases
Figure VI-2 Reduced
Graph per Year
A 0.25






B 0.27








C 0.34








D 0.51














Percent Avg. Cost per Incremental
Reduction in Total Case Reduced Cost per
Cancer Aromatized from Current Incremental Pollution Controls Implemented
Incidence from Cost Control Case Reduced




Current Control (tlOOO/vearl (11000/case) (SlOQO/case) Source Control Potion
7 -{195 -J780 $39,000 Point Source A-benzene controls
Point Source 6
Perc degreasing
TCE degreasing
Methylene chloride degreasing
Misc. industrial
methylene chloride use
7.5 J73 $268 {13,400 Point Source A-benzene controls
Point Source B
Point Source C
Point Source 0
Perc degreasing
TCE degreasing
Methylene chlonde-degreasing
Misc. industrial
methylene chloride use
9.5 $183.4 J534.9 $1,572 Point Source A-chrone controls
Point Source A-benzene controls
Point Source B
Point Source C
Perc degreasing
TCE degreasing
Methylene degreasing
Misc. industrial
methylene chloride use
14 Jl. 416.2 JZ.800.4 $7,573 Light duty diesel vehicles
Point Source A-coke ovens
controls
Point Source A-chrome controls
Point Source A-benzene controls
Point Source B
Point Source C
Point Source D
Perc degreasing
TCE degreasing
Methylene chloride degreasing
Dryc lean ing
Misc. industrial
methylene chloride use
Woodstoves
3
3
2
2
1
1

2
3
4
1
2
2
1
1

1
3
3
4
2
2
1
1

L

2
1
1
3
I
1
2
2
1
2
2

L
                           VI-27

-------
                                                 Table  Vl-9  (continued)

                                            BALTIMORE IEMP AIR TOXICS STUDY

                                    PHASE II RESULTS INTENDED  FOR POLICY DEVELOPMENT

                             SCHEDULE OF CONTROL STRATEGIES FOR REDUCING CANCER INCIDENCE1'2
  Point
   on
Figure VI-2
   Graph
  Cases
 Reduced
per Year

   0.69
    Percent
Reduction in
    Cancer
Incidence from
Current Control

       19
    Total
 Annualized
    Cost
fllQOO/vear)

$7.730
Aug. Cost per Incremental
Case Reduced    Cost per
from Current  Incremental    Pollution Controls Implemented
   Control    Case Reduced   	
 (SlOOO/case)  (tlOOO/case)  Source
                               Control  Option
                                                           $11.273
$39.463
                            Light duty diesel vehicles        1
                            Point Source A-coke oven controls  2
                            Point Source A-chrome controls    1
                            Point Source A-benzene controls   1
                            Point Source B                    3
                            Point Source C                    <
                            Point Source 0                    1
                            Gasoline marketing                1
                            Perc degrees ing                   2
                            TCE degreasing                    2
                            Methylene chloride degreasing     1
                            Dryc lean ing                       3
                            Misc. industrial                  2
                                methylene chloride use
                            woodstoves                        1
                            Residential coal                  1
                 0.72
                    20
                                               $15.592
                                               $21.570
                               $211.344
                                                           Light  duty diesel  vehicles        1
                                                           Point  Source A-coke oven  controls 2
                                                           Point  Source A-chrome  controls
                                                           Point  Source A-benzene controls
                                                           Point  Source B
                                                           Point  Source C
                                                           Point  Source D
                                                           Perc degreasing
                                                           TCE degreasing
                                                           Methylene chloride degreasing
                                                           Dryc lean ing
                                                           Misc.  industrial
                                                               methylene chloride use
                                                           Woodstoves
                                                           Commercial oil controls
                                                           Residential coal
                 0.76
                   21.0
                  $45,084
                                                            $59.433
                                $826.106  Heavy duty diesel vehicles
                                          Light duty diesel vehicles
                                                              1

                             PoTnt Source A-coke oven controls 1
                             Point Source A-chrome controls    1
                             Point Source A-benzene controls   1
                             Point Source B                   3
                             Point Source C                   2
                             Point Source 0                   1
                             Gasoline marketing               1
                             Perc degreasing                  2
                             TCE degreasing                   3
                             Methylene chloride  degreasing    1
                             Drycleaning                      2
                             Misc. industrial                 2
                                  methylene chloride use
                             Woodstoves                       1
                             Conmercial oil controls          2
                             Residential coal                 1
                                                           VI-28

-------
                                                 Table VI-9 (continued)

                                            BALTIMORE IEMP AIR TOXICS STUDY

                                    PHASE II RESULTS INTENDED  FOR POLICY DEVELOPMENT

                              SCHEDULE OF CONTROL STRATEGIES FOR REDUCING CANCER INCIDENCE2
                               Percent                    Avg.  Cost per Incremental
  Point                    Reduction  in         Total     Case Reduced    Cost per
   on           Cases         Cancer        Annualized   from Current  Incremental   Pollution Controls Implemented
Figure VI-2    Reduced     Incidence  from       Cost         Control    Case Reduced
   Graph      per Year     Current Control  (SlOOO/vear)   (SlOOO/case)  (SlOOO/case) SourceControl Option

     H          0.80            22.0     199.976       $130.000        $1.479.569   Heavy duty diesel vehicles        L
                                                                                    Light duty diesel vehicles        1
                                                                                    Point Source A-coke oven controls 1
                                                                                    Point Source A-chrome controls    1
                                                                                    Point Source A-benzene controls   1
                                                                                    Point Source B                    3
                                                                                    Point Source C                    3
                                                                                    Point Source 0                    I
                                                                                    Gasoline marketing                1
                                                                                    Perc degreasing                   3
                                                                                    TCE decreasing                    3
                                                                                    Methylene chloride  degreasing    1
                                                                                    Dryclean ing                       3
                                                                                    Misc. industrial                  2
                                                                                        methylene chloride use
                                                                                    Woodstoves                        1
                                                                                    Conrercial oil controls           2
                                                                                    Residential coal                  1

     I          0.83            23.0    S191.800       1230.000        $2.572.101   Light duty diesel vehicles        1
                                                                                    Point Source A-coke oven controls 1
                                                                                    Point Source A-chrome controls    1
                                                                                    Point Source A-benzene controls   1
                                                                                    Point Source B                    3
                                                                                    Point Source C                    2
                                                                                    Point Source 0                    1
                                                                                    Perc degreasing                   2
                                                                                    TCE degreasing                    3
                                                                                    Methylene chloride degreasing     1
                                                                                    Drycleaning                       2
                                                                                    Misc. industrial                  2
                                                                                        methylene chloride use
                                                                                    Woodstoves                        1
                                                                                    Comnercial oil controls           2
                                                                                    Residential coal
                                                                                    Residential oil

     J          0.84            23.3    $195,876       $230,000       $407,600       Heavy duty diesel vehicles
                                                                                    Light duty diesel vehicles
                                                                                    Point Source A-coke oven controls
                                                                                    Point Source A-chrome controls
                                                                                    Point Source A-benzene controls
                                                                                    Point Source B
                                                                                    Point Source C
                                                                                    Point Source 0
                                                                                    Gasoline marketing
                                                                                    Perc degreasing                   3
                                                                                    TCE degreasing                    3
                                                                                    Methylene chloride degreasing     1
                                                                                    Orycleaning                       3
                                                                                    Misc. industrial                  2
                                                                                        methylene chloride use
                                                                                    Woodstoves                        1
                                                                                    Comnercial oil controls           2
                                                                                    Residential coal                  1
                                                                                    Residential oil                   1
                                                          VI-29

-------
                                                 Table VI-9 (continued)

                                             BALTIMORE  IEHP AIR TOXICS STUDY

                                    PHASE II RESULTS INTENDED FOR POLICY DEVELOPMENT

                              SCHEDULE OF CONTROL STRATEGIES FOR REDUCING CANCER INCIDENCE1'2
  Point
   on
Figure VI-2
   Graph
  Cases
 Reduced
per Year
                0.85
    Percent
Reduction in
    Cancer
Incidence from
Current Control
    Total
 Annualized
    Cost
(SlOOO/vear)
Avg. Cost per Incremental
Case Reduced    Cost per
from Current  Incremental    Pollution Controls Implemented
   Control    Case Reduced
 HlOOO/case)  HlOOO/case)  Source
Control Option
                   23.5     5210,916
                             1250.000         (1.744.186   Heavy duty diesel vehicles        1
                                                          Light duty diesel vehicles        1
                                                          Point Source A-coke oven controls 1
                                                          Point Source A-chrome controls    1
                                                          Point Source A-benzene controls   1
                                                          Point Source B                    3
                                                          Point Source C                    3
                                                          Point Source D                    1
                                                          Gasoline marketing                1
                                                          Perc degreasing                   3
                                                          TCE degreasing                    3
                                                          Methylene chloride degreasing     1
                                                          Drycleaning                       3
                                                          Misc. industrial                  3
                                                              methylene chloride use
                                                          Woodstoves                        1
                                                          Comnercial oil controls           1
                                                          Residential coal                  1
                                                          Residential oil                   1
   This study uses conservative estimates of  increased  cancer risk from ambient (i.e.. outdoor) exposure to establish
   priorities among pollutants and sources.   The risk estimates are calculated using modelled or monitored concentrations
   and EPA unit cancer risk factors.   There  is considerable uncertainty in the estimated concentrations, which could either
   overstate or understate the true concentrations  (see Chapter IV).  Unit cancer risk factors combine CAG potency
   estimates with EPA exposure assumptions.   The CAG potency estimates provide a plausible upper limit to the cancer risk
   of a compound (see Appendix A); however,  the true value of the risk is unknown and may be as low as zero.  The exposure
   assumptions are extremely conservative in  that they  assume continuous exposure to outdoor air for 70 years.  Because of
   the generally conservative bias in the information,  it  is highly unlikely that the true risks would be as high as the
   estimates, and they could be considerably  lower.

   Cost estimates were based on engineering estimates and best professional judgment and may significantly over- or
   uhderstimate real costs.
                                                          VI-30

-------
               FIGURE VI-2

TOTAL ANNUALIZED COST VERSUS REDUCTION IN
    ANNUAL EXCESS CANCER INCIDENCE




i_
-------
area source controls, such as the control options for woodstoves
and dry cleaners, and controls at Point Source D that are costly
given the amount of risk reduction that they afford.

     Overall, we can identify control strategies that can reduce
only about 24 percent of the estimated cancer incidence in the
Baltimore area.  Implementation of the most expensive strategy
identified would reduce the estimated annual excess cancer
incidence from approximately 3.6 to only about 2.8.*

     The data suggest that any cost-effective control strategy
would need to include controls on area-wide uses of solvents and
relatively less expensive controls on emissions from Point
Sources A, B, and C.  To achieve any modest reductions in risk
(14 to 19 percent reduction from current baseline levels posed by
the study pollutants), we would also need to impose controls on
most area sources, such as light-duty trucks, dry-cleaning
establishments, and the use of woodstoves for home heating.
Significant reductions in incidence (greater than 20 percent)
would require relatively expensive controls on all the point
sources included in the analysis, as well as controls on area
sources.

     The results in Table VI-9 are presented only as part of our
demonstration project.  There is considerable uncertainty
surrounding our estimates of emissions, risks/ and cost.  A more
detailed analysis would be necessary to validate these results.

               (2)  Control Strategies for the Grid Cell of
                    Maximum Annual Excess Cancer Incidence

     We also explored the impact that each of the strategies
identified in Table VI-9 would have on the grid cell of maximum
incidence—the location of the highest population-weighted
average individual risk.  As discussed in the previous chapter,
the estimated annual excess cancer incidence in this section of
the study area is approximately 0.15, with three sources
accounting for most of the quantified incidence (approximately 91
percent):  Point Source B, Point Source A, and road vehicles.
Metals emissions at Point Source B account for 58 percent of the
total;  POM releases from Point Source B and road vehicles
accounts for the remainder.

     At all points, the combination of control options would
reduce the annual excess cancer incidence from 0.15 to 0.06 (a
percentage reduction of approximately 43 percent) by controlling
emissions of metals, primarily hexavalent chromium, from Point
     'Transportation control measures  were not  included in the
analysis.  Transportation control measures seek to reduce
emissions by reducing single-passenger automotive use.

                              Vl-32

-------
Source B (Option 3).  Although the more costly control strategies
also include controls for Point Source A and light and heavy duty
diesel vehicles (a subset of road vehicles), these controls would
not result in significant reductions in POM ambient air releases.

     The results for the grid cell of highest predicted annual
excess cancer incidence are presented only as part of our
demonstration project.  There is considerable uncertainty
surrounding our estimates of emissions, risks, and cost.  A more
detailed analysis would be necessary to validate these results.

               (3)  Noncarcinogenic Pollutants

     Although we did not explicitly evaluate control options to
reduce the emissions of noncarcinogenic pollutants, we did
explore the extent to which our control strategies to reduce
annual excess cancer incidence also lower the concern for
noncancer effects.  In the previous chapter, we indicated a need
to further investigate ambient air concentrations at two
locations because of a concern for noncancer effects.  At one
location, identified using the refined 2.5 km grid system (UTM
coordinates 4342.25 369.75), benzene exposures exceeded the no-
effect threshold for blood effects, thus indicated a need to
further explore these exposure levels.  Point Source A accounted
for 94 percent of the benzene ambient air concentrations at this
location.  All control strategies selected at points D and higher
on Table VI-9 include Option 1 for benzene at Point Source A;
Option 1 has a benzene removal efficiency of approximately 97.7
percent.  If Option 1 were implemented at Point Source A, the
ratio of benzene ambient concentrations to the no-effect
threshold for blood effects would drop from approximately 5.1 to
0.4, thus indicating that there is less need to investigate these
exposures.

     If the control strategies associated with points A through C
on Table VI-9 were implemented, the need to further examine
benzene exposures would still exist.  Although some benzene
controls at Point Source A are part of the control strategies at
points A through C/ the recommended control strategies include
either Option 3, which has a removal efficiency for benzene of
only 50.5 percent, or Option 2, which has a benzene removal
efficiency of approximately 79 percent.  Option 3 would result in
a decrease in the ratio of benzene ambient air concentrations to
the no-effect threshold for blood effects from approximately 5.1
to 2.7.  A ratio value of one or more indicates a need to further
examine these exposures.  Similarly, Option 2 would result in a
ratio above 1.0.

     At the second highest location of maximum increased lifetime
individual cancer risk, the pollutant of concern was xylene from
Point Source E.  None of the control strategies considered Point
Source E.  Thus the results of our analysis cannot be used to

                              VI-33

-------
identify strategies for reducing xylene exposures at this
location.

     These results are presented only as part of our
demonstration project.  There is considerable uncertainty
surrounding our estimates of emissions, risks, and cost.  A more
detailed analysis would be necessary to validate these results.

          ii.  Results;  Maximum Increased Lifetime Individual
               Cancer Risk

     For the control options analysis, we limited our evaluation
of cost-effective control options to the site within the
Baltimore study area with the maximum increased lifetime cancer
risk.  As discussed in the previous chapter, the maximum
increased lifetime individual cancer risk posed by the study
pollutants at this location is approximately 1.3 x 10" .  Similar
to other risk findings, Point Source A accounts for approximately
91 percent of the estimated maximum increased lifetime individual
risk at this site.

     Table VI-10 arrays the least-cost control strategies for
reducing the estimated increased lifetime individual cancer risk.
This table also displays the total annualized cost and
incremental cost (that is, the fraction of the total cost
expended to move from one risk reduction scenario to the next)
for each control strategy.  Figure Vl-3 graphically portrays the
total cost of achieving specified levels of individual risk
reduction.

     Unlike the situation for reducing annual excess cancer
incidence posed by the study pollutants, few strategies exist  for
reducing the maximum increased lifetime individual cancer risks
at this maximum increased lifetime individual cancer risk site.
Point Source A dominates because (1) it is the largest
contributor to the maximum increased lifetime individual cancer
risk and (2) the potential control costs at this source would  be
significant.  [Note, however, that the costs of control are
likely to be overestimated to the extent that construction of  new
equipment has a utility to this producer separate from its value
in reducing pollutant emissions,  we were not able to factor in
the savings to Point Source A resulting from the greater
production efficiency afforded by the control options.]

     Implementation of control strategy A, shown as point A on
Figure VI-3, would result in a very minor (3 percent) reduction
in risk at possible cost-savings.  Control strategy A combines
relatively low-cost options at Point Source A and Point Source C,
and cost-saving options that require improved solvent recovery by
industrial and commercial solvent users.  Control strategy B
would provide a significant reduction in risk (an additional 45
percent) by requiring  large outlays for construction of new plant

                              VI-34

-------
                                          Table  VI-10

                                BALTIMORE  IEHP AIR TOXICS STUDY

          SCHEDULE OF CONTROL  STRATEGIES FOR REDUCING LIFETIME INDIVIDUAL CANCER RISK
                          AT SITE OF MAXIMUM INDIVIDUAL CANCER RISK1

                       PHASE II RESULTS INTENDED FOR POLICY DEVELOPMENT
Point
on
Figure VI -3
Graph
A
B
Percent
Reduction in Total
Expected Cancer Risk Annual Ized
Lifeline from Cost
Cancer Risk Current Control ($1000/year)
1.3 x 10'3 3 -$85
7.4 x ID'4 45 $26,000
Incremental Pollution Controls Implemented
Cost
($1000/yr) Source Control
Potion
-$85 Point Source A-chrome controls 1
Point Source C 4
Perc degreasing 2
TCE degreasing 2
Methylene chloride degreasing 1
Misc. industrial 1
methylene chloride use
$26,085 Point Source A-coke oven controls 1
Point Source A-chrome controls 1
Perc degreasing 2
TCE degreasing 2
Methylene chloride degreasing 1
Misc. industrial 1
methylene chloride use
7.2 x 10
        -4
46.9
$42,000
7.1 x 10
        ,-4
47.1
$211,000
 $16,000    Light duty  diesel  vehicles       1
           Point Source  A-coke oven  controls  1
           Point Source  A-chrome  controls   1
           Point Source  A-benzene controls  1
           Point Source  B                   3
           Point Source  C                   3
           Perc  degreasing                  3
           TCE degreasing                   3
           Methylene chloride degreasing    1
           Dry cleaning                     2
           Misc. industrial                 3
             methylene chloride use
           Woodstoves                       1
           Comnercial  oil heating          2
           Residential coal heating         1

$169,000   Heavy duty  diesel  vehicles       1
           Light duty  diesel  vehicles       1
           Point Source  A-coke oven  controls  1
           Point Source  A-chrome  controls   1
           Point Source  A-benzene controls  1
           Point Source  B                   3
           Point Source  C                   3
           Point Source  0                   1
           Gasoline  marketing              1
           Perc  degreasing                  3
           TCE degreasing                   3
           Methylene chloride degreasing    1
           Drycleaning                     3
           Misc. industrial                 3
             methylene chloride use
           Woodstoves                       1
           Conrercial  oil controls         1
           Residential coal                 1
           Residential oil                  1
                                           (continued)
                                             VI-35

-------
                                             Table VI-10  (continued)

                                                     NOTES
This study uses conservative estimates of increased  cancer risk from ambient (i.e.. outdoor) exposure to establish
priorities among pollutants and sources.   The risk estimates are calculated using modelled or monitored concentrations
and EPA unit cancer risk factors.  There  is considerable uncertainty in the estimated concentrations, which could
either overstate or understate the true concentrations  (see Chapter IV).  Unit cancer risk factors combine CAG potency
estimates with EPA exposure assumptions.   The CAG potency estimates provide a plausible upper limit to the cancer risk
of a compound (see Appendix A); however,  the true value of the risk is unknown and may be as low as zero.  The exposure
assumptions are extremely conservative in that they  assume continuous exposure to outdoor air for 70 years.  Because or
the generally conservative bias in the information,  it  is highly unlikely that the true risks would be as high as the
estimates, and they could be considerably lower.

Cost estimates were based on engineering  estimates and  best professional judgment and may significantly over - or
underestimate real costs.
                                                       VI-36

-------
                   FIGURE VI-3

 TOTAL ANNUALIZED COST VERSUS REDUCTION IN
  MAXIMUM LIFETIME INDIVIDUAL CANCER RISK





ZED COST
DOLLARS)
D h-
3?
ZfA

-------
equipment.  We can achieve only minor additional reductions in
risk and large increases in cost through imposition of strategies
C and D.  The resulting risk, about 7 x 10' , is still
significantly above the area-wide individual risk of about 1 to
2 x 10'*.

     Regarding noncancer effects, the control strategies for
reducing the maximum increased lifetime individual cancer risk
that include high removal of benzene releases from Point Source A
(e.g., Option 1) will succeed in lowering the benzene ambient
concentrations to levels that require less investigation.  Again,
the strategies considered in this analysis will not reduce xylene
ambient air concentrations at its maximum point (identified by
the discrete point analysis) because the control options analysis
did not consider the main contributor of xylene releases at this
location, Point Source E.

     These results are presented only as part of our
demonstration project.  There is considerable uncertainty
surrounding our estimates of emissions, risks, and cost.  A more
detailed analysis would be necessary to validate these results.

          iii. Limitations

     We emphasize that the cost-effectiveness analysis was
designed primarily to demonstrate a methodology for evaluating
alternative pollution control strategies.  Although the results
of the analysis can provide  the risk manager with an  indication
of possible remedial programs to pursue, additional analysis
needs to be performed before any strategy can be implemented.
Specifically, some of the major limitations of the analysis
include:

     •    The cost-effectiveness analysis could not consider all
          sources and pollutants because of gaps in data.

     •    The analysis  is based on preliminary estimates of
          emissions, exposures, and  risks by  source.  As we
          discussed in  Chapters  IV and V, we  believe  that we have
          overstated the exposures associated with Point Source A
        .  because of limitations in  our ability to adequately
          model  this site.

     •    The analysis  did  not  consider all possible  control
          options.  A detailed  engineering  analysis is  needed  to
          determine if  a technology  is, in  fact,  feasible.

     •    We overestimate  the cost of  certain control options  to
          both  industry and area sources  to the extent  that we do
          not take  into account the  age or  usefulness of equip-
          ment  that  is  replaced under  the proposed  option  or  the
          income from  the  sale  of  replaced  equipment  or scrap.

                              vi-38

-------
           The cost estimates were rough approximations  based on
           limited data and were developed to  compare  relative.
           and not absolute, costs.   Additional  work is  needed in
           this area.
 4.    MAXIMIZING THE BENEFITS TO SOCIETY IN CONTROLLING AIR  TOXICS

      Significant economic  benefits  can  result  from control  of
 toxic air pollutants by co-controlling  emissions  of VOCs, which
 impair visibility and cause  smog, and particulates, which cause
 crop  and  materials damage.   Consideration  of these additional
 effects could  greatly alter  how the decision maker views the
 relative  effectiveness and environmental impact of alternative
 control strategies directed  at  toxic air pollutants.  In this
 section,  we  present the results of  our  preliminary evaluation of
 least-cost control strategies for reducing air toxics emissions
 that  will also achieve the highest  level of economic benefits.

      a.    Results of the Analysis of Strategies to Maximize
           Benefits

      In identifying control  strategies,  we chose  the combinations
 of control options that provided the most  economic benefits at
 arbitrarily  chosen maximum costs to the  regulated sources.  We
 summarize these results in Table VI-11.

      As shown,  we can achieve $3.6  million in  economic benefits
 at zero net  cost.   At point  A in Table VI-11,  costs of controls
 for all point  sources,  dry cleaners, residential  heating
 (woodstoves  and bans on the  use of  coal  for residential heating),
 and trucks are offset by cost savings resulting from the control
 and recovery of solvents.  The  greatest  total  benefit for the
 levels  of  costs that we have chosen occurs  at  point J.  However,
 the total  annualized control costs  for the  selected control
 strategy  exceed total benefits.  Thus, the  net benefits are
 negative.  The point that maximizes the  net benefits occurs at
 point D.   At an annualized cost to  the affected sources of  $5
 million,  we  achieve estimated net benefits  of  $8.5 million  and
 about a 16 percent reduction in annual excess  cancer incidence.

      b. Comparison of Benefit and Cost-Effectiveness Analyses

     We compared  control strategies (the combination of control
 options)  to  achieve a requisite reduction of cancer risk against
 control strategies  that maximized economic  benefits subject to
 cost constraints.   The  cost-effectiveness analysis generated a
 set of control  strategies that  provides  selected  reductions of
 cancer risk—measured as either annual excess  cancer incidence or
maximum increased  lifetime individual cancer risk—at the lowest
annualized implementation cost  to sources.   For the benefit-cost

                              VI-39

-------
                                           TABLE VI-11

                                      CONTROL STRATEGIES
                                  BASED ON BENEFITS ANALYSIS (1)

                     Percent
                  Reduction  in     Total       Total
           Cases       Cancer    Annualued  Annuahzed     Net    Pollution Control  Implemented
         Reduced   Incidence  from    Cost     Benefits   Benefits
 Point   per Year  Current  Cntrl   (Smil./yr) (Jmil./yr) (Jmil./yr) SourceOption

—X	JT2	STTO	JO       JTB       J3.6 Point Source A-coke oven  cntr
                                                                              chrome cntrIs
                                                                              benzene cntrIs      3
                                                                 Point Source B                  1
                                                                 Point Source C                  4
                                                                 Point Source 0                  1
                                                                 Gasoline marketing
                                                                 Solvent usage-perc               2
                                                                               TCE                2
                                                                               methyl,  chl.       1
                                                                 Dry Cleaners
                                                                 Misc. methylene chloride  use    2
                                                                 Woodstoves                      1
                                                                 Heating-comnercial  oil
                                                                         residential coal
                                                                         residential oil
                                                                 Heavy duty trucks
                                                                 Light duty trucks

   B         0.51          14.00%       $1.0       IS.4       $4.4 Point Source A-coke oven  cntr
                                                                              chrome cntrIs       1
                                                                              benzene cntrIs      2
                                                                 Point Source B                  3
                                                                 Point Source C                  4
                                                                 Chemical Plant                  1
                                                                 Gasoline marketing
                                                                 Solvent usage-perc               2
                                                                               TCE                2
                                                                               methylene  chlor    1
                                                                 Dry Cleaners                    2
                                                                 Misc. methylene chloride  use    3
                                                                 Woodstoves                      1
                                                                 Heating-connercial  oil
                                                                         residential coal
                                                                         residential oil
                                                                 Heavy duty trucks
                                                                 Light duty trucks                1
             0.54         14.90%       $2.6       $6.3       $3.7 Point Source A-coke oven cntr    2
                                                                              chrome cntrIs       1
                                                                              benzene cntrIs      1
                                                                 Point Source B                   3
                                                                 Point Source C                   4
                                                                 Point Source 0                   1
                                                                 Gasoline marketing
                                                                 Solvent usage-perc               2
                                                                               TCE                3
                                                                               methylene chlor    1
                                                                 Dry Cleaners                     2
                                                                 Misc. methylene chloride use     3
                                                                 Hoodstoves                       I
                                                                 Heating-connercial oil
                                                                         residential coal
                                                                         residential oil
                                                                 Heavy duty trucks                1
                                                                 Light duty trucks                1
                                                  VI-40

-------
                                           TABLE VI-11

                                      CONTROL STRATEGIES
                                   BASED ON BENEFITS ANALYSIS (1)

                     Percent
                  Reduction in     Total       Total
           Cases      Cancer     Annual)zed  Annual*zed     Net    Pollution Control Implemented
         Reduced  Incidence from    Cost     Benefits   Benefits
 Point   per Year Current Cntrl   (Jmll./yr)  ($mil./yr) (Jrail./yr) SourceOption

—B	Q757	rSTTH	JO	JITS	SB.5 Point Source A-coke oven cntr
                                                                              chrome cntrIs       1
                                                                              benzene cntrIs      2
                                                                 Point Source B                   3
                                                                 Point Source C                   4
                                                                 Point Source 0
                                                                 Gasoline marketing
                                                                 Solvent usage-perc               2
                                                                               TCE                2
                                                                               methylene en lor    1
                                                                 Dry Cleaners                     2
                                                                 Nisc. methylene chloride use     1
                                                                 Hoodstoves                       1
                                                                 Heating-coimercial oil
                                                                         residential coal         1
                                                                         residential oil
                                                                 Heavy duty trucks                1
                                                                 Light duty trucks                1
             0.59         16.32*       $5.4      S13.8       S3.4 Point Source A-coke oven cntr    2
                                                                              chrome cntrIs       1
                                                                              benzene cntrIs      1
                                                                 Point Source B                   3
                                                                 Point Source C                   4
                                                                 Point Source D
                                                                 Gasoline marketing
                                                                 Solvent usage-perc               2
                                                                               TCE                2
                                                                               methylene chlor    1
                                                                 Dry Cleaners                     2
                                                                 Misc. methylene chloride use     1
                                                                 Hoodstoves                       1
                                                                 Heating-conmercial oil
                                                                         residential coal         1
                                                                         residential oil
                                                                 Heavy duty trucks                1
                                                                 Light duty trucks                1

              0.6         16.69%       $7.5      S14.7       S7.2 Point Source A-coke oven cntr
                                                                              chrome cntrIs       1
                                                                              benzene cntrIs      1
                                                                 Point Source B                   3
                                                                 Point Source C                   4
                                                                 Point Source 0
                                                                 Gasoline marketing               1
                                                                 Solvent usage-perc               2
                                                                               TCE                2
                                                                               methylene chlor    1
                                                                 Dry Cleaners                     2
                                                                 Misc. methylene chloride use     2
                                                                 Woodstoves                       1
                                                                 Heatlng-cornnercial oil
                                                                         residential coal         1
                                                                         residential oil
                                                                 Heavy duty trucks                1
                                                                 Light duty trucks                1
                                                 VI-41

-------
                                            TABLE VI-11

                                       CONTROL  STRATEGIES
                                   BASED ON  BENEFITS ANALYSIS (1)

                     Percent
                  Reduction in     Total       Total
           Cases      Cancer    Annual!zed  Annual!zed     Net    Pollution Control Implemented
         Reduced  Incidence from    Cost     Benefits   Benefits   	
 Point   per Year Current Cntrl  (Smil./yr)  ($mil./yr) ($m11./yr) SourceOption

—5	ITW	17^61	5577	STO	56.2 Point Source A-coke oven cntr    T
                                                                              chrome cntrIs       1
                                                                              benzene cntr Is      1
                                                                 Point Source B                   3
                                                                 Point Source C                   4
                                                                 Point Source D                   1
                                                                 Gasoline marketing               1
                                                                 Solvent usage-perc               3
                                                                               TCE                3
                                                                               methylene chlor    1
                                                                 Dry Cleaners                     3
                                                                 Hisc. methylene chloride use     3
                                                                 Hoodstoves                       1
                                                                 Heating-commercial oil
                                                                         residential coal         1
                                                                         residential oil
                                                                 Heavy duty trucks                1
                                                                 Light duty trucks                1

   H         0.64         17.66%       $8.7      $14.9       $6.2 Point Source A-coke oven cntr    2
                                                                              chrome cntrIs       1
                                                                              benzene cntrIs      1
                                                                 Point Source B                   3
                                                                 Point Source C                   4
                                                                 Point Source 0                   1
                                                                 Gasoline marketing               1
                                                                 Solvent usage-perc               3
                                                                               TCE                3
                                                                               methylene chlor    1
                                                                 Dry Cleaners                     3
                                                                 Misc. methylene chloride use     3
                                                                 Hoodstoves                       1
                                                                 Heating-commercial oil
                                                                         residential coal         1
                                                                         residential oil
                                                                 Heavy duty trucks                1
                                                                 Light duty trucks                1
             0.71         19.60%      $19.5      $14.0       ($5.5)Po1nt Source A-coke oven cntr    2
                                                                              chrome cntrIs       1
                                                                              benzene cntrIs      1
                                                                 Point Source B                   3
                                                                 Point Source C                   4
                                                                 Point Source 0                   1
                                                                 Gasoline marketing               1
                                                                 Solvent usage-perc               3
                                                                               TCE                3
                                                                               methylene chlor    1
                                                                 Dry Cleaners                     3
                                                                 Misc. methylene chloride use     3
                                                                 Hoodstoves                       1
                                                                 Heating-commercial oil           2
                                                                         residential coal         1
                                                                         residential oil
                                                                 Heavy duty  trucks                1
                                                                 Light duty  trucks                1
                                                  VI-42

-------
                                            TABLE  VI-11

                                       CONTROL  STRATEGIES
                                   BASED ON  BENEFITS ANALYSIS (1)

                     Percent
                  Reduction in     Total-       Total
           Cases      Cancer    Annualized  Annualized     Net    Pollution Control Implemented
         Reduced  Incidence from    Cost     Benefits   Benefits   	
 Point   per Year Current Cntrl   (Smil./yr)  ($mil./yr) (Jmil./yr) SourceOption

—]	OI	ZTTSftJ3TB      JT7T2     (J28.4)Point Source A-coke oven cntr    T
                                                                              chrome cntrIs       1
                                                                              benzene cntrIs      1
                                                                 Point Source B                   3
                                                                 Point Source C                   4
                                                                 Point Source 0                   1
                                                                 Gasoline marketing               1
                                                                 Solvent usage-perc               3
                                                                               TCE                3
                                                                               methylene chlor    1
                                                                 Dry Cleaners                     3
                                                                 Misc. methylene chloride use     3
                                                                 Uoodstoves                       1
                                                                 Heatlng-cormercial oil           2
                                                                         residential coal         1
                                                                         residential oil
                                                                 Heavy duty trucks                1
                                                                 Light duty trucks                1
  This study uses conservative estimates of increased cancer  risk  from ambient  (i.e.. outdoor)
   exposure to establish priorities among pollutants and sources.   The risk estimates are calculated
   using modelled or monitored concentrations and EPA unit  cancer  risk factors.   There  is
   considerable uncertainty in the estimated concentrations,  which could either overstate or
   understate the true risk (see Chapter IV).  Unit cancer  risk  factors combine CAG  potency
   estimates with EPA exposure assumptions.  The CAG potency  estimates provide a  plausible upper
   limit to the cancer risk of a compound (see Appendix A); however,  the true value  of  the risk  is
   unknown and may be as low as zero.  The exposure assumptions  are extremely conservative in that
   they assume continuous exposure to outdoor air for 70 years.  Because of the generally
   conservative bias in the information, it is highly unlikely that the true risks would be as high
   as the estimates, and they could be considerably lower.
                                                  VI-43

-------
analysis, we sought control strategies that provided optimal
monetary benefits (including both health, cancer and non-cancer,
and welfare benefits) subject to an arbitrarily chosen set of
maximum expenditures that could be required of the regulated
sources.

     We compared the two approaches by asking whether optimal
strategies for obtaining equivalent reductions in annual excess
cancer incidence under the two different approaches result in
different combinations of pollution control options.  In other
words, does the omission of noncancer health and welfare benefits
from the cost-effectiveness scenarios result in strategies that
obtain the desired reduction in cancer risks, but do not provide
the greatest net economic benefits?  Or, does the least cost
(i.e., the optimal) set of control options to achieve a desired
level of excess cancer risk result in less net economic benefit
to residents of the Baltimore area than an alternative set that
imposes greater cost on the regulated sources?

     In Figure VI-4, we display the total costs of controlling
air toxics associated with a reduction in risk under each
approach.  As shown, the control costs to reduce cancer risk rise
gradually up to about the 20 percent reduction level.  Above this
point, the costs of reducing additional cancer risk rise sharply.
The selection of control options using either the benefit-cost
method or cost-effectiveness method appear to be fairly
consistent when looking at the relative positions of the point on
the figure.

     Figure VI-5 displays the total benefits for each risk
reduction strategy under the benefit-cost and cost-effectiveness
approaches.  The points in the lower left quadrant indicate that
for low levels of risk reduction, there are tradeoffs between
maximizing benefits and minimizing cancer risks.  Some controls
that provide greater total benefits do a better job of reducing
noncancer risks than other options providing a greater reduction
in the estimated excess cancer incidence.  For cancer incidence
reduction rates of 14 to 20 percent, the selection of control
options is, again, fairly consistent across the two methods.

     The final figure, Figure VI-6, combines the information from
the total cost and total benefit figures in order to discern
which control options provide the greatest net benefits.  The
rise in net benefits is relatively small up to a 16 percent
reduction in the estimated annual excess cancer incidence.  After
this point, net benefits begin to fall, reaching zero around the
19 percent risk reduction level.  The selection of options
reducing annual excess cancer incidence beyond this point are
costly, and the benefits from reducing annual excess cancer
incidence and the risk of noncancer effects fail to keep pace
with the cost increases.
                              VI-44

-------
                         FIGURE VI-4

 SUMMARY OF COSTS OF CONTROLLING AIR TOXICS IN BALTIMORE
COMPARISON OF BENEFIT METHOD AND COST-EFFECTIVE SCENARIOS









1
•£•
01












-------
                                 FIGURE VI-5

        SUMMARY OF BENEFITS OF CONTROLLING AIR TOXICS IN BALTIMORE
        COMPARISON OF BENEFIT METHOD AND COST-EFFECTIVE SCENARIOS
0\



i^^
CO
5
3
0
Q
1
IL
O
CO
z
o
3
I







4i -
20 -
19 -
18 -
17 -

16 -
15 -
14 -
13 -
12 -
11 -
10 -

9 -
w
8 -
7 -
6 -
5 -
4 -

3 -
2 -
1 -
+f

+
+ o


o +
o + a +







a

at-

a

+ + +

IIIIIMIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIMIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIlllllllll
               2    4    6    8    10   12   14   16   18   20   22   24
                PERCENT REDUCTION IN ANNUAL EXCESS CANCER INCIDENCE
              a   BENEFIT METHOD
CE METHOD

-------
                                  FIGURE VI-6
SUMMARY OF NET ECONOMIC BENEFITS OF CONTROLLING AIR TOXICS IN BALTIMORE

       COMPARISON OF BENEFITS METHOD AND COST-EFFECTIVE SCENARIOS
     
-------
     The peak in Figure VI-6 is the point where the additional
benefits from control exceed the incremental costs by the
greatest amount.  Additional reductions in risk cost more than
the cancer and noncancer benefits afforded by their
implementation.  An examination of the net benefits column in
Table VI-8 shows which options can provide benefits that exceed
their implementation costs.  Presentation of information in this
format allows decision-makers to consider the additional benefits
and costs of adopting each control option, and can help in the
efficient selection of controls where the costs borne by the
affected facilities may face constraints, or where a particular
level of risk reduction is a goal for the community.  Even in
those instances where controls need to be undertaken to meet
required reductions in risk—and the costs of these controls
exceed the benefits they provide—the use of benefit information
can lead to more efficient uses of society's resources than
decisions made in the absence of this information.  These
economic frameworks not only provide an opportunity to address
the multiple benefits of controlling toxic emissions, but also a
framework to compare the benefits from environmental controls
with other risk reduction actions that may be available.
                               VI-48

-------
                APPENDIX  A








            GENERAL METHODOLOGY




                   FOR AN




INTEGRATED ENVIRONMENTAL MANAGEMENT PROJECT

-------
               APPENDIX A. GENERAL METHODOLOGY
      This  Appendix  provides  a  general overview of the analytical
 steps which  EPA employs  in Phase I of an Integrated Environmental
 Management Project  (IEMP)  in the  assessment  of  risks.   It also
 defines  the  concepts of risk assessment and  risk management and
 describes  the  analytical methods  used  for risk  assessment.   By
 and  large,  the Baltimore  IEMP followed  these steps,  with some
 variations.
 APPLICATION  OF METHODOLOGY

      This  section  presents  a comprehensive  description of EPA's
 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 know-
•ledge) 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
 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 be-
 fore  examining  the actual procedures.    First, the  risk estimates
 are based  primarily  on existing knowledge about pollutant potency,
 releases,  and ambient  conditions; these data,- however,  vary wide-
 ly  in quality and are almost always incomplete.   Second,  the ex-
 posure estimates  incorporate a  series  of simplifying  assumptions;
 although these  assumptions  are  necessary,  they  they remain  open
 to  question  and may be  controversial.  Third, the potency esti-
 mates are  necessarily  based on  current knowledge of the toxicolo-
 gical 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  Appendix,  the range  of
 potential  environmental  issues at  any  site  is  so large  that  it
                                A-l

-------
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  orga-
nized 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

          —Establishing risk assessment approaches,  e.g., selec-
            ting  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
                               A-2

-------
 allows  like  comparisons rather than having to confront the dilem-
 ma of comparing  different  subjects  for which there is no obvious
 common  denominator.  This  aspect  of  scoping  as it applies to the
 Baltimore  IEMP is  discussed  in  more  detail  in Chapter IV.  [Some
 issues  can be placed into  more than one reference category? toxic
 contamination of fish, for example,  can not only be an ecological
 problem, but at  the same time a public health threat.]

     After issues  have  been grouped  in  reference  categories,
 project participants may decide to  consider  only one, or perhaps
 just a subset, of  these groupings of issues.  Such was the case in
 both Santa Clara and  Philadelphia.   At the  outset  of these pro-
 ject participants  decided  to study  only issues relating to human
 health risks.

     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 pri-
 mary sources of  these data,  which should  be  collected and evalu-
 ated 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.   (Intermedia  transfers  are  the  relocation of pollu-
 tants from one medium to another.   For example, the incineration
 of waste  reduces  —  but  does  not  eliminate.  —  the volume  of
 solid material requiring  landfilling.   However,  the  process  may
 also produce air-borne pollutants.)   For  certain common sources,
 such as  dry cleaners, degreasers, mobile  sources (air),  and non-
point run-off  (water),  EPA's Regulatory  Integration  Division in
the Office of  Policy  Analysis has developed  algorithms  to esti-
mate pollutant loadings  to  all  relevant media that  are easily
adapted  to different geographic areas.  In some situations,  some
new monitoring may be  initiated  at this  stage if it  is deemed
necessary  for setting priorities.

     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.
                                A-4

-------
     The  screening  process   involves  two  complementary  steps.
 The  first step quantifies risks to human  health  or the environ-
 ment to  identify the sources,  pollutants,  and exposure pathways
 of greatest  concern.  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 use  a  second step which focuses on
 issues that  are not readily  quantified.   This approach qualita-
 tively assesses  an  issue's  analytical feasibility;  its relevance
 to EPA,  state, and local  program  objectives;  and  its potential
 for  effective  response  (i.e., controllability).

     Step 1;   Risk Screen.  The  risk  screen  is  a  preliminary
 quantitative assessment of the  risks to human health or the envi-
 ronment,  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  pollution in  one medium,  or  we can  address  the
 sources of pollution in  several media,  each of which contributes
 to the same health effect.

     It may take  several  cuts at the  entire set of issues within
 a reference category to develop the best common denominator.  Once
 this common  denominator is  developed,  a  subset  of  issues  will
 become particularly significant.   For  example,  if the study par-
 ticipants determine  that  carcinogenic  effects  should  be  the
 "common currency" of evaluation, those issues or conditions that
 might be  expected  to  cause greater levels  of cancer will become
more important  than those that  might  cause other types of health
 consequences.

     3ecause the  focus  of the  IEMP  analysis is  for  policy  pur-
poses and due  to  inevitable   resource  constraints,  the IEMP pro-
 ject 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 deter-
mining when enough data have  been  gathered to support a success-
ful risk  screen.
                               A-5

-------
      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 quantitative 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 inte-
 grated studies may  want to modify or add to these criteria.  For
 example,  the Baltimore IEMP added  an explicit criterion of avoid-
 ing duplication  of  existing analyses of control programs.

     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 avoid-
 ed because of  the complexity  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 develop-
 ing 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 IEMP projects
 to indicate where shifts in EPA priorities and methods are appro-
priate,  so the relationship of the issue to national ^PA 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

     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.
                                A-6

-------
 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 useful  candidate for
 study.

     Risk  Management Challenges in Interpreting Results and
 Applying the Process.

     In the  process  of  establishing priorities, study  partici-
 pants are  likely to  encounter  several  basic  problems involving
 the  interpretation and application of environmental risk informa-
 tion.  Some specific issues are discussed below.

     Maximum Exposed  Individual  Risks  Versus  Total  Population
     Risks

     The management  or  project  advisory  group  selecting  study
 topics will have to  confront  one of  the classic  trade  offs in
 risk management--that of  maximum  exposed individual versus cumu-
 lative 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 population;  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 substantial 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.

     Comparing Across Effects

     Another difficult  issue  involves   the  comparison of  risks
 across different health  effects  (e.g.,  kidney damage versus can-
 cer).  Typically, this is done on the basis of informed judgment.
 RID is developing 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 con-
 sidering local  concern  for  the severity of  different  effects.

     Comparing Human  Health  Effects  to  Environmental  Effects

     In cases where  the scope of the  study,  as  in  the Baltimore
 IEMP, includes human health as  well  as ecological issues, diffi-
 cult value  judgments  must  be  made.   These  judgments  are  not
generally made within  the same policy  decision  analysis  because
 the quantitative means  of equating  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
                               A-7

-------
human health is not the same as that placed on pur other environ-
mental goals.   Decisions  concerning  the  setting of  priorities
between these  two  types   of  issues   (because  the  quantitative
measures used to assess adverse impacts for each are not scienti-
fically or  analytically comparable)  involve  the difficult  and
necessarily subjective  balancing  of  both  public policy  objec-
tives.

     Phase I Product

     This Report contains information  about the risks  we quanti-
fied, identifies priority issues  based on  our  risk assessments,
and describes the  methods  we used to  assess  exposure  and risks.
Furthermore, it  describes  the  process by  which  these  products
were made possible.  In this regard,  we elaborate on the institu-
tional arrangements necessary  for the  making of key policy deci-
sions and management  judgements  and  the  manner  in which  these
decisions were made.

Phase II;  Risk Management Control Options
     After Xfcfcselecting Phase I issues warranting further analy-
sis, we begin Phase II.  Phase  II  consists of three major tasks:
additional data  gathering  to  improve  our  exposure  and  risk
assessments, a  scientific  review  of  our  risk  assessments,  and
analysis of control options.

     Additional 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 as
accurate as possible,  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 poten-
tial control alternatives,  where effectiveness is measured by the
reduction in  health risk — either  Most Exposed  Individual  (MEl)
(defined on p. A. 15) risk or total  population  incidence.  In these
                               A-8

-------
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 in-
dicate 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.

     Developing Pollution Control Options

     The next step 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  risks  to health  (total population as  well as to
the maximum  exposed  individual)  or the environment  (the measure
of effectiveness)   at  the  minimum cost.  This  occurs  through the
lEMP's cost-effectiveness analysis, which  presents the tradeoffs
of costs  and risk  reductions  that  decision makers  usually take
into consideration  when formulating regulatory  strategies,  The
analysis considers  risks  that  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  ma^or  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   discuss  later
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
                                A-9

-------
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.   Largely because of
lack of  sophistication  with  the  methodology,  we  chose  not  to
examine other  issues  which would  involve  ecological and  non-
cancer health effects.

     In the  Baltimore  IEMP, we  have  expanded  the scope  of  the
project to include ecological impacts  and have decided to examine
non-cancer health effects in Phase II.  We will again be treading
new ground in  developing analytic methodologies  to  examine eco-
logical impacts and the risk of non-cancer effects.
RISK ASSESSMENT AND RISK MANAGEMENT

     The two key organizing concepts of an integrated environmen-
tal management project are  risk assessment  and  risk management.
Risk assessment is the central task of Phase  I;  risk management
is the central  task  of Phase  II.   Conceptually,  risk assessment
should be done independently of risk management.1   In IEMP
studies,  we  separate   assessment  from  management   as  much  as
possible; however, this  is  not always  possible  so  that  some of
the activities in Phase I and Phase II overlap.

     We emphasize that our discussion of  the IEMP methodology is
presented at a  conceptual level;  our  actual  application  of  the
methodology may vary  from one geographic project  to the next.
Furthermore, we may  use  different  methods  in  response  to  the
needs of a  particular  IEMP.   For  example,  project  participants
in the Baltimore  IEMP  had identified issues pertaining  to local
ecology and groundwater resources,  in addition to health  issues,
for study in Phase I.   To assess the significance of these issues
and to decide  which of   them  warranted  further examination  in
Phase II, we needed different analytical tools from those we used
in setting  priorities  among  human health  issues  in Philadelphia
and the  Santa  Clara Valley.   (Note that  Philadelphia and Santa
Clara Valley  are  areas  where  we  conducted  earlier  geographic
studies.)  We describe  these other (nonhealth related) methods in
Chapters VI (Analysis of  sources with Potential Adverse Impacts on
Ground-water) and VII (Analysis of Ecological Impact) and discuss
only our  methods  for  analyzing  risks  to human health  in  this
Appendix.

Phase I Risk Assessment and Priority-Setting

     There are  always  more  environmental issues  to study  than
resources with which to  address them in a manner  that  will lead
to their successful resolution.  A key task of Phase I of an IEMP
is, therefore,  this: to choose which issues  demand the  immediate
                               A-10

-------
attention of  environmental  officials—and  the  expenditure  of
their time and resources—and  which can be addressed at  a  later
date.  Such comparing  of issues  and setting  of priorities  re-
quires that all  issues be  reduced to  a  common  element.   (Note
that an integrated approach to all issues  means we try to examine
the same issue across media.   An IEMP is not a comprehensive look
at all the environmental issues of the study area.)   Risk, either
to human health or the environment  (or  both),  is  the  common ele-
ment we  use  for  comparing  environmental  issues in  geographic
studies.  Risk can be  categorized  in  several  different  ways.  As
appropriate, we can  examine  risk associated with exposures from
a specific  pollutant  source.   We  can  examine risks  associated
with exposure  to  pollutants,  irrespective  of  source.   We  can
look at risk  associated with  exposures  in a single environmental
medium, such as air,  water, or  food.  We  can examine risks asso-
ciated with exposure pathways.   In the absence  of  other impor-
tant considerations,  the  more  significant  the  risk,  the  more
likely the  issue  will  rank high  in priority  for examination in
Phase II,  the risk  management  phase.  Important considerations
that may   alter   the   priority  ranking   include,   for  example,
resource limitations  or  the   fact  that  a particular   issue  is
already being handled by  the  Federal, state of local government.

     The  risks we examine are  associated with exposure  to toxic
chemicals in the environment  and represent estimates of the poten-
tial for these exposures to cause adverse effects.  Chemicals can
adversely affect  human health,  and thus pose  a risk, when people
are exposed to toxic levels in  the  air, water,  food,  or  contami-
nated soil.  Chemical exposures can adversely affect the  environ-
ment by disrupting the ecology  of surface waters or the  dynamics
of biological  communities.   Furthermore,  toxic chemicals can so
contaminate a  natural  resource that  future  generations  can, for
all practical  purposes,  no longer  use  or  enjoy  it  or must pay
greatly to  do so.

     In order  to  examine potential  impacts  to human health  from
environmental exposures  in  a  particular  geographical   location,
we perform  a  risk  assessment.   Based  on the  National  Research
Council's recommendations,1 an  IEMP risk  assessment contains the
following elements:

    o Hazard  Identification
      Does  the agent cause the  adverse  effect?

      For the  chemicals  of interest, RID toxicologists  and  con-
      tractors prepare individual  Profile reports that  summarize
      the information  on  animal  and  human health effects.   Each
      report  examines  both oral and inhalation exposure  data  for
      the  following  ten health  effect categories:  cancer,  liver,
      kidney,  reproductive,  fetal  developmental,  neurobehavioral,
      mutagenicity,  respiratory,   cardiovascular,  and  other.   A
                                A-ll

-------
 weight-of-evidence evaluation  is  provided  for  each health
 effect category.   For  cancer,  we use  the EPA  weight-of-
 evidence evaluations  that  are prepared   by  the  Agency's
 Cancer Assessment Group  (CAG)  according  to the  1986 cancer
 risk assessment guidelines.3   EPA  does  not have weight-of-
 evidence evaluation  schemes  for  the  other health  effect
 categories, so we use ranking  schemes developed specifically
 for our  work.   Our  ranking  schemes  are  currently  being
 reviewed by scientists within and outside EPA.

o Dose-Response Assessment
  (What is  the  relationship between  dose  and incidence  of
  effect in humans?)

  This information is also  provided in  the chemical profile
  reports.  With the  exception of  mutagenicity,  a  separate
  evaluation is provided for each of the health effect cate-
  gories.  For cancer, we use the dose-response evaluations,
  or "unit  risk  factors"  that  are prepared, by EPA's Cancer
  Assessment Group  (CAG).   For  the  remaining   non-cancer
  health effect  categories,  EPA does  not  have a  satisfac-
  tory method  for  quantifying  dose-response relationships.
  Therefore, as  an alternative  we  use the EPA  Reference
  Doses (RfDs)  as  benchmarks  to  estimate  exposure  levels
  that are  "safe"  and for  which we assume  the incidence  of
  effects is insignificant.*

o Exposure Assessment
  (What exposuresare  currently experienced or  anticipated
  under different conditions?)

  The IEMP  studies include  an  assessment  of  environmental
  exposure levels  for  selected  chemicals   in  the media  of
  interest,  usually air and  water.   Environmental exposures
  are estimated primarily by modelling.  As feasible,  field
  monitoring is  conducted  and  the  collected data  are  used
  to validate  the  modelling.    In  addition  estimates  of
  exposed populations  are  made  by examining  census  data.

0 Risk Characterization
  (What is the estimated incidence  of the  adverse effect  in
  a  given population?)

  For the environmental media of  interest,  we  use our expo-
  sure assessment  to  estimate  both  the  concentrations  of
 *In so doing,  IEMP follows  EPA practice.   However,  other
  authorities may  use  other  methods.   For  example,  for
  airborne contaminants, Threshold Limit Values (TLVs)
  may be more suitable under certain circumstances.
                          A-12

-------
       specific chemicals  and the  number  of  people  exposed  at
       those concentrations.   We  then  link this  information  to
       our chemical-specific  dose-response  assessments  in  order
       to make  rough  estimates of the  incidence  of disease that
       may occur  above  the  baseline   rate  in  the study  area.

         Note that  because  of  limitations  in   our  ability  to
       quantify dose-response   relationships  for  each  of  the
       eight health  effect  categories,  we  must  analyze  cancer
       differently from  the  non-cancer  health  effects.   For can-
       cer, we  treat  the CAG  unit  risk  factor  as the  slope  of
       the dose-response curve; therefore,  we  can use this value
       to estimate a  crude  incidence.   For  the non-cancer health
       effect categories, we  do not have  estimates of the slopes
       of the dose-response  curves.   Instead, we have  the RfDs,
       which we  use  as  estimates  of  exposure  levels  that  are
       "safe", in terms  of  non-cancer health  risks.   So for non-
       cancer effects,  we  do  not  calculate  the incidence  of
       effects.  Rather, we calculate  the  number  of  people  ex-
       posed to  levels  above the RfDs  in order  to estimate  how
       many may be at some increased risk.

         The risk characterization also includes a discussion of
       the uncertainties  in  the  hazard  identification,  dose-
       response assessment,  and  exposure  assessment.    For  the
       hazard identification,  this  discussion   focuses  on  the
       weight-of-evidence evaluations.  For the dose-response and
       exposure assessments, the discussion focuses on uncertain-
       ties in the databases and models used.

     During Phase I  of our first geographic project in Philadelphia
and then again  in Baltimore,  we also  relied  on priority-setting
methods other than quantitative risk assessment;  e.g., the use of
expert judgement to establish relative  ranking of  issues.

     In order  to  perform  an IEMP  risk assessment,  as outlined
above,  we rely heavily on  approved  EPA quantitative methods,  CAG
un-.t risk values, RfDs and other appropriate standards and crite-
ria.  In  addition,  we  often use other approaches  and criteria
to refine our priority ranking of issues.  These criteria include
analytical feasibility,  relevance  to  EPA,  state,  and local pro-
gram objectives, and the potential for  effective  response.  These
criteria (which  incorporate  risk  management perspectives)  are
discussed in Chapter VIII of this report.

     Bear in mind that  the objectives  of the  IEMP in the use of
risk assessment procedures  are to allow for relative comparisons
of potential problems and  to provide  a general  sense of the re-
lative significance of  an  issue,  rather  than to make definitive
                               A-13

-------
 statements  about  the absolute  risks  posed by a  particular sub-
 stance, source, or pathway.  It must be  emphasized that we conduct
 risk  assessments;  we  do  not  conduct  epidemiological  studies.
 Thus, we  do not,  and cannot  examine  the  disease rates  in  the
 local population in order to correlate health effects with current
 or past environmental  exposures.   Moreover,  the  exposure assess-
 ments conducted as part of an IEMP are  not designed to be used in
 nor are they appropriate for an epidemiological study.

     Although different in their objectives, design and interpre-
 tation, risk assessments and epidemiological studies are actually
 complementary.  Risk  assessment  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  studies  increase  the scientific  under-
 standing of the relationship between exposure and health effects,
 thereby strengthening  the  basis  of risk assessment.  Epidemiolo-
 gical studies may  also,  in some cases, be useful for confirming
 specific hypotheses suggested by risk assessment.

     Other  qualifications regarding  the  use of  quantitative and
 other health risk  assessment  methods  are  discussed later in this
 appendix.

 Phase II Risk Management

     In Phase II,  we will focus the  resources of the project on
 considerations of  how best  to manage  risk  in   selected  areas.
 Risk management  is  the  process  of   evaluating  and  selecting
 approaches to reduce the  risks  identified  through  risk assess-
 ment.  Risk  management  considers   not  only  the   level  of  risk
 posed by a  particular  pollutant or source  of pollution  but also
 factors such as the  feasibility  and cost  of control, public pre-
 ferences,  and  institutional  capabilities.   Setting  priorities
 for research or  government action  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; in the Baltimore IEMP it also includes devel-
opment of analytical  tools  (i.e., a model for helping local govern-
ments establish priorities for  addressing potential  threats  to
ground-water resources  from  underground  storage  tanks)  as- well
as a  research plan (the Harbor blueprint).   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.

     The criterion of cost-effectiveness (i.e., cost  per change in
some unit  of effectiveness)  is  generally  expressed as the amount
of human health risk reduced by alternative controls in relation
                               A-14

-------
 co  che  cost of  implementing  those  controls.   However,  in  some
 cases, such comparisons cannot  be made and other economic crite-
 ria or definitions  of  cost-effectiveness  must  be used (e.g., the
 costs of preventing damage to a resource).

     Even where  mechanisms  are already in  place  for controlling
 chemical exposures,  an  analysis   of  the   cost-effectiveness  of
 these control strategies  can  be useful to local communities.  An
 analysis of the  cost-effectiveness  of  control options  may  lead
 to  the identification  of  control  strategies that attain the  same
 level of control for less cost.
GENERATION OF RISK ESTIMATES

     The following  overview of  methods  for quantifying  health
risks theoretically applies  to both  cancer and non-cancer health
effects.  In  practice,  appropriate  quantitative  methods  have
only been  developed   for  estimating carcinogenic risks.   There-
fore, unless  noted  otherwise,   the   following  discussion  deals
only with estimation  of carcinogenic  risk  to  an exposed  popula-
tion.

     In IEMP  studies,  we calculate  risk  using  three  measures of
exposure:  risk  to the  most exposed  individual  (MEI),  risk to
the average  exposed individual  (AEI),  and  the  excess aggregate
population incidence.  We define risk to the MEI as the increased
probability that an individual chronically exposed to the highest
concentration of one or more chemicals will have exposure-related
cancer during the  course of his  or  her  lifetime.  For the MEI,
the exposure is calculated as either the highest modelled concen-
tration or the  average  concentration obtained  for the monitoring
site with the highest daily or annual average monitored value for
one or more chemicals.  We define risk to the AEI as the increas-
ed probability  that an  individual  exposed to the area-wide aver-
age concentration  of  one  or more chemicals will  have exposure-
related cancer  during  the  course of  his or her lifetime.   Aggre-
gate population  risk  is the estimate  of  the increased incidence
of cancer, above the  background rate,  in an exposed population.
We use the standard EPA assumption that exposure is for 70 years.
Thus, elsewhere when we refer to  "lifetime risk" or "70-year life-
time", this means 70 years of exposure.  A  more detailed discussion
of our assumptions  and approach  for calculating  these  risks is
provided later  in this Appendix.

     For a given population, the  IEMP risk  assessment involves
linking estimates  of  chemical potency,  derived  from  animal and
human health  effects  data,  with  estimates  or  measurements of
area-wide contamination  or exposure  levels.  A  relatively high
degree of uncertainty unavoidably  underlies this  approach.   In
                               A-15

-------
 most cases, the risk estimates cannot easily be  verified  or  dis-
 proven by observation.  The approach is  useful,  however,  because
 it is straightforward and allows application of  the  most  current
 scientific information to the analysis of  adverse health  effects
 from current and projected exposures.

      The two key  elements  in  estimating  health  risk are:  1)  the
 calculation of the potency of a  chemical,  and 2) the  determina-
 tion of the level  of exposure  to  that chemical.

 1.  Estimating Potency

    A.  Carcinogens

    Potency  is estimated  from  an  analysis of  relevant animal   and
 human   (i.e.,  occupational,  epidemiological)  data concerning  the
 toxicity of a chemical.   At EPA,  data are  obtained  from  the  open
 literature and from proprietary  studies   submitted to che  Agency.

    Chemical potency is determined by: 1)  evaluating qualitatively
 the preponderance  (or weight)  of  evidence that a chemical  causes
 the adverse health effect  in question; and, if  so, (2)  estimating
 quantitatively the relationship   between  the  dose  (that   is,   the
•amount  of chemical  an animal  or person  is  to exposed  over  a
'period  of time)  and the incidence of the  effect within  an  exposed
•population of  a given size.   The latter  "dose-response relation-
•ship" describes  the potency of the chemical.   That  is, the lower
•the threshold  and  the steeper  the  slope  of  the dose-response
 curve,  the more potent the chemical.   For assessing  carcinogenic
 risks,  potency estimates  can  be  used to  relate the levels  of  ex-
 posure  to the probability  that  individual(s)  will  have  cancer.

    The  EPA Cancer  Assessment  Group  (CAG)  calculates  chemical-
 specific cancer  potency values called  "unit risk" estimates.  We
 use these unit risk estimates in our IEMP  studies and many  Pro-
gram Offices  throughout EPA also  use these values  for  a  variety
of  risk  assessment  and risk management activities.

    o Qualitative evaluation

     The qualitative,  or  weight-of-evidence  evaluation  involves
     reviewing  the  scientific  literature, both human and  animal,
     to  establish  the sufficiency of  evidence  for  a  chemical's
     carcinogenic potential  in humans.   This evaluation  of   the
     weight-of-evidence for  a  causal  association   between   the
     chemical  exposure and cancer is  an  important part  of  risk
     assessment.   It  is necessary because the extent and  quality
     of  chemical-specific  databases  rarely  allow for an  unambi-
     guous determination of carcinogenic  potential.
                                A-16

-------
      Two  evaluation  schemes are  available for  assessing  cancer
weight-of-evidence.   One was  developed by the International
Agency  for  Research  on  Cancer (IARC)  and the other was developed
by EPA.   The EPA  scheme  is a modification  of the  IARC  system.
All  chemicals evaluated  for carcinogenicity  by EPA's CAG are now
classified  according to the EPA scheme.

      Briefly, the IARC scheme has  the  following three categories:2

      Group  I;   Known Human Carcinogen.  This  category  is  used
only  when   there  is  sufficient  evidence  from  epidemiological
studies to  support a causal association between the exposure and
cancer.

      Group  II;   Probable  Human  Carcinogen.  This category
includes exposures for  which,  at  one extreme, the  evidence of
human carcinogenicity  is almost  "sufficient" and  at  the  other
extreme, it  is "inadequate".  To reflect this range, the category
is divided  into  Group IIA  and Group I IB  to  indicate  higher and
lower degrees of evidence, respectively.

      Group  III;   This category includes exposures that can not be
classified  as to their carcinogenicity in humans.

The EPA scheme has the following five categories; 3

     Group A;  Human Carcinogen.   This category is used only when
there is  sufficient  evidence from  epidemiological   studies  to
support a  causal  association between exposure to  the agent and
cancer in humans.

     Group B;  Probable Human Carcinogen.  This category includes
agents for  which  the evidence based on  epidemiological  studies
is "limited" and  agents for which  the evidence based on animal
studies is  "sufficient".  To  reflect  this  range,  the category is
divided into Group  Bl  and Group B2,  indicating  higher and  lower
degrees of  evidence, respectively.   Group  Bl has agents  with
"limited" evidence  from epidemiological  studies.   Group  B2 has
agents with  "sufficient" evidence from animal studies and "inade-
quate" evidence or "no data" from epidemiological studies.

     Group C;  Possible Human Carcinogen.  This category  is used
for agents  with  "limited" evidence  of carcinogenicity in animals
and an absence of human data.

     Group  D;  Not Classifiable as to Human Carcinogenicity.  This
category is  used  for agents  with  "inadequate" human  and animal
evidence of  carcinogenicity  or  for which  no data are available.
                               A-17

-------
      Group  E:    Evidence  of Non-Carcinoqenicity for  Humans.

      This category  is  used for  agents  that show  no  evidence of
 carcinoqenicity  in at least two adequate  animal  tests in different
 soecies  or  in both adequate eoidemiological  and animal studies.
 This  designation is  based  on  the available evidence  ind  should
 not be  interpreted as a definitive conclusion that the agent will
••not be  a carcinogen under any  circumstances.

-Pr3sentlv,  there is  no  similar  scheme  Cor assessing  the weight-
'of-evidance for  agents causing non-cancer  health effects.3

      0 Quantitative evaluation

        In general,  the  EPA Carcinogen  Assessment  Group   (CAG)
       estimates the cancer potency of agents that are classified
       Group B  and  higher.  Potency is  expressed as  the   "unit
       risk". This unit risk estimate for an air or water oollu-
       tant  is defined  as  "the  incremental  Vifetime  cancer risk
       occurring in a hypothetical population  in which all  indi-
       viduals are  exposed  continouslv   from  birth  throughout
       their lifetimes to a concentration  of 1 uq/m3 of the  agent
       in the  air they  breathe,  or  to  1  uq/L  in  the  water they
       drink.  -This calculation  is done  to estimate in quantita-
       tive terms the impact of  the  agent as a carcinogen.  Unit
       risk estimates are  used  for  two  purposes:  (1)  to compare
       the  carcinogenic potency of several agents with each other,
       and  (2) to give a  crude indication of the population risk
       that might be  associated  with air  or  water  exposure to
       these agents, if the actual exposures are known."4

      0 Key  simplifying assumptions of the  quantitative evaluation

       In most cases, CAG  uses a so-called linearized multistage
       model to  calculate  chemical-specific  cancer  potency  or
       unit risk  estimates.   A  mathematical  model is  needed in
       order to  calculate the unit risk  factor because animal and
       human carcinogenicity data are always associated with  exoo-
       sure levels that are much higher  than those normally  found
       in environmental settinqs.   Therefore,  in  order  to  esti-
       mate potential cancer risk for the low exposure levels of
       concern,  it is necessary  to  extrapolate from data associ-
       ated with  high  exposures.  This  is done through  the use
       of aporopriate models.

       In explaining their choice of model, CAG notes the follow-
       ing  important considerations:

            "The unit risk  estimate   represents  an extrapolation
       below the doss range of experimental data.   There is cur-
       rently no  solid   scientific   basis  for   any  mathematical
                               A-18

-------
       extrapolation model that relates exposure to cancer risk
       at the extremely low concentrations,... that must be dealt
       with in evaluating environmental hazards.  For practical
       reasons the correspondingly low levels of risk cannot be
       measured directly either by animal experiments or by epi-
       demiological studies.  Low-dose extrapolation must, there-
       fore, be based on current understanding of the mechanisms
       of carcinogenesis.  At the present time the dominant view
       of the carcinogenic process involves the concept that most
       cancer-causing agents also cause irreversible damage to
       DNA.  This position is based in part on the fact that a
       very large proportion of agents that cause cancer are
       also mutagenic.  There is reason to expect that the quan-
       tal response that is characteristic of mutagenesis is
       associated with a linear (at low doses) nonthreshold
       dose-response relationship.  Indeed, there is substantial
       evidence from mutagenicity studies with both ionizing
       radiation and a wide variety of chemicals that this type
       of dose-response model is the appropriate one to use.  This
       is particularly true at the lower end of the dose-response
       curve; at high doses, there can be an upward curvature,
       probably reflecting the effects of multistage processes
       on the mutagenic response.  The low-dose linear nonthres-
       hold dose-response relationship is also consistent with
       the relatively few epidemiologic studies of cancer respon-
       ses to specific agents that contain enough information to
       make the evaluation possible (e.g., radiation-induced leu-
       kemia, breast and thyroid cancer, skin cancer induced by
       arsenic in drinking water, liver cancer induced by afla-
       toxins in the diet).  Some supporting evidence also exists
       from animal experiments (e.g., the initiation stage of the
       two-stage carcinogenesis model in rat liver and mouse skin).

     Because its scientific basis, although limited, is the best
of any of the current mathematical models, the nonthreshold model,
which is linear at low doses, has been adopted as the primary
basis for risk extrapolation to low levels of the dose-response
relationship.  The risk estimates made with such a model  should
be regarded as conservative, representing a plausible upper limit
for the risk; i.e., the true risk is not likely to be higher than
the estimate, but it could be, and probably is, lower.

     For several reasons, the risk estimate based on animal bioas-
says is only an approximate indication of the absolute risk in
populations exposed to known carcinogen concentrations.   First,
there are important species differences in uptake, metabolism,
and organ distribution of carcinogens, as well as species diffe-
rences in target site susceptibility, immunological response,
hormone function, dietary factors, and disease.  Second,  the
concept of equivalent doses for humans compared to animals on a
                              A-19

-------
 mg/surface area basis  is  virtually  without  experimental  verifica-
 tion as regards carcinogenic  response.   Finally, human popula-
 tions are variable with  respect  to  genetic  constitution  and
 diet, living environment,  activity  patterns, and other cultural
 factors.

      The  risk estimate can give  a rough  indication of the  rela-
 tive potency of a  given agent compared with other carcinogens.
 Such estimates are,  of course, more reliable when the comparisons
 are  based on studies in which the test species, strain,  sex,
 and  routes of exposure are similar.

      The  mathematical  formulation chosen to describe the linear
 (at   low   dose)  nonthreshold dose-response  relationship  is the
 linearized multistage  model.  This  model employs enough  arbitrary
 constants to be able to fit almost  any monotonically increasing
 dose-response data,  and it incorporates a procedure for  estima-
 ting the  largest possible  linear slope (in  the 95% confidence
 limit sense)  at  low  extrapolated doses that is consistent with
 the  data  at all  dose levels of the  experiment." 4

      The  EPA Carcinogen Risk Assessment Guidelines indicate  the
 following concerning the choice of  extrapolation model:

      "The Agency will  review each assessment as to the evidence
 on carcinogenesis  mechanisms and other biological or statistical
 extrapolation model.   ...A rationale will be included to justify
 the  use of  the chosen  model.  In the absence of adequate informa-
 tion to the contrary,  the  linearized multistage procedure  leads
 to a plausible upper limit  to the risk that is consistent with
 some proposed mechanisms of carcinogenesis.  Such an estimate,
 however,  does not  necessarily give  a realistic prediction of the
 risk.  The  true  value  of the risk is unknown, and may be as  low
 as zero."3

      B. Mutagens

      In the  IEMP studies,  we restrict our evaluation of mutagen-
 icity  to  a  qualitative assessment.   We simply indicate if the
 chemical  is considered to  be a mutagen or if there is insuffi-
 cient  dat'a  to make such a  determination.  Most of the experimen-
 tal  dates addressing muta'genicity involve in vitro testing of
 mammalian cell cultures and non-mammalian organisms such as
bacteria.   At  present,  we  do not have methodologiess for using
 these  kinds of data  to calculate dose-response curves amd quanti-
 fy risk to humans.    EPA  assumes that a dose-response relation-
 ship  for mutagenicity,  like carcinogenicity, would have no thres-
hold;  therefore, any exposure Level would be associated with
 some degree of risk.  Evidence of mutagenicity often, but not
always, suggests carcinogenic potential.
                               A-20

-------
      C.  Non-Carcinogens

      Chemicals that give rise to toxic endpoints other than cancer
 and  gene  mutations are  often  referred to  by EPA  as  "systemic
 toxicants" because  they  affect the  functioning of  various organ
 system, such  as  liver,  kidney, cardiovascular,  and respiratory
 systems.  Generally, based  on our understanding of physiological
 mechanisms, systemic  toxicants  are  treated  as   if  there  is  an
 identifiable exposure  threshold (both for the individual and for
 the  population)  below  which  adverse  effects  are  not observable.
 From the  viewpoint of risk  assessment  and risk management, this
 characteristic distinguishes  systemic toxicity from carcinogeni-
 city and  mutagenicity,  since the latter  two  are  usually treated
 as non-threshold processes.

      Because of this concept  of  identifiable  thresholds, systemic
 effects traditionally have been  evaluated by  EPA through the cal-
 culation of  "safe" exposure  levels.  This  is unlike  EPA's  ap-
 proach for  evaluating  carcinogens,  for which all  exposures  are
 assumed to  involve some  measure of  risk.   For  many  years,  the
 concepts of "acceptable daily intake"  (ADI)  and  "margin of safe-
 ty"  have  been  at the heart  of  EPA's approach to risk assessment
 for  non-cancer health  effects.  Although there  are  limits  to
 some  of these approaches, EPA is often called upon to apply these
 concepts when  making  and  explaining  decisions  concerning  the
 significance to human health  of  certain chemical exposures  in the
 environment.  Thus, the threshold  concept for non-cancer effects
 is extremely important in the regulatory and  risk management con-
 text.

      More recently, the  Agency has  tried to come  to  grips with
 some  of the technical and philosophical issues inherent in defin-
 ing  a "safe"  exposure  level  for systemic  toxicants.   As  an out-
 growth of this effort,  the  concept of the "reference dose" (RfD)
has been  recommended  to  replace that  of the  ADI.   The  RfD  is
 calculated in the  same manner  as  an ADI; however,  it  is  viewed
 as a  benchmark  level  without  the  value-laden   connotations  of
 absolute safety or acceptability versus absolute unacceptability.
Thus, exposures that are  less than the RfD  are  not likely to be
 associated with non-cancer  health  effects and are therefore less
 likely to be of regulatory concern.  Conversely, as  the frequency
of exposures exceeding the  RfD increases and as  the  size of the
 excess increases,  the probability   also  increases  that  adverse
 effects may be  observed  in  a human  population.   Nonetheless,  a
 clear conclusion  cannot  be  categorically  drawn  that  all doses
below the RfD  are  "acceptable" and  that  all  doses  in excess  of
 the RfD are "unacceptable."^

      To date,  the  Agency's   calculation and use of ADIs and RfDs
has been applied only  to oral  exposure  routes,  such as drinking
                               A-21

-------
water and  food.  EPA has an established methodology for calculat-
ing oral  RfDs  and  there are values  for approximately  400 chemi-
cals.  In  contrast,  the Agency does  not  have a  methodology for
calculating similar  benchmark  levels  for  inhalation  exposures.

     EPA has not yet  established  a procedure for estimating such
benchmark  values for  chronic inhalation exposures;  a methodology
is under  development  by  a special  committee in FY'87  and the
values will be  called inhalation  RfDs.  Thus,  with the exception
of the  six  or  seven heavily  studied criteria air  pollutants
(ozone, carbon monoxide, sulfur oxides,  etc.) there are no bench-
mark values for chronic inhalation exposures that are appropriate
for use in risk assessment.  Because the Agency's RfDs are calcu-
lated for  lifetime oral exposures  only, and IEMP studies examine
both oral  and  inhalation  exposure  routes, it has  been necessary
for us to estimate  IEMP  benchmark  levels for inhalation exposures.

     IEMP studies rely on  EPA oral RfDs and on other "threshold"
values computed by  our toxicologists and consultants.  Our "thres-
holds" for various  non-cancer effects have been calculated employ-
ing the same methodology  used to  estimate oral RfDs.  Highlights
of the RfD methodology are as follows:

       RfDs are calculated by collecting the available
    animal and human data, noting the various dose
    levels (in milligrams per Kg body weight per day)
    at which different non-cancer health effects  are
    seen,  and identifying the highest NOEL (No Observed
    Effect Level).   The  highest NOEL represents the
    highest dose at which biologically or  statistically
    significant effects  were not seen.   The scientists
    also try to identify the LOEL  (Lowest   Observed
    Effect Level),   which is the lowest dose at which a
    biologically   or  statistically  significant   non
    cancer effect was seen.

       Many different NOELs and LOELs can  be identified
    for   each  chemical,     depending  on  the   doses
    selected,   the   spacing  between the doses and  the
    health   effects  examined   and  reported  by   the
    researchers.   Toxicological   research   is    very
    expensive and each experiment cannot be  exhaustive
    in  terms  of  the number of doses tested  and  the
    number    of    effects    studied.      Similarily,
    epidemiological  research is expensive and  subject
    to many problems with both  measurement of exposures
    and  identification   of  exposure-related  effects.
    This  problem introduces a degree of uncertainty  in
    the identification of the "true" NOEL  and LOEL of a
    chemical  and  this  uncertainty is carried  through
    the calculation of the RfD.
                               A-22

-------
        The  RfD is calculated by dividing the  highest
    reliable   NOEL  (or  lowest  reliable   LOEL,    if
    appropriate) by uncertainty factors.   The selection
    of   appropriate  uncertainty  factors  is   based,
    primarily,  on  the nature of the study from  which
    the NOEL, or LOEL,   was derived.   Each factor has a
    value  from 1-10 and represents an extra degree  of
    uncertainty  to account for:  1)  extrapolation from
    the average human to sensitive members of the human
    population,   2)  extrapolation  from  the  average
    animal to the average human,  3)  extrapolation from
    subchronic  exposures to chronic exposures  and  4)
    extrapolation from the LOEL to the NOEL.5

     By definition, the RfD is based on the critical non-cancer
health effect.  This is the effect first seen as exposures in-
crease above zero.  Since IEMP studies often examine exposures
that are above the RfD for a particular chemical, it would be
valuable to know what kind of effect(s) might be seen as expo-
sures get higher and higher above the RfD.  This type of infor-
mation is not obtainable from the single RfD that is calculated
for a chemical.  To help solve this problem, we have taken the
following approach in our IEMP studies:

       1)  non-cancer health effects are divided into the six
    broad health effect categories of liver, kidney, reproduc-
    tive, neurobehavioral, fetal developmental, and other.  The
    category "other" includes a variety of effects such as cardi
    ovascular, respiratory, and gastrointestinal.

       2) for each chemical of interest, examine both the oral
    and inhalation data for humans and animals.

       3) using the RfD methodology, and route-specific toxicity
    data, we calculate a separate inhalation and oral  "threshold
    for each of the six health effect categories.  Obviously, the
    more toxic a chemical, and the more it has been studied, the
    more  health effect   categories  for  which  we  can esti-
    mate "thresholds".

2. ) Assessing the Level of Exposure

     The second key element  (see p. A-16) in  estimating health  risk
is determining the level of  exposure to that  chemical.  Most of
our exposure assessments attempt to estimate  ambient concentra-
tions of substances in the air and drinking water.  We then make
certain assumptions about how ambient concentrations relate to
actual human exposure or dose level.  For example, we take no
account of population mobility.  These assumptions — the exposure
constants regarding how much air a person breathes or water he
or she drinks--are described below.
A-23
                                  "

-------
      We  estimate ambient concentrations in two ways:   direct moni-
 toring or  simulation modelling.  There are a  number of advantages
 to  the modelling of ambient concentrations over direct monitoring
 alone:

      o  Modelling  may  provide the only  way  to estimate  ambient
         concentrations under alternative  exposure scenarios.

      o  Modelling  can  take into account the  geographic variabi-
         lity of a  large  area.   Because monitoring data are often
         from  a  few  specific  points,   they  best  serve  as refe-
         rence points for evaluating the  performance  of  air dis-
         persion models.

      o  Modelling  is often less costly than extensive monitoring.

      o  Modelling  enables  predictions  of exposures  in any loca-
         tion (and,  in  particular,  the  location of  the  most ex-
         posed individual),  whereas monitoring can  only provide
         an indication  of the  exposures  in  the vicinity  of the
         sampling sites.

      o  Modelling  links  concentration  estimates,  and hence expo-
         sures, to  sources.  Such source information is important
         as a  risk  management tool in  that  it allows us to esti-
         mate the  impact of  various  pollution control  options.

      o  For  some  pollutants,   there  are  no accepted methods for
         monitoring them in ambient air.

     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
receptor.  Building such  a model thus  requires an understanding
of those processes,  which  is not  necessary  if  one  can simply
monitor  ambient concentrations directly.

     When carrying out an IEMP, we first employ  existing data-
bases which have  been  reviewed for   currency and  quality.   We
conduct monitoring  as   an  adjunct as   resources  allow  and use
monitoring data to verify ambient concentrations  estimated from
our models.   Under  certain  circumstances,   such  as  situations
where we do not  know sources  and  emissions,  monitoring data are
crucial  for estimating  exposures.   Furthermore, an  advantage of
reliable, long-term 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 contami-
nation is suspected.
                               A-24

-------
     The  key steps in modelling exposures are:

       (1) estimating the number of sources of emissions;

       (2) gathering data on emissions (in g/sec.) of target
           compounds from selected sources;

       (3) gathering release specifications for chemical emis-
           sions, such as building dimensions, stack height,
           location of release points;

       (4) characterizing the processes and pathways by which a
           pollutant is transported in air, soil, and water,
           including the speed of transport, the extent of dilu-
           tion or dispersion, and any chemical transformation
           the pollutant might undergo (such as degradation to
           a nontoxic, or possibly even more toxic form).

       (5) locating receptor sites; and

       (6) assigning populations to receptor sites.

     We try to estimate resulting ambient concentrations at
various distances from the source.   Estimating long-term average
concentration levels,  which are of concern for evaluating chronic
health impacts,  is simpler than short-term modeling, which must
take greater account of variations in meteorological conditions.


RISK CHARACTERIZATION

     As noted on pages A-12-13, risk characterization is the part
of risk assessment that brings together the exposure assessment
and the dose-response information in order to estimate the inci-
dence of  an adverse effect in a given population.  As further
noted,  because of limitations in our ability to quantify dose-
response  relationships for each of our health effect categories,
we  must  analyze cancer differently from the non-cancer health
effects.   For cancer,  we treat the CAG unit risk factor as the
slope of  the dose-response curve; therefore, we can use this
value with the results of our exposure assessment to estimate a
crude incidence.  For the non-cancer health effect categories,
we do not have estimates of the slopes of the dose-response
curves.  Instead, we have RfDs and our own "thresholds" for each
of the effect categories.  Therefore, for non-cancer, instead of
calculating the incidence of effects, we calculate the number of
people exposed to levels of chemicals that are above the RfDs and
other "thresholds" in order to estimate how many may be at some
increased health risk.  For each chemical of interest, we examine
each of the six non-cancer health effect categories separately.
                               A-25

-------
      If  exposures to a  substance  are below the threshold  for a
 particular non-cancer  health  effect  category,  then  we  expect
 little increased  risk  of effect from that substance.   If  expo-
 sures exceed  the  estimated  threshold  for a  particular  health
 effect category,  we  then indicate  the  possibility of  increased
 risk of that effect  and  try  to estimate the size  of  the popula-
 tion exposed  to  such  concentrations.   Exposures  exceeding  an
 estimated threshold  are  generally  more  appropriate subjects  for
• further investigation.

      Note that the use of GAG  potency estimates to evaluate can-
 cer risk and the  use of  RfDs to identify exposures of  potential
 concern for non-cancer  health effects  are relatively straightfor-
 ward practices and have been used throughout EPA for  many years.

 Risk Characterization as  Applied to an IEMP

      In IEM studies,  we calculate cancer risk using  three measures
 of exposures:    risk  to  the most exposed  individual  (MEI),  risk
 to the  average exposed  individual (AEI),  and the excess  aggregate
 population incidence.  For all  calculations, we use the  following
 standard assumptions:  1) the  average lifetime is  70 years,  2)
 the average adult  breathes 20  cubic  meters of  air per  day,  and
 3) the  average adult  drinks 2 liters of  water per day.

      We express individual risk  as  either  MEI  risk or  AEI  risk.
 We define risk  to  the  MEI as  the  increased probability  that  an
.individual exposed to the highest  concentration of one  or  more
 chemicals  will  have exposure-related cancer during  the  course  of
 his or  her lifetime.  For the MEI,  the exposure is  calculated  as
 either  the highest modelled concentration or the average concen-
 tration obtained for the monitoring site with the  highest  daily
 or annual  average monitored value for  one or more  chemicals.   We
 define  risk to  the AEI as the increased  probability that an  indi-
 vidual  exposed  to the areawide average  concentration  of one  or
 more  chemicals  will  contract cancer during the course of his  or
her lifetime.   Aggregate  population risk  is the  estimate of  the
 increased  incidence  of  cancer,   above  the  background rate,  in  an
 exposed population.

      To  estimate the risk to  the   most exposed  individual,  we
•typically  need  to know how far  an individual lives  from  the  maxi-
mum pollutant  concentration  near a  source.   To estimate average
 individual  risk,  we estimate  an average pollutant  concentration
to determine exposure.  To estimate population risk, we must  iden-
tify  the number  of people exposed to a given pollutant concentra-
tion.   In  our exposure estimates, we are assuming that people are
exposed to  outdoor ambient air concentrations  24 hours a day  for
a  lifetime.  In  reality,  people spend  most of their time  indoors,
either at  home  or at  work.    While  our  assumption  overstates
                               A-26

-------
 actual  exposure  to  outdoor air,  the bias introduced by this pro-
 cedure  may  not  be  too  great,  since  recent EPA  (TEAM)  studies
 show  that  many  outdoor  air contaminants are  also found indoors
 at equal  or  greater  concentrations.    This  represents  current
 standard practice at EPA.

      Under these assumptions  and  exposure  scenarios,  lifetime
 cancer  risk  to the exposed  individual is  simply  the  product of
 exposure and potency:

             R                 E        x       P              (1)
    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 constants (i.e., the  standard assump-
 tions of  body  weight,   breathing  rate  and  water consumption):

    E          •             Y          x             Z         (2)
 exposure        ambient concentration      exposure constants
                 in medium  of  concern

 and

       R       =         Y       x      Z        x          P      (3)
   individual        ambient        exposure           potency
      risk        cone,  in medium    constants           factor
                    of concern

                       SAMPLE CALCULATION

     The following is a simple example to illustrate how we would
 calculate the  lifetime  risk to the MEI, as well  as the lifetime
 risk to a population, associated  with an ambient air exposure of
 ten micrograms  per  cubic   meter  (ug/m3)  of  benzene.   We  have
 divided the  geographic  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 air concentrations and popula-
 tion data:

                          Annual Average


Grid *
1
2
3
4
Benzene
Concentration
(ug/m3 )
10
8
8
5

Exposed
Population
20,000
30,000
25,000
25,000
                               A-27

-------
     The MEI risk of cancer for an individual exposed to 10 ug/m3
of benzene in the air (the highest benzene exposure) is calculat-
ed by  using  equation (3).  The  ambient concentration  (Y)  is  10
ug/rH, which is  equivalent to 0.01 milligrams  per  cubic -neter  or
0.01 mg/m3.  The  potency  factor  (P)  for  inhaled  benzene,  devel-
oped by  EPA's  Carcinogenic  Assessment  Group  (GAG),  is  0.029
(mg/kg/day)-1.  This value means that  an  individual inhaling one
milligram  (mg) of benzene per kilogram body weight per day for a
lifetime has  an   estimated  increased  probability   of  developing
cancer of  about  three  in  one  hundred  (upper-bound  estimate).

     Using equation  (3)  and  the  maximum  ambient  concentration
value of 0.01 mg/m3, we calculate the risk to the MEI as follows:

 R      =     Y         x       z          x        P          (4)
MEI       Avg.  annual       exposure           potency
Risk     ambient cone.       constants          factor

where, the exposure  constants (Z)  are  adult body  weight (70 kg.)
and adult breathing rate (20 m3/day)

therefore,

 R  = 0.01 mg/m3   x  (20 n3/day x 1/70 kg)  x 0.029 (mg/kg/day)-1
    = 8.3 x 10-5

The lifetime upper-bound estimate  of risk to the MEI (R) in this
example is 8.3 "x -10-5  or roughly  eight  chances   in  100,000  of
developing cancer over a lifetime given constant exposure, to 0.01
mg/m3 per day.

     The risk to  the average  exposed  individual  (AEI)  is  calcu-
lated the  same  way, using  an average exposure determined  from
either monitoring data or air dispersion models.

     The final step  in  our  example  is  to estimate the increased
incidence of  cancer in  the  total population  in   our  four  grid
geographic area.   To do this,  we multiply the  risk to the  AEI  in
each grid  by  the number  of people  in that  grid.   We  then sum
these estimates  of  incidence  across all  grids to  calculate the
lifetime aggregate  incidence  of  cancer   for  our four  grids.

     Generally,  EPA presents incidence  as the  expected number  of
excess cancer cases  per year.  Dividing the upper-bound lifetime
estimate by  70,  we  arrive  at  the  upper-bound  estimate  of the
annual average number of excess cancer cases in the population of
concern.

     We estimate   individual and  population  risks   from ingestion
of drinking water in exactly  the same way  we  estimate the risks
from inhalation.   The only difference is that the potency factors
                               A-2 8

-------
 have different  values and we  use  the water consumption exposure
 constant  of  2 Liters/  day.   Because  the  potency  of a chemical may
 differ  for different  routes  of exposure, we must use the potency
 factor  that  is  appropriate for ingestion.   This  potency factor is
 provided  to  us by the Carcinogen Assessment Group.  Concentrations
 of  selected  chemicals  in  the drinking water are  measured or esti-
 mated and then  used  with the  oral  potency factors  in equation
 (3)  to  estimate risk.   We calculate  both maximum individual risk
 and  aggregate excess  incidence  in  a population in  the same way
 indicated in our  four  grid example.

 Risk Characterization  as  Applied to  the  Baltimore IEMP

     The  Baltimore  Phase  I  activities   emphasized  cancer  as the
 primary health  effect  of  concern but some  attention was given to
 non-cancer effects  of chemicals that were  addressed as carcino-
 gens.   In Phase II, we will  more broadly consider noncarcinogenic
 effects to the  extent  that available data  will allow.   We empha-
 sized cancer in Phase  I because  the  public  has expressed frequent
 concern about the possible link  between  exposure to environmental
 pollution and the incidence  of cancer Also,  the  quantitative risk
 assessment methodology for cancer has more  scientific credibility
 within EPA   than  do  similar  methods  for  assessing  non-cancer
 effects.
INTERPRETING RISK ASSESSMENT RESULTS

     The  estimates of  individual health  risk and aggregate inc-
idence from  exposure to toxics should not be interpreted as pre-
cise or absolute estimates of future health  effects.  The simpli-
fying assumptions  and uncertainties  in both  the toxicology and
exposure components are simply  too great to justify a high level
of confidence in the precision of the results.

     The potency and  the  threshold estimates  used  in this study
are consistently  conservative in the direction of overestimating
risk; they may  overestimate  the  likely  effects of chemical expo-
sure but are  unlikely to underestimate  them.   Such an estimate,
however, does not  necessarily  give a realistic prediction of the
risk.  The true value  of  the risk is unknown,  and  may be as low
as zero."  This  leads  to  assessments 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 expo-
sure assessments,  as  is appropriate  in a priority-setting exer-
cise.  It is important to read the detailed  chapters carefully to
understand what confidence to place in a particular  risk estimate.
                               A-29

-------
     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 identi-
fy the  sources and  pollutants  of  greatest  concern  and address
those that  data  and methods  allow  us to  cover.   We  acknowledge
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  indistinguishable.   On  the other hand,  for example,
it is reasonable to conclude that a  source projected to result in
one case per year  represents less risk to the  population than a
source projected  to  result in ten  cases  per year.   Despite the
uncertainties,  our risk estimates are useful for roughly assess-
ing the potential magnitude  of the  overall risks from particular
pollutants,  sources,  and  pathways;   comparing  issues  with  one
another; and setting  priorities  among  environmental  issues and
concerns.
                               A-30

-------
                 REFERENCES FOR APPENDIX A


 1.    National Research Council (1983).  Risk Assessment in the
 Federal  government:   Managing  the  Process.    Committee  on  the
 Institutional Means  for Assessment  of  Risks  to  Public  Health,
 Commission on Life Sciences, national Academy of Sciences.
 Rational Academy Press, Washington, D.C.

 2.    International Agency for Research on Cancer (1982).  IARC
 MONOGRAPHS on the  EVALUATION  OF  THE CARCINOGENIC  RISK  OF CHEMI-
 CALS  TO  HUMANS.   World  Health  Organization,   Volumes  1  to 29,
 Supplement 4, pp. 13, 14.

 3.  U.S. Environmental  Protection  Agency  (1986).   Guidelines for
 Carcingenic Risk Assessment.  Federa'l Register, Vol. 51, Mo. 135,
 33993 -  34003.

 4.  U.S. Environmental  Protection  Agency  (1985).   Health Assess-
 ment Document for 1,2-Dichloroethane (Ethylene Dichloride).
 Office of Health  and Environmental Assessment;  Washington,  D.C.
 pp. 9-235 to 9-241.

 5.  U.S. Environmental Protection Agency (1986).  Reference Doses
 (RfD):  Description  and Use  in  Health Risk  Assessments.   Draft
prepared by  the  RfD  Workgroup   for  the  Risk  Assessment  Forum,
May 1986.
                               A-31

-------
           APPENDIX B








SUMMARY OF AMBIENT AIR MONITORING






   PROVIDED BY THE MARYLAND AMA

-------
                            MONITORING SITES



                 BALTIMORE AND ANNE ARUNDEL COUNTIES



KEY                       SITE NAME



CG                        Coast Guard



0                          Dundalk



P                          Parkville



20                         Glen Burnie



23-                         Riviera Beach



28                         Essex



32                         Dundalk



33                         Chesapeake Terrace Elementary

-------
FIG. 1 - MONITOR SITES - BALTIMORE AND ANNE ARUNDEL COUNTIES

-------
                               MONITORING SITES



                                BALTIMORE CITY



KEY                          SITE NAME



BM                 Baltimore Museum of Industry



BP                 Baltimore Polytechnic Institute



LF                 Lacy's Foundry



P7                  Pier 7



SG                 Spring Gardens



35                  Fire Dept. Headquarters



36                  Old Town Fire Station



37                  Allegany Pepsi



38                  N. E. Police Station



40                  S. E. Police Station



41                  S. W. Police Station



42                  Sun Sreet



46                  Holabird Elementry School



47                  Canton Pier II



48                  Guilford



50                  Fort McHenry



54                  1-95 & Moravia Road

-------
                         FIG.  2 MONITOR SITES - BALTIMORE CITY
                                                                 'HERRING   .GaraenvTiie  :j  ,
                                                                                     «-•> I
to»J.  Park,.0!ll, 'il^'^LS
n I      	'   ^—S  '^~^T««,« v. t
                                 ^^^^^2^M^"
                                                               Fairficlcl


                                                           ^  'V-Curlis Qny
                                                           —   i •   , %i   p

-------
                        DOCUMENTATION OF DATA (Cont'd)
 Fire Dept. - Headquarters


 Old Town Fire Station



 Allegany Pepsi


 N. E. Police Station


 S. £. Police Station


 S. W. Police Station
 Sun Street
Holabird Elementary
Canton Pier II
Guilford
Fort McHenry
1-95 and Moravia Road
 Maryland Air Management, 1986 Maryland Air
 Quality Data Report, 1986, p 21.

 Maryland Air Management, 1986 Summary of
 Baltimore, Maryland Vinyl Chloride Air Sampling,
 September - October 1986, p. 3.

 Maryland Air Management, 1986 Maryland Air
 Quality Data Report, 1986, p. 21.

 EPA, 1987, Baltimore Integrated Environmental
 Management Project, Phase I Report, Table V-3

 Maryland Air Management, 1986 Maryland Air
 Quality Data Report, 1986, p. 22

 Maryland Air Management, 1986 Maryland Air
 Quality Data Report, 1986, p. 22

 EPA, 1987 Baltimore Integrated Environmental
 Management Project, Phase I Report, Table V-3

 Maryland Air Management, 1986 Summary of
 Baltimore, Maryland Vinyl Chloride In Air
 Sampling, September - October 1986, p. 3.

 Maryland Air Management, 1986 Summary of
 Baltimore, Maryland Hydrocarbon and Chlorinated
 Hydrocarbon Air Sampling February - September,
 1986, Table 2

 EPA, 1987 Baltimore Integrated Environmental
 Management Project, Phase I Report, Table V-3

 EPA, 1987 Baltimore Integrate Environemtnal
 Management Project, Phase I Report, Table V-3

 Maryland Air Management, 1985, Maryland Air
 Quality Data Report, 1985, p. 32

 EPA, 1987 Baltimore Integrated Environmental
 Management Project, Phase I Report, Table V-3

 Maryland Air Management, 1986 Maryland Air
Quality Data Report, 1986, p. 21

EPA, 1987 Baltimore Integrated Environmental
Management Project, Phase I Report, Table V-3

Maryland Air Management, 1986 Maryland Air
Quality Data report, 1986 p. 21

-------
                            DOCDMENTION OF DATA
Baltimore Museum of Industry




Baltimore Polytechnic Institute



Coast Guard


Dundalk
Lacy's Foundry
Parkville
Pier - 7
Spring Gardens


Glen Burnie


Riviera Beach


Essex


Chesapeake Terrace Elementary
Maryland Air Management, 1987 Determination of
Ambient Background Concentrations of Total
Suspended Paniculate and Chromium in the
Vicinity of the Allied Chemical Plant, Table 2.

Conney, Maryland Air  Management  1986 sorbent
tube sampling at Polytechnic Institute for acetic
acid and other volatile organics, Table 1

EPA, 1987, Baltimore Integrated Environmental
Management Project, Phase I Report, Table V-3.

EPA, 1987 Baltimore Integrated Environmental
Management Project, Phase I Report, Table V-3.

EPA, 1988 Baltimore Integrated Environmental
Management Project. Phase n Draft Final Report,
Ambient Air Toxics, Table IV-9.

Maryland Air Management, 1987 Determination of
Ambient Background Concentrations of Total
Suspended Particulate and Chromium in the
Vicinity of the Allied Chemical Plant, Table 2.

EPA 1988  Baltimore Integrated Environmental
Management Project, Phase II Draft Final Report,
Ambient Air Toxics, Table IV-9

Maryland Air Management, 1987 Determination of
Ambient Background Concentrations of Total
Suspended Particlate and  Chromium in the
Vicinity of the Allied Chemical Plant, Table 2.

O'Melia, 1987 Sampling Results from BGdcE Spring
Gardens facility.

Maryland Air Managment, 1986, Maryland Air
Quality Data Report, 1986, p 20.

EPA, 1987 Baltimore Integrated Environmental
Management Project, Phase I Report, Table V-3

Maryland Air Management, 1986 Maryland Air
Quality Data Report, 1986, p 22.

EPA, 1987 Baltimore Integrated Environmental
Management Project, Phase I Report, Table V-3

-------
                                 AVERAGE MBASUKKU AMUIENT CONCENTRATIONS: VOLATILE OKGANICS (ug/m3)






CHEMICAL                  BN[    BP     CGD      LF£      P7SG2023283233353637



ARSENIC                                                                             .0017          .0016                 .0016




BENZENE                                   7.8     3.9            5.2                          10.3           12.6    5.5



WiNZO-A-PYRENE                                                             .002    .92            .92                   .95




CADMIUM                                                                            .001           .001                  .001




CARUONTETRACIILORIDE                   .7      .2             1.1                          .6             .9      .9




CHLOROFORM                              .6      .1             ND                          I.I            |.      3.2




CHROMIUM                 .002                         .018           .004           .007           .007                  .007




1,2  IJICI1LOUOETIIANE                      1.0                                               .2             .4      2.6




1,2  DICIILOROPROPANE                     .4                                                .2             .3      .1




ETHYL BENZENE                           5.2                                               6.3            4.7     2.9




LEAD                                                                                .058           .060                  .084           .074



METHYLI-NE CHLORIDE                     4.l->




MI-THYL ISOUUTYL KETONE




PEHCIILOROETIIYLENE                     2.9     .6             2.1                          2.4            9.3     1.5




TOLUENE                          5.4     5.5                                               7.4            9.3     7.5



1,1,1 TRICIILOROETIIANE                          1.5            5.9




TRICIILOROETHYLENE                     .2      .4             .01                          .3             1.4     .4




VINYL CHLORIDE                                               ND                                                             NL>




XYLENES                          8.35    13.6   5.5            15.4                         19.2           13.7    5.7

-------
                             AVERAGE MEASURED AMBIENT CONCENTRATIONS: VOLATILE OKGANICS (ug/m3)
CHEMICAL



ARSENIC



DENZENE



HENZO-A-PYRENE



CADMIUM




CAUDONTETRACHLORIDE




CIII.OROFROM




CHROMIUM




1.2 DICULOUETIIANE




1,2 DICHLOROPROHANE




ETHYL BENZENE




I,I-AO




METIIYI.ENE CHLORIDE




METHYL ISO8UTYI, KETONE




PERCIILOROETIIYLENE




TUI.UKNE




 1,1,1 TKICIILORETUANE




TIUCtll.OUOl-THYLENIi




 VINYL CHLORIDE




 XYLENES
38 40

9.5
.96
.001
.9
.4
.009 .018
.2
.2
6.4
.111 .143


5.4
9.0

.5

21.1
41

10.2


.6
.2
.011
.3
.2
8.0
.057


7.0
9.9

.5

20.8
42 46
12.1 12.9


1.2 1.3
2.6 .7

.7 .5
2.0 .7
6.2 7.4

.7
.3
3.2 6.0
7.6 16
.6
l.l 3.9
ND
17.2 20.1
NO. of
47 48 SO 54 Sites
3
12.0 10.6 12
5
.002 5
I.I 1.4 12
2.1 4.7 12
.016 10
.2 .2 10
.3 .4 10
9.3 5.7 10
.079 .059 .081 10
2
1
4.8 3.9 12
16.4 4.8 1 1
3
.9 1.0 12
3
30.4 16.4 13
Total
.0049
112.6
3.752
.006
10.9
16.7
.099
6.3
4.8
62. t
.806
4.8
.3
49.1
98.8
8
10.61
-
207.45
AVG. COMENTS
.00163
9.38
.75
.0012
.91
1.39 ND: NO DETECTABLE LEVEL
.01
.63
.48
6.2
.081
2.4
.3
4.1
8.98
2.7
.88
ND: NO DETECTABLE LEVEL
15.90

-------
            APPENDIX C








SUMMARY OF CONTROL OPTIONS COST AND



       REMOVAL  EFFICIENCIES

-------
TABLE C-1 DESCRIPTION Of CONTROL OPTIONS for POINT SOURCES
Source
Point Source A








Plant Area
A&B Plants



Benzene/Litol
Plani


Storage Tanks
Mills
Release Pt.
Tar decanter, tar intercepting
sump, flushing liquor
circulation tank
Tar storage tanks, tar
dewatering tanks
Excess ammonia liquor
storage tanks
Light oil condenser
Light oil sump
Benzene condensor
Light oil benzene mixture
storage tanks
Benzene storage tanks
Chrome plating line
Option/Alternative
Current Control
Alternative 1
Current Control
Alternative 2
Alternative 3
Cui rent Control
Alternative 4
Alternative 5
Onrrnni Conjrol
.-:6
Current Control
Alternative 7
Current Control
Alternative 8
Current Control
Alternative 9
Alternative 10
Current Control
Alternative 11
Alternative 12
Current control
Alternative 13
Option Description
Uncontrolled
Coke oven gas blanketing system
Uncontrolled
Wash Oil Scrubber
Coke oven gas blanketing system
Uncontrolled
Wash Oil Scrubber
Coke oven gas blanketing system
Uncontrolled
Coke oven gas blanketing system
Uncontrolled
Cover
Uncontrolled
Coke oven gas blanketing system
Uncontrolled
Wash Oil Scrubber
Coke oven gas blanketing system
Uncontrolled
Wash Oil Scrubber
Coke oven gas blanketing system
Packed bed wet scrubber
Packed bed wet scrubber
Efficiency
97
90
97.9
90
97.7
97.9
98.2
97.9
89.9
98.2
90
97.9
50

-------
TABLE  C-1 DESCRIPTION Of CONTROL OPTIONS for POINT SOURCES (continued)
Source
Point Source A
(continued)



Plant Area
Coke Ovens



Release Pt.
11 and 12 Batteries


A Battery
Option/Alternative
Current Control
Option 1
Option 2
Current Control
Option Description
Meets proposed NESHAP limitations for
coke oven doors, I.e., 10% leaking
doors, for 75% control. Charges are
currently every 32 seconds for 97%
control. Four percent of lids leak.
Construct two new batteries to replace
the existing batteries. The new
batteries wouild be designed to meet
the propoosed NESHAP.
Modify charging to 16 second intervals.
for 98% control. Limit leaking lids to 3%
of topside ports through maintenance
and inspection
Meets proposed NESHAP
Efficiency
"
1.3
13
—

-------
                  TABLE  C-1 DESCRIPTION of CONTROL OPTIONS for POINT SOURCES (continued)
Source
Point Source B
Point Source C
Point Source D
Plant Area



Release Pt.
Electric arc furnace (EAF)
Argon-oxygen decarburization
vessel (AOD)
Electric arc furnace (EAF)
Argon-oxygen decarburization
vessel (AOD)
EAF and AOD
Process Line
Option/Alternative
Current Control
Option 1
Current Control
Option 2
Current Control
Option 1
Current Control
Optbn 2
Option 3
Current Control
Option 1
Option Description
Direct evacuation control, segmented
canopy hood
Scavenger ducts, cross draft partitions,
closed roof
Close Fitting Hoods
Segmented canopy hood, scavenger
ducts, cross draft partitions, and
closed roof
Segmented canopy hood, closed roof
Direct evacuation control, scavenger
ducts, and cross draft partitions
Close Fitting Hoods
Segmented canopy hood, scavenger
ducts, and cross draft partitions
Vinyl strip doors to block wind
circulation
Horizontal packed bed wet scrubber,
venturi scrubber
Second packed bed wet scrubber
Efficiency
99%
99%
99%
99%
50%
40%
* Control alternatives for the A&B Plant. Benzene/Litol Plant, and Storage Tanks were grouped into a series of Options as shown in Table vi-6
The selection of control options and the estimation of capital and annual costs and removal efficiencies are projections and estimations based
on discussions with AM A staff, literature review and engineering judgement. Actual costs as well as feasibility and effectiveness could differ
considerably from our estimates.

-------
                                  TABLE  C-2  DESCRIPTION of CONTROL OPTIONS for AREA SOURCES
Area Source
Current
 Control
             Option 1
            Option  2
             Option 3
DRY CLEANING
  None
Quickly identify and repair leaks.

 Degenerative filter wastes not to
exceed 25kg of solvent per 100 kg
of wet waste material.

Distillation wastes not to exceed
 60 kg of solvent per 100 kg of wet
waste materials.

Better housekeeping

Overall control efficiency - 20%
Quickly identify and repair leaks

Regenerative filter wastes not to
exceed 25kg of solvent per 100 kg
of wet waste material

Distillation wastes not to exceed
60 kg of solvent per 100 kg of wet
waste materials

Better housekeeping

Carbon adsorption systems for all
commercial dry cleaners

Overall control efficiency - 58%
Quickly identify and repair leaks

Regenerative filter wastes not to
exceed 25kg of solvent per 100 kg
of wet waste material

Distillation wastes not to exceed
 60 kg of solvent per 100 kg of wet
waste materials

Better housekeeping

Carbon adsorption  systems for all
coin-op and commercial dry cleaners

Overall control efficiency - 65%
DECREASING
  None
Cold Cleaners
Cover during idle time (control
efficiency of 90%)

Drain racks with 30 second drain
and control efficiecy of 50%
during operation of degreaser

Open Top Vapor Degroasers
Cover during ida time (control
efficiency of 90%)

Increase freeboard ratio to 0.75
during operation (control
efficiency of 27%)
Cold Cleaners
Cover during idle time (control
efficiency of 90%)

Drain racks with 30 second drain
and control efficiecy of 50%
ddurmg operation of degreaser

Open Top Vapor Degreasers
Cover during ide time (control
efficiency of 90%)

Use refrigerated freeboard device
(control efficiency of 60% for
vaporization losses and 29% of
carryout losses)
Cold Cleaners
Cover during idle time (control
efficiency of 90%)

Dram racks with 30 second drain
and control efficiecy of 50%
dduring operation of degreaser

Open Top Vapor Degreasers
Cover during ide time (control
efficiency of 90%)

Use carbon adsorption unit
(control efficiency of 70% for
vaporization losses and 30% of
carryout losses)

-------
                             TABLE  C-2 DESCRIPTION of CONTROL OPTIONS for AREA SOURCES  (continued)
Area  Source
Current
 Control
            Option  1
            Option 2
             Option 3
OTHER INDUSTRIAL
  None
Enlarged condensation zone

Waste recovery facility

Manual enclosure

Overall control efficiency: 21%
Enlarged condensation zone

Waste recovery facility

Automatic enclosure

Overall control efficiency: 32.8%
Enlarged condensation zone

Waste recovery facility

Refrigerated condenser

Overall control efficiency: 36.6%
GAS MARKETING
 Stage I
Stage II (vapor balance system at pumps)

Overall control efficiency: 70.5%
ROAD VEHICLES
Light Duty Diesel
Road Vehicles

Heavy Duty Diesel
Road Vehicles
  None
  None
Catalytic Converter
Catalytic Converter

-------
                             TABLE  C-2  DESCRIPTION of CONTROL OPTIONS for AREA SOURCES  (continued)
Area Source
HEATING
Residential
Wood Stoves
Commercial Oil
Heating
Residential Oil
Heating
Residential Coal
Heating
Current
Control
None
None
None
None*
Option 1
Stop using wood stoves. Instead
use main fossil fuel heating system
(gas,coal or oil)
Replace oil-fired furnaces with
natural gas systems
Replace oil-f iredfurnaces with
natural gas sytems
Replace coal-fired units with
natural gas systems
Option 2

Replace existing burner with dual
burner and burn natural gas


Option 3




* The burning of coal is illegal, but does occur

 The selection of control options and the estimation of capital and annual costs and removal efficiencies are projections and estimations based
 on discussions with AMA staff, literature review and engineering judgement.  Actual costs as well as feasibility and effectiveness could differ
 considerably from our estimates.

-------
                                                         TABLE  C-3
                                 SUMMARY OF CONTROL OPTIONS FOR  POINT SOURCE  A
Benzene  Sources

Capital Costs ($)
Annual Costs ($/yr)
TSP Emissions (kkg/yr)
VOC Emissions (kkg/yr)
Benzene Emissions (kkg/yr)
Toluene Emissions (kkg/yr)
Xylene Emissions (kkg/yr)
Percent Benzene Emitted
Percent Toluene Emitted
Percent Xylene Emitted
Percent VOC Emitted

Chrome  Plating  (Mills)
Current  Control  Option 1   Option 2    Option  3   Option  4   Option 5
      664
      565
       2.0
       0.9
      100
      100
      100
      100
Current  Control  Option 1
770.000
158,000
16
13
0.05
0.02
2.3
2.5
2.2
2.3
410.000
64.000
142
117
0.4
0.6
20.7
20.0
66.7
21.4
200.000
4.000
365
291
1.0
0.7
51.5
50.0
77.8
55.0
150.000
54.000
583
497
1.8
0.3
88.0
90.0
33.3
87.8
210.000
60.000
441
391
1.5
0.8
69.2
75.0
88.9
66.4
Capital Costs ($)
Annual Costs ($/yr)
TSP Emissions (kkg/yr)
VOC Emissions (kkg/yr)
Chromium-6 Emissions (kkg/yr)
Percent Emitted
Coke Ovens
Capital Costs ($)
Annual Costs ($/yr)
TSP Emissions (kkg/yr)
VOC Emissions (kkg/yr)
POMCO Emissions (kkg/yr)
Percent Emitted
—
—
—
—
1
100
Current Control
—
—
864
—
864
100
460.000
200.000
—
—
0.5
50
Option 1
238,000,000
243,000
787
—
787
91






Option 2
433,000
243.000
857
—
857
99
 The selection of control options and the estimation of capital and annual costs and removal efficiencies are projections and
 estimations based on discussions with MA A staff, literature reviews and engineering judgement.  Actual costs as well as
 feasibility and effectiveness could differ considerably from our estimates.

-------
                       TABLE C-4
    DESCRIPTION Of POINT SOURCE A BENZENE CONTROL  OPTIONS
             Benzene                           Control
         Control  Options                     Alternatives
                                              (see Table VI-4)
           Current Control

                #1                          1,3,5,6,7,8,10,12

                #2                                1,6,8

                #3                                 6,8

                #4                              4.7,9.12

                #5                                  1
The selection of control options and the estimation of capital and annual costs and
removal efficiencies are projections and estimations based on discussions with AMA
staff, literature review and engineering judgement. Actual costs as well as feasibility
and effectiveness could differ considerably from our estimates.

-------
                   TABLE C-5 SUMMARY of CONTROL OPTIONS for POINT SOURCE B
                               Current Control      Option 1         Option 2           Options
                                               (Alternative 1)    (Alternative 2)    (Alternatives 1&2)
Capital Costs ($)
Annual Costs ($/yr)
TSP Emissions (kkg/yr)
VOC Emissions (kkg/yr)
Chromium-6 Emissions (kkg/yr)
Nickel Emissions (kkg/yr)
Cadmium Emissions (kkg/yr)
Arsenic Emissions (kkg/yr)
Percent TSP'emittad
Percent Chromium emitted
Percent Nickel emitted
Percent Cadmium emitted
Percent Arsenic emitted
193E-01
2.25E+00
100E-01
100E-01
                 70.000
                 20.000
   191E-01
   1 49E+00
   109E-02
   1.09E-02
     51
     98
     66
     11
     11
                  240.000
                   60,000
   1.35E-02
   931E-01
   1 OOE-01
   1 OOE-01
     57
      7
      4
     100
     100
     310.000
     80.000
        7

    1 17E-02
    1.78E-01
    109E-02
    109E-02
        9
        6
        B
       11
       11
The selection of control options and the estimation of capital and annual costs and removal efficiencies are projections and
estimations based on discussions with AMA staff, literature review and engineering judgement  Actual costs as well as
feasibility and effectiveness could differ considerably from our estimates.
                    TABLE C-6  SUMMARY of CONTROL OPTIONS  for POINT SOURCE C
Capital Costs ($)
Annual Costs ($/yr)
TSP Emissions (kkg/yr)
VOC Emissions (kkg/yr)
Chromium-6 Emissions (kkg/yr)
Nickel Emissions (kkg/yr)
Arsenic Emissions (kkg/yr)
Percent TSP emitted
Percent Chromium-6 emitted
Percent Nickel emitted
Percent Arsenic emitted
                               Current Control
  247

2.82E-01
382E+00
6.50E-02
   Option 1
(Alternative 1)

   120.000
   30.000
     239

   282E-01
   379E+00
   608E-02
     97
     100
     99
     94
   Option 2
(Alternative 2)

   160.000
    40,000
     240

   2.68E-01
   364E+00
   6 50E-02
     97
     95
     96
     100
    Options
(Alternatives 1&2)

    280.000
     70,000
       232

    266E-01
    3.61E+00
    608E-02
       94
       95
       94
       94
   Opflon4
(Alternative 3)

    5.000
     200
     230

   2.73E-01
  3.69E+00
   5 98E-02
     93
     97
     97
     92
The selection of control options and the estimation of capital and annual costs and removal efficiencies are projections and
 estimations based on discussions with AMA staff, literature review and engineenng ludgement.  Actual costs as well as
 feasibility and effectiveness could differ considerably from our estimates.

-------
 TABLE  C-7  SUMMARY Of CONTROL OPTIONS  for POINT SOURCE D
                             CURRENT CONTROL          OPTION 1

CAPITAL COST ($)                      —                      —
ANNUAL COST ($/YR)                   —                   220,000
CHROMIUM-6TOAIR(kko/yr)            0.19                    0.11
TSPTOAIR(kkg/yr)                    2.36                    1.42
PERCENT EMITTED                   100                     60
7779 selection of control options and the estimation of capital and annual costs and
removal efficiencies are projections and estimations based on discussions with AMA
staff, literature review and engineering judgement. Actual costs as well as
feasibility and effectiveness could differ considerably from our estimates.

-------
        TABLE C-8  SUMMARY  Of DECREASING CONTROL OPTIONS
                      PERCHLOROETHYLENE DECREASING

                                CURRENT  OPTION 1  OPTION 2  OPTIONS
                                CONTROL
CAPITAL COSTS ($)
ANNUAL COSTS ($/yr)
PERCENT EMITTED
PERCLOROETHYLENE TO AIR (kkg/yr)
VOC TO AIR (kkg/yr)
TSPTOAIR(kkg/yr)
   100
   315
   315
60,000
(50,000)
75
235
235
140,000
(80,000)
57
180
180
390,000
70,000
51
160
160
                      TRICHLOROETHYLENE DECREASING
CAPITAL COSTS ($)
ANNUAL COSTS ($/yr)
PERCENT EMITTED
TRICHLOROETHYLENE TO AIR (kkg/yr)
VOCTOAIR(kkg/yr)
TSP TO AIR (kkg/yr)
CURRENT OPTION 1  OPTION 2  OPTIONS
CONTROL

   —     120,000   330.000    900,000
   —    (130,000)  (220,000)  (220,000)
   100      77        56        49
   660      505      370       325
   660      505      370       325
                      METHYLENE CHLORIDE  DECREASING
                                 CURRENT OPTION 1
                                 CONTROL
CAPITAL COSTS ($)
ANNUAL COSTS ($/yr)
PERCENT EMITTED
METHYLENE CHLORIDE TO AIR (kkg/yr)
VOC TO AIR (kkg/yr)
TSP TO AIR (kkg/yr)
   100
   172
    0
 6,000
(50,000)
  64
  110
   0
 The selection of control options and the estimation of capital and annual costs and removal
 efficiencies are projections and estimations based on discussions with AMA staff, literature
 review and engineering judgement. Actual costs as well as feasibility and effectiveness
 could differ considerably from our estimates.

-------
       TABLE  C-9  SUMMARY of DRY CLEANING CONTROL OPTIONS
                  PERCHLOROETHYLENE DRY CLEANERS

                                CURRENT  OPTION 1  OPTION 2  OPTION 3
                                CONTROL

CAPITAL COSTS ($)                   —        0     2.750.000  4.470.000
ANNUAL COSTS ($/yr)                  —     110.000   (170,000)   110,000
PERCENT EMITTED                   100      80       42        35
PERCLOROETHYLENETOAIR(kkg/yr)    1,825    1,460      765       640
VOCTOAIR(kkg/yr)                  1,825    1,460      765       640
TSPTOAIR(kkgtyr)                    —       —       —        —
       MISCELLANEOUS INDUSTRIAL  METHYLENE  CHLORIDE  USAGE

                                CURRENT  OPTION 1  OPTION2  OPTIONS
                                CONTROL

CAPITAL COSTS ($)                    —     390,000     830,000  121,000
ANNUAL COSTS ($/yr)                   —     (120,000)   (130,000) (130,000)
PERCENT EMITTED                   100       79            67     63
METHYLENE CHLORIDE TO AIR (kkg/yr)    753      595          505    475
VOCTOAIR(kkg/yr)                    00             00
TSPTOAIR(Kkgtyr)                     —        —        —         —


The selection of control options and the estimation of capital and annual costs and removal
 efficiencies are projections and estimations based on discussions with AMA staff, literature
 review and engineering judgement. Actual costs as well as feasibility and effectiveness
 could differ considerably from our estimates.

-------
     TABLE  C-10 SUMMARY Of COMMERCIAL OIL CONTROL OPTIONS
                                 CURRENT CONTROL   OPTION 1
CAPITAL COST ($)                         —
ANNUAL COST ($/YR)                      —
ARSENIC TO AIR (kkg/yr)                  3.37E-01
CADMIUM TO AIR (kko/yr)                  5.12E-01
CHROMIUM-TTO AIR (kkg/yr)               2.04E-01
CHROMIUM-6 TO AIR (kkg/yr)               7.45E-04
NICKEL TO AIR (kkg/yr)                  2.24E+00
FORMALDEHYDE TO AIR (kkg/yr)           5.97E-01
TOLUENE TO AIR {kkg/yr)
BENZENE TO AIR (kkg/yr)
POMCDO TO AIR (kkg/yr)                 1.26E+01
POMCRO TO AIR (kkg/yr)                 1.40E+01
TSPTOAIR         •                   8.06E+01
VOC TO AIR (kkg/yr)                     1.13E+01
PERCENT EMITTED*                       100
PERCENT FORMALDEHYDE EMITTED        100
$163,110,000
 $8,920,000
  1.96E+00
  9.79E-01
  4.88E-01
  O.OOE+00
  O.OOE+00
  9.51 E+00
  1.68E+01
     0
     33
 OPTION 2

$20,040,000
 $8,920,000
  1.96E+00
  9.79E-01
  4.88E-01
  O.OOE+00
  O.OOE+00
  9.51 E+00
  1.68E+01
     0
     33
* arsenic, cadmium, chromium, chromium, chromium-6, nickel, toluene, benzene, TSP,
  VOC. POMCDO, POMCRO

77?e selection of control options and the estimation of capital and annual costs and removal
 efficiencies are projections and estimations based on discussions with AMA staff, literature
 review and engineering judgement. Actual costs as well as feasibility and effectiveness
 could differ considerably from our estimates.

-------
                TABLE C-11  SUMMARY of RESIDENTIAL COAL
                   and RESIDENTIAL OIL CONTROL OPTIONS
                           Residential Coal Controls

                                     CURRENT CONTROL
OPTION 1
CAPITAL COST ($)                             -           $18.100,000
ANNUAL COST ($/YR)                           -            $1.300.000
ARSENIC TO AIR (kkg/yr)                      2.38E-01           0 OOE+00
CADMIUM TO AIR (kkg/yr)                      7.75E-03           O.OOE+00
CHROMIUM-TTOAIR(kKg/yr)                   6.03E-02           O.OOE+00
CHROMIUMS TO AIR (Kkg/yr)                   448E-05           0 OOE+00
NICKEL TO AIR (kkg^r)                       6 03E-02           0 OOE+00
FORMALDEHYDE TO AIR (kkg/yr)               387E-02           2.65E+00
TOLUENE TO AIR (kkg/yr)                      0 OOE+00          1 33E+00
BENZENE TO AIR (kkg/yr)                      O.OOE+00          661E-01
POMRAN TO AIR (kkg/yr)                      998E+01           0 OOE+00
POMRBT TO AIR (kkg/yr)                      345E+01           O.OOE+00
TSP TO AIR (kkg/yr)                          5 36E+02          1.06E+00
VOC TO AIR (kkg/yr)                          148E+02          1.88E+00
                                  3         100               0
PERCENT FORMALDEHYDE EMITTED              100              680
PERCENT BENZENE & TOLUENE EMITTED           0               100
                            Residential Oil Controls

                                     CURRENTCONTROL     OPTION 1

CAPITAL COST ($)                              —           $699.270.000
ANNUAL COST ($/YR)                           -            $77.700.000
ARSENIC TO AIR (kkg/yr)                      4 51E-02          0 OOE+00
CADMIUM TO AIR (kkg/yr)                     329E-01          0 OOE+00
CHROMIUM-T TO AIR (kkg/yr)                  329E-02          0 OOE+00
CHROMIUMS TO AIR (kkg/yr)                  2.82E-03          0 OOE+00
NICKEL TO AIR (kkg/yr)                       3.09E+00          0 OOE+00
FORMALDEHYDE TO AIR (kkg/yr)               7.80E+00          103E+02
TOLUENE TO AIR (kkg/yr)                     O.OOE+00          5.13E+01
BENZENE TO AIR (kkg/yr)                     0 OOE+00          2S6E+01
POMRDOTOAIR(kkg/yr>                      836E*01          0 OOE+00
POMRRO TO AIR (kkg/yr)                      3.73E-04          0 OOE+00
TSP TO AIR (kkg/yr)                          2 55E+02          411E+01
VOC TO AIR (kkg/yr)                          7.30E+01          7 26E+01
PERCENT EMITTED*                            100                0
PERCENT FORMALDEHYDE EMITTED             100              1300
PERCENT BENZENE & TOLUENE EMITTED           0                100

•except benzene, toluene, and formaldehyde

The selection al control options and the estimation of capital and annual costs and
removal efficiencies an protections ana" estimations based on discussions with AMA
staff, literature review and engineering ludgement. Actual costs as well as
feasibility and effectiveness could differ considerably from our estimates.

-------
     TABLE C-12  SUMMARY of RESIDENTIAL WOOD STOVE CONTROLS
CAPITAL COST ($)
ANNUAL COST ($/YR)
ARSENIC TO AIR (kkg/yr)
CADMIUM TO AIR (kkg/yr)
CHROMIUM-T TO AIR (kkg/yr)
CHROMIUM-6 TO AIR (kkg/yr)
FORMALDEHYDE TO AIR (kkg/yr)
BENZENE TO AIR (kkg/yr)
TOLUENE TO AIR (kkg/yr)
NICKEL TO AIR (kkg/yr)
PHENOL TO AIR  (kkg/yr)
POMRBT TO AIR (kkg/yr)
POMRAN TO AIR (kkg/yr)
POMRDO TO AIR (kkg/yr)
POMRRO TO AIR (kkg/yr)
POMRWS TO AIR (kkg/yr)
VOC TO AIR (kkg/yr)
TSP TO AIR (kkg/yr)
                                CURRENT CONTROL
9.00E-04
2.50E-04
7.00E-03
   NA
1.50E+00
1 .OOE-02
   112
  93.9
   149
                  OPTION 1
 220,000
3.00E-04
3.50E-04
1 .OOE-04
3.00E-06
1.40E-01
6.50E-02
3.00E-02
8.00E-03
1.00E+00
4.00E-02
1.00E-01
9.00E-02
5.50E-04
O.OOE+00
  0.25
    1
The selection of control options and the estimation of capital and annual costs and
removal efficiencies are projections and estimations based on discussions with AM A
staff, literature review and engineering judgement. Actual costs as well as
feasibility and effectiveness could differ considerably from our estimates.

-------
         TABLE C-13 SUMMARY of ROAD VEHICLE CONTROL OPTIONS
                             Light Duty Diesel

                                     CURRENT CONTROL      OPTION 1

CAPITAL COST ($)                              _               956.320
ANNUAL COST ($/YR)                           —                 _
CADMIUM TO AIR (Kkg/yr)                     6 OOE-09            6.00E-09
FORMALDEHYDE TO AIR (kkg/yr)                 32.5                49
POMLDD (kkg/yr)                              24.5                2
TSP TO AIR (Kkg/yr)                           122.5               110
VOC TO AIR (Kkg/yr)                            364                55
PERCENT EMITTTED (cadmium))                  100                100
PERCENT EMITTTED (POMLDD. TSP)              100                90
PERCENT EMITTTED (formaldehyde. VOC)          100                15
                            Heavy Duty Dlmal

                                     CURRENTCONTROL      OPTION 1

CAPITAL COST ($)                              _              5,941.000
ANNUAL COST (S/YR)                            —                 —
CADMIUM TO AIR (kkg/yr)                     6.00E-09           6.00E-09
FORMALDEHYDE TO AIR (kkg/yr)                  10                  2
POMHDD (kkg/yr)                               59                 S3
TSP TO AIR (kkg/yr)                            197                 177
VOC TO AIR (Kkg/yr)                           1316                197
PERCENT EMITTTED (cadrmum))                  100                 100
PERCENT EMITTTED (POMLDD. TSP)              100                 90
PERCENT EMITTTED (formaldehyde. VOC)          100                 15
 77»e selection of control options and the estimation of capital and annual costs and
 removal efficiencies are projections and estimations based on discussions with AMA
 staff, literature review and engineering lodgement Actual costs as well as feasibility
 and effectiveness could differ considerably from our estimates.

-------
  TABLE  C-14  SUMMARY Of GASOLINE MARKETING CONTROL OPTIONS
CAPITAL COSTS ($)
ANNUAL COSTS ($/yr)
PERCENT EMITTED
BENZENE TO AIR (kkgfyr)
ETHYLENE CHLORIDE (EDC) (kkg/yr)
ETHYLENE DIBROMIDE (EDB) (kkg/yr)
XYLENE (kkg/yr)
TOLUENE (kkg/yr)
ETHYL BENZENE (kkg/yr)
VOC TO AIR (kkg/yr)
TSPTOAIR (kkg/yr)
                                            CURRENT
                                           CONTROLS
 100
 25.1
 2.1
 0.2
 8.4
 46.2
 0.4
4,185
OPTION 1

8,300,000
 970.000
   29
    7
   0.6
   0.06
   2.4
   13.4
   0.1
  1,230
The selection of control options and the estimation of capital and annual costs and removal
efficiencies are projections and estimations based on discussions with AMA staff, literature
review and engineering judgement. Actual costs as well as feasibility and effectiveness could
differ considerably from our estimates.

-------
            TABLE  C-15    GLOSSARY OF TERMS
POMRDO
POMRRO
POMCDO
POMCRO
POMRAN
POMRBT
Polycyclic organic matter from residential distillate oil
Polycyclic organic matter from residential residual oil
Polycyclic organic matter from commercial distillate oil
Polycyclic organic matter from commercial residual oil
Polycyclic organic matter from residential anthracite coal
Polycyclic organic matter from residential bituminous coal

-------
                  APPENDIX  D






BACKGROUND DOCUMENTATION FOR BENEFITS ESTIMATES

-------
                           APPENDIX D

         BACKGROUND DOCUMENTATION FOR BENEFITS ESTIMATES


     This appendix describes the sources of available benefits
numbers for TSP and VOC and how these benefits numbers were used
in the Baltimore IBMP.  As discussed in Chapter VI, it is
desirable to know the economic value of an outcome to put into
perspective the benefit and costs of changing pollution levels
when devising pollution control programs.  In many situations,
considering the benefits, and not just cost-effectiveness, could
results in different the risk management decisions.

     We begin by describing the general techniques available for
quantifying benefits (i.e., valuation using direct and indirect
economic techniques).  We then present the sources and benefits
results for the TSF and VOC effects considered in the Baltimore
IBMP.  We conclude with a discussion of how the benefits
estimates were used to calculate the overall benefits afforded by
individual control options evaluated in the Baltimore IEMP.

GENERAL VALUATION TECHNIQUES
Hedonic Property Method

     One approach frequently used to value environmental
conditions that are related to the price of goods is the hedonic
property method.  Hedonic property studies assume that property
values reflect the economic value of environmental quality
characteristics in the vicinity of the property relative to other
areas.  Given that pollutant levels vary within most urban areas,
and property owner preferences are affected by pollutants, the
market prices for properties will capture these qualities.  Using
data collected over a range of property prices, pollution
conditions and other attributes related to property values, and
conducting the appropriate statistical tests, can provide
estimates of the impacts of pollution on market prices and
individual preferences for different pollution conditions.  The
pollution price will reflect any health and welfare risks
perceived by the individual, but there is little empirical
evidence to suggest how the total price should be attributed to
specific outcomes.  Therefore, this technique is not well-suited
for use in instances where control programs are evaluated on the
basis of their ability to reduce specific risks.
                               D-l

-------
Hedonic Wage Method

     Hedonic wage studies are based upon the assumption that
wage rates vary across occupations for a variety of reasons, one
of which is the risk of injury and death inherent in an
occupational environment.  Each occupation can be characterized
as having a probability of death or lesser injury associated
with the nature of work.  The risk of injury is presumed to be
known by individuals and employers across different occupations.
Given this information, individuals attempt to select jobs where
the marginal adjustment in the wage rate is equated with the
incremental level of risk they expect to face when choosing an
occupation.  Combining the marginal change in wage accepted with
the marginal change in risk of injury yields an implicit value
for the acceptance of various risks.  By observing the variations
in wages across different occupational risks, a functional
relationship can be estimated that describes how much wages must
change to accept incremental changes in risk.

Travel-Cost Method

     A third market-oriented method of valuing changes in
pollution condition is the travel-cost method.  This method
assumes individuals produce activities that require inputs, for
example, travelling to a beach to swim in order to enjoy some
level of satisfaction.  The individual, in selecting which beach
to visit, considers the time and resources necessary to get to
each beach, plus the environmental and social amenities provided
at each beach.  If there are sufficient numbers of observations
on beach visits and variations in environmental amenities at the
beaches, then it is possible to infer the value of the
environmental amenities through the participation rates and costs
(e.g., travel and time costs).  Because most of the expenses are
incurred in travelling to these sites, the method is referred to
as the travel-cost method.

Resource Cost Method

     The health and welfare outcomes of pollution exposures are
often observed in market transactions for goods other than homes
and salaries directly affected by the pollutant, or goods used to
mitigate the risks posed by pollutants.  After calculating a
dose-response function to measure the change in outcomes (e.g.,
health effects, agricultural effects) for a given change in
exposure to pollution, outcomes are valued using either the price
individuals must pay to recover from the damages once they have
occurred, or preferably, the price they are willing to pay to
protect against suffering the damaging outcomes of pollution.
This approach can be used in those instances where the outcomes
are observed, but recognition of the contribution of pollution to
the outcome need not be well understood by the individual.  The
price of preventing or rectifying the effects are then used as

                               D-2

-------
implicit values for benefits of reducing pollutants.  The
relevant proportion pollutants contribute to the estimated
outcomes can be used to estimate the economic benefits of
changing the contribution of pollution to these outcomes.  For
human health outcomes, the technique of assigning a value to
human health outcomes based on the cost of medical care resources
is referred to as the cost-of-illness approach.

     The use of cost-of-illness and other resource cost methods
may fail to capture the total extent of damages attributable to
pollution.  Concentration-response functions do not fully account
for defensive behavioral responses that may occur in polluted
environments.  Estimating concentration-response functions using
health statistics would not capture the actions of individuals
taken to avoid these damages.  Therefore, the valuation of
pollutant outcomes should look beyond the direct effects and
include the costs of additional actions taken to reduce pollution
risks.  In addition, there is a failure to account for the pain
and suffering caused by pollutants that are not captured in the
costs of providing medical care.  The magnitude of these other
economic values will be a function of the ability of individuals
to recognize the relationship between pollution and effects, and
their ability to take personal actions to prevent the damages
from occurring.  Therefore, rather than rely solely on the
opportunity cost of medical resources, the individual's
preferences should also be considered when valuing the benefits
of controlling pollutants.

Contingent Valuation Method

     The potential inability of the labor, housing, and other
markets to reflect preferences for risk reductions and the values
of reducing risks have led to other alternative mechanisms.  The
contingent valuation methods uses a survey format to gain
information from individuals on their preferences for the
perceived benefits from reductions in pollution levels.
Individuals respond to questions that pose the existence of a
market where they can exchange goods for dollars, even though
such a market may not exist in reality.  If the questions are
posed in a manner so that there are no particular gains to be had
on the part of the individual from revealing other than their
true preferences, then this method can provide values that are
both meaningful and useful for evaluating environmental benefits.
values obtained using this method are capable of including the
value of reduced pain and suffering, and the value of maintaining
activities in accordance with their preferences and not those
associated with mitigating behavior.
                               D-3

-------
PARTICULATE MATTER BENEFITS ESTIMATES

Health Effects

     Mortality Effects

     Mazumdar (1982) investigated the mortality rates in the
greater London area from the period 1958-1972 and tested whether
the daily variations in these rates were correlated with
variations in British Smoke (BS) and S02 concentrations Schwartz
(1986) reanalyzed the data and established more robust
statistical relationships between daily mortality figures and BS
levels less than 200 ug/m3.  After controlling for the effects of
temperature and humidity, the relationship reads:

     change in daily mortality rates in London

     = 0.138 x (change in daily BS concentrations).

     To convert BS measures into total suspended particulate
concentrations, several studies have been conducted with BS and
TSP measures taken in London during the same period when
mortality rates were analyzed.  The relationship between BS and
TSP in London is that a 1 ug/m3 change in BS is likely to result
in a 1.41 ug/m3 change in TSP, for BS levels less than 200 ug/m3.
Using this relationship to convert BS to TSP, the mortality
equation now is:

     change in daily mortality rates in London

     = (0.138)/(1.41) x (change in daily TSP concentrations).

     Given the London population exposed to TSP concentrations
was about 8 million persons, the change in daily mortality rates
per 100,000 persons exposed to TSP would be 1.22 x 10~3 mortality
cases per day per 100,000 people exposed.  On an annual basis:

    change in annual mortality cases per 100,000 persons

     = 0.446 x (change in annual average TSP concentration).

     For example, if annual average TSP concentrations fall from
75 to 70 ug/m' in a city with population of 1 million persons,
the estimated annual mortality rate will fall by 22.3 cases per
year. It is likely that many of these estimated cases will be
reduced for persons aged 65 and over, or those suffering from
chronic illnesses that put them at greater risk.

     Mortality Values

     Most of the studies that deal with valuing changes in
mortality rates have not dealt with the value of preventing the

                               D-4

-------
death of a particular individual.  Instead, most studies examine
the value of a relatively small reduction in the probability of a
death occurring, and then extrapolate this to value the total
number of cases reduced.  Most studies have looked at wage
premiums associated with different Job-related risks, the theory
being that higher risk occupations will be rewarded with greater
wages to induce individuals to participate in riskier occupa-
tions.  Summaries of the empirical analyses performed to date
(ERC: 1983, 1986) suggest that individuals are willing to accept
an additional risk of 1 x 10~4 on an annual basis for $100 to
$700 per year.  Extrapolating this to one case yields a
willingness to accept this risk for $1 to $7 million per case.
The general rule is to use $2 million per case, or $2 per 1 x
10~*> reduction in premature mortality risk.

     Acute Morbidity Effects

     Ostro (1983,1987) used data collected in the Health
Interview Survey on restricted activity days (RADs) and
information on fine particulate matter (FP) to develop
concentration-response relationships between RADs and FP
concentrations.  The form of the estimated relationship was as
follows:

     change in RAD per 2-week period

     = 0.0048 x  (baseline RADs for 2-weeks)(change in 2-week FP).

     To convert the relationship to one based on annual average
number of RADs and changes in FP,

     change in RADs per person per year

     •  (0.0048) x  (baseline RADs per year)(change in annual FP)

     For example, if the baseline annual average number of RADs
is 20 days per year, and the change in FP is from 40 ug/mj to 35
ug/nr, then:

     change in RADs per person per year

     =  (0.0048)(20)(5) =0.48 fewer RADs per person exposed.

     Given that  TSP is a more standard measure of particulate
matter, it may be desirable to express the above relationship in
terms of TSP  rather than FP.  The relationship between TSP and FP
varies depending on the emission sources of TSP.  For example,
should  the relationship be 1 ug/m3 TSP equals approximately 0.25
ug/m^ FP, and TSP  concentrations drop from 90 ug/mj to 75 ug/m-3,
then:
                                D-5

-------
     change in RADs per person per year

     = (0.0048)(20)(15)(0.25) =0.26 fewer RADs per person.

     Samet (1981) examined how changes in the number of emergency
room visits varied with changes in ambient concentrations of TSP,
S02/ and N02 during early spring and fall months of 1974-1979.
The estimated relationship between daily emergency room visits in
Steubenville, Ohio (exposed population group) and TSP is:

     change in visits to emergency rooms per person per year

     =    fO.Oin    (365)(change in annual average TSP levels)
        (31000 persons in Steubenville),

     = (1.29 x 10~4) (change in annual average TSP).

     For example, if TSP levels fell from 90 ug/m3 to 75 ug/m3 in
a city having a population of 1 million persons, the expected
change in the number of emergency room visits would be:

     change in number of emergency room visits per year

     = (1.29 x 10-4)(15)(1 x 106),

     = 1940 fewer emergency room visits per year.


     Chronic Morbidity Effects

     Ferris (1978) examined how the  number of chronic
respiratory disease cases in Berlin, New Hampshire varied in
accordance with variations in ambient TSP concentrations.  Ferris
did not report concentration-response functions, but rather
reported on age and sex-standardized rates of prevalence for
chronic non-specific respiratory diseases for the two years
examined (1961 and 1967).  The ratios suggest that TSP reductions
from 180 ug/m3 to 130 ug/m3 result in significant changes in the
number of chronic respiratory disease incidents, but changes
below 130 ug/m3 did not result in measurable changes in
respiratory disease.  For those instances where Ferris found a
relationship between TSP and chronic respiratory disease cases,
the equations are as follow:

     change in number of chronic respiratory disease
     incidents per person per year

     = (b)(change in TSP)(average annual rate of chronic
                         respiratory disease incidents per
                         person),


                               D-6

-------
where b = 0.0027 for males and b = 0.0019 for females. For
example, if TSP levels fall from 145 to 130 ug/m^, average annual
incident rates for males are 0.247, and rates for females are
0.281, then:


     change in number of incidents for males per year

     = (0.0027)(15)(0.247) =0.01 fewer incidents per year,
and

     change in number of incidents for females per year

     = (0.0019)(15)(0.281) = 0.008 fewer incidents per year.

     Morbidity Values

     Most dollar estimates for the value of preventing morbidity
effects related to PM have been based on a cost-of-illness
approach.  The cost of providing care to correct for the health
effect and the opportunity cost of time and productivity lost
during the illness are used to estimate the value of reducing
health effects attributable to PM.  Economic theory suggests that
these estimates are underestimates of the willingness to pay on
the part of exposed individuals to avoid incurring the estimated
health effects.  Cost-of-illness studies fail to capture the
discomfort and inconvenience imposed upon individuals.  Efforts
to estimate willingness to pay using surveys has been successful
for some morbidity effects, primarily those related to ozone.
Some of the effects attributed to PM may be of a similar nature
to those caused by ozone, however, most RADs attributable to PM
are considered to be of a more severe  nature than RADs caused by
ozone.

     Dollar values for chronic and acute health effects that lead
to RADs are estimated using wage information for working persons
and medical expenditures.  If a person suffers a work loss day,
the economic value of the day lost is estimated to be equivalent
to the productive value of the individual.  Economic theory
suggests that the productive value is equal to the individual's
wages.  Therefore, if an individual is removed from the work
force due to illness caused by PM exposure, the benefits of
reducing the health damages should be at least equal to the value
in foregone productivity as measured by wages.  As stated above,
this figure does not include the inconveniences imposed on the
individual and other persons at work or in the home, so it is
likely to underestimate the true value of reducing health
effects.  For the U.S., the average wage rate is $10/hour, so a
lost work day would be valued at $80/day for full-time persons.
Persons working part time, persons working in services that do
not receive hourly wages  (self-employed sectors of the economy,
homemakers) and retired persons are also assumed to have their

                               D-7

-------
more severe RADs valued using average wage rates.  Convention
dictates that these RADs be valued at the same rate.  For those
RADs of a less severe nature that allow the individual to
continue working/ but presumably at some diminished level or in a
state of mild discomfort, the value is often assumed to be some
fraction of the average daily wage.  Most studies use one-half to
one-quarter the daily wage rate, although these values are
arbitrarily chosen.

     An additional cost included in the value of RADs is the
direct medical expenditures necessary to combat the health
effects of PM.  Based upon national health statistics for medical
expenditures and estimated number of RAD incidents, the average
cost per incident is around $2 for acute effects.  The average
medical cost incurred attributable to chronic health effects of
PM are $10 per RAD incident.  Severe acute and chronic attacks
will result in significantly higher medical expenditures than
those presented here, but these figures are intended to represent
the average expense aggregated over a considerable number of
cases.  In many instances, less severe attacks will not require
that individuals spend money for medical services.  In those
cases where more severe episodes occur, it Is likely that the
individual will have to take advantage of acute care provided by
emergency rooms.

     Dollar value estimates for emergency room visits use the
same theory as RADs, but supplement the daily wage estimate with
the average cost of providing emergency room care to the
individual.  Average medical expenditures for an emergency room
visit amount to approximately $180 per out-patient visit.  This
value includes costs covered through medical health insurance.
Assuming that an emergency room visit results in a lost work day
or its equivalent for non-full time employees, then the total
value for an emergency room visit would be $260  ($180 + $80).  As
with other less severe health effects, this dollar estimate does
not include the values attributed to reducing potential pain and
anxiety associated with the estimated health effects.  Therefore,
this estimate is expected to be an underestimate of individual's
willingness to pay to avoid the outcome.  Few studies have
examined the magnitude of the underestimation (ERC, 1986), but
for those conducted to date, the willingness-to-pay estimate
tends to be 2-3 times larger than the cost-of-illness value.

Welfare Effects

     Materials Effects and Values

     Estimates of materials damages due to TSP are developed from
using aggregate consumer purchase data on household goods and
changes in these purchases as a function of TSP  concentrations.
The relationship is directly between TSP concentrations and
expenditures, so there is no intermediate relationship between

                               D-8

-------
TSF and household material damages.  The data was collected on 24
Standard Metropolitan  Statistical Areas (SMS As) and expenditures
on a set of commodities.  Demand equations are estimated for
seven different commodity groups: food, shelter, household
operations, home furnishings, clothing, transportation, and
personal care.  Air quality data concentrations and expenditures
data was analyzed using conditions present during 1972 and 1973.
A system of demand equations are estimated, each having pollutant
variables that are introduced to explain variations in purchases
for goods on the basis of different pollution conditions in the
24 SMS As.

     The TSP variable was significant (at the 95% level) in the
demand equations for home repair activities, and laundry and
cleaning operations.  The most robust measure of TSP was based on
the second-highest monitored 24-hour concentration of TSP for all
monitors in the SMSA.  Through a somewhat complicated series of
estimation procedures, the welfare estimates for changes in TSP
are estimated using compensated demand functions and expenditure
functions.  Therefore, it is difficult to express the benefit
estimate procedure as a simple formula.

     A similar procedure was followed to estimate the impacts of
particulate matter on the manufacturing industry.  Examination of
cost functions and TSP concentrations showed that TSP was a
significant variable in estimating cost variations across
manufacturers of fabricated metals and machinery.  As with the
household sector model developed by Mathtech, there is no simple
equation that can be used to estimate effects.  However, attempts
to quantify benefits on the basis of TSP emissions have shown
that the benefits to household and manufacturing sectors of
reducing TSP emissions in urban areas may range between $0 - $50
per ton, with an average benefit of $10 per ton.  The range is a
function of the proximity and density of household and industrial
operations to TSP emission sources.

     Visibility Effects

     Trijonis (1982) developed an equation that described the
relationship between visibility as measured by visual range, and
the presence of TSP, NOX, sulfates (804), and relative humidity
(RH).  He estimated the following relationship for California:
     Visual range in  miles
                               24.3
       _
         (.12 +  .04(S04/(1-(RH/100)) +  .03(NOX/(1-(RH/100))
                        +  .003((TSP-S04-NOX)/(1-(RH/100))),

where RH = relative humidity. As this  relationship relies on
knowing several different  pollution variables, it is somewhat
cumbersome.  However, to illustrate an example, if we assume that

                               D-9

-------
at the nearest ambient monitoring station to the individual's
place of residence.  The statistical analysis was conducted using
the number of respiratory restricted activity days (RRADs)
experienced over the two-week period as the dependent variable,
and the pollution and socioeconomic variables as the independent
variables.  The ozone variable was positively correlated with the
number of RRADs and the coefficient on the ozone variable was
significant.  Two different functional forms were examined that
described the variability of RRADs and ozone.  The first
regression was:

     number of RRADs per person per 2-week period

     = exp(aZ + b(ozone)),

where Z = other independent variables, and a and b are estimated
parameters.  When using this functional form to estimate a change
in RRADs for a change in ozone, one must know the baseline
number of RRADs,

     change in number of RRADs per person per 2-week period

     -   (average number of RRADs){b(change in ozone)).

     The second functional form used the square root of the
observed ozone concentration as an independent variable, so that
the regression equation read:


     number of RRADs per person per 2-week period

     = exp(aZ + b(ozone)-5),
and

     change in number of RRADs per person per 2-week period

     =  (average number of RRADsU.5HbWchange in ozone} .
                  (beginning ozone level)'3


The coefficients for the above equations were b = 6.883 for the
linear equation, and b = 4.926 for the square root function.  The
average number of RRADs = .162 days per person during a two-week
period, and E(ozone) = .042 ppm in the study.  Using these as the
baseline conditions, the calculated change in RRADS for a 10%
reduction in the two-week arithmetic average of the daily high
hourly ozone levels would be:
                               D-ll

-------
     linear model: change in expected number of RRADs
                   per person per 2-week period

         =  (.162)(6.883){.042 - (.9)(.042)>

         =  .005 fewer cases per person during a two-week period.
and
     quadratic model: change in expected number of RRADs
                       per person per 2-week period

         = f .162W.5W4.926U0.42 - f.9W0.421V
                         (.042)-3

         = .008 fewer cases per person during a two-week period.


These health effects were estimated for adult populations,
including both asthmatic and non-asthmatic individuals.  However,
a different techniques is used to estimate ozone effects for
asthmatic individuals.

     Holguin (1984) developed a prospective epidemiological study
to determine the impacts of ozone on asthmatics.  Information on
pollutant levels as measured at monitors in proximity to their
place of residence were used to estimate exposure levels.
Holguin used information on ozone, N02, pollen, and weather data
to supplement socioeconomic characteristics that would explain
variations in asthma symptoms across his sample.  Seventy-five
percent of his sample were individuals between the ages of 10 and
19 years and all individuals were non-smokers.  The ozone
exposure  variable related to the presence or absence of an
asthma attack over a 12-hour period was the maximum 12-hour
period exposure concentration.  Holguin used a weighted logistic
model to  determine the change in probability of suffering asthma
attack given a change in the maximum exposure concentration for
ozone.  The estimated equation is:

     probability of experiencing an asthma attack per asthmatic
     per 12-hour period

          	1	
            (l-i- exp(az + b(ozone))

where a = vector of coefficients for other independent variables,
z = other independent variables, b = estimated coefficient for
ozone (6.20 in Hoguin's sample), and ozone = ozone concentration
(12-hour period 1-hour maximum concentration in ppm).  For
changes in ozone, the change in expected asthma attacks would be:
                               D-12

-------
     change in probability of having an asthma attack during a
     per asthmatic per 12-hour period

     =         (l - pi fexofbfOzone^-Ozonein - 11
             (!+((!- p)/p)exp(b(0zone2 -
where p - baseline attack rate per 12-hour period, Ozone^ =
initial 1-hour maximum ozone concentration, and Ozone2 = final 1-
hour maximum ozone concentration.   If the asthmatic population
constitutes 8 percent of the population exposed, the baseline
attack rate is 0.10, and ozone the maximum ozone concentration
falls from 0.13 ppm to 0.12 ppm, then the number of attacks
reduced per 12 -hour period would be:

     change in number of asthma attacks per asthmatic per
     12 -hour period

     = ri - 0.10 Wexpf 6.20U0.12 - 0.1311 - llf population HO. 081
              (1 + ((1 - 0. 10)/0.10)exp(6. 20(0. 12-0. 13)) )

     = (- 4.576 x 10~4) (population exposed).

     If the same 0.01 ppm reduction in the 12-hour period maximum
1-hour ozone concentration occurs over the entire ozone season
then the change in number of asthma attacks per ozone season is :

       = (- 4.576 x 10~4) (exposed population) x
              (number of 12-hour periods in the ozone season) .

     McDonnell (1983) used six groups of exercising male
volunteers in a clinical human exposure study to test whether
exposures to different ozone concentrations over short (2.5
hours) time periods led to changes in pulmonary function,
ventilatory function and other health symptoms.  Using a sigmoid
function, regression equations were used to estimate to what
extent different ozone levels could explain the variations in
pulmonary, ventilatory and symptomatic effects across individuals
in the sample.  The tests showed small differences from the
baseline ozone levels and levels around 0.12-0.18 ppm.  Larger
differences were observed at the upper levels tested (0.24-0.40
ppm).

     The functional form of the regression leads to the following
relationship for coughing symptoms is:

     number of coughing incidents per 3-hour period per
     exercising adult non-smoking male
        	1	 (non-smoking male population) x
        (1 + exp(-a-b(ozone))(% exercising at moderate level),

                               D-13

-------
where a and b are estimated coefficients (a = - 1.527 and b =
12.78) and ozone = 3-hour period average ozone exposure level.
To determine the change in the number of coughing incidents for
exercising non-smoking males when ozone exposure concentrations
are reduced, use the following:

     change in number of number of coughing incidents per 3-hour
     period per exercising adult non-smoking male

     =         (I - pUexpfbfOzoneg-Ozonein - 11     x
             (!+((!- p)/p)exp(b(0zone2 -Ozonei))

          (percent time spent exercising heavily) x

          (adult male non-smoking population),

with p = initial probability of coughing at initial ozone level,
Ozonei = initial ozone concentration and Ozone2 • new
concentration.

     If adult males spend approximately 1.5 percent of their
waking hours (15 hours per day) exercising heavily, and ozone
levels decrease from 0.10 to 0.09 ppm over the 3-hour period they
are exercising (one-fifth of all waking hours) then:

     change in number of coughing incidents per 3-hour period per
     exercising non-smoking adult male

     = fl - Q.438Wexofl2.78fO.09-0.10n - 1UO.015/51     x
         (1 + ((1 - 0.438)/0.438)exp(12.78(0.09 - 0.10))

               (adult male non-smoking population)

     = ( - 9.50 x 10~5)(adult male non-smoking population).

     McDonnell also determined how periods of shortness in breath
changed with ozone exposures.  He used the same functional
relationship as that for coughing and estimated new coefficients
(a = -0.182 and b = 8.106).  Therefore, the change in number of
shortness of breath incidents, under the same assumed changes in
ozone as above would lead to the following:

     change in shortness of breath incidents per 3-hour period
     per exercising non-smoking adult male

     = il - Q.652Hexpf8.106(0.09-0.10n - 1HO.015/51    x
         (1 + ((1 - 0.652)/0.652)exp(8.106(0.09 - 0.10))

          (adult male non-smoking population).

     = - 5.45 x 10"5 (adult male non-smoking population).

                               D-14

-------
     Hammer (1974) conducted a prospective epidemiological study
of 112 nurses in the Los Angeles area.  The nurses were asked to
keep a daily diary of their activity patterns and medical
history.  The nurses were asked to report on the presence/absence
and severity of respiratory disease symptoms.  He supplemented
their diaries with pollution data for oxidants, CO, NC>2/ and
weather data.

     Hasselblad and Svendsgaard (1975) reanalyzed the Hammer data
and estimated a relationship between eye irritation and daily
maximum one-hour concentration of ozone levels.  The relationship
is:

     probability of experiencing eye irritation per day  per
     person

     =  (0.0407) +  _ fl - 0.04071 _   .
                    (1 + exp(4.96 - 9.07(ozone))

The change in the probability of experiencing eye irritation for
a change in ozone levels would be:

     change in probability of having an eye irritation case per
     day per person

           (I - 0.0407U1 - pi fexof 9. 07 f Ozones-Ozone! 11 - 11 .
         (!+((!- p)/(p - 0.0407) )exp(9.07(0zone2 -
If the initial daily maximum ozone level is 0.13 ppm, and the new
daily maximum is 0.12 ppm, then the change in probability is:

     change in probability of having an eye irritation case per
     day per person

     =   (I - 0.0407U1 - O.Q621Uexpf9. 07/0. 12-0.1311 - 11
       (1 +(( 1-0. 0621 )/( 0.0621-0.0407) )exp(9. 07(0.12-0.13)))

     = - 1.90 x 10~3.


To determine the change in number of cases, the change in
probability would be multiplied by the number of persons exposed
to the assumed ozone concentrations .

     Likewise, Hammer estimated the relationship between ozone
levels and changes in chest discomfort.  Using the same
functional form, Hasselblad and Svensgaard estimated the
following relationship:
                               D-15

-------
     probability of suffering chest discomfort per day per person

     = (0.017) + ___	(1 - 0.0171	  .
                  (1 + exp(3.53 - 0.23(initial ozone level)))


Therefore, the change in the probability of suffering chest
discomfort, given the above scenario (the reduction of 0.13 to
0.12 ppm in daily maximum ozone concentration) yields the
following:

     change in probability of suffering chest discomfort per day
     per person

     =     II - 0.017W1 - Plfexpf0.23(0zone2-0zonein - 11
         (!+((!- p)/(P - 0.017))exp(0.23(0zone2 -Ozone!)))

     = fl - Q.017W1 - Q.045BUexpf0.23(0.12 - 0.1311 - 11   ,
      (1 + ((1 - 0.0458)/(0.0458 - 0.017))exp(0.23(0.12 - 0.13)))

     = - 6.33 x 10~5.

     Lastly, Hammer with Hasselblad and Svendsgaard estimated the
relationship between headaches and ozone exposures.  Using the
same functional relationship,

     probability of experiencing a headache per day per person

   = (0.0976) + 	fl - 0.09761	
                   (1 + exp(4.88 - 4.7(initial ozone))).


If we assume a change in daily maximum ozone concentrations  from
0.13 to 0.12 ppm, then the change in probability of having a
headache will be:


     change in probability of a headache per day per person

     =    fl - 0.0976H1 - O.lllfexpf4.7f0.12 - 0.1311 - 11— ,
       (1 + ((1 -  0.11)7(0.11 - 0.0976))exp(4.7(0.12 - 0.13)))

     - -  5.24 x 10~4.


     Schwartz et al.  (1987) also  re-examined data  collected  in
the prospective study by Hammer  (1974) to  test for the effects of
ozone on  young female adults.  Using a logistic functional form
for the concentration-response function, Schwartz  estimated  the
following:
                               D-16

-------
     Probability of incurring a cough per day per person

     = 	1	    +  dv,
       ( 1 + exp(az + b(Ozone))

where z = other variables, a = estimated coefficients for z
variables, b = coefficient for ozone, Ozone = maximum 1-hour
observation, v = lagged error variables, and d = coefficients for
lagged error variables.  If the baseline probability of incurring
a cough is 0.095 and ozone levels decrease from 0.13 to 0.12 ppm,
given b = 0.61, the change in coughing would be the following:

     change in probability of coughing per person per day

     =    11 - O.Q95UexpfO.61(0.13-0.12n - 11    ,
         (1 + ((1 - 0.095)/0.095)exp(0.61(0.13 - 0.12))

     = - 5.26 x 10~4.

     Schwartz also examined the effects of ozone on eye
discomfort.  Using the same functional form, with b = 2.02 and
the probability of suffering eye irritation = 0.063, a change in
maximum ozone concentrations from 0.13 to 0.12 would yield the
following:

     change in probability of suffering from eye irritation
     per person per day

     =    (i - 0.063WexDf2.02fO. 13-0.121 - 11      ,
          (1 + ((1 - 0.063)/(0.063)exp(2.02(0.13 - 0.12))

     = - 1.20 x lO'3.

This estimate is smaller than the estimated change in probability
given by Hasselblad and Svensgaard in their analysis of the same
data set.  The difference is due in part to Schwartz's attempts
to control for autocorrelation in the data given the time-series
aspects of the data.  Additional analysis of the effects of ozone
on chest discomfort and headaches found that ozone was not a
significant explanatory variable in these regression equations.


     Chronic Health Effects

     The relationship between exposure to ozone concentrations
and chronic respiratory disease was established for humans on the
basis of animal studies conducted by Fujinaka(1985) and
extrapolated to the estimate likely impacts on humans.  The
concentration-response function developed by McKee (1987) using
the animal study data is:
                               D-17

-------
     Proportion of individuals contracting chronic respiratory
     disease per year

     = (exposed population)(l/.64){(7/12)(O.OSppm ozone) +

         (5/12)(annual average daily ozone concentration)},


where the annual average daily ozone level is based on
measurements taken between 10:00 a.m. and 7:00 p.m.  For example,
if one million persons are exposed to ozone concentrations that
fall from 0.07 ppm to 0.06 ppm, and of the one million persons
exposed one in 160 is at risk of contracting chronic respiratory
disease, then the estimated number of fewer chronic respiratory
cases would be:

     change in number of chronic respiratory disease cases per
  year

     = (1 x 106 persons)(l/160)(l/.64){(5/12)(0.06) -
          (5/12)(0.07)>,

     = - 40.7, or 40.7 fewer chronic cases per year.


     Morbidity Values

     For the non-asthmatic population, the health effects
literature suggested that individuals would suffer from coughing,
shortness of breath, eye irritations, and headaches when exposed
to larger ozone concentrations. In order to attribute values to
these effects, a series of contingent valuation studies were
conducted to ascertain how much individuals would be willing to
pay to avoid suffering from each of the above symptoms.  The
average values for each of the willingness-to-pay estimates have
been summarized by Krupnick (1986) and are listed below for each
symptom:


        Symptom             Willingness to Pay Value
                              	Per Incident

          cough                        $4.00
          shortness of breath           8.00
          chest discomfort              6.00
          eye irritation                5.00
          headache                      5.00


     For those individuals expected to suffer from a respiratory
restricted activity day  (RRAD), the unit value was estimated to
be $21.00 per day.  Since the  symptoms given above can

                               D-18

-------
individually, or collectively, contribute to a respiratory
reduced activity day, it is necessary to develop some protocol to
avoid double-counting of health effects and economic benefits.
Although there is no "correct" way to do this, it is suggested
that the number of RRADs and symptom days be calculated
individually using the relevant exposure groups and concentration
response functions.  The sum of the symptom effects excluding eye
irritation effects can be compared with the estimated number of
RRADs.  It expected that individuals experiencing a RRAD will
have one or more of the above symptoms.  Therefore, it is not
clear what relationship may exist between the total number of
cases estimated using the concentration-response relationships
for the symptomatic effects and the number of RRADs. Therefore,
it is suggested that the values for the two estimates be
individually calculated and the arithmetic mean of the two be
used to calculate the benefits of changing ozone concentrations
for all symptoms other than eye irritation, which should be
calculated separately.

     Most of the health effect studies have focused on male adult
populations.  It is likely that some of these same health effects
will occur for females and children when exposed to ozone,
although not necessarily to the same extent.  Several
epidemiological studies have failed to show that ozone
concentrations are statistically correlated with bed rest days or
lost school days experienced by children due to respiratory
illness (Portney, 1984).  The analyst should use some discretion
in determining which populations should be fitted to each
concentration-response function.

     The average value of reducing asthma incidents is estimated
to be $25.00 per incident. This value is developed using
contingent valuation techniques that capture both the cost of
mitigating the effects plus the pain and suffering avoided for a
moderate asthma attack.  This value should be used for both adult
and children asthmatic populations.

     The average value of reducing chronic disease associated
with ozone exposures is based on cost-of-illness estimates for
expenditures undertaken for medical expenses and the value of
lost work days.  Data on national expenditures for chronic
respiratory diseases suggest that the average annual cost-of-
illness for individuals suffering from chronic respiratory
diseases is $740 per year.

     Since the cost-of-illness estimate fails to include the
value of avoiding the pain and suffering that individuals would
be willing to pay to avoid having chronic respiratory diseases,
Rowe  (1985) attempted to estimate the ratio of cost-of-illness
values to willingness-to-pay values for respiratory diseases.
Their study suggested that willingness-to-pay values are
approximately.two times the values estimated using cost-of-

                               D-19

-------
illness measures  for asthmatics.  Therefore, doubling the chronic
respiratory disease value to  $1480 per year may yield a more
accurate estimate of the value of reducing chronic diseases from
ozone.

     Since the  individuals suffer from the effects of ozone
induced chronic diseases not  only for the year in which the
disease's effects first appear, but also in subsequent years, the
costs should be estimated over the entire time period that the
health effects  occur.  Therefore, a present value estimate of the
cost-of-illness should be used.  When estimating present values,
it is appropriate to use a discount value on future expenditures
that reflects the opportunity cost of resources used to combat
the disease.  The current convention suggests that rates between
five and ten percent are appropriate, depending on the
opportunity cost  of capital used in the supply of medical
services.

Welfare Effects

     Agricultural Effects

     Heck (1982,1983) studied the impacts of ozone on yields for
several major types of crops  using ambient and experimental
concentrations  of ozone.  Dose-response functions for each crop
were estimated  using several different functional forms.  The
Weibull function was used to  combine test results conducted at
different test  sites for the  same species of plant in order to
arrive at a common proportional yield response to ozone effects.
The functional  form is:

     Yield = (Maximum yield)(exp(-(ozone concentration/b)c)),


where b = estimated ozone concentration when yield is
(0.37)(maximum  yield) and c = estimated dimensionless shape
parameter for the function.  Using this form, the percentage
change in yield for a given change in ozone relative to the
background ozone  level (average measured at 0.025 ppm) would be:

     percent reduction in yield relative to 0.025 ppm per year

     = 1 -  .


     The individual or common response ozone concentration
parameters estimated by Heck are:
                               D-20

-------
         Crop             b (ppmV            c

          Soybeans         0.150            1.28
          Corn (Coker 16)  0.221            4.46
          Corn (others)    0.158            3.53
          Wheat            0.174            2.90
          Peanut           0.111            2.11
          Cotton           0.197            1.12
          Kidney Bean      0.287            1.77
          Lettuce          0.098            1.22
          Turnip           0.093            2.75
          Spinach          0.135            2.08


     As an example, if the average ozone concentrations during
the growing season falls from 0.040 ppm to 0.030 ppm, then for
soybeans crops, the estimated percent change in yield rates would
be:

          percent reduction in soybean yield per year

     = 1 - {exp((0.040/0.ISO)1-28 - (0.025/0.ISO)1-28)}

     = 1 - {1.087}

     = -0.087,  or an increase in yield of 8.7 percent relative
                to that harvested at 0.040 ppm ozone.


     Agricultural Values

     Kopp (1983) took these dose-response functions and, with a
behavioral model of farming practices, determined how aggregate
crop yields and revenues would change under different ambient
ozone concentration scenarios.  In his original analysis, Kopp
assumed that agricultural markets and prices were established in
a competitive market structure.   Subsequent reviews suggested
that government price support structures may lead to inefficient
and distorted agricultural markets.  When this alternative view
of agricultural markets is considered, then the expected changes
in behavior and agricultural revenues from changes in ozone
concentrations will be different from the more simplistic model
of the agricultural sector.

     The model provides estimated changes in revenues by
agricultural producing area and crop type for different ozone
concentration scenarios.   In order to estimate welfare changes,
one must have some knowledge of the agricultural market
structure.
                               D-21

-------
     Forest Effects

     Tree growth rates have been shown to be sensitive to ambient
ozone concentrations, with reduced growth rates observed at
higher ozone concentrations.  Several studies have been conducted
on controlled exposures to young seedlings/ comparisons of growth
of young seedlings in uncontrolled environments over different
time periods and ozone concentrations/ and radial growth patterns
for sensitive and insensitive species.

     There is a limited amount of data available to draw upon and
construct dose-response functions for different tree species
exposed to ozone concentrations.  The controlled exposure studies
are considered to have the most robust dose-response
relationships/ although the exposure patterns varied between 6
and 12 hours per day, so it is difficult to easily draw a
relationship between percent changes in growth patterns and daily
exposure levels.

     Wilhour (1986) summarized the empirical results developed
using experimental studies exposing seedlings to controlled ozone
concentrations.  He concluded that the average change in percent
growth was 1.8 percent for softwood trees per 0.01 ppm change in
ambient ozone concentration level.  For hardwood species, the
change in growth rate was 1.9 percent per  .01 ppm change in
ozone.

     Forest Values

     The changes in growth patterns can be transformed into
changes in optimal harvest rates, given there exists a particular
size of tree that is desirable to harvest when maximizing
revenues.  Producers may delay harvests to allow for the tree to
mature to the size it would reach if no ozone effects were
present.  Producers may also find that they cannot wait for the
tree to meet full maturity and therefore harvest a smaller tree.
In either event the effects of ozone can result in a reduction in
social welfare over time.  As with agricultural crops, the
existence of market distortions in the lumber business, should
they exist, would need to be reflected in the calculation of
welfare changes attributable to changes in ozone concentrations.

     Materials Effects and Values

     Ozone has been shown to damage products made with certain
elastomers.  Tires exposed to ozone concentrations observed today
in urban areas can lead to cracking and a  reduction in
flexibility that leads to premature changing or retreading of
tires, or attempts on the part of tire manufacturers to introduce
anti-oxidant materials into tires.   The McCarthy analysis
calculated the required levels of anti-oxidants that would need
to be introduced into rubber tires for alternative annual average

                               D-22

-------
ozone concentrations.  McCarthy considered tire-use life, oxidant
use per tire, and annual tire production based on national
statistics.  In addition, several studies have shown that larger
ozone concentrations may influence the expected life of dyes,
fabrics, and other materials.  Collectively, these damages are
reflected in the concentration-response function reported below:

     change in annual per capita materials damages due to ozone

 = $ 28.8 x (change in annual average ozone concentration (ppm)).


BENEFITS OF REDUCING TSP AND VOC EMISSIONS IN BALTIMORE

Introduction

     Based on the methodologies used to estimate risks from TSP
and ozone, and the subsequent values of reducing those risks, we
estimated the potential benefits of reducing TSP and ozone
concentrations in the Baltimore area.  Because the cost analysis
was prepared in a manner that allowed comparisons of costs of
pollution reduction on a per ton basis, it was determined that
the best way to incorporate economic benefit information into the
analysis would also require that benefits be calculated on a per
ton basis.

     However, most of the equations used to calculate benefits of
TSP and ozone reduction required that benefits first be
calculated using changes in ambient concentrations.  Therefore,
information on current ambient concentrations and EPA standards
were used to formulate the necessary changes in ambient
concentrations.  After calculating the necessary percentage
changes in ambient concentrations that must occur to meet the TSP
and ozone standards, baseline information on particulate matter
and volatile organic compound (VOC) emissions from sources in the
Baltimore area were used to calculate the necessary changes in
emissions to meet the desired change in ambient concentrations.
Combining the dollar benefits estimated with changes in ambient
concentrations with the emission reductions necessary to achieve
the change in ambient concentration, provided a dollar benefit
per ton estimate for both TSP and ozone.  The techniques use to
calculate changes in ambient concentration and emissions are
described below in greater detail.


Benefits  from Changes in Total Suspended Particulate Matter

     Methodology

     To calculate dollar benefits per ton for TSP reduction, the
analysis made several assumptions about emissions and ambient
concentrations in the Baltimore area.  The  first assumption was

                               D-23

-------
that a given percentage reduction in ambient concentration could
be achieved through an equal percentage reduction in emissions.
For example, if it was necessary to reduce average annual ambient
concentrations by !%/ then a reduction in annual emissions of 1%
would achieve this objective.  Information on the total number of
particulate emissions in the Baltimore area were used to
calculate the baseline emissions, and a 1% change in the baseline
was used to calculate the number of tons of particulate matter
that would have to be reduced to meet that objective.  In this
case, we used an annual emissions estimate of 17,789 metric tons
per year (mtpy) provided by Versar.  Therefore, a 1% reduction in
emissions would require a reduction of 178 mtpy.

     Using the 1% reduction in ambient concentrations with the
health and welfare effects equations and valuation methods
outlined earlier, we then calculated the total expected benefits
that would result from a 1% reduction in TSP concentrations.  The
total benefits of a 1% reduction in TSP in Baltimore was
estimated to be $2.5 million per year.  Dividing this dollar
benefit by the number of tone necessary to achieve these benefits
(178 mtpy) gave us an estimated dollar benefit per ton of $13,800
per mtpy.  This benefit estimate was then applied to any
projected reduction in particulate matter emissions discussed in
the emission control scenarios.

     The benefit per ton value does not include the benefits from
reduced carcinogenic risks.  In order to calculate the additional
benefits of reducing carcinogenic risks, the expected reduction
in cancer risks, measured as fewer expected cases per year
(cumulative cancer risk), was multiplied by $2 million per cancer
case avoided.  As discussed in the valuation section, this value
is estimated using willingness-to-pay estimates of small changes
in risks, which are then extrapolated to cases (i.e., $2 for 1 in
1 million risk leads to $2 million for 1 in 1 risk).  As these
cancer risks are statistical in nature (i.e., we have not
identified the particular individuals expected to have cancer),
this method is suitable to use.  If we can identify the
individual(s) who will contract cancer, then this method is
inappropriate.

     Limitations

     Because of the relative simplicity of the method used to
ascribe benefits to particulate matter emission reductions, we
should note that there are several issues to consider when using
this approach.  Some of the benefits equations used to draw
connections between TSP and effects were estimated under
conditions atypical to those experienced in Baltimore.  Several
of the relationships have been drawn during high TSP
concentration episodes.  There remains some uncertainty as to
whether there are threshold concentrations, below which health
and welfare effects are reduced.  We used threshold information

                               D-24

-------
where pertinent, given Baltimore's ambient conditions, but some
uncertainty remains.

     We also did not directly link the location of emission
reductions to exposed populations, except for the carcinogenic
benefit estimates.  Most of the benefit estimates are based upon
aggregate (area-wide) average changes in emissions and ambient
concentrations.  Most of the equations were not developed using
detailed emission source information.  Therefore, drawing
relationships between average changes in ambient concentrations,
and site-specific changes in emissions may serve to bias the
benefit estimates.  We should note, that the benefits could be
either understated or overstated for any of the measured effects
(e.g., mortality, morbidity).  The location of the sources
relative to populations, while taking account of dispersion
characteristics and behavioral patterns, will provide some
indication of the direction of the error.

Benefits from Changes in Ozone

     Methodology

     The method used to draw the connection between changes in
ozone concentrations and economic benefits is somewhat comparable
to that used for TSP, although there are several important
differences.  The benefit equations required changes in ambient
ozone concentrations, but the control strategies reported changes
in VOC emissions.  In order to reconcile these different units,
we were forced to make several assumptions.

     Previous work prepared by EPA had drawn relationships
between changes in VOC emissions and changes in ozone
concentrations.  Work prepared by Ostro and McGartland (1985)
estimated that a 1% reduction in VOC emissions in an urban area
would, on average, reduce peak ambient ozone concentrations in
the urban area by 0.6%. Therefore, in order to obtain a 1%
reduction in peak ozone readings, VOC emissions would have to be
reduced by about 1.7%.  Using Versar's estimate on the current
amount of VOC emissions in the area  (94,017 mtpy), a 1% reduction
in peak ozone readings requires a 1570 mtpy reduction in VOC
emissions.  Using the benefits equations outlined earlier in the
study, we then calculated the total benefits resulting from a 1%
reduction in peak ozone concentrations.  We estimated that the
total benefits of a 1% reduction in ozone concentrations would be
$540,000.  Dividing this benefit estimate into the fewer tons of
VOCs emitted necessary to achieve this benefit, resulted in an
estimated dollar benefit per ton of  $345 per mtpy.  This benefit
estimate was then applied to VOC emission reductions analyzed in
the control scenarios.

     The carcinogenic risks of VOC emissions were not included in
this benefit per ton estimate.  The procedure used to calculate

                               D-25

-------
cancer reduction benefits used with TSP emissions was duplicated
for VOC emissions.  A value of $2 million per cancer case avoided
was used in conjunction with the non-carcinogenic benefits
accounted for in the $345 per mtpy value.

     Limitations

     The approach used to draw connections between VOC emissions
and ozone concentrations suffer from the same problems
experienced for particulate matter emissions and ambient TSP
concentrations.  Some of the health effects studies were
conducted in settings that are not normally present in the
Baltimore area.  The existence of threshold values was considered
in the analysis, but the specific conditions of Baltimore may
result in an overestimate or underestimate of the benefits of
reducing ozone concentrations.  The relationship between VOC
emissions and ozone concentrations is dependent upon many factors
which were not controlled for to capture Baltimore's specific
conditions.  The relationship between VOC emissions and ozone
concentrations depends upon many variables, but the relationship
used in the analysis was felt to be representative of average
conditions found in urban settings.

     The contribution of the specific VOC emission sources
examined in the analysis to ozone concentrations and risks may be
more direct than that attributed to the TSP emissions and risks.
Since the VOC emissions are likely to affect a broader area given
the atmospheric characteristics of ozone formation, the
relationship between VOC emissions and ozone health and welfare
effects is likely to be stronger than that assumed to exist
between particulate matter emissions and TSP health and welfare
effects.
Conclusions

     To calculate the economic benefits of reducing particulate
matter and VOC emissions, we made use of many established
methodologies that attempt to draw linkages between exposures to
pollutants and effects.  After establishing the extent of the
effect, we then use economic theory and techniques to calculate
the economic value of preventing these effects from occurring.
This approach allows us to estimate the economic benefits of
reducing pollution emissions, which can then be compared against
the economic costs of installing and operating equipment, or
making other changes in production processes.

     We have estimated that each ton of TSP that is prevented
from entering Baltimore's air results in an average savings of
$13,700.  Similarly, each ton of VOC that is not emitted into the
air results in an average savings of $345.  Both of these
estimates do not include the value of reducing carcinogenic risks

                               D-26

-------
identified with these emissions.  A method is available to add
these benefits to the non-carcinogenic benefits in order to
calculate the total benefits of reducing pollution emissions from
each of the sources identified in the air toxics study.

     To conclude, we illustrate an example of using this
information to calculate the economic benefits of reducing
emissions from one of the sources identified in the study.
Woodstoves used for residential heating emit particulate and
VOCs, for which some of these components contribute cancer risks.
The control option proposed would be to discontinue using
woodstoves for heat, and switch to other conventional forms of
heat (e.g., gas, electric).

     By eliminating residential woodstove emissions, annual TSP
emissions would be reduced by 93.6 mtpy and VOC emissions would
be reduced by 148 mtpy.  The cancer risks associated with these
emissions is calculated to contribute 0.033 cancer cases per
year.  Using the TSP and VOC benefit per mtpy values ($13,700 and
$345, respectively), the annual non-carcinogenic benefits of
reducing TSP emissions are $2,127,769, and the annual non-
carcinogenic benefits of reducing VOC emissions are $32,292.
Using $2,000,000 per reduced annual cancer case, the cancer
benefits are valued at $66,877.  The total benefits are
$2,126,769.  This can be compared against the annual costs of
limiting the use of woodstoves, which is estimated to be $220,000
per year.  Therefore, the net benefits (benefit minus costs) are
$1,906,769 per year.

     To compare this control option with other possible control
options, it is desirable to calculate the benefit-cost ratio
(benefits divided by cost).  Comparing ratios across control
options allows one to examine which additional expenditure
(control cost) provides the greatest additional benefit (benefit
estimate).  If there is no limit on the amount that can be spent
to control emissions, it is economically efficient to  first adopt
those control options that provide the greatest incremental
benefits  (i.e., have the highest benefit-cost ratio).  The
benefit-cost ratio for the control of residential woodstoves does
not have the largest benefit-cost ratio, but its ratio of 9.9 is
one of the higher ones among the control options examined in the
study.
                               D-27

-------
                            REFERENCES
Chestnut, L.G., R.D. Rowe, and J. Murdoch.  1987.  "Review of
     Establishing and Valuing the Effects of Improved Visibility
     in Eastern United States."  Energy and Resource Consultants,
     Inc. prepared for the Office of Policy Analysis, U.S.
     Environmental Protection Agency.  Contract No. 68-01-7033,
     October.

Ferris, B.6. Jr.  1978.  "Health Effects of Exposure to Low
     Levels of Regulated Air Pollutants: A Critical Review,"
     Journal of the Air Pollution Control Association.  28 (May):
     482-97.

Fujinaka, L.E., D.M. Hyde, C.G. Plopper, W.S. Tyler, D.L.
     Dungworth, L.O. Lollini.  1985.  "Respiratory Bronchitis
     Following Long-Term Ozone Exposure in Bonnet Monkeys: A
     Morphometric Study," Experimental Luna Research. 8:167-190.

Hammer, D. I., V. Hasselblad, B. Portnoy, and P.F. Wehrle.  1974.
     "Los Angeles Student Nurse Study: Daily Symptom Reporting
     and Photochemical Oxidants," Archives of Environmental
     Health.  28: 255-260.

Hasselblad, V., and D. Svendsgaard.   1975.  "Reanalysis of the
     Los Angeles Student Nurse Study," staff report, Health
     Effects Research Laboratory, U.S. Environmental Protection
     Agency, Research Triangle Park, N.C.

Heck, W.W., O.C. Taylor, R. Adams, G. Bingham, J. Miller, E.
     Preston,  and L. Weinstein.  1982.  "Assessment of Crop Loss
     for Ozone," Journal of the Air Pollution Control
     Association. 32 (4): 353-61.

Heck, W.W., R. Adams, W.W. Cure, A.S. Heagle, H.E. Heggestad,
     R.J. Kohut, L.W. Kress, J.O. Rawlings, O.C. Taylor. 1983. "A
     Reassessment of Crop Loss from Ozone," Environmental Science
     and Technology. 17 (12): 572-81.

Holguin, A.H., P. fluffier, C. Constant, T. Stock, D. Kotchmar, B.
     Hsi, D. Jenkins, B. Gehan, L. Noel, and M. Mei.  1984.   "The
     Effects of Ozone on Asthmatics in the Houston Area,"  Air
     Pollution control Association Transactions on Ozone/Oxidants
     Standards.  Houston, Texas, November, 1984: 262-280.

Kopp, R.J., W.J. Vaughan, and M. Hazilla. 1983.  Agricultural
     Sector Benefits Analysis of Ozone: Methods Evaluation and
     Demonstration.  Prepared by Resources for the Future for
     Office of Air Quality Planning and Standards, U.S.
     Environmental Protection Agency, Research Triangle Park,
     N.C. September.

                               D-28

-------
                       REFERENCES (CONTD.)


Krupnick, A.J., J.R. Kurland, and T. Narel.  1986.  A Preliminary
     Analvaia for the Control of Photochemical Oxidants.
     Prepared by Resources for the Future for Office of Air
     Quality Planning and Standards, U.S. Environmental
     Protection Agency, Research Triangle Park, N.C., Contract
     No. 68-02-3844.

Mazumdar, S., H. Schimmel. and I.T.T. Higgins.  1982.
     "Relationship of Daily Mortality to Air Pollution: An
     Analysis of 14 London Winters, 1958/59 - 1971/72." Archives
     of Environmental Health  37 (July/August): 213-20.

McDonnell, W.F., D.H. Horstman, M.J. Hazucha, E. Seal, E.D. Haak,
     S. Salaam, and D.E. House.  1983.   "Pulmonary Effects of
     Ozone Exposure During Exercise: Dose-Response
     Characteristics,"  Journal of Applied Physiology.  54 (5)i
     1345-1352.

McCarthy, E.F., et al.  1983.  Damage Coat Model for Pollution
     Effects on Material.  U.S. Environmental Sciences Research
     Laboratory, Research Triangle Park, N.C.

McKee, D. and J. Graham. 1987. "Estimated Dose-Response Function
     for Chronic Respiratory Disease Associated with Ozone
     Exposures," Staff Paper, Office of Air Quality Planning and
     Standards, U.S. Environmental Protection Agency, Research
     Triangle Park, N.C.

Ostro, B.D.  1983.   "Urban Air Pollution and Morbidity: A
     Retrospective Approach,"  Journal of Environmental Economics
     and Management.  10  (December). 371-82.

Ostro, B.D.  1987.   "Air Pollution and Morbidity Revisited: A
     Specification Test'"  Journal of Environmental Economics and
     Management.  13  (December).

Portney, P.R. and J. Mullahy.  1986.  "Urban Air Quality and Acute
     Respiratory Illness,"   Journal of Urban Economics. 20
      (July): 21-38.

Rowe,  R.D. and L.G.  Chestnut.  1985.  Qxidants and Asthmatics in
     Los Anaeles: A  Benefits Analysis.   Prepared  by Energy and
     Resource Consultants, Inc. for Office of  Policy Analysis,
     U.S. Environmental Protection Agency, EPA-230-07-85-010.
     Wa s hington, D.C.
                               D-29

-------
                       REFERENCES (CONTD.)


Samet, J.M., Y. Bishop, F.E. Speizer, J.D. Spongier, and E.G.
     Ferris.  1981.  "The Relationship Between Air Pollution and
     Emergency Room visits in an Industrial Community,"  Journal
     of the Air Pollution Control Association.  31 (March): 236-
     40.

Schwartz, J. and A. Marcus.  1986.  "Statistical Reanalysis of
     Data Relating Mortality to Air Pollution During London
     Winters  1958-72."  Working Paper, U.S. Environmental
     Protection Agency, Washington, D.C., October.

Trijonis, J.  1982.  "Visibility In California,"  Journal of the
     Air Pollution Control Association.  32 (February): 165-69.

Violette, D.M., and L.G. Chestnut.  1983.  Valuing Reductions in
     Risks; A Review of the Empirical Estimates.  EPA-230-05-83-
     003.  Prepared by Energy and Resource Consultants to Office
     of Policy Analysis, U.S. Environmental Protection Agency,
     Washington, D.C., June.

Violette, D.M., L.G. Chestnut, and A. Fisher.  1986.  "Valuing
     Risks to Human Health,"  Toxics Law Reporter.  September.

Wilhour, R. 1986. "Benefit to Forests of Reducing Ambient Ozone
     Concentrations," Memorandum to Office of Air Quality
     Planning and Standards and Office of Policy Analysis, U.S.
     Environmental Protection Agency, Washington D.C. September
     22, 1986.
                               D-30

-------