TD886.5
.03
                                                      OOOR87104

                                Technical Report
                    Air Toxics Emissions From Motor Vehicles
                                       By

                                 Penny M. Carey



                                 September,  1987
                                     NOTICE

              Technical  Reports  do  not  necessarily  represent  final  EPA
              decisions  or  positions.   They  are  intended  to  present
              technical  analysis  of  issues  using data  which are currently
              available.   The  purpose  in the  release  of  such  reports is
              to  facilitate  the exchange of technical  information  and to
              inform  the public of technical developments  which  may form
              the  basis  for  a  final  EPA  decision,  position or regulatory
              action.

                             Technical Support  Staff
                      Emission Control Technology Division
                            Office of Mobile Sources
                           Office of  Air  and Radiation
                      U. S.  Environmental Protection Agency

-------
                            Preface

     This  report   titled   "Air   Toxics  Emissions  from  Motor
Vehicles"  is   designed  to  be  a   compilation  of  available
information  on  emission  levels  of  potentially  carcinogenic
substances  from  motor  vehicles.   While  earlier  EPA  reports
(e.g.,  "The  Air  Toxics  Problem   in  the  United  States;  An
Analysis  of  Cancer  Risks  for  Selected  Pollutants")  discussed
air  toxics  emissions  in general,  their  main emphasis  was not
mobile sources such as  in this report.

     EPA  currently  plans   no  further  regulatory  action   as  a
result of this  report.   EPA invites comments on this report and
plans  to update  this   information  as  additional  data  become
available.

     A draft version  of this  report was circulated  prior to the
release  of  this  final  version.   Users  are  cautioned  that
Section  7.0  dealing  with  gasoline  PIC/POM/organics  has  been
revised to correct an  error due  to  a misinterpretation  of some
of the  references.  The estimated  risks  from  gasoline organics
are  now  much   lower  although  still  in  the  range of  100-200
cancers/year.    The    summary    tables    have   been    changed
accordingly.   This is the only substantive  change  of  note from
the draft version.
                                          , II

-------
                       TABLE OF CONTENTS
PREFACE 	 i

EXECUTIVE SUMMARY 	   v

1.0  INTRODUCTION 	   1

2.0  DIESEL PARTICULATE 	   2
     2.1   Formation and Composition  	   2
     2.2   Control Technology 	   3
     2.3   Emissions	   4
           2.3.1 Emission Standards 	   4
           2.3.2 Emission Factors by Model Year 	   6
           2.3.3 Emission Factors for Calendar Years
                 1986 and 1995	   6
           2.3.4 Nationwide Diesel Particulate Emissions.  .  12
           2.3.5 Other Sources of Particulate 	  17
     2.4.  Ambient Concentrations of Diesel Particulate .  .  17
           2.4.1 Estimation of Urban and Rural Exposure .  .  17
           2.4.2 Comparison to Other Exposure Estimates .  .  24
                 2.4.2.1 1983 QMS Exposure Estimate ....  24
                 2.4.2.2 1981 Lovelace ITRI Exposure
                           Estimate	25
                 2.4.2.3 Exposure Estimate Based on a
                           Lead Surrogate Approach  ....  26
           2.4.3 Comparison to Particulate Monitoring Data.  27
     2.5   Health Effects of Diesel Particulate and Unit
                 Risk Estimates	27
     2.6   Current and Projected Health Risk	28

3.0  FORMALDEHYDE	30
     3.1   Formation, Composition and Control Technology.  .  30
     3.2   Emissions	30
           3.2.1 Emission Factors for Calendar Years
                 1986 and 1995	30
           3.2.2 Nationwide Mobile Source Formaldehyde
                 Emissions	33
           3.2.3 Other Sources of Formaldehyde  	  33
     3.3   Ambient Concentrations of Formaldehyde
                 Emitted by Mobile Sources  	  34
           3.3.1 Estimation of Urban and Rural Exposure .  .  34
           3.3.2 Contribution of Mobile Sources to Ambient
                 Formaldehyde Levels  	  36
     3.4   Health Effects of Formaldehyde and Unit Risk
                 Estimates	38
     3.5   Current and Projected Health Risk	39
     3.6   Current Activities 	  41

4.0  BENZENE	43
     4.1   Formation, Composition and Control Technology  .  43
     4.2   Emissions	43
           4.2.1 Emission Factors for Calendar Years
                 1986 and 1995	43

                              ii

-------
                  TABLE OF CONTENTS  (Continued)
           4.2.2 Contribution of Mobile Sources to
                 Nationwide Benzene Emissions 	  45
     4.3   Health Effects of Benzene and the Unit Risk
                 Estimate	45
     4.4   Current and Projected Health Risk	46
     4.5   CARB Analysis of Current and Projected
                 Health Risk	48
     4.6   Comparison of EPA and CARB Health Risks  ....  49
     4.7   Current Activities 	  50

5.0  GASOLINE VAPORS	51

6.0  GAS PHASE ORGANICS	54
     6.1   Formation and Control Technology 	  54
     6.2   Composition	54
     6.3   Mutagenicity of VOC	56
     6.4   Risk Associated with Individual VOC	56
     6.5   Reactivity of VOC	63
     6.6   Current Activities 	  65

7.0  ORGANICS ASSOCIATED WITH NON-DIESEL PARTICULATE  ...  66
     7.1   Emission Rates and Composition 	  66
     7.2   Risk from Particle-Associated Organics 	  66

8.0  DIOXINS	73
     8.1   Composition	73
     8.2   Emissions	73
     8^3   Concentrations of Dioxins  	  73
     8»4   Current Activities 	  73

9.0  ASBESTOS	74
     9.1   Emissions and Ambient Concentrations 	  74
     9.2   Cancer Risk	74
     9.3   Current Activities 	  74

10.0 VEHICLE INTERIOR EMISSIONS 	  76
     10.1  Composition and Concentration  	  76
     10.2  Cancer Risk and Current Activities	76

11.0 METALS	77
     11.1  Lead	77
           11.1.1 Source and Emission Factors 	  77
           11.1.2 Health Effects  	  78
     11.2  Manganese	80
           11.2.1 Source and Emission Factors 	  80
           11.2.2 Health Effects  	  81

                              iii

-------
                  TABLE OF CONTENTS (Continued)
     11.3  Platinum	82
           11.3.1 Source and Emission Factors 	  82
           11.3.2 Health Effects   	  82
     11.4  Cadmium	83
           11.4.1 Source and Emission Factors 	  83
           11.4.2 Health Effects and Risk Estimate  ....  83

12.0 SIX MONTH STUDY:  SUMMARY AND COMPARISON OF RESULTS   .  84
     12.1  Purpose of Six Month Study and Summary of Results 84
     12.2  Comparison of Six Month Study Results with Results
                 of This Study	86
          12.2.1 Formaldehyde 	  87
          12.2.2 Benzene  	  89
          12.2.3 PIC, B(a)P, Diesel Particulate and
                   Gasoline PIC/POM 	  90
          12.2.4 Gasoline Vapors   	  91
          12.2.5 1,3-Butadiene  	  92
          12.2.6 Ethylene 	  92
          12.2.7 Asbestos 	  93
          12.2.8 Ethylene Dibromide (EDB) and Cadmium ...  93
          12.2.9 Total Aggregate Risk 	  94

13.0 SUMMARY AND LIMITATIONS	95

REFERENCES	97

GLOSSARY OF TERMS	105
                               IV

-------
EXECUTIVE SUMMARY

     EPA  completed a  report in  1985,  commonly referred  to as
the Six Month Study,  which contains estimated cancer  risks for
a  number  of  toxic air  pollutants.   The report  indicated that
mobile sources  may be  responsible  for  a large  portion of the
aggregate cancer  incidence.   The Six Month  Study,  however, was
broad  in nature with  the goal  to  obtain a  quick  assessment of
the  air  toxics  problem  in the  United  States  and to  guide
further studies.

     The  purpose  of  this  report  is  to focus  on  air  toxics
emissions   from  mobile   sources.    Specific   pollutants   and
pollutant   categories   which   are   discussed   include   diesel
particulate,  formaldehyde,  benzene,  gasoline vapors,  gas  phase
organics,    organics  associated  with   non-diesel   particulate,
dioxins, asbestos,  vehicle  interior emissions and  metals.   The
report considers  all  air  carcinogens  for  which EPA has  unit
risk  estimates  and  are  emitted  from  motor vehicles.   Where
adequate  information  was  available,  quantitative estimates  of
cancer  incidence  were  made  for  calendar  years  1986  and  1995.
The results were  then compared to  the  results obtained in the
Six Month Study.

     The  unit  risks  used   in  this paper  are  defined  as  the
individual  life   time   excess   cancer   risk  from   continuous
exposure to 1  ug  carcinogen per  m3 inhaled air.   Assuming  a
life time is 70 years, the excess lung cancer risk in  1  year is
derived by  simply  dividing  the  unit  risk   by  70.   Using  this
approach,   latency  is  ignored.    The unit  risks used in  this
paper  are  95  percent upper  confidence  limits rather  than  best
estimates.   This is consistent  with current  EPA practice.   The
risk estimates  presented  should  therefore  be considered  upper
bound estimates.

     The risks  obtained in  this  study  are  summarized  in  Table
S-l.   The  aggregate risk in 1986 for the total  U.S.  population
is estimated  to range  from  385 to  2286  cancer  incidences  and
drops  roughly 40  percent  by 1995.   Reasons for the  projected
decrease  in  risk  in  1995  include 1)  more  stringent  diesel
particulate standards  for both light-  and  heavy-duty vehicles
and 2) the  increasing use  of  3-way catalyst-equipped  vehicles
coupled with the phase out of non-catalyst-equipped vehicles.

     As seen   in   Table  S-l,  there is  a wide  range of  risk
estimates    associated   with   each   pollutant.     For   diesel
particulate, the range  is  due  to the range  of potency (or  unit
risk)   estimates which  were used and,   for   1995,  a  range  of
assumptions  regarding  future  diesel   sales.   The  range  for
formaldehyde can be attributed  to the uncertainty  regarding the

-------
                                   Table  S-l

              Summary of Risk Estimates Contained in This Study3

                          U.S. Cancer Incidences/Year"  Section of Report
Motor Vehicle Pollutant          1986

Diesel Particulate              178- 860
Formaldehyde                     46- 131
Benzene                          92- 223
Gasoline Vapors                   65
Other Gas Phase Organics
   1,3-Butadiene
   Ethylene
Gasoline PIC/POM
Dioxins
Asbestos
Vehicle Interior Emissions
Cadmium
Ethylene Dibromide

Total:                          385-2286
 1995

 92-576
 29- 77
 57-145
  NDd
0-
0-
1.3 -
ND
0.41-
ND
0.18
1.8
656
60
176

113.4



0-460
0-31
0.78-136
ND
ND
ND
0
0.54
Discussing Pollutant

            2
            3
            4
            5

            6
            6
            7
            8
            9
            10
            11
            12
179-1426
     The risk estimates are 95°o upper confidence limits.

     The risk  estimates for  gasoline  vapors,  asbestos,  cadmium and ethylene
     dibromide are  for  urban  exposure  only.  Risks  for the  other  pollutants
     include both urban and rural exposure.

     The total  risk in  1995  is  slightly  underestimated.   Due  to  inadequate
     information and the  sensitivity of 1995 risk to  control decisions which
     have not yet  been  made,  projected risk  estimates  were  not made for some
     of the pollutants.
     ND=Not Determined.

-------
contribution  of photochemically  formed formaldehyde.   The low
end  of  the range  attempts  to  account  only  for  formaldehyde
directly  emitted  from the exhaust of motor vehicles.   The high
end  of  the range  attempts  to  account  for  both  formaldehyde
directly  emitted and  formaldehyde  formed  in the atmosphere from
other  mobile  source volatile organic  compound  (VOC) emissions.
Both  ends use  a  single unit risk estimate  which  is  based  on
formation of malignant tumors.

     For   benzene,   the   lower   limit  is   based   on  ambient
concentrations predicted by a model, whereas  the  upper limit is
based  on actual monitoring data,  with a  mobile source fraction
assigned  based on  the  mobile   source emissions  contribution.
The   ranges   for   1,3-butadiene   and  asbestos   account  for
uncertainties   in   emission   factors;   for   ethylene,   the
uncertainty is  based on the unit  risk estimate.  The range for
gasoline  products  of incomplete  combustion   (PIC)  is due  to  a
number of different  assumptions  regarding both  emission factors
and unit  risk estimates.

     Mobile source  emissions are  extremely  complex.   Hundreds
of  compounds,  both  in  the  gas  phase  and  associated  with
particles  are  present.    The lack  of  emissions  data  and/or
health  data  and/or  exposure data prevented  quantitative risk
estimates for any additional  pollutants.   Of  particular concern
are  pollutants which  are  formed  photochemically  from  mobile
source  emissions.   This  category  of  pollutants   could  have
considerable   impact   but  not   enough   is   known   to  make   a
quantitative estimate.

     A comparison  of the  results  obtained in  this  study with
those  obtained  in the  Six Month  Study  is given in  Table S-2.
In the Six  Month  Study, the  mobile source contribution  to the
air toxics  risk was determined based  on  dispersion modeling  of
emissions in 35 highly  populated counties (total population  of
45  million).    They  represent   a  variety  of   industrial  and
population   distributions,   but    are   not    considered   a
statistically  representative  sample   of  the   country.    When
comparing the results  of the Six  Month Study with  the results
of  this  study,   cancer  incidences  are  expressed  per  million
urban people exposed.

     The  aggregate  risk from mobile  sources  in  the  Six  Month
Study  is  2.65  per million.   In  this  study,  the  aggregate risk
ranges from 1.80  to  10.58  per urban million.   As seen in Table
S-2,   a   few   of   the   unit   risk  estimates  have   increased
considerably since  the  release  of the  Six  Month  Study,  most
notably  formaldehyde and  1,3-butadiene.    If the  formaldehyde
risk in the Six Month Study is increased to reflect  the updated
unit  risk,  the  aggregate  risk  in  the Six  Month  Study  would

                              vii

-------
 I

00
        CO
        a;
        o
        01
        a

        §
        u
0) J5
U *•» >i
G C T)
03 O 3
O -H 4J
rH C/J
(!) r-H
U -rl 01
1-1 6 -rt
3 JS
O C H
O
r-
o
i
(N

.
0
I/I 03
CU U
rH 3
O 0) X
S 044J
— c
rH O
03 X S
3 w
C ••-* \o


[-^
O

o
c a

•^
3
4-1
I/I

01
•H
X

i
0
r— t
^
•
v£)
in
^
— -o
0) 3
rH 4->
•rt M
E
*^ 01
Cn -rt
*•* jC
CO EH
^*
o
•
o
1
01
^i
o

0
4-1 U
• rt 0
0} 4-1
RO
03

O »C
CJ C -M
o a
• rt O
01 S
01
• rt v£)


CO
co
0
•
0
1





CO
TJ

4->
C

^J
3
r— i
r— 1
O
Ol
f^
0)
-a

03
Ui
o
fo
O
rH
1
^
Q
0 2






i— i r~
co o
. .
O OJ




<£
f
O
r-H
^
O
Q
oo 2

to —
1 l
O O
rH rH
<£ OJ
• •
v£» «#






0
O
rH -0
Q
0 2



Ul
)
O
i-l
rH X
ro VJD
i-l vD
* •
O rH











CO
C
CO
N u
G O
CO M
CQ P[



0)






OJ
o

o




n
1
o
r-H
X
oo
.
CO



0) 0)
G C
•rt 'rt
r-H r-H
0 0
01 01
OJ 03
O O
vO
r-H
CO
1
O









O




IT
I
O
r-l
X
00
.
01

r-
1
O
r-H
X
VO
.
«*
\o
01
o

o
1

01
o

0




m
1
0
rH
X
0
.
r-l






CO
C
CO
•rt
T3
0)

3
n
i
CO
V
r-l
01
o
1
o








a
2




f
i
o
rH
X
r-
•
01






P
2
i-H
VO
01

0
1

co
01

o










Q
J2»










0)
C
CO
r— 1
£
^j
w
co
VO
o
1
01
O
o

o








Q
^

«
o
r-l
X
o
{Q
01
1
*JD
•
vO






Q
2


to
1
o
i-H
X
00
01
1
^J*










Q
£5










Ol
o

01
CO
01

i-H
o
o

o






1-H
0

o




M
1
o
r-l
X
00
.
i-H


1
o
i-H
X
CO
•
01



to
1
o
rH
X
rj\
.
rH




to
i
0
rH
X
in
•
00













•rt
1
TJ
CJ
rH
O

O






01
0

o




T
1
O
i-H
X
rH
.
in

•r

O
rH
X
rH
.
in



in
i
o
rH
X
i-H
.
f.^




T
1
0
rH
X
CO
.
rH

CO
T3
•rt
E
0

•rt
Q

CO
C
CO
1— 1
J51
4-)

OO
in
o
i-H
1
O
oo
.
rH






in
vO
.
01






. .
1— 1
03
4-1
O
H
















































limits.
CO
o
C
co
-d
• H
•a
o
o

ui
CO
cu
3
oN°
m

0)
t-i
m
01
CO

ft)
£~
• H
j i
01
0)

*^
01
•H
Ul

(1)
.c
H
ro

TJ
0)
01
3

01
03

C
O

r-H
l~^
•rt
£
O
OO
rH

|[ |
O
C
o
•rl
4J
03
rH
3
CU
a

G
(0

J_|
3

£
J3
organics from gasoline-fueled vehicles were considered to
PIC risk was not determined in this study.

'O flj
(U
4-1 -
(0 CO
•H S-l
0 O
O MH
01 0)
01 U
(0 (U
1 JC
CO H
u
•r-l
Ui 'O
Ol 4J
01
•a
c 01
0) -H
JC
(1) 4J
4J
03 C
r-H -r-|
3
U O
•H M
4J Cli
fO 4-)
Oj G
0)
r-l 01
 u
CU Qj
•H CO
a u
u






















T3
0)
C

'e

0)
4J
0)
Q

4J
o
2
II
Q
•o
















C*
O
•H
4_)
8
03
t-H
Qj
X
CO

U
0

4J
X
CO
4->

(U
0)
00
01

-------
increase to  2.73  per million.  If the 1,3-butadiene risk in the
Six Month  Study is increased to reflect the  updated  unit risk,
the risk from 1,3-butadiene would no longer be negligible.

     The Six Month Study  did not explicitly  include  diesel or
gasoline-fueled particulate or  gas  phase exhaust organics (with
the exception  of  formaldehyde  and  benzene).   It did  include a
broad  category  of  pollutants   referred   to  as  products  of
incomplete   combustion  (PIC).    PIC    are   postulated   to  be
primarily  polynuclear  aromatic  hydrocarbons.   PIC  therefore
includes most  of  the  organics  on  motor  vehicle particulate and
some  gas  phase  exhaust organics.   The  unit  risk  for  PIC was
derived from epidemiological  studies  of the general population
and studies  of  occupational  exposure  to  PIC (e.g.,  coke  oven
emissions).   The  PIC  unit  risk  is   expressed  per  unit  of
exposure of benzo(a)pyrene (B(a)P).

     In  the Six   Month  Study,  the  B(a)P  emission factor  for
mobile  sources   was  used   to   calculate  an  annual  average
exposure.   The  B(a)P exposure  was  then multiplied by  the PIC
unit  risk  to estimate  annual cancer incidence  due  to PIC from
mobile sources.   The same process  was performed for  all other
sources  emitting  B(a)P to determine  cumulative annual  cancer
incidence  from  PIC.   Since  roughly  75 percent of   the  total
B(a)P emissions were  said  to  be attributable to mobile sources,
75 percent of the total PIC  cancer  incidence is due  to mobile
sources.    The  PIC  risk from mobile sources  in the  Six Month
Study is estimated to be 2.07 per urban million.

     In this study,  risks  from diesel particulate  and gasoline
particle-associated  organics  (referred  to as  gasoline  PIC/POM
where  POM  is   polycyclic   organic  matter)   were   analyzed
separately.   The  sum  of  the diesel  particulate  and gasoline
PIC/POM risks will be  considered  to represent PIC.   Using  this
approach,   the  total PIC  risk  ranges  from 0.77-4.43  per urban
million.    The  PIC  risk obtained  in the  Six Month Study  lies
within this range.

     When  reviewing the results  of this  study, the  following
are  a  few  of   the  important  limitations  that   should  be
considered.  The  report only  accounts  for  a small number of the
mobile source  pollutants  known  to  be  emitted, and  does  not
consider    reactions   of   mobile  source   pollutants   in   the
atmosphere.  Resulting secondary pollutants may  be  more  or  less
carcinogenic than  what  was  originally  emitted.   The report also
does  not  fully   and  accurately  take  into  account  seasonal
variations  in  emissions.   This may result  in  underestimating
the risk  since  it is  likely some  air  toxics  emissions  will
increase   with  decreasing  temperature.    The  emission  factors
were  developed  using  75°F as  the  ambient  temperature.   The
risks  presented   are  assumed   to   be    additive.     The  risk
projections  for   1995  are  based  on  the  emission  standards

                               ix

-------
currently  in  place.   Changes  in  fuel  composition  are  not
considered.   The  risk  projections  do  not  consider  use  of
alternative fuels, such as methanol.  Neither do the  current or
projected  risk   estimates  consider  use   of   alcohol/gasoline
blends.

     The  following  is  a  brief  summary  of   each  pollutant/
pollutant category covered  in  this  report.   It includes current
research  and  regulatory activity.   It  should  be  read  when
reviewing   the    tables,    since   it   describes    important
uncertainties.    Literature  references  are  not  given  in  this
summary, but are in the body of the report.

     Diesel  Particulate   -   Diesel   particulate  exhaust   is
composed of an elemental  carbon  core with hundreds of  adsorbed
organic   compounds  ranging  from  C14  to  about C40.   Over  90
percent  of  diesel particulate is  less  than 1  micron  in  size.
The   light-duty   vehicle   and  truck  emissions   standard   is
currently 0.6 gram/mile.  New  standards, effective  in 1987,  are
0.20  gram/mile  for  light-duty vehicles  and 0.26 gram/mile  for
light-duty  trucks.   For  heavy-duty diesel  engines,   there  is
currently no diesel  particulate emission standard.   A standard
of  0.6  gram/brake horsepower-hour  (g/bhp-hr)  begins  in  1988,
with  increasingly more  stringent  standards  effective  in  1991
and 1994.   These  increasingly  stringent  standards  are accounted
for in the 1995 projections.

     Diesel particulate  was found to be  mutagenic in  the  late
1970's.    Subsequent  studies  revealed  that   nitropolynuclear
aromatic hydrocarbons  (nitro-PAH),   specifically,  nitropyrenes,
dinitropyrenes   and  nitrohydroxypyrenes  together  account  for
much of the mutagenicity  observed.   The organics extracted from
diesel  particulate  and  other  known  carcinogens  such  as  coke
oven  emissions  were tested in a  battery of bioassays.   These
included bacteria and mammalian  cell bioassays,  and  one  skin
painting  study  with SENCAR  mice.   Animal  inhalation  studies
were  conducted at that time but  gave negative  or  inconclusive
results.  The  unit  risk  for  diesel particulate was  determined
by  comparing  the  potency  of  diesel  particulate   with  the
potencies of  the other  carcinogens  determined in  these tests.
The range  of upper  confidence limit  unit   risks  used  in  this
study,  0.2-1.OxlO~4,  is  based  on  various  analyses  of  the
comparative   potency    data.     Several    animal    inhalation
experiments have  been  recently completed which,  in contrast to
the  earlier studies,   show that  diesel  exhaust  causes  lung
tumors  in  rats.   After  being analyzed,  these experiments  may
help narrow the range of unit  risks.

     Total diesel particulate  emissions  in 1986 were estimated
to be 274,000 metric tons,  or roughly 3.9 percent  of the total
suspended  particulate  (TSP)  emissions.    Diesel  particulate
emissions  are   projected  to   drop  to  125,000-154,000  metric

-------
tons/year  in 1995.   This is  due  to the  more  stringent diesel
particulate  standards.   The  range  of  emissions  estimates  in
1995 is due to the range  of  diesel sales assumed.

     The annual  mean exposure level, estimated using a modified
version  of  the  NAAQS  exposure model   (NEM)  for  CO,  was  2.6
ug/m3  in  1986.   This exposure  level accounts  for  both indoor
and  outdoor  exposure  to diesel  particulate.    This  represents
roughly  5  percent  of   the   1984   annual  geometric  mean  TSP
concentration.   This  level  drops  to  1.2-1.6   ug/m3  in  1995.
The  resulting annual  lung  cancer  risk  from  diesel  particulate
exposure for  the U.S.  population is  178-860 in  1986  and 92-443
in 1995.

     Formaldehyde  -  Formaldehyde  is emitted in the  exhaust of
both gasoline- and diesel-fueled vehicles.   It  has the chemical
formula  CH2O.    Formaldehyde   is   of  interest  due   to   its
photochemical  reactivity in   ozone  formation  and  suspected
carcinogenicity.    An  upper  confidence  limit  unit  risk  of
1.3x10   was  used.   It   is  based  on  a  single  study in  which
rats  exposed  to  formaldehyde developed  malignant   and  benign
tumors  in  the  nasal  cavities;  the  unit  risk  is  based  on
malignant  tumor  formation only.   The current  consensus  in  EPA
favors  use of malignant tumors only.   EPA's  Office of  Toxic
Substances is  currently  using the unit  risk  based on malignant
tumors only  in  assessing the  risk  to garment workers  and  home
residents.    The  consideration  of  benign  tumors would increase
the formaldehyde  risk  presented in  this  report by a  factor of
15.

     Formaldehyde   exhaust   emissions   from   motor   vehicles
correlate  well  with  exhaust hydrocarbon  (HC)  emissions.   For
this  analysis,  formaldehyde  emissions  were  expressed  as  a
weight percentage  of exhaust  HC.   These percentages  were  then
applied  to the  exhaust  HC  output  from the MOBILES  emissions
model  for  1986  and  1995 to obtain  the  formaldehyde  emission
factors.   In  this  way,  deterioration  and  other effects  are
included.  The percentages generally vary from 1  to  4 percent,
depending  on  the vehicle  class.    Mobile source formaldehyde
emissions  in  1986 were  estimated  to  be roughly  71,000  metric
tons, or roughly 28  percent  of the total formaldehyde emissions
in the U.S.   Mobile  source  formaldehyde emissions are expected
to drop  to  41,000  metric tons  in  1995.   This is  due to  the
increasing   use   of    3-way   and   3-way    plus    oxidation
catalyst-equipped  gasoline-fueled   vehicles together  with  the
phase out  of  non-catalyst-equipped  vehicles.   The result is  a
marked   decrease   in   projected   HC   and,   by   association,
formaldehyde emissions.

     Nationwide exposure levels, using the  modified CO NEM,  are
roughly  1.04-1.13  ug/m3  and  0.59-0.65  ug/m3  for  1986   and
1995, respectively.   The resulting  risk is 46-50 in 1986  and
29-31  in  1995.    The range  accounts  for both the presence  and
                               XI

-------
absence of an  Inspection/Maintenance  program.   This  attempts to
account  for  direct  emissions  only  and  does  not  account  for
either   the   destruction   or   photochemical   formation   of
formaldehyde in the atmosphere.

     Another  approach intended  to  include  photochemistry  was
also  developed.   With this  approach,  a mobile  source  fraction
was  applied  to  an  annual  average  formaldehyde  concentration
developed  by  OAQPS  using  available  ambient  monitoring  data.
Mobile  sources  account  for  28  percent  of   the   total  VOC
emissions and  30  percent of  the  formaldehyde  emitted directly.
Assuming  that  the  VOC  from  all  sources have the  equivalent
potential to  form  formaldehyde,   a  mobile  source fraction  of
0.30  was  selected.   This  fraction   was  applied  to  an  urban
population  weighted  average  of   12.71  ug/mj   (based  on  data
obtained  in  4  cities)   and  a  rural  concentration  of  1.50
ug/m1.  Since  the summer concentrations  used to  calculate  the
urban  concentration  probably  represent  maximum  rather  than
average values, the  risk estimates  can  be used to  represent a
plausible upper limit.  Using this approach,  the  annual  cancer
risk  from mobile  source formaldehyde  is  131  in 1986  and 77 in
1995.  Combining both  approaches,  the cancer  incidences  due to
mobile sources range from 46-131 in 1986 and 29-77 in 1995.

     Research  activity is planned or  underway  in three  areas:
1) emissions characterization,  2)  photochemistry,  specifically,
factors affecting formaldehyde  formation  in the atmosphere,  and
3) ambient monitoring.

     Benzene  - Benzene  is   an aromatic  hydrocarbon  with  the
formula   C6H6.    It   is   present    in   both   exhaust   and
evaporative  emissions.   Several  epidemiologic   studies  have
associated  benzene  with  an  increased  incidence  of  leukemia.
The  upper confidence limit  unit  risk estimate  determined from
these studies is 8.0 x 10"6.

     Mobile  sources   (including  refueling  emissions)  dominate
the  nationwide benzene  emission  inventory.    In  1982,  mobile
source benzene emissions were  roughly 250,000 metric  tons,  or
85  percent   of  the  total benzene  emissions.   Of  the  mobile
source contribution,  70  percent  comes from exhaust,  14 percent
from  evaporative  emissions   and  1  percent from motor  vehicle
refueling.

     Nationwide exposure  levels  from  exhaust   and  evaporative
emissions were estimated using  the modified  CO NEM.   Benzene
emissions  were   expressed    as   percentages    of  exhaust  and
evaporative  HC.   Percentages  of  exhaust  HC  vary  from  1.1-5.12
percent for  the vehicle  classes;  for evaporative emissions,  the
percentage varies  from 0.35-1.53  percent.  Nationwide exposure
levels from  both  exhaust and  evaporative  emissions  are roughly
3.1-3.2   ug/m3  and   1.7-1.8  ug/mj   for   1985   and   1995,
                              XII

-------
respectively.   (A previous  analysis was  relied on  which used
1985  as  the base year.  It was  assumed  in this report that the
calendar  year  emission factors  for  1985 and  1986  would not
differ   significantly.)   This  assumed   the  presence   of   a
standard,   minimum   I/M  program.    The  range   is   due  to
consideration   of  both  a  low  and  high  range  evaporative
emissions   estimate   for   light-duty  gasoline-fueled  vehicles.
Annual cancer  incidences  from exhaust  and evaporative emissions
are  estimated  to  be  84-89  in  1985  and  50-52  in  1995.   The
reason for  this marked  decrease  is  the decrease in projected HC
in 1995 and, thus, benzene emissions.

     Exposure  to benzene during  refueling includes self-service
refueling,  occupational exposure  (service station  attendants)
and   community  exposure   in  an  urban   area.    Exposure  to
self-service refueling  and occupational  exposure was determined
by  measuring  benzene  levels  in the  region  of the  face  of  a
person  refueling a  vehicle tank.   The  exposure  in  a  typical
urban  area  was  estimated using  a  dispersion model.   Annual
cancer incidences from benzene refueling are estimated  to  be  8
in  1985  and  7 in 1995.   Total  cancer  incidences  from benzene
exhaust, evaporative  and refueling  emissions  are 92-97  in 1985
and 57-59 in 1995.

     An   alternative   approach,    similar   to  that   used  for
formaldehyde, was also  developed.  With  this  approach,  a mobile
source  fraction  was   applied  to  estimated  urban  and  rural
concentrations  developed  using  available  ambient  monitoring
data.   Mobile  sources  account  for  85  percent  of   the  total
benzene emissions.   Therefore, a fraction of 0.85 was  applied
to  an urban population weighted average  of  10.24 ug/m1  and  a
rural  concentration  of  7.52  ug/m3.  Using this approach,  the
annual  lung cancer  risk from mobile  source  benzene  is  223  in
1985.  Based on  the NEM  modeling,  emissions  of  benzene  from
mobile  sources  are  projected to  decrease  roughly  40  percent
from  1985  to   1995.    Accounting  for  this  decrease  and  the
projected population increase,  the  annual lung cancer  risk  is
145  in  1995.    Combining  both  approaches,   the  lung  cancer
incidences  due  to  mobile sources range  from  92-223  in 1985 and
57-145 in 1995.

     The  California  Air   Resources  Board  (CARB)   has   also
attempted to determine  the risk  posed by benzene emissions from
various sources  in  1984 and  2000.   The  vehicular  contribution
is  estimated to be  21-166 cancer  cases in California  in 1984
and 20-154 cancer cases in 2000.  Expressed as  individual  risk,
the vehicular  contribution is 8.1-64.3x10"7  cancers/person  in
1984  and  6.4-49.0x10"7  cancers/person in  2000.   In  this  case,
the range  is due to  a range  of unit  risks used by  CARB.   The
range of risk estimates given  in this report, when  expressed  as
individual  risk,  are  3.8-9.3x10"7  cancers/year  in  1986  and
2.2-5.6xlO"7 cancers/year  in  1995.   The  upper  bound of  these
                              Xlll

-------
risk estimates  are  roughly  equivalent  to the lower bound of the
GARB  estimates  because,   for  these   estimates,   the  ambient
concentrations  and  unit  risks   are  roughly  equivalent.   The
upper bound CARB estimates use an upper bound unit  risk  that is
roughly  7  times higher than the EPA-CAG unit risk  used  in this
report.

     The California  Air Resources Board  (CARB)  is  considering
implementing  regulations  requiring  control  of  motor  vehicle
benzene emissions.  OAQPS has designated benzene as a hazardous
air  pollutant  under  Section  112 of  the Clean  Air Act  and is
implementing necessary controls for stationary sources.   QMS is
including  benzene  impacts  in  its  assessment  of  VOC  control
options.

     Gasoline  Vapors  -  Totally  vaporized  gasoline  has  been
found  to cause a  statistically  significant increase  in kidney
tumors in male  rats  and liver tumors  in female mice.  An upper
confidence limit based on the rat data is 1.18 x  10"s.

     Exposure to gasoline vapors  during refueling  was estimated
based  on  an American Petroleum  Institute  (API)  study  that
involved measuring  gasoline and vapor  levels  in the  region of
the face of a person refueling a vehicle tank.  The exposure in
a  typical  urban  area  for  these refueling  emissions was  also
estimated using the  Industrial  Source Complex dispersion model
to calculate  annual  concentrations.   Based  on these exposures,
the risk from gasoline vapors (excluding benzene)  was estimated
as 65 lung cancer incidences per year.

     EPA has  not made a decision  on what  controls  should be
proposed for gasoline vapors.

     Gas  Phase  Organics  - Gas  phase  organics,   or  volatile
organic  compounds  (VOC),  are  present  in  both   exhaust  and
evaporative emissions.  Over 300 VOC have been identified.   The
majority   of   VOC    consist   of  unsaturated   and   saturated
hydrocarbons  along  with  benzene,   alkyl   benzenes,  aliphatic
aldehydes  and   a  variety of  polycyclic aromatic  hydrocarbons.
Most of  the  known mutagenicity  of  motor  vehicle   emissions is
associated with the particulate phase,  however.

     Of  all  the VOC emitted from motor  vehicles,  only benzene,
formaldehyde,    benzo(a)pyrene    (B(a)P),    ethylene,     and
1,3-butadiene have  unit  risks.   Gas phase  B(a)P was considered
with particle-associated  B(a)P  since  the majority of B(a)P is
in the  particulate  phase.   Ethylene  emissions  are  present in
vehicle  exhaust and  constitute  from   6  to  13  percent   of  the
exhaust  HC  emissions.   Exposures  were  estimated  using  the
modified NEM model.   Based on the  upper confidence  limit  unit
risk  provided  in  the  Six  Month  Study   (2.7  x  10"6),  risk
estimates of lung cancer  incidence  for  1986 and 1995  are 55-60
and  29-31,  respectively.    The range  accounts  for  the presence
and  absence  of an  I/M  program.   The  unit risk   is  extremely
                              xiv

-------
tentative,  however,  since there is no available direct evidence
that  ethylene is carcinogenic.   The unit risk  for  ethylene was
estimated  based on  assumptions regarding  its  potency relative
to  ethylene  oxide,   a  metabolite  of  ethylene and  an  animal
carcinogen.   For this reason,  a lower  risk  estimate  of zero is
used.

      1,3-Butadiene   is   a  photochemically   reactive  compound
present  in vehicle exhaust.  The  Six Month  Study  found no risk
associated with  emissions  of  1,3-butadiene  from mobile sources;
however, since  the  release of the Six Month Study the unit risk
has  increased by a  factor of  1000.   Updated  emission factors
for   1,3-butadiene also  appear higher  but  determination  of an
accurate emission factor  is difficult  because 1,3-butadiene and
n-butane coelute and thus  have the  same  retention  point  on the
gas  chromatograph.    Emission  characterization  studies to  date
have  not  attempted to determine the percentage  of  the peak due
to  1,3-butadiene.   Therefore,  assumptions  must be  made  about
the  percentage  each  compound contributes to  this  peak.   It was
assumed  in  this report  that 15 percent of  the  peak  was  due to
1,3-butadiene,  based on  data collected  in  New York's  Lincoln
Tunnel,  although   ambient  data   indicate   that   the   actual
percentage  could  well  be  much  lower.   Fifteen  percent  was
chosen   as   an  upper   limit.    Based   on   data   from  in-use
gasoline-fueled  vehicles,  1,3-butadiene  is  roughly 0.94 percent
of  the total  exhaust HC  as  measured  by  the  flame  ionization
detector (FID).   Due to  the lack  of data for the  other vehicle
classes,  this  percentage   was  simply  applied  to  the  MOBILES
composite  exhaust  HC emission  factor.   It was  further assumed
that the percentage would remain the same from 1986 to 1995.

     The   modified   NEM   was   used  to   estimate   exposures.
Nationwide urban exposure  in 1986 is estimated  to  be 0.60-0.66
ug/ms.  The  range  accounts for the  presence and absence of  an
I/M program.  These  exposure  estimates  are  for  direct emissions
of   1,3-butadiene   and   do  not  account   for  reactions   of
1,3-butadiene in the atmosphere.  Available  ambient  monitoring
data  were  reviewed  and  compared  to  the  exposure  estimates.
Average  mean values  in  urban  settings  range   from  0.24-24.23
ug/ms,  although the  accuracy  of  the  analytical  methods  used
is uncertain.   The NEM urban exposure estimate  lies within this
range.

     Using the exposure estimates  in conjunction with the  upper
confidence  limit unit  risk  estimate  (2.8xlO~4),   estimates  of
lung  cancer  incidence  for  1986  and  1995  are   593-656   and
391-460,  respectively.    Preliminary emission  characterization
results indicate the  presence of  1,3-butadiene, but  the  amount
has not  yet  been quantified.   Therefore, a  lower  risk estimate
of  zero  will also  be used.    The  resulting ranges  of  cancer
incidences for  1986  and 1995 given  in Table S-l  are 0-656 and
0-460,  respectively.   OAQPS  is currently working  on  a  source
assessment document for  1,3-butadiene which  should be completed
sometime in late 1987.

                               xv

-------
     The  atmospheric  photochemical  reaction products  of  mobile
source VOC  are largely unknown.  Smog  chamber  experiments  have
produced  unidentified  gas phase mutagens  starting with  simple
VOC  compounds  that  are  present  in  vehicle  emissions.    It
appears that,  as  non-catalyst-equipped  vehicles are phased out
of the  fleet,  the reactivity of vehicle  exhaust  will  decrease.
This  is  evidenced  by the  decreasing  percentages  of  reactive
olefins and aromatics, coupled  with the  increasing  percentage
of  less  reactive  paraffins   (particularly  methane)   in   the
exhaust.   The  Integrated  Air   Cancer  Project   (IACP)   is  a
long-term effort  by EPA ORD to  identify the principal  airborne
carcinogens  and their sources,  and  may clarify the risk  posed
by reaction products.

     Organics   Associated   with   Non-Diesel    Particulate
Gasoline-fueled vehicles  emit far  less particulate than  their
diesel  counterparts.    It   is   thought   that   a  number   of
nitro-polycyclic  aromatic hydrocarbon  (PAH)  compounds  account
for much  of the  mutagenicity of  diesel particulate  emissions.
Particulate  emissions  from  gasoline-fueled   vehicles  contain
significantly   less   of    these   nitro-PAH's;   however,   the
mutagenicity  of   the  gasoline soluble  organic fraction  (SOF),
expressed as revertants/ug SOF,  is  greater  than  diesel  SOF.
Also,  unlike diesel SOF,  the  mutagenic  activity of gasoline SOF
increases  with   the  addition  of   S9   activation,   indicating
indirect-acting  activity.  This  suggests  that  the  classical
PAH's may be responsible  for  the  mutagenicity of gasoline  SOF,
rather than the nitro-PAH's.

     The  organics  associated with  gasoline  particulate  were
considered   to  represent  gasoline   products  of   incomplete
combustion  (PIC).  Three different  approaches were  taken  to
estimate  the  risk  from  gasoline  PIC.    The  first   approach
assumes the  risk  from B(a)P emissions adequately represents the
risk  of   all  gasoline PIC.   B(a)P  is  emitted  primarily  from
gasoline-fueled vehicles.   The annual cancer  risk of B(a)P from
gasoline-fueled vehicles was  determined by adjusting  the  upper
confidence  limit  B(a)P  risk obtained in the Six Month Study to
account for  the difference  in emission factors.  The  resulting
annual cancer  risk in 1986 is  0.007 per urban million,  or 1.3
cancer  incidences,   assuming  an  urban   population  of   180
million.   B(a)P emissions should  decrease  with  the phase-out of
leaded fuel.  In 1995, the cancer risk  is  projected  to decrease
to 0.004 per urban million,  or  0.78 cancer incidences, assuming
an urban population of 195 million.

     The  second  approach  uses  B(a)P  emission  factors  from
gasoline-fueled vehicles  together with  the PIC  upper confidence
limit unit risk used in the Six Month Study (which  is  expressed
per unit  of exposure  of  B(a)P)  to  estimate  the  annual  cancer
risk  of  PIC  from  gasoline-fueled  vehicles.    This   approach

                              xvi

-------
assumes  that  B(a)P   is  an  adequate  surrogate  for  all  PIC
compounds.   It  should be  noted that  this approach  is  rather
uncertain  as  the mix  of  PIC  compounds  differs among  source
types.   The annual cancer  risk  was  determined by adjusting the
PIC  risk obtained in  the  Six  Month Study  to account  for  the
difference  in  B(a)P  emission  factors.   The  resulting  annual
cancer risk is 0.68 per urban million,  or 122 cancer incidences
in  1986.   In 1995, the cancer  risk  is  projected  to  decrease to
0.37 per urban million, or  72 cancer incidences.

     The  third   approach  uses   estimated  emission  rates  of
gasoline   particle-associated   organics    (as  an  unspeciated
mixture) together with an upper confidence limit  unit  risk for
these  organics.    Exposures were  estimated using the  modified
NEM model.

     Estimated    composite   emission   factors   for   gasoline
particle-associated organics,  with and without an I/M program,
are  0.0075-0.0082 g/mile  in 1986  and  0.0048-0.0058 g/mile in
1995.  Total organic  emissions  are not projected to decrease as
much  as  B(a)P   emissions.   This   is   because  emission  data
indicate that  use  of a  catalyst  reduces B(a)P  emissions  to  a
greater extent than total organic emissions.

     A  unit  risk  estimate  for  gasoline  particle-associated
organics  has  been  estimated,  based  on  data   from  the  only
catalyst-equipped   vehicle  tested    for   particle    organic
mutagenicity.   The   vehicle  had  exceptionally   high  exhaust
emissions,  comparable to  those  from  a   non-catalyst  equipped
vehicle.   It  was originally chosen for testing in 1979  on this
basis  since  it  was  easier   to  collect  enough  extractable
organics   for    analysis.    The   mutagenic  activity  of   the
particle-associated organics from this  vehicle, as indicated by
the Ames Salmonella strain  TA-98 bioassay, is on  the low end of
the   range,   when   compared   with   other   catalyst-equipped
vehicles.  As  a  result,  the vehicle should be considered to be
of  uncertain  representativeness.   An  upper  confidence  limit
unit risk estimate based on this vehicle is 2.5xlO~4.

     The  total  risk  in  1986,  accounting for both urban  and
rural  exposure,   is   163-176  cancer  incidences  and  drops  to
115-136 cancer incidences in 1995.

     For this  report, a  range  of  risk estimates for  gasoline
PIC  will be  reported, which  encompasses the results  of  all
three approaches.   The resulting range of cancer incidences is
1.3-176 in 1986 and 0.78-136 in 1995.

     Dioxins - The major dioxin  compound  of  interest and  the
one  considered  in  this  report   is  2,3,7,8-tetrachlorodibenzo-
p-dioxin.  This dioxin compound exists  in  the  particulate state
                              xvn

-------
or  is  adsorbed onto particulates.   Some qualitative analytical
measurements  have  found  dioxin  to  be  present  in  the  muffler
scrapings   of  vehicles   using   either   leaded  or   unleaded
gasoline.   It  appears  to  be  emitted in only trace  quantities
(e.g.,  10~9  g/mile) in  vehicle  exhaust.   OAQPS plans  to  make
a decision within  the  coming  year on whether to  list  dioxin as
a hazardous air pollutant.

     Asbestos  -  Asbestos  is used  in  brake  linings,  clutch
facings  and  automatic  transmissions.  About  22  percent of  the
total  asbestos used  in  the  U.S.  in  1984  was  used  in  motor
vehicles.   Asbestos emissions  from  vehicles  with  front  disc
brakes and rear drum brakes  ranged from 4-28 ug/mile.   Based on
these emission rates,  maximum annual  average  asbestos  levels in
urban  areas  due to motor  vehicles are estimated to range from
0.25  to  1.75  nanograms   per  cubic  meter  (ng/m3)-   Asbestos
from mobile  sources appears to  account  for roughly 1-10 percent
of urban asbestos  concentrations,  although mobile sources  could
be  responsible for  as  much  as  70  percent,  under worst  case
emissions conditions.

     EPA has not  attempted  to  quantify  the  excess  mortality
from asbestos  exposure.   The  National Academy of Sciences  (NAS)
has, however,  estimated  life time risks  for  persons  in  urban
areas.   Based on  the  data in  the  NAS  report,   the  individual
annual cancer  risk from urban levels of asbestos  is  estimated
to  range  from 9   x  10 "9  - 3.6 x  10"7 per  ng/m3  exposure.
Assuming  an urban population of  180  million,  mobile  source
asbestos  emissions could  be responsible for  as many as  113
cancer incidences per  year.

     EPA's Office of Pesticides and  Toxic  Substances (OPTS)  has
proposed  regulations  under  Section  6  of TSCA  to  ban certain
uses  of  asbestos  and  to  allocate permits to  mine and import
asbestos which would  restrict its remaining uses.   EPA is also
considering  a  ban  on asbestos friction  products about  5  years
after the final rules  are promulgated.   The fact that there are
not  yet  available good   substitutes  to  replace  asbestos  in
certain  automobile and  truck brakes,  however,  may  push  back
EPA's goal for banning and phasing out asbestos for such uses.

     Vehicle Interior  Emissions  - A total  of  147  compounds have
been identified in vehicle interiors.   In closed vehicles under
high temperatures,  vinyl chloride  was present  at  levels ranging
from  below   2  ppb  to  7  ppb.   A  number  of other  carcinogenic
compounds  were  identified  qualitatively.   Due  to   the  low
exposure  level, no significant   risk  should be present  from
vehicle interior emissions.

     Ethylene  Dibromide  (EDB) and Cadmium  -  Updated emission
factors for  EDB and cadmium  were estimated.   The risk estimates
for EDB  and  cadmium were  then determined by adjusting the risk
obtained in  the Six Month Study to  account  for  the difference
in emission  factors.  As  seen in Tables S-l and  S-2,  the  risks
from  these   pollutants  are negligible   in  comparison  to  other
mobile source pollutants.

                              xviii

-------
1.0  INTRODUCTION

     Considerable  effort is  underway within  the Environmental
Protection Agency  (EPA)  to  determine the magnitude  of  the air
toxics problem  in  the United States.  For  the purposes of this
report,  air  toxics  are  defined as  carcinogens  associated with
air   pollution.    The   overall   objective   is  to   develop
quantitative   estimates   of   the   upper-bound   cancer   risk
associated  with air  toxics  emissions,  and  to   determine  the
relative contribution of specific pollutants and  sources.

     In  1985, EPA  completed  a report which is commonly referred
to  as  the  Six Month  Study.   This  study contains estimated
cancer risks  due to  15  to  45 toxic  air  pollutants   (the number
of pollutants examined  varied with the three different analyses
that  were  performed).[1]*   The estimates  of  cancer incidence
from  these analyses  ranged   from  1300  to  1700  cases  annually
nationwide  for  all  pollutants  combined.  These  estimates were
based  on use of upper-bound  unit  risks.  The  analyses further
indicated  that  mobile  sources  may  be  responsible for  a  large
portion  of  the aggregate  cancer  incidence.    The  Six  Month
Study, however,  was broad in  nature with the goal to  obtain a
quick  assessment of  the  air  toxics problem in the United States
and to guide further studies.

     The  purpose  of  this  report  is to  focus  on   air  toxics
emissions   from   mobile   sources.    Specific   pollutants   or
pollutant  categories  which   will  be  discussed  include  diesel
particulate, formaldehyde, benzene,  gasoline vapors,  gas  phase
organics,  organics   associated  with  non-diesel  particulate,
dioxins, asbestos,  vehicle interior  emissions,  and metals.  For
each pollutant  or  pollutant  category, available  information is
given  on  the   formation  and  composition,  control  technology,
emissions,  ambient  concentrations,  unit  risk  estimate,  the
current  and  projected  public  health  impact  (or   risk),  and
current EPA regulatory and/or research activity.

     The results obtained  are then  compared  to the  earlier EPA
report.[1]   Following  this  is a summary of  the total risk from
mobile   source   air   toxics  emissions,   together   with  the
limitations inherent in the estimate.
     Numbers in brackets  designate  references  at the end of the
     report.
                              -1-

-------
 2.0  DIESEL PARTICULATE

     It should be noted that, since there  has  not been a recent
 analysis  by  EPA  of  diesel  particulate  emission  factors  and
 exposure, a detailed analysis was  performed  during preparation
 of this report.  Other pollutants are not treated as thoroughly.

 2.1  Formation and Composition

     Particulate  emissions  from  diesel-equipped  vehicles  are
 formed as a  result  of  incomplete  combustion of  the fuel.   The
 particles  in  diesel   exhaust   differ  both  in  quantity  and
 composition   from   particles  in   gasoline  engine   exhaust.
 Dies el-equipped  vehicles  can emit  from 30  to  100  times  more
 particulate   mass,   on   a  grams   per   mile   basis,   than
 gasoline-powered,   catalyst-equipped   vehicles.[2]    Over   90
 percent of diesel  particulates  are  less  than 1 micron in  size
 and are therefore  small  enough  to be inhaled and deposited deep
 within the lungs.[3]  Virtually all are less than 10 microns.

     Gasoline  particulate   emissions   from   catalyst-equipped
 vehicles using  unleaded fuel are  very low in mass and consist
 largely of sulfates  and  also  some  organics.   Diesel particulate
 emissions, on  the other hand, are  very complex,  being composed
 of carbonaceous  matter  with condensed and/or absorbed  fuel  and
 lubricant components and  other varied combustion products.   The
 soluble organic  fraction  (SOF) of  the  total  diesel particulate
 mass  (i.e.,  organics   extractable  with  methylene  chloride)
 varies  significantly.    Typically,  diesel  passenger  car  SOF
 ranges  between  5   and  50  percent  of the  total  particulate
 mass.[4]

     The  chemical   composition   of  diesel  SOF   is   complex.
 Generally,  diesel   SOF   ranges   from  Ci4  to  about  C40.[4]
 Hundreds of compounds are  present  and the analytical capability
 does not exist to  identify every compound.  Instead, effort has
 been  focused  to  identify  the  chemical  classes  and  specific
 compounds  associated with the SOF that  are  mutagenic  in  the
 Ames bioassay test.

     The mutagenicity of diesel SOF decreases upon  the .addition
 of S9 activation,  indicating  direct-acting frameshift mutagenic
 activity.   The SOF  was solvent-partitioned into  organic  acids,
 bases,  and  neutral components;   the neutral  components  were
 further  fractionated  and  the   mutagenic  activity   of  each
 fraction   was    determined   using    the   Ames    Salmonella
 typhimurium/microsome  assay.   The  moderately  and  highly polar
neutral compounds  account  for  89-94 percent  of the  mutagenic
 activity   and   only  32   percent   of   the   mass. [5]     Gas
 chromatography/mass    spectroscopy   identified    nonmutagenic
 fluorenones and methylated  fluorenones  as  major  constituents of
                              -2-

-------
the  moderately  and highly polar neutral fractions but  they did
not  account  for the direct-acting  activity observed.   Studies
with nitroreductase-deficient strains of Salmonella  typhimurium
suggested  that  nitrated  compounds  are  present.    Subsequent
studies    show    that    nitropyrenes,    dinitropyrenes    and
nitrohydroxypyrenes   together   account   for    much   of   the
mutagenicity observed.[5, 6]

     EPA  conducted a  large   research  program  to evaluate  the
health  effects  associated with  exposure  to  diesel  emissions,
with   particular   emphasis   on   the  organic   extracts.    The
culmination of  this effort was  an estimation of the  unit risk*
for diesel particulate.[7]

2.2  Control Technology

     A   variety   of   approaches    are   being   developed   by
manufacturers  to   control diesel  particulate  emissions.   One
approach  is  the use  of   a  catalytic converter  to  oxidize  ar.d
remove  the  soluble organics   absorbed on  the particulate.   T .e
other,  more popular approach  is  the use of a particulate filter
or  trap.   Reduction  of   particulate formation  via  combustion
chamber modifications  is also being investigated.   The primary
traps  being evaluated  are:   1)  a  catalyzed  ceramic  monolith
trap, 2)  a  ceramic  monolith  trap,  and  3)  a catalyst-coated wire
mesh trap.  These traps are being evaluated  for  both light-duty
and heavy-duty applications.

     For  light-duty applications,   Daimler-Benz  introduced  in
the  1985  model  year a  catalyzed  ceramic  monolith trap  in some
of  its  turbocharged diesels.   Most  manufacturers,  however,  are
investigating   the  feasibility   of   non-catalyzed,   ceramic
monolith traps.

     A  trap  must be periodically  regenerated by  oxidizing the
collected particulates.   Otherwise,  particulates  collected on a
trap  can  cause  the  exhaust back pressure  to  increase  and
adversely affect fuel  economy and vehicle  performance.   Active
and  passive  regeneration techniques  have  been assessed.   An
active  technique would  use a diesel-fueled burner  or  electric
resistance  heater   to   raise  the  temperature  of  the  engine
*Unit risk is defined as the individual  life  time  excess cancer
risk  from  continuous  exposure  to  1  ug  carcinogen  per  m3
inhaled air.   Assuming a life time is 70 years,  the excess lung
cancer risk  in 1  year  is derived  by  simply dividing  the unit
risk by 70.   Using this approach,  latency is ignored.
                              -3-

-------
 exhaust  gas  flowing into the  trap or to  raise  the temperature
 of  the trap itself to the  ignition temperature of the collected
 particulates,  thus combusting  the particulate which regenerates
 the  trap.   It  should  be noted that  the  organic emissions from
 trap regeneration  have not  been fully characterized.

     A passive technique uses  catalytic  material to  lower  the
 ignition temperature of the collected  particulate.   One method
 is  to  apply   the  catalytic  coating  to  the  trap  substrate
 itself.   Daimler-Benz   introduced  in  the  1985 model  year  a
 catalyzed  ceramic monolith trap  in  some  of its  turbocharged
 diesels.   Johnson-Matthey,  a catalyst manufacturer  involved in
 trap   development,   is    investigating   the   merits   of   a
 catalyst-coated stainless steel wire mesh trap.

     The other  passive  method  is  the  use  of  a metal  fuel
 additive in  conjunction  with  a ceramic  wall  flow  monolith
 trap.  This  yields the lowest  ignition temperature of any other
 passive  method.   The  most  promising  metal  fuel additives  to
 date are manganese and copper compounds.

 2.3  Emissions

     In  this section,  the  Federal  diesel particulate standards
 are  presented.   This  is followed by an  estimation of  in-use
 emission factors   by model  year  for  each vehicle  class  (i.e.,
 light-duty   car,   light-duty  truck   and   heavy-duty  vehicle).
 These  model  year   emission  factors are  then  used in conjunction
 with  information   on diesel  sales fractions and  fraction  of
 diesel  travel   by  model  year  to  obtain  calendar  year  emission
 factors  for  each  vehicle   class.  For  this  analysis,  calendar
 years   1986   and    1995   were   selected.    Nationwide   diesel
 particulate  emissions  (metric  tons/year)  for 1986 and  1995  are
 then   calculated   by   combining   the  emission   factors   with
 projected vehicle  miles  traveled  (VMT)  data for  1986  and  1995.
 The  results  are   then  compared  to  the  most  recent  national
 particulate emission estimate.

 2.3.1  Emission Standards

     Table  2-1  provides   a   summary  of  the  Federal   diesel
 particulate  emission standards  for  light-duty cars  and  trucks
 and  heavy-duty  engines.   The  test   procedure  specified  for
 light-duty vehicles  is the CVS-75 or FTP procedure,  a constant
volume  sample  test  which   includes both  cold  and  hot  starts.
For  heavy-duty  vehicles,  a  transient   test   procedure   is
 currently used.   This  is an engine  dynamometer  procedure  with
 starts,  stops,  and  speed/load changes.    In order to meet  the
 future Federal  standards  (1987  and later),  some  light-duty  and
heavy-duty vehicles  will need  to  be  equipped with a particulate
aftertreatment system,  such as  a trap.
                               -4-

-------
                           Table  2-1

                  Diesel Particulate Standards

                       Light-Duty  Vehicles

                                     Standard
                                  (grams per mile)
           Year                   Cars      Trucks

           1981 and prior
           1982-1986              0.60       0.60
           1987 and later         0.20*      0.26*
                       Heavy-Duty Engines
                                      Standard
           Year             (grams per brake horsepower-hour)

           1987 and prior
           1988-1990                  0.60
           1991-1993                  0.25**
                                      0.10***
           1994 and later             0.10**
*   Emissions averaging may be used to meet this standard.
    For trucks, some restrictions apply.

**  Emissions averaging may be used to meet this standard.  Some
    restrictions apply.

*** For urban bus engines, the standard is 0.10 g/Bhp-hr
    beginning in 1991.   Particulate averaging is not allowed
    with this standard.

NOTE:   In  an  emissions  averaging  program,  the  manufacturer
determines  emission  limits  for  each  vehicle/engine  family.
Family  emission limits  are allowed  to  exceed the  standards;
however,  the weighted  average  of the  family  emission  limits
myst be  in compliance with  the  applicable standard.   For  cars
and  trucks,  the   average   emission   level  is  based  upon  a
production-weighted average of the family emission limits.   For
heavy-duty engines, the  average  emission level is determined by
calculating  a  production-  and  horsepower-weighted average  of
the family emission limits.
                               -5-

-------
 2.3.2  Emission  Factors by Model Year

     The  light-duty and heavy-duty  diesel  particulate emission
 factors  are given  in  Table  2-2.   The heavy-duty sub-groups are
 defined by  gross vehicle weight rating (GVWR) as follows:

                 Class 2B = 8,500  to  10,000  Ibs.
               Classes 3-5 = 10,001 to 19,500 Ibs.
                Classes 6 = 19,501  to 26,000 Ibs.
              Classes  7 and  8 = 26,001 Ibs  and up.


     The  emission  factors for  light-duty  cars  and  trucks were
 obtained   from   an  EPA   report   titled,   "Diesel  Particulate
 Study."[8]   For  1987  and  later,  these emission   factors  are
 considered   to   be,   for  the  purpose  of  this  report,  the
 respective  emission  standards.   For  heavy-duty vehicles,  the
 emission  factor assumed  for 1983 and prior is  1.0  grams  per
 brake  horsepower-hour  (g/Bhp-hr).   For  1984-1987,  the emission
 factor  assumed   is  0.75  g/Bhg-hr.   This is  based  on transient
 test data  for  a variety of  in-use heavy-duty vehicles.[9]  For
 1988-1993,  the  emission  factor assumed is the  standard.   For
 1994  and 1995,  the standard was  adjusted  upward  slightly  to
 account  for projected  trap failure.  Within classes  7-8,  the
 emission  factor  for buses  was  adjusted upward  to  account  for
 projected trap  failure beginning  in  1991.   The equation used to
 make  this  adjustment  is   given   and  discussed  in  detail  in
 reference 8.

     It  was  necessary   to  convert  the   heavy-duty  emission
 factors  from g/Bhp-hr  to  g/mile.    As  seen  in  Table  2-2,  the
 g/mile  emission  factors  for  the heavy-duty subgroups  differ
 even though they are subject to the same g/Bhp-hr  standard  for
 1988 and  later  and  similar  in-use g/Bhp-hr  emissions  prior  to
 1988.  The  conversion of g/Bhp-hr  to  grams/mile  is  dependent  on
 the  engine  brake-specific  fuel consumption  (BSFC,  Ib/Bhp-hr),
 the fuel density (Ib/gallon) and the fuel  economy (mile/gallon)
 of the particular  vehicle/engine  configuration.   Since the BSFC
 and  fuel   economy  varies  among  the  different   heavy-duty
 subgroups,  estimated g/mile  emissions  vary  as well.  Conversion
 factors  for the heavy-duty  subgroups  by model  year  were based
 on the  information provided in reference  10.   These conversion
 factors are consistent with those used in the MOBILE3 emissions
model.[11]

 2.3.3  Emission Factors for Calendar  Years  1986  and 1995

     The  next  step  is  to combine  the   model  year  emission
 factors for  each vehicle  type  into a single, weighted calendar
year emission  factor  for  each vehicle  type.  Model  year data
for the previous 20  years  are  used,  i.e.,   for 1986,  model year
data back  to 1967  are  used;  for  1995, model year  data back  to
 1976 are used.   Each model year's  emission  factor  is multiplied

                              -6-

-------
                            Table 2-2
               Diesel  Particulate Emission Factors

                       Light-Duty Vehicles
                         grams per mile
Model Year

1978 and prior
1979
1980
1981-1986
1987 and later
Model Year

1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
            Cars

            0.70
            0.80
            0.50
            0.27
            0.20

    Heavy-Duty  Engines*

         grams per mile

2B       3-5      6
                   Trucks

                   0.90
                   0.90
                   0.50
                   0.28
                   0.26
                  7-8
1
1
1
1
1
1
1
1
1
1
1
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
.00
.00
.00
.00
.00
.00
.00
.00
.00
.00
.00
.00
.99
.98
.98
.97
.97
.73
.73
.73
.72
.58
.57
.57
.24
.24
.24
.09
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
0
0
0
0
0
0
0
0
0
0
0
0
0
.71
.71
.71
.71
.71
.71
.71
.71
.71
.71
.71
.71
.53
.34
. 16
.97
.97
.73
.73
.73
.72
.58
.57
.57
.24
.24
.24
.09
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
0
0
0
0
.71
.73
.75
.75
.79
.82
.84
.85
.86
.87
.87
.86
.86
.86
.86
.87
.85
.37
.36
.34
.33
.06
.06
.06
.44
.44
.44
.19
2
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
2
2
2
2
1
1
1
0
0
0
0
.99
.02
.06
.09
.14
.20
.24
.29
.33
.24
.28
.23
.18
. 16
.13
.09
.08
.30
.29
.28
.28
.81
.79
.77
.69
.69
.67
.30
0.09
0.09
0.18
0.29
   See text for explanation of the g/Bhp-hr emission  factors
   used.
                               -7-

-------
by  that  model  year's  fraction  of  calendar  year  VMT  and the
diesel  sales  fraction  for  that model  year,  and then summed
across  all 20 model years.

     The  fraction of  travel by  model  year  for  the  light-duty
vehicles  and  trucks  is   given  in  Table  2-3.   These  travel
fractions  are  dependent on  vehicle  age  rather than model year;
therefore,  the same  travel  fractions  were  used  to  determine
both   the  1986  and   1995   emission   factors.   These  travel
fractions were obtained from the MOBILES emissions model.[11]

     The  fraction of travel  by model year  for  the  heavy-duty
subclasses for  calendar years 1986  and 1995 are given in Tables
2-4 and 2-5,  respectively.   For 1986,  the travel  fractions for
heavy-duty gasoline-fueled trucks were used  for  subclasses 2B,
3-5 and 6.   The  travel  fractions  for  heavy-duty  diesels  were
used  for  subclasses  7-8,  since  this  combined  subclass  is
dominated  by   heavy-duty  diesels.   For   1995,  the   travel
fractions  for  heavy-duty  gasoline-fueled  trucks were  used for
subclasses 2B  and 3-5;  travel fractions for  heavy-duty diesels
were  used for  subclasses  6 and 7-8.   Unlike the  heavy-duty
gasoline-fueled trucks, the  mileage  distributions  and resulting
travel  fractions  for heavy-duty  diesels are  dependent  on model
year  and  reflect  the increasing  penetration  of diesels  in the
lower  mileage,  lighter weight  classes  of the  heavy-duty truck
category.  These  travel fractions were  calculated  based  on the
information provided in Reference 11.

     Projecting   future   diesel  sales  is   rather  uncertain.
Light-duty diesel sales  in  particular have  dropped off quite
dramatically  and  should  fall  well  short  of  projections  made
just a  few years  ago that light-duty diesels  would account for
25 percent of  light-duty  vehicle sales  in 1995.  To account for
these uncertainties,  low  and high  diesel  sales  scenarios  were
developed.  Both  scenarios  incorporate  actual sales data to the
extent  available.

     For  light-duty vehicles,  actual diesel  sales  fractions up
to and  including  1985 are available and were used. [12,13]   For
1986 and beyond, the low diesel sales scenario assumes  that the
light-duty vehicle  diesel  sales  fraction will  remain  .constant
at the  1985 level (0.009).  The high diesel sales  scenario  uses
MOBILES  projections  of  diesel  sales  for  1995 (0.009   in  1986
with a gradual increase to 0.115 in  1995).

     For  light-duty  trucks and heavy-duty  subclass 2B,  actual
diesel  sales fractions through  1984  were used.[12,13]   For  1985
and beyond,  the  low diesel- sales  scenario   assumes  that  the
diesel  sales  fraction  will  remain  constant  at the 1984 level.
The  high  diesel  sales   scenario   uses  the   latest   EPA/OMS
projections for 1985-1995.[14]
                              -8-

-------
                           Table 2-3

              Fraction of Travel  by Model  Year  for
                 Light-Duty Vehicles and Trucks

Vehicle               LDV                          LDT
  Age             Travel  Fraction            Travel Fraction

   1                 .038                         .035
   2                 .142                         .129
   3                 .125                         .114
   4                 .111                         .101
   5                 .098                         .088
   6                 .084                         .078
   7                 .075 .                        .067
   8                 .065                         .058
   9                 .055                         .050
  10                 .047 ,                        .043
  11                 .040                         .037
  12                 .032                         .031
  13                 .026                         .026
  14                 .021                         .022
  15                 .015                         .018
  16                 .011                         .015
  17                 .007                         .012
  18                 .003                         .009
  19                 .003                         .006
  20+                .004                         .009
                              -9-

-------
                Table 2-4

   Fraction  of  Travel by  Model Year  for
Heavy-Duty Vehicles in Calendar Year 1986
Model
Year
1986
1985
1984
1983
1982
1981
1980
1979
1978
1977
1976
1975
1974
1973
1972
1971
1970
1969
1968
1967
2B, 3-5
and 6
.000
.227
.175
.134
.105
.080
.062
.049
.037
.028
.023
.017
.013
.010
.009
.006
. 005
.004
.003
.013

7-8
.000
.255
. 186
.136
.105
.078
.057
.043
.033
.026
.019
.014
.011
.008
.006
.005
.004
.003
.002
.009
                  -10-

-------
                Table 2-5

   Fraction  of Travel by  Model Year  for
Heavy-Duty Vehicles in Calendar  Year 1995
Model
Year
1995
1994
1993
1992
1991
1990
1989
1988
1987
1986
1985
1984
1983
1982
1981
1980
1979
1978
1977
1976

2B, 3-5
.000
.227
.175
. 134
.105
. 080
.062
.049
.037
.028
.023
.017
.013
.010
. 009
.006
.005
.004
.003
.013

6
.000
.292
.208
.146
.105
.074
.052
.037
.026
.019
.013
.008
.006
.004
.002
.002
.001
.001
.001
.003

7-8
.000
.272
.199
.144
. 106
.076
.055
.040
.029
.021
.015
.011
.008
.006
.004
.003
.002
.002
.001
.006
                  -11-

-------
     The  diesel  sales  fractions  for heavy-duty  classes  3-5,  6
 and  7-8 were based  on actual data  through  1985.  For  1986  to
 1995,   the  latest  EPA/OMS  projections were  used.[14]   These
 sales  fractions  were  used  for  both the  low  and high  diesel
 sales scenarios.

     The  diesel  sales  fractions  for the light-duty vehicles and
 trucks  are  presented in Table 2-6.  The diesel  sales fractions
 for the heavy-duty vehicle classes are given in Table 2-7.

     The  resulting weighted  emission factors for calendar years
 1986 and  1995 are given below:

                                     (g/mile)
                                               HDV
                        LDV     LPT     2B     3-5   	6_   7-8
Calendar Year 1986
  Low Diesel Sales       .008    .0105   .089   .040   .479   2.35
  High Diesel Sales      .008    .0105   .091   .040   .479   2.35

Calendar Year 1995
  Low Diesel Sales       .003    .008    .061   .082   .330   .844
  High Diesel Sales      .013    .041    .084   .082   .330   .844


2.3.4  Nationwide Diesel Particulate Emissions

     In  this section,  nationwide diesel  particulate  emissions
(metric  tons/year)   for  1986   and   1995  are   calculated  by
combining  the  calendar  year  emission factors  with  projected
vehicle miles traveled (VMT) data for 1986 and 1995.

     The VMT  fractions  and resulting projected VMT for 1986 and
1995 is  given  in Table 2-8.   These data were obtained  from the
MOBILE3  Fuel  Consumption Model. [15]   In  this  model,  heavy-duty
subgroups  2B  and 3-5 were treated as one subgroup.    To  split
the  VMT fraction  for  these subgroups,  data  from EEA's  Motor
Fuel Consumption, Tenth  Periodical  Report  were  used.[16]   Total
VMT is projected to increase 19 percent from 1986 to 1995.

     The projected  nationwide diesel particulate  emissions for
1986 and 1995  are  given in Table 2-9.  As seen in this  table,
nationwide  diesel   particulate  emissions  are   projected  to
decrease roughly 44 to  54  percent  from  1986  to  1995,  despite
the 19  percent  projected  increase  in VMT.  This  is due  to the
rather  small  projected  further  infiltration  of  the   fleet  by
diesels   (with  the   exception  of   the  heavier   heavy-duty
subclasses 7-8)  together with  increasingly  stringent  standards
in future years.
                              -12-

-------
                           Table 2-6

             Low and High Diesel Sales Fractions by
         Model  Year  for  Light-Duty Vehicles  and Trucks
Model
Year
1995
1994
1993
1992
1991
1990
1989
1988
1987
1986
1985
1984
1983
1982
1981
1980
1979
1978
1977
1976
1975
1974
1973
1972
1971
1970 +
Light-Duty Vehicles*

   Low          High
   .009         .115
   .009         .102
   .009         .089
   .009         .076
   .009         .063
   .009         .050
   .009         .044
   .009         .034
   .009         .019
   .009         .009
   .009
   .023
   .019
   .039
   .060
   .045
   .026
   .009
   .003
   .003
   .003
   .003
   .002
   .002
   .001
   .000
Light-Duty Trucks**

 Low          High
 .026         .339
 .026         .300
 .026         .263
 .026         .226
 .026         .188
 .026         .150
 .026         .119
 .026         .088
 .026         .057
 .026         .026
 .026         .026
 .026
 .042
 .092
 .082
 .048
 .015
 .010
 .005
 .003
 .002
 .000
 .000
 .000
 .000
 .000
*1970-1985 sales fractions are based on actual data and are the
 same for both scenarios.   These fractions are not repeated in
 the high sales column in order to highlight the differences
 between scenarios.

*1970-1984 sales fractions are based on actual data and are the
 same for both scenarios.
                              -13-

-------
               Table  2-7

Low and High Diesel Sales Fractions by
  Model Year for Heavy-Duty Vehicles
Model
Year
1995
1994
1993
1992
1991
1990
1989
1988
1987
1986
1985
1984
1983
1982
1981
1980
1979
1978
1977
1976
1975
1974
1973
1972
1971
1970
1969
1968
1967
2B
Low
.190
.190
.190
.190
.190
.190
. 190
. 190
. 190
.190
. 190
. 190
.153
.127
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000

High
.300
.300
.300
.300
.286
.273
.259
.245
.231
.218
.204


















3-5
Best
.300
.300
.300
.300
.291
.282
.273
.265
.256
.247
.239
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
6
Best
.712
.693
.-674
.655
.635
.616
.597
.578
.559
.540
.521
.382
.312
.289
.322
.183
. 114
.078
.070
.042
.032
.016
.016
.016
.015
.016
.000
.000
.000
7-8
Best
.883
.882
.879
.878
.877
.875
.874
.872
.871
.868
.867
.849
.862
.845
.881
.871
.820
.815
.878
.823
.733
.770
.780
.760
.750
.750
.750
.750
.750
                 -14-

-------
                           Table 2-8

            VMT Fractions  and VMT  for  1986 and  1995


                             VMT Fractions
1986
1995
1986
1995

LDV
1098.
1298.
LDV
.675
.671

53
75
LDT
.225
.227
Projected
LDT
365.25 35
438.59 49

2B
.022
.025
VMT (

2B
. 16
. 19
HDV
3-5 6 7-8
.004 .009 .065
.003 .006 .068
9
10 miles)
HDV
3-5 6 7-8
6.11 15.34 106.48
5.68 10.64 131.48
TOTAL
1626.87
1934.33
                              -15-

-------
                           Table 2-9

            Nationwide Diesel Particulate Emissions
            	(metric tons per year)	

                       1986                         1995
             Low Sales    High Sales     Low Sales   High Sales

LDV            8,788         8,788         3,896      16,884
LOT            3,835         3,835         3,509      17,982
HDV
  2B           3,129         3,200         3,001       4,132
  3-5            244           244           466         466
   6           7,348         7,348         3,511       3,511
  7-8        250,228       250,228       110,969     110,969


TOTAL        273,572       273,643       125,352     153,944
                             -16-

-------
2.3.5  Other Sources of Particulate

     To put  the preceding  emission estimates into  perspective
the  results  are  compared  to  national   particulate  emission
estimates for 1984. [17]  1984  is  the  most recent  year for which
data are available.

     Total  particulate emissions   in  the  U.S.   in  1984  were
projected to be  7.0  million  metric tons.   In comparison,  diesel
particulate  emissions  in  1986 were  projected  to  be  roughly
274,000  metric  tons,   or  3.9  percent  of  the   total  1984
emissions.    Transportation  sources in  1984  were  estimated  to
contribute  18.6  percent  of  the  total,  or  1.3  million  metric
tons.  Diesel particulate  emissions in turn  currently appear  to
account   for   about  21   percent   of    the   emissions   from
transportation sources.

2.4  Ambient Concentrations of  Diesel  Particulate

     In this section,  urban  and  rural concentrations  of  diesel
particulate  are  estimated  for  1986 and  1995,  using  a modified
version of  the  EPA NAAQS  Exposure Model   (NEM)  for  CO.   These
concentrations are then compared  to previous exposure estimates
and to monitoring data for  total suspended particulate.

2.4.1  Estimation of Urban and  Rural Exposure

     The model  used provides an  estimate  of nationwide  annual
person-hours  of  exposure  to  any non-reactive  mobile  source
pollutant of interest.[ 18]   It  is  based  upon the NEM developed
originally   by   EPA's   Office   of  Air  Quality  Planning  and
Standards  (OAQPS).   The  NEM  approach  relies  on  an  activity
pattern model that simulates a set of population groups  called
cohorts as they  go  about  their day-to-day activities.  Each  of
these cohorts are  assigned to  a  specific location  type  during
each hour of the day.   Each  of several specific  location types
in the  urban area are  assigned  a particular ambient pollutant
concentration based on  fixed   site monitor  data.    The  model
computes the hourly exposures  for each cohort and  then sums  up
these values over the desired average  time to arrive at average
population   exposure   and  exposure   distributions.    Annual
averages are possible because  a  full  year's data  from fixed
site monitors is an input  to the model.

     The NEM approach was designed to determine an integrated
exposure to  a  pollutant by  estimating  both indoor  and outdoor
exposures to the pollutant.   It  was  not  designed  to determine
exposure  from  a  particular  source,  such   as  mobile  sources.
Also, because  its  basic   time unit  is  an  hour,  it  does  not
account well for  short  periods  spent  in  locations with  high
exposure such as an  on-road  vehicle.   Hence, Southwest Research
Institute  (SwRI),  under EPA contract,  developed  a  new model
based on the NEM for CO which  could be used to better determine

                              -17-

-------
exposures    resulting   specifically    from   mobile    source
pollutants. [18]  The  CO  NEM was used since outdoor CO is almost
exclusively mobile  source  related.   Since the CO  monitor data,
on which the  CO NEM was based, can  be  assumed  to  be related to
mobile  source emission  rates,  exposure  to other  non-reactive
mobile   source   pollutants   can   be    modeled    using   this
relationship.

     The CO  monitor data  are  used to  provide  CO  concentration
data  for  each  neighborhood and most of  the microenvironments.
In each  neighborhood or microenvironment,  CO  emission  factors
(in grams/minute)  are chosen which  are felt to best represent
vehicle conditions  in that  neighborhood/microenvironment.   For
example,  in  an  urban  commercial  neighborhood,   an  emission
factor at  10  mph steady state  is  chosen since this  is  thought
to best  represent  the vehicle  conditions in this  neighborhood.
The emission  factor is  a  fleet  average  emission  factor,  thus
weighting  emissions   from  both  light-duty  and  heavy-duty
vehicles.     The  model  ratios  the   CO   concentrations   and
appropriate      CO      emission      factors       for      each
neighborhood/microenvironment,               so              each
neighborhood/microenvironment   contains  a   ug/m3/grams/minute
factor.  Emission factors  in grams/minute for the  pollutant of
interest  for each  neighborhood/microenvironment  are input  to
the model.    The model  simply multiplies   the  input  emission
factor  (grams/minute)  by   the  factor  (ug/m3/grams/minute)  to
obtain concentrations  in each  neighborhood/microenvironment for
the pollutant of interest.

     It  should  be  noted  that   indoor  concentrations  (and
therefore,  exposure)  due  to ambient mobile source  pollutants
are also accounted  for in  the model.   A  scaling factor  of 0.85
was applied to the appropriate  neighborhood  CO monitor  data to
estimate indoor  exposures  to the pollutant  of  interest  in each
neighborhood.   The  scaling factor was  based on comparisons  of
indoor  and  outdoor  CO  levels  of  homes  with no  indoor   CO
sources (e.g., gas stove, smokers).

     The model  does  not  account  for   photochemical  reactions.
The exposure  levels predicted by the model  are those resulting
from direct  exhaust emissions, and  do not  account  for  either
the destruction  or  photochemical formation  of  the pollutant in
the atmosphere.  The model  also  assumes  that the  pollutant of
interest has  emission formation  and dispersion characteristics
similar to that of CO.

     In the SwRI model,  the relatively insignificant  indoor CO
exposures were  set  to zero.   Exposures in  three  mobile  source
microenvironments   (street   canyons,    tunnels   and   parking
garages),   where  elevated  concentrations  of  mobile   source
pollutants  could  be  experienced,  were  added  to  the  OAQPS
version  of   the model.    Finally,   a   national   extrapolation
procedure designed expressly for mobile sources was devised.

                              -18-

-------
     There  are  three  required   inputs.    The   first   is  the
nationwide  urban   and  rural  populations   for   the  year  of
interest.  The  second  is  a series of 33 emission  factors  which
are  described  in Table 2-10.   The third input is  a  list  of 25
ambient    pollutant   concentrations,    specifying   the    24
concentration   intervals   or   bins   for   which   cumulative
person-hours of exposure are to be calculated.

     The output lists the  total annual  person-hours of  exposure
found  in  each of the  specified concentration intervals.   Using
this information, a mean exposure level may be calculated.

     Urban  and  rural  populations   for   1986   and  1995   were
estimated  based on U.S.   Department  of Commerce  data and are
given below.[19]

                 Urban               Rural            Total
1986          180,000,000          60,000,000      240,000,000
1995          195,000,000          65,000,000      260,000,000

     As  seen  in  Table  2-10,  the  emissions  input  required for
the model are FTP emission factors and an  emission  factor at an
average  speed  of  10  mph,  expressed  in grams/minute.   For the
tunnel microenvironment and  rural  areas,  a 35 mph  steady state
emission factor will  be  used.   For light-duty vehicles, an idle
emission  factor  is  also  required.   As  mentioned  previously,
these  emission  factors  were  chosen  to  best  represent  the
vehicle    operating   conditions     in    each    neighborhood/
microenvironment.

     The weighted emission  factors  given  in Section  2.3.3 are
considered FTP  (or  transient)  emission factors.  VMT  fractions
are  then  used  to  determine  FTP emission  factors  for  each
vehicle  class  and a  composite emission factor  for the fleet.
The LDV/LDT FTP has an average speed of 19.6 mph;  this average
speed was used  for  both  LDV/LDT and  HDV  to  convert  the g/mile
factor to  g/min,  as  required  for  the model.   For  LDV/LDT this
is certainly correct.  For  HDV it is less  clearly  so but  is a
reasonable approximation.

     The VMT  fractions  in  Table  2-8  represent VMT  fractions
nationwide.   Urban and rural VMT  fractions differ,  particularly
for  the  heavy-duty  vehicle   classes   7   and  8.    For  these
heavy-duty subclasses, rural VMT  is  estimated  to  exceed urban
VMT by  a factor of 2.7.[15]   Since  these  heavy-duty subclasses
are  responsible   for  the  majority  of  diesel   particulate
emissions,  urban  and  rural VMT  fractions will  be used.   The
urban and rural VMT fractions  for each vehicle class  are given
below.[15]
                              -19-

-------
                                   Table 2-10
                          Model Emission Factor Inputs

Neighborhood or Microenvironment        Assumed Emission Factor (g/min)
Weekday urban residential
Weekday urban commercial
Weekday urban industrial
Weekday suburban residential
Weekday suburban commercial
Weekday suburban industrial
Weekday street canyon
Weekday tunnel
Weekday parking garage

Dummy mobile source emission factor

Weekday rural
Saturday urban residential
Saturday urban commercial
Saturday urban industrial
Saturday suburban residential
Saturday suburban commercial
Saturday suburban industrial
Saturday street canyon
Saturday tunnel
Saturday parking garage

Dummy mobile source emission factor
Saturday rural
Sunday emission factors
         FTP
       10 mph
       10 mph
         FTP
         FTP
         FTP
       10 mph
       35 mph
0.5 times idle factor plus 0.5
  times 10 mph (LDV only)
Use tunnel factor as dummy (to
  allow for future expansion)
       35 mph
0.72* times weekday FTP factor
0.72 times 10 mph factor
0.72 times 10 mph factor
0.72 times FTP factor
0.72 times FTP factor
0.72 times FTP factor
0.72 times 10 mph factor
       35 mph
0.5 times idle plus 0.5 times
  10 mph (LDV only)
Use tunnel factor as dummy
0.72 times 35 mph factor
In current version of program,
  Sunday emission factors are
  equal to Saturday factors.
*  The factor of 0.72 is used to adjust for relative traffic volume.
                                      -20-

-------
                            VMT Fractions
                                           HDV
               LDV     LOT     2B     3-5      6      7-8
1986
  Urban        .721    .206    .026   .005    .010    .032
  Rural        .617    .247    .016   .003    .008    .109

1995
  Urban        .717    .209    .031   .004    .006    .033
  Rural        .613    .250    .018   .002    .005    .112
     To determine  the  10 mph and  35  mph emissions  factors  and
the  idle  emission  factor  for  the  light-duty diesel  vehicle
fleet,  it  was  necessary to  calculate  ratios  of  emissions  at
these  speeds  to  FTP emissions.    For  the  light-duty  vehicles,
data from the New York City Cycle  (NYCC), with  an  average speed
of  7.07 mph,  were used to represent the  10  mph cyclic emission
factor.    The  results   of   two   studies   were   used   which
collectively  contained  data  at idle, steady  state  speeds of 31
mph, 50 mph,  and  53 mph as well  as FTP  and  NYCC  data. [20,21]
Ratios of  emissions at the steady  state  speeds  (g/min)  to  FTP
emissions   (g/min)  were   plotted   and  found   to   increase
exponentially.  The ratio at 35 mph was estimated from the plot.

     For   heavy-duty  vehicles,    emissions   data   from   three
in-service  buses   over  various   chassis  dynamometer  driving
cycles were used  to calculate the 35 mph steady state ratio for
the  tunnel microenvironment  and   urban  areas.[22]   Emissions
data were  taken at idle,  steady  state speeds of 12.5 mph and 25
mph  and  over  the  heavy-duty  chassis  driving cycle.   To  obtain
the  ratio  at  35 mph,  it was assumed that the ratio  from 25 mph
to 35  mph  would continue to  increase  linearly.  Data  were  not
available  at   higher  speeds  to  determine  whether  the  ratios
would  increase exponentially at   some  point;  therefore,  the
linear assumption  may  result  in  a slight underprediction of the
35 mph ratio.

     Very  little   data  exist  to  calculate  cyclic  emission
factors  at 10  mph  for  heavy-duty vehicles.   Recent  EPA test
programs have  focused  on emissions characterization  of  in-use
transit  buses.    These  programs   involved  buses  which  were
temporarily  removed  from  operating  service  and  which  were
tested  without   additional   maintenance   in  their   chassis
configurations  over   test   cycles  designed  specifically   to
simulate transit  bus operation.    The  test cycles  have average
speeds ranging from 8.8 to 12.4 mph.   Information  regarding the
test programs can  be found  in reference 23.  An overall average
transit bus emission factor was estimated to  be 5.52 g/mile, or
roughly 0.92 g/min,  based  on  an  average speed  of  10  mph.  A 10
mph/FTP ratio  was  estimated  such  that  the  resulting emission
factor would  equal  0.92  g/min,  based  on  the  1986  FTP  data.

                              -21-

-------
 This ratio will be used for both  1986  and  1995 to calculate the
 heavy-duty  contribution   in  the  street  canyon  neighborhood,
 where  public  exposure  to transit  bus emissions  is  relatively
 high.

     For   the   remaining  urban   industrial   and   commercial
 neighborhoods,  a  10 mph  cyclic  emission factor  for  heavy-duty
 trucks  is  required.   For these  neighborhoods,  the  composite
 heavy-duty FTP  g/mile  emission  factor will be  used.   The ratio
 at  10  mph is thus  equal  to the ratio at  this  speed  to the FTP
 average  speed of 19.6  mph.   The resulting ratio  at  10  mph is
 0.51.

     The  resulting  ratios for  the  light-duty  and  heavy-duty
 vehicles are given in Table 2-11.   These  ratios were  applied to
 the FTP  g/min calculated  for  each vehicle class and then summed
 to  obtain  VMT-weighted  10  mph  and  35  mph  g/min  composite
 emission  factors.   The emission factor  inputs  for the 1986 low
 and high  sales  scenarios  are  the same.  The  reason is  that the
 sales fractions used for  the  scenarios only differ for the 1985
 and  1986  model  years.   This  difference   was  found  to  be
 negligible when computing  the model emission factor inputs.

     It  should  be noted  that the  absolute emission factors and
 the  assumptions used  to   estimate  the off-FTP  values  directly
 influence the  estimated  absolute particulate  exposures in  1986
 and 1995, but that the  percent decrease  in particulate exposure
 from  1986 to 1995  is sensitive only  to the more reliable FTP
 emission factors for the  two years.

     The  final  model  input  required  is  a list of 25  ambient
 pollutant  concentrations   defining  the  concentration  intervals
 for which  cumulative person-hours  of  exposure  are  calculated.
 The  chosen concentrations are  given  in  Table 2-11   and range
 from  0.00  to  50.00  ug/m3 for  1986   and  0.00  to 31.47  ug/m3
 for 1995.

     The  modified  NEM does  not  take  into   account  projected
 increases in VMT.  Since  most  of the  modified NEM is  based on
 1981 monitoring  data,  projected VMT  for  1986  and 1995 will be
 compared  to  1981  VMT   and  the  exposures  predicted  by  the
modified NEM  adjusted upward accordingly.  Since  VMT  from  1981
to 1986  is expected  to increase  10 percent, the  1986  exposures
were adjusted upward by  1.10.   Similarly,  since  VMT  from  1981
to 1995  is expected  to increase  31 percent, the  1995  exposures
were  adjusted  upward  by a   factor  of  1.31. [15]   This   VMT
 adjustment is likely to be somewhat conservative  on top  of the
 adjustment that  is  made   for  population  growth,  since  some of
both types of growth will be outward  at urban fringes,  rather
than  upward  in  the   areas   of  current  population  and   VMT
concentration.   The  caveat  applies  to  many  of   the  other
pollutants, whose exposure is  estimated in the same manner.
                              -22-

-------
                           Table  2-11

          Diesel  Particulate  Ratios and Model  Inputs

                            Ratios
                              g/min/g/min
             Idle/FTP         10 mph/FTP        35 mph/FTP

LDV/LDT        0.33              0.64               0.79
HDV            N/A*              0.51**            0.63


              25 Pollutant Concentrations (ug/m3)

                              1986

     0.00, 0.50, 0.61, 0.75, 0.91,  1.11, 1.36,  1.66, 2.00, 2.50,
     3.00, 3.70, 4.50, 5.50, 6.80,  8.20, 10.10, 12.30, 15.00,
     18.40,  22.00, 27.00, 33.00, 41.00, 50.00

                              1995

     0.00,  0.13,   0.16,  0.21,  0.26,   0.33,  0.42,  0.54,  0.68,
     0.86,  1.10,   1.40,  1.77,  2.25,   2.87,  3.64,  4.63,  5.88,
     7.47, 9.50, 12.07, 15.34, 19.49,  24.77,  31.47
*Anidle emission factor  is  used  to  simulate emissions in a
     parking garage.  It  is assumed there  are  no HDV in parking
     garages; therefore, a HDV idle/FTP ratio is not applicable.

**   For the street canyon microenvironment, a 10 mph/FTP ratio
     of  2.50  is  used.   For   1986,  this  ratio  results   in  an
     emission  factor   of 0.92   g/min,   equal   to   a   current
     estimated transit  bus emission  factor.   The  same  ratio  is
     applied in 1995.
                             -23-

-------
     The  mean exposure  levels  predicted by the model, adjusted
 to  take  into account  projected  increases in  VMT,  are  given
 below.

                         Exposure  (ug/m3)

                   Urban          Rural         Nationwide

 1986               2.63            2.38            2.56
 1995 Low Sales     1.27            1.06            1.22
 1995 High Sales    1.69            1.27            1.58


     As  seen above,  the rural  exposures  are  quite  similar  to
 the  urban exposures.   This is due  to the  greater  fraction  of
 heavy-duty vehicle classes 7-8 in rural areas, resulting  in a
 higher  composite  emission factor.   In addition,   the modified
 NEM  calculates  rural   exposures  quite  crudely.    This  is  a
 weakness  of   the  model   and  should be  taken into  consideration
 when reviewing the results.

 2.4.2  Comparison to Other Exposure Estimates

     This  section  compares the mean exposure  levels  predicted
 by the  modified NEM to  three other  diesel particulate exposure
 estimates made  previously.  These  are:   1) diesel  particulate
 exposures  estimated  by  OMS  in  a somewhat  similar  manner  in
 1983, 2) an estimate made  by  the  Lovelace Inhalation Toxicology
 Research  Institute (ITRI)  in  1981,  and  3)  exposures  estimated
 by OMS in 1983 using a  lead surrogate  approach.   The comparison
 among  these   methods  also  holds   for  other pollutants  in  this
 report for which the modified NEM was used.

 2.4.2.1  1983 QMS Exposure Estimate

     In  1983,  OMS  projected   diesel  particulate  exposure  in
 urban  areas   for  1995. [8]   OMS used  the original  CO  NEM,  with
 the only modification being the removal of  all  indoor sources.
 The annual average CO  exposure  was 2.12 ppm.  An average annual
 exposure for  diesel  particulate for  1995 was then  estimated  by
 ratioing  CO   and  diesel  particulate   emission   factors   and
multiplying the result by 2.12 ppm.  The  CO emission factor  was
 62.3 g/mile.   This is the CO national  average  emission factor
 for 1978, which  is the  same year  as the CO NEM  data base.   The
 composite diesel particulate  emission  factor used  (for the best
 estimate  base sales  scenario)  was  0.0554  g/mile.    Since  VMT
 from  1978 to 1995  was  expected  to  increase  45  percent,  the
diesel particulate emission factor was adjusted upward by 1.45.
                              -24-

-------
     The projected  exposure  in 1995 for the best  estimate base
sales  scenario  is  1.5-1.6  ug/m3  from  LDDV  and  1.6-2.1  ug/mj
for  HDDV.    The  total  projected  exposure  is  3.1-3.7  ug/m3.
This  is  considerably  higher than  the  urban exposure  range of
1.3-1.7 ug/m3  estimated in  this  study.   The urban  exposure of
1.7  ug/m3  estimated  in  this  study  corresponds  to  an  FTP
composite  diesel  particulate emission  factor  of  0.053  g/mile.
This   is   quite  similar  to   the   composite   emission  factor
calculated   in  1983.    It  appears   that   the   modified  NEM
calculates  lower exposures  per  g/mile that  the  CO NEM used
previously.    The  modified   NEM   does   include  more   recent
monitoring  data,  including  an  expanded  number  of  monitors,
relative to the CO NEM, which may account for the difference.

2.4.2.2  1981  Lovelace  Inhalation Toxicology Research Institute
         (ITRI) Exposure Estimate

     ITRI's  exposure  estimate  was confined to  LDDV.   ITRI used
a  particle  dispersion  model  to  project  future  atmospheric
concentrations  of LDDV particulate in urban and rural  areas of
the U.S.  in 1995.[24,25]  The atmosphere is  represented  by a
grid of cells  with  variable  heights that move with the velocity
and direction  of  wind.   Atmospheric  concentrations of particles
were calculated by  sweeping  the grid of cells, beginning on the
upwind  side of  the  urban  area  and  passing  them  across  the
city.   The model  accounts for  emissions  as well  as turbulent
mixing,  particle  diffusion,  particle  settling  and  particle
resuspension.   Assumptions  include:   1)  the  population  was
distributed  with  the  same density  as  diesel  vehicles, 2)  the
typical U.S. city has 40 percent of the population  in  the city
center during  work hours  and  has  an  average wind  speed  of 5
m/sec, and  3)  particles  entering a  cell are assumed to be mixed
uniformly throughout the cell.

     Information  used  included the  current  (1981)  land  areas,
populations  and gasoline  consumption  rates  for all  major U.S.
cities and  standard metropolitan  statistical areas.   These data
were scaled to represent  conditions after 1995  when 20 percent
of all  LDV were  assumed  to  be diesel-powered.   ITRI projected
average  concentrations  of  LDDV  particulate  based  on  assumed
LDDV  emission  rates  of  0.2  g/mile  and  0.5  g/mile.    This
comparison will be  limited to  the results obtained with the 0.2
g/mile emission rate,  since this emission rate  agrees with that
used in this study and the previous QMS study.

     Average concentrations  of  LDDV particulate  in urban and
rural   areas were projected  to  be  0.2  ug/m3   and  0.02  ug/m3,
respectively,  based on  an  emission rate  of   0.2 g/mile.   In
addition   to   projecting  these    "background"    atmospheric
concentrations in urban  and  rural  areas, calculations were also
made of  the higher  concentrations  that  may  occur   near  urban
street  canyons   and   expressways.    The  exposure  estimates
developed by ITRI are given below:

                              -25-

-------
                                  Particulate Exposure (ug/m3)
                                       Based on 0.2 g/mile

     Typical urban resident                     0.2
     Urban residents near freeway               2.0
     Workers on urban freeway                  15.0
     Workers in urban street canyon            15.0

The  exposures  in  urban  areas  and  near  street  canyons  and
expressways were  weighted,  with  the bulk of the resulting total
exposure being due  to  the urban background, and  used  in ITRI's
risk assessment.  The  composite  LDDV urban particulate exposure
was  not given by ITRI  but has  been calculated  by EPA  to be
roughly 0.50 ug/m5.

     The exposure estimate projected by  ITRI  is  lower than the
estimates  projected  in  this  study;  however,  the  different
assumptions  used  in  both  studies  make  a  direct  comparison
difficult.    ITRI  assumed 20  percent  of  all  LDV  were diesels.
In comparison, this  study assumed 0.9-11.5 percent of LDV sales
were   diesels.    ITRI   therefore   assumed  a   much   greater
penetration  of  LDDV.   Heavy-duty  vehicles were  not considered
by  ITRI.   In this  study,  heavy-duty vehicles are  responsible
for  66-90   percent   of   the 1995  urban  exposure depending on
diesel sales.  (These percentages were  derived  using the g/mile
emission factors  in Section 2.3.3  together with  the 1995 urban
VMT fractions in Section 2.4.1.)

2.4.2.3  Exposure Estimate Based on a Lead Surrogate Approach

     For comparison,  in  1983,  EPA  also  used  1975  atmospheric
lead monitoring   data  as  a surrogate  to  estimate  atmospheric
levels  of  diesel  particulate   in  1995.[8]    Estimates  were
provided of ambient  diesel  particulate  concentrations  at one or
two  particular  monitor  locations  in  a  large number of  U.S.
cities.  The monitors  were  chosen  in  areas  having  no  large
stationary sources of lead.

     For this  analysis,  an estimate was  made of  the  fleet's
automotive  lead  emission  factor  which  caused  the  observed
ambient  lead  levels, and is  compared  to  the  expected  diesel
particulate emission factor.   The  1995  projected  ambient diesel
particulate concentration  in 1995  was  set equal  to  the  urban
ambient lead  concentration in 1975 multiplied by the  ratios of
1995   diesel  particulate  to   1975   lead  emission   factors,
dispersion factors and  VMT.   Even  though the 1995 fleet average
(best  estimate  sales)  diesel particulate emission  factor  is
less than  half  the 1975  fleet  average  lead emission  factor,
ambient  concentrations  of  diesel  particulate  in  1995  were
projected   to   be   1.63   times   the   urban   ambient   lead
concentrations in 1975.   This  is  due to the projected increase
                              -26-

-------
 in VMT  (60  percent)  in 1995 and the higher estimated dispersion
 factor  for  diesel  particulate  relative   to  lead  (1.0  versus
 0.43).   Resulting  ambient  diesel  particulate  concentrations
 varied  from  0.7  to   3.9  ug/m3,  depending  on  the  city  and
 particular  monitor.   The estimates projected in  this  study,  as
 well as the earlier QMS estimates, fall within this range.

 2.4.3  Comparison to Particulate Monitoring Data

     The  annual  geometric  mean  total  suspended  particulate
 (TSP)   concentration   in   1984   was   50   ug/m3. [17]    It  has
 remained  fairly  constant  at  this   level   since  1982.   The
 nationwide  average  diesel  particulate  concentration  in  1986
 calculated  above  is  roughly  5.1 percent  of  the 1984  total.
 This percentage  is  similar  to  that estimated for emissions (3.9
 percent) in Section 2.3.5.
        o  denotes  particulate matter  less than  10 microns  in
size.   Since  virtually  all  diesel  particulate  falls  under
PM10,  it  would be  useful  to  estimate  the  contribution  of
diesel   particulate   to   PM10   concentrations.    A   PMi0/TSP
ratio  of  0.50  is  used,  based on  an  examination  of  1982-1983
monitoring  data. [26]   Applying  this   ratio  to  the  1984  TSP
concentration results  in  an annual  mean PMio concentration  of
25   ug/m3.    Thus,   diesel  particulate   emissions  appear  to
account  for  roughly  11  percent   of   the   annual  mean  PMio
concentrations.   It  should be  pointed out,  however,  that this
comparison  is probably misleading  since  the  TSP and  PM10 data
are  all  at  fixed  site monitors.   At   these  sites, the  diesel
levels  are  probably   much  higher  than  the   mean   personal
estimates given in this report.

2. 5  Health  Effects   of   Diesel   Particulate   and  Unit  Risk
     Estimates

     The  mutagenic   activity  of  organic extracts  from  diesel
particulate  was first  reported  in 1979.   At  that  time,  the
limited  data available  from  animal  and  epidemiology  studies
were not  sufficient  for  a cancer risk assessment.  As a result,
EPA  conducted a large  research  program to evaluate  the health
effects associated with exposure to diesel emissions.

     A comparative potency method was used to determine a unit
risk  estimate   for  diesel  particulate.    In  this  method,  the
potency  of  the  organic  extracts  from  diesel particulate are
compared with the potencies  of extracts from sources  for which
epidemiological   data   are  available.    A  large   number  of
mutagenesis  and  carcinogenesis  studies  were  performed  to
determine  relative potencies.   The  methodology  and results are
discussed  in detail   in reference  7.   The unit  risks  for the
light-duty    diesel     particulate    sources    ranged    from
                              -27-

-------
 2. OxlO"s-3.5x10"s.   These  represent  the  means  of  the  lower
 and  upper  95  percent  confidence  limits.   The  upper confidence
 limit unit  risks  ranged  from 3.3x10"s-6.3xlO~s.

     Two  other analyses  of EPA's  comparative potency  data were
 performed by  Dr.  Harris  for  NAS   [27]  and  by  the  Lovelace
 Inhalation  Toxicology Research  Institute  (ITRI).[24]   Harris1
 mean  estimate  was  a  0.0035  percent proportional  increase in
 risk  per year per  ug/m3  exposure  assuming  lifetime  exposure.
 Harris'  relative  risk was translated into  an  absolute measure
 of  lung  cancer  incidence  using  the  methodology  described in
 reference 28.   The  result is  a unit risk  of  IxlO"4.   Harris'
 upper  confidence  limit   estimate  (0.0252  percent  proportional
 increase  in risk)  is roughly equivalent to a unit risk of 7.5 x
 lO'4.

     ITRI   calculated  a   range   of   unit  risks   which  differ
 depending on which  comparative  source was used.   The resulting
 range  was  4.9xlO~s  to   2.1x10"4.   These  risks  appear  to  be
 upper  confidence  limits.  ITRI  chose  7.OxlO"5  as being  most
 representative.

     The  range of  unit risks that  will be  used  in this  paper is
 2. OxlO"s  to 1.0  x  10"4.   Assuming  an  average  lifetime  of 70
 years, the  range of  risk  estimates,  on an  annual  basis,  is 0.29
 x  10~6   to  1.4   x  10 "6  lung  cancers  per  person  per  ug/m3
 particulate.   The lower  end  of  the  range  is  the lowest  EPA
 estimate.   The upper end of the range is Harris' mean estimate
 which is  also  in  rough  agreement  with the  ITRI  estimate.   The
 range incorporates both  EPA's  and ITRI's upper confidence limit
 estimates.   Harris'  upper  confidence  limit estimate  was  not
 included  due  to   the  uncertainty of  translating  his  relative
 risk into an absolute measure of risk.

     The  comparative potency  studies that are the  basis  of  the
 risk   estimates   used   the  organics   extracted  from  diesel
 particulate.   Inhalation studies  using whole  diesel  emissions
 (i.e., particulate and gas phase emissions) were also performed
 concurrently but the results were  negative or inconclusive.[29]

     Long-term animal  inhalation  studies  are  presently  being
 conducted    by    ITRI,     Fraunhofer,    Battelle-Geneva    and
 Battelle-Northwest.   In   contrast  to  the  previous   studies,
 preliminary  results  indicate  that lung tumors have been found
 at concentrations  no  higher   than those  tested  with  negative
 results in the previous studies.  Reports are expected shortly.

 2.6  Current and Projected Health Risk

     The  annual  risk estimates  are  combined with  the  exposure
 estimates (given  in  Section 2.4.1) and  population  estimates to
 obtain  estimates   of  lung cancer  incidence  for  1986  and 1995.
The results are given in  Table 2-12.

     As seen in this table,  the total risk in  1986 ranges  from
 178-860 cancer  incidences and  drops  from 33-48 percent in 1995,
depending on projected diesel sales.

                              -28-

-------
                          Table 2-12



   Annual Lung Cancer Risk from Diesel Particulate Exposure
1986



1995  Low Sales



1995  High Sales
 Urban



137-661



 72-346



 95-461
Rural



41-199



20- 97



24-115
 Total



178-860



 92-443



119-576
                              -29-

-------
 3.0   FORMALDEHYDE

 3.1   Formation,  Composition  and Control Technology

      Formaldehyde  is  the most  prevalent  aldehyde  in  vehicle
 exhaust  and is  formed  as a  result  of incomplete combustion of
 the  fuel.   Formaldehyde  is  emitted  in  the  exhaust   of  both
 gasoline  and diesel-fueled  vehicles.   It  is  not a component of
 evaporative emissions.   Formaldehyde is of interest  due to its
 photochemical reactivity and  suspected carcinogenicity.

      Formaldehyde  has  the chemical formula HCHO.   Its structure
 is shown  below.

                                H
                            H - C  = 0

      Use  of  a  catalyst  has  been  found  to  be  effective for
 controlling formaldehyde emissions.   Formaldehyde emissions are
 controlled  to  roughly  the  same extent  as   total  hydrocarbon
 emissions with a catalyst.

 3.2   Emissions

      In  this section,   formaldehyde  emission  factors   (g/mile)
 for  calendar years  1986  and  1995  are  presented.   Nationwide
 mobile  source  formaldehyde  emissions  (metric  tons/year)  for
 1986  and  1995   are  then  calculated  by combining  the   emission
 factors with VMT data  for 1986 and  1995.   The results  are then
 compared   to   estimates  of   nationwide  total  formaldehyde
 emissions.

 3.2.1.  Emission Factors  for Calendar Years 1986  and 1995

     For  this analysis,  formaldehyde emissions  for  the various
 classes  are expressed  as a  percentage of total hydrocarbons.
 These  percentages  were  then   applied   to   MOBILES   exhaust
 hydrocarbon  (HC)  output  for  1986  and  1995   to  obtain  the
 formaldehyde emission  factors.  This was  done  for  two   reasons:
 1)  formaldehyde   emissions   can   vary  considerably  within  a
 vehicle  class  but are  more  consistent   when  expressed   as  a
 percentage  of  total  hydrocarbons, and 2)  use  of the MOBILE3 HC
 output should more accurately represent in-use emissions  since
 deterioration      and      effects     of     malfunction     and
 tampering/misfuel ing are  accounted for in MOBILES.

     An  important  issue to  address  is whether  the percentages
 chosen  are  adequate  to  use for  the  excess  hydrocarbons  that
 come  from deterioration, malfunction  and  tampering/misfueling.
Effects  of  deterioration, malfunction  and tampering/misfueling
have been studied  to the greatest extent  with LDGV;  therefore,
 LDGV data are examined here.
                              -30-

-------
     The percentages  for  LDGV were  based  on data  for  both new
vehicles  and  in-use  low  and  high   mileage   vehicles.    The
percentages   do,   therefore,    factor   in   the   effects   of
deterioration  and  minor  malfunctions.   The  percentages  are
remarkably  similar  for non-catalyst-equipped  vehicles  as  well
as  vehicles  equipped  with  oxidation catalysts  and 3-way  plus
oxidation  catalysts.   Overall,  formaldehyde  emissions  varied
from roughly 1-3 percent  of  total  HC while HC  emissions varied
from roughly 0.2 -6.0  g/mile.   Even within a particular vehicle
category, wide variations in HC  emissions  generally have little
effect  on  the  formaldehyde   percentages.    For   example,  HC
emissions from  two  in-use oxidation catalyst-equipped  vehicles
varied  from 0.4  g/mile   to  4.7  g/mile,   yet  the  formaldehyde
percentage for  both vehicles was  1.5  percent. [303   It  is  true
that  the  formaldehyde percentage   itself  is  subject   to  some
variation,   but  the   variation   is   slight  compared   to   the
variations in the absolute level of formaldehyde emissions.

     Studies   have   specifically   examined   the   effects   of
misfueling and  malfunctions  on  formaldehyde  and other  exhaust
emissions.    In  a  misfueling  study,  exhaust  emissions from  a
catalyst-equipped vehicle were measured initially with unleaded
fuel.[31]   Total aliphatic aldehydes were measured but can be
used  as  an  indicator  of formaldehyde  emissions.   The  vehicle
was  then  driven   5000   miles   on  various  commercial  leaded
gasolines.   Total  aliphatic  aldehyde emissions  initially  (with
unleaded  fuel)   were  2.1 percent  of  the  total HC  emissions.
Following  5000  miles  of  misfueling, the  percentage  increased
only slightly to 2.5 percent.

     Malfunctions that have been evaluated include, but are not
limited to,  12 percent misfire,  disabled EGR,  rich best  idle
and  high  oil  consumption.   Results with  non-catalyst-equipped
vehicles indicate roughly similar  formaldehyde  percentages with
and  without malfunctions.[32]   With catalyst-equipped vehicles,
formaldehyde  emissions tend  to be reduced  with  malfunctions
while total  hydrocarbon  emissions  increase,  resulting in lower
formaldehyde percentages  with malfunctions.[33-35]

     In  summary,  formaldehyde  percentages  appear  relatively
stable  over  a  wide  range  of  operating  conditions  and  HC
emissions.   The data support the use of expressing  formaldehyde
emissions as a percentage of total hydrocarbons.

     Formaldehyde emissions, expressed  as  a percentage of total
hydrocarbons,   for   the  vehicle  classes in  1986 and  1995  are
given below:
                              -31-

-------
                                   % of Total Exhaust HC
               Vehicle Class          1986     1995
               LDGV                   1.3      1.0
               LDGT  1,2               1.3      1.1
               LDDV                   4.1      4.1
               LDDT                   4.1      4.1
               HDGV                   3.1      3.1
               HDDV-TRUCK             3.0      3.0
               HDDV-COMM. BUS        10.0     10.0
     Data  for  the   light-duty  gasoline-fueled  vehicles  were
taken  from  references 30 and 32-36.  These data were assumed to
apply  to  light-duty  gasoline-fueled  trucks  as  well.   These
references  include data for both  new and in-use vehicles.   The
decreasing  percentages  for  these  vehicles  from  1986  to  1995
reflect the  phase-out of non-catalyst-eguipped vehicles.   Since
a  greater   percentage  of  non-catalyst-equipped  trucks  than
vehicles  are projected  in  1995,  the percentages  for trucks  do
not decrease as much  in  1995 as the vehicles.

     Data  for  the  light-duty  diesel  vehicles were taken  from
references  37  and 38.   Again, these  data were assumed  to apply
to   light-duty   diesel  trucks   as  well.    The  formaldehyde
percentage  of  total HC  is  assumed  to remain constant  from 1986
to 1995.

     Data   for   the  heavy-duty   gasoline-   and  diesel-fueled
engines  were  taken  from  references  22  and  38.   Since  the
heavy-duty   diesel  bus  engines   appeared   to   have   greater
percentages  than  the  other heavy-duty diesel engines,  they were
treated separately, and  assigned a VMT fraction of 0.002.[17]

     Applying  these  percentages  to  the  MOBILES  exhaust  HC
emission  factors  by vehicle class  for 1986 and  1995  and using
MOBILES  VMT  fractions,  resulting  composite  FTP  formaldehyde
emission factors are given below.

                           FTP g/mile
                     1986               1995

with I/M           0.0418              0.0201
without I/M        0.0453              0.0224


     Note that  MOBILES  runs were  used which  assumed  both  the
presence   and   absence   of  an   Inspection/Maintenance   (I/M)
program.  An I/M  p-ogram should  have a  beneficial  impact  on
formaldehyde   emissions   since    formaldehyde   emissions   are
correlated with HC emissions.  The  I/M  program selected has the
following characteristics:
                              -32-

-------
     Start year (January 1):               1983
     Pre 1981 MYR stringency rate:         20%
     Mechanic training program?:             No
     First MYR covered:                   1951
     Last MYR covered:                    2020
     Vehicle types covered:                LDGV
     1981 and later MYR test type:         Idle
     1981 and later MYR test cutpoints:    1.2% ICO/220 ppm IHC

     This represents the minimum EPA requires of I/M programs.

3.2.2  Nationwide Mobile Source Formaldehyde Emissions

     Nationwide  mobile  source  formaldehyde  emissions  (metric
tons/year)  for  1986  and 1995 are  calculated by  combining  the
calendar  year  emission  factors  in  the  previous  section  with
projected VMT  data for  1986  and  1995.   The  VMT data are given
in Table 2-8.

     The  nationwide mobile  source  formaldehyde  emissions  for
1986 and 1995 are given below.

              Mobile Source Formaldehyde Emissions
              	(metric tons per year)	

                  1986                 1995

with I/M          68,003               38,880
without I/M       73,697               43,329

     As seen, formaldehyde emissions  decrease from 1986 to 1995
due to the projected HC reductions.

     It should  be noted that  these  estimates only  account  for
formaldehyde directly emitted.   Formaldehyde  formed indirectly
from photooxidation of mobile source  volatile organic compounds
(VOC) is not included but will be addressed  in a later section.

3.2.3  Other Sources of Formaldehyde

     Formaldehyde  is  produced  in  the  U.S.  by   14  chemical
companies in 48  locations encompassing  21 states.   Formaldehyde
is  used  in the  manufacture  of  four  major  types   of  resins:
urea-formaldehyde,  melamine-formaldehyde,   phenol-formaldehyde
and polyacetal  resins.   These  resins  are used in a wide variety
of products, such  as  plywood,  particle  board  and  counter tops.
Formaldehyde  is   also   used   as   a  raw material  in  several
synthetic organic chemical production processes.

     In addition,  formaldehyde  is produced  as a  by-product  in
the   following   types  of   processes:   combustion   (mobile,
stationary  and  natural  sources),  petroleum  refinery catalytic


                              -33-

-------
 cracking   and   coking,   phthalic   anhydride   production  and
 atmospheric photooxidation  of  unburned hydrocarbons.   Table 3-1
 provides  a  summary of  1985 estimated formaldehyde emissions by
 source    category.[39]    The    contribution   of   atmospheric
 photooxidation   is  extremely   uncertain  and   therefore  not
 included  in Table  3-1.

     The  estimates of mobile  source formaldehyde  emissions in
 1986  with  and  without  I/M  were  68,003  and  73,697  metric
 tons/year, respectively.  This  represents 26-29  percent  of the
 total  formaldehyde emissions  given  in Table  3-1.   Therefore,
 roughly  26-29 percent  of  the  formaldehyde  emissions  emitted
 directly  into the  air  (versus  formed photochemically) appear to
 be attributable to mobile sources.

 3.3  Ambient  Concentrations of  Formaldehyde Emitted  by  Mobile
     Sources

     In   this   section,   urban   and  rural   concentrations  of
 formaldehyde  emitted  from  vehicle  exhaust   are  estimated  for
 1986  and  1995,   using  the  modified  version   of  the  CO  NEM
 discussed previously.   The  contribution of  mobile sources  to
 ambient formaldehyde levels is then discussed.

 3.3.1  Estimation of Urban and Rural Exposure

     The  three   required  model  inputs  are:  1)  the  nationwide
 urban  and rural populations  for  the  year(s)  of  interest,  2)
 FTP,  10  mph,  35 mph, and  LDGV idle  emission  factors (g/min),
 and 3) a  list of 25 ambient pollutant  concentrations,  defining
 the  concentration  intervals for which  cumulative  person-hours
 of exposure are  to be calculated.  The  first input is  given in
 Section  2.4.1.   Similarly,  the  FTP emission factors  have also
 already been  provided and  are  simply converted  from  g/mile to
 g/min  using  the average  FTP speed  of  19.6 mph.   To  calculate
 emission  factors at 10  mph  and 35  mph, MOBILES  runs  were made
 at  these  speeds  and  the  ratios  of  the  total  exhaust  HC
 emissions at  these speeds  to  those at  the  FTP  average  speed
 were  determined.   These speed  correction  factors  were  then
 multiplied by the  composite FTP emission factors to  obtain the
 emission  factors  at  10  mph and  35 mph.  The  speed  correction
 factors were 1.75 for  10 mph and 0.54 for 35 mph.

     The  idle  LDGV  emission  factors  for   1986  and  1995  are
 0.0009 and  0.0006 g/min,  respectively.  Idle  emission factors
 of both HC and formaldehyde have been measured  less  often than
 those for the  FTP, and  the formaldehyde:HC ratio  approach was
not used for idle.   They were instead  determined directly using
 the formaldehyde  emission  rate  data  in reference 40  together
with projected LDGV VMT  fractions of  non-catalyst-eguipped and
 catalyst-equipped  vehicles   in  1986  and 1995.    In  1986,  12
                              -34-

-------
                           Table 3-1

            Summary of Estimated  1985 Formaldehyde
                 Emissions by Source Category

                                     Formaldehyde Emissions
Source                                 (metric tons/year)

Direct Producers                                      910

Resin Manufacture                                   2,789

Synthetic Chemical Production                         655

Indirect Producers*
     Combustion (Mobile,  Stationary, Natural)      250,000
     Petroleum Catalytic  Cracking                   3,200
     Phthalic Anhydride Production                      1

                                        TOTAL:    257,555
*  Excludes atmospheric photooxidation due to the
   uncertainties inherent in the estimate.
                              -35-

-------
 percent of the LDGV VMT is projected  to  be due to non-catalyst-
 eguipped vehicles; in 1995, the percentage drops to 0.2 percent.

     The  25  ambient  pollutant  concentrations for  1986  ranged
 from  0.0  to  24.0 ug/m3 .   For  1995,  they ranged  from  0.0  to
 10.0  ug/m1.   Logarithmic  intervals  were chosen  to  maximize
 resolution over the entire range.

     The mean  exposure  levels  predicted by the  model,  adjusted
 to  account  for increased VMT,  are  given below.   These exposure
 levels account for direct emissions from mobile sources only.

                             Formaldehyde Exposure (ug/m3)
                        Urban         Rural        Nationwide

 1986
  With I/M               1.21          0.56          1.04
  Without I/M            1.30          0.60          1.13

 1995
  With I/M               0.68          0.31          0.59
  Without I/M            0.76          0.35          0.65

     It should  be noted that these  model  predictions  were made
 assuming   that   formaldehyde    has    emission  formation   and
 dispersion  characteristics  similar  to that  of  CO.   The model
 also  does  not   account   for  photochemical   reactions.   The
 exposure levels predicted by  the model are those resulting from
 direct exhaust  emissions, and  do  not account  for  either  the
 destruction or  photochemical formation  of formaldehyde  in the
 atmosphere.

 3.3.2  Contribution of  Mobile  Sources to Ambient  Formaldehyde
       Levels

     Ambient  formaldehyde  levels are a result  of  formaldehyde
 directly  emitted  by  sources   and   formaldehyde   formed  from
 photooxidation  of  VOC.    The  mobile  source  contribution  to
 ambient formaldehyde levels also  contains  both components.   The
 previous  sections  have  attempted  to  quantitate  the  directly
 emitted component.   It appears  that  roughly 26-29 percent  of
 directly  emitted  formaldehyde  may  be  attributable  to  mobile
 sources.

     Formaldehyde formed photochemically  is much more  difficult
to  quantify.   One  approach  being used  by EPA  is  to  determine
the relative contribution  of various  sources  based  on  estimates
of  annual U.S.  VOC emissions for each source.  These  estimates
 are  given  in  Table 3-2.[41]   Mobile  sources  account  for  30
percent of the  total VOC  emissions.  If it is assumed  that the
VOC  from  all  sources  have  the  equivalent  potential  to  form
 formaldehyde,   then  30  percent  of  the  formaldehyde  formed
photochemically is due to mobile  sources.   It should be  noted,

                              -36-

-------
                           Table 3-2

           Sources  of  VOC  Emissions  and  1985 National
                  VOC Air Emission Estimates*
                              VOC Emissions
Source                      (metric tons/year)         % of Total

Mobile Sources                  7,200,000                30

Misc. Solvent Uses              3,600,000                15

Hazardous Waste Treatment,
  Storage and Disposal
  Facilities                    3,500,000                14

Surface Coating                 3,160,000                13

Petroleum Marketing             2,230,000                 9

Petroleum Refining                740,000                 3

Chemical Manufacture              500,000                 2

Industrial Processes              365,000                 2

Miscellaneous Sources           3,020,000                12

               TOTAL:          24,315,000
*  Reference 41.
                              -37-

-------
however,  that  mobile sources  are estimated  to account  for a
higher percentage of  the total  VOC in urban areas.   Examination
of  1983  VOC  emissions in  61  urban  ozone  non-attainment areas
showed  the  average  mobile  source  VOC  contribution  to  fall
between 40-50 percent. [42]

     Since  the  mobile  source  contribution  to both  directly
emitted and photochemically produced  formaldehyde nationwide is
roughly equivalent,  as a  rough approximation,  30 percent of the
ambient  formaldehyde  concentration  could  be  due  to  mobile
sources  on a  national basis.   The  mobile  source  contribution
will vary depending on vehicle  use in  a particular  area, season
and  a  variety of other meteorological  factors.   In the winter,
for  example,   formaldehyde  emissions  from  mobile  sources  are
expected   to   increase  whereas   photochemical   reactions  are
expected  to  be  minimized.   The  mobile source  contribution  to
ambient formaldehyde  emissions in the  winter  could,  therefore,
be greater.

     Photochemical   modeling   is  one   approach   to   use   in
attempting  to  determine  the  relative  contributions  of mobile
and  stationary  sources to  ambient  formaldehyde concentrations.
Photochemical  modeling has been  done  to  simulate meteorology
and  photochemistry  occurring  during  summer months due  to  the
interest  in  ozone.   As part of  one modeling study, designed to
determine   the   ozone   impact   of   methanol-fueled   vehicle
substitution  in  California's  South  Coast  Air  Basin  (SCAB)
during a severe ozone  episode,  mobile  source VOC emissions were
removed    and    the   resulting   formaldehyde   concentrations
determined.[43]  The result  indicated  that  23  percent  of  the
ambient   formaldehyde   concentration    is    due   to   direct
formaldehyde  emissions  from  mobile  sources  and  formaldehyde
formed  photochemically  from mobile  source VOC.   This  result
lends some support to  the previous estimate.   Mobile  source VOC
in  the  SCAB   is  roughly   50  percent  of  the  total   which  is
somewhat  higher  than  most   areas   of  the  country.    Thirty
percent,  therefore,  can be considered an upper estimate.

3.4  Health Effects of Formaldehyde and Unit Risk Estimates

     Formaldehyde can cause  a number  of  acute  adverse health
effects  such   as   eye,   nose,   throat  and   skin   irritation,
headaches and  nausea,  as well  as  death.   Formaldehyde has also
been found  to  cause  nasal  cancer  by  inhalation  in males  and
females of  one strain  of  rat  and  in males of  another strain,
and  there  is  evidence of  its  carcinogenicity in  mice.   Human
data are  more  limited.   EPA  has  classified  formaldehyde  as a
probable (Bl)  carcinogen in humans.

     The unit  risk  estimates  derived  by  EPA are based on  a
single  animal  inhalation   study  conducted  by  the  Chemical
Industry  Institute  of Toxicology  (CUT).[44]    In  this  study,


                              -38-

-------
statistically  significant  increased  levels  of  squamous  cell
carcinomas (malignant tumors) were  found  in the  nasal  cavities
of  rats  at   14.3  ppm.    (This  is  roughly  1000  times  ambient
levels.)  The smaller increase  in these carcinomas  observed in
mice  was not  statistically  significant.   In  addition  to  the
squamous  cell  carcinomas,   small   numbers  of   benign  tumors
characterized  as  polypoid  adenomas  were  observed  in  rats at
each dose level.

     The  upper  confidence  limit  unit  risk  estimates  based on
these  data  are 1.7  x  10"4  for  benign tumors and  1.3  x  10"s
for malignant  tumors.[45]   The unit  risk  for  malignant tumors
of   1.3xlO"s   was  used   for  this   analysis.   The  current
consensus  in EPA  favors  use of  malignant data only.   EPA's
Office  of Toxic  Substances  is   using  the unit  risk  based on
malignant tumors  in  assessing the  risk to garment  workers  and
home  residents.[45]   Assuming an average  lifetime of 70 years,
the annual risk estimate is 1.9xlO~s.

3.5  Current and Projected Health Risk

     The  annual  risk estimate  is  combined with the  exposure
estimates (given  in  Section  3.3.1)  and population estimates to
obtain  estimates  of  cancer  incidence for   1986  and   1995.   The
results are given in Table 3-3.

     As  seen  in this  table,  the total  risk  in  1986  ranges from
46-50 cancer incidences  and drops to  29-31  cancer incidences in
1995.   This  decrease  is  accounted  for   by  the  anticipated
decrease in HC emissions.

     These risk estimates are for formaldehyde  emitted directly
from  vehicle  exhaust.    Unfortunately,   the  model  used  to
determine  ambient concentrations  could  not  account  for  the
destruction and photochemical formation of  formaldehyde in the
atmosphere.

     An  approach  to  include photochemistry  is  to   use actual
ambient  monitoring  data  and assign  a mobile source fraction.
This  approach  accounts  for  formaldehyde  directly   emitted  and
formed or destroyed photochemically.   OAQPS calculated an urban
population weighted  average  of 12.71  ug/m1  based on data from
four  cities.[46]   It represents  the average of  the population
weighted  summer   average   in Baltimore,  Los  Angeles  and  New
Jersey  with  the  winter  average  in  Philadelphia.    The summer
concentrations  used  represent   maximum   rather   than  average
values.   As  a  result,  12.71  ug/m3  is   closer to  an  urban
population  weighted   maximum concentration.   It raay  also  be
biased  high  since  these  cities  are  in  ozone  non-attainment
areas.   It  will  be  used here  to  represent  a plausible upper
limit.    A   concentration   of   1.50  ug/m3  was selected  to
represent rural areas.[46]
                              -39-

-------
                           Table 3-3

           Annual Cancer Risk From  Direct Emissions
              of Formaldehyde from Mobile Sources

                        Urban          Rural      Total

1986
  With I/M                 40             6           46
  Without I/M              43             7           50

1995
  With I/M                 25             4           29
  Without I/M              27             4           31
                             -40-

-------
     In  Section  3.3.2,  it  was  indicated that  mobile  sources
constituted roughly 30 percent  of  formaldehyde emitted directly
as well  as  30  percent  of the total  VOC emitted.  Assuming  the
VOC  from all  sources have  the  equivalent  potential  to  form
formaldehyde,  a mobile source fraction  of  0.30 is selected  and
applied  to  the estimated urban and  rural  concentrations  above.
Resulting urban  and rural  formaldehyde concentrations  due  to
mobile sources are 3.81 and 0.45 ug/m ,  respectively.

     In   comparison,    the   urban   and   rural   formaldehyde
concentrations  obtained   from   modeling   are  1.21-1.30   and
0.56-0.60   ug/m3,   respectively.    The   urban   concentration
estimated from  ambient monitoring data  is  roughly three  times
greater  than  that obtained by  modeling.   Two possible  reasons
are:   1) the  ambient monitoring data accounts for  formaldehyde
formed  photochemically,  and  2) the  ambient  data  may be  from
fixed site monitors that overrepresent 24-hour exposures  of  the
population.   In contrast, the rural  concentration  obtained from
modeling  is  slightly   higher  than that  estimated  from  ambient
monitoring  data.    The  rural  concentrations  were   estimated
rather crudely with both approaches.

     The  exposure  estimates  based  on  ambient monitoring  data
are then  combined  with the annual risk estimates  and  urban  and
rural population estimates  to obtain estimates  of lung  cancer
incidence from current mobile  sources.   The  results  are  given
below:

                           Annual  Lung Cancer Risk
                   (accounts  for photochemical  reactions)

                          Urban      Rural     Total

Current Mobile Sources    126          5         131

     Mobile sources currently could be responsible for  as many
as  131  cancer  deat.is  from   formaldehyde   exposure.    Direct
emissions of formaldehyde were  projected  to  decrease  roughly 42
percent from 1986  to 1995.   Mobile source VOC  follows  a  similar
trend.    If  it is  assumed that  the photochemical component  is
also  reduced  similarly,  the  mobile source  risk  in  1995  is
estimated be roughly 77 cancer deaths.

3.6  Current Activities

     Research  activity is planned or  underway in three  areas:
1)   emissions  characterization,   2)  photochemistry,   and   3)
ambient monitoring.  Formaldehyde emissions  from mobile  sources
are  being  characterized  under  cold  temperature  conditions,
since  it appears  formaldehyde  emissions  could increase  under
these conditions.    In addition,  work  is  planned in FY87  to
                              -41-

-------
investigate  factors  affecting  formaldehyde  formation  in  the
atmosphere.   Specifically,  smog chamber  data  will  be  analyzed
to  rank  order  the  various  organic  species  based  on  their
ability to generate  formaldehyde.   Modeling  simulations  will be
conducted  to  assess  the  effect  of  varying   light  intensity,
temperature and  other  meteorological  parameters on formaldehyde
yields.   Smog chamber  experiments  will  also  be  conducted to
determine  how   well   two  major  chemical   mechanisms   predict
formaldehyde formation.

     In the  monitoring area,  formaldehyde measurements  will be
included  in  EPA's Toxic  Air  Monitoring System  (TAMS)  network.
Measurements  will initially  be  made in  three  cities  (Houston,
Boston  and Chicago)  with  a  single  monitoring   site  in  each
city.  TAMS  should eventually be  operational  in  three  to  five
cities with up to three to four monitors in each city.

     It  should  be  noted  that,  in  the  case   of  formaldehyde,
formaldehyde  levels  inside  many  homes are significantly greater
than outside  levels.   Recent  monitoring  results  indicate  that
formaldehyde   levels   in  new   (less   than   one  year   old)
conventional  homes  generally  fall  within  the  range  of  62
ug/m3  to  250  ug/m3;   few  measurements  exceeded  375  ug/m3.
In new mobile  homes,  formaldehyde  levels generally  fall  within
the  range of  250 ug/m1  to 375 ug/m3  with the  highest  levels
measured  near   500  ug/m3.[45]  EPA   has  calculated   expected
10-year averages  for formaldehyde  levels in homes built today,
acknowledging that there  is a  significant source  of uncertainty
associated with  the  estimates.   The calculated  10-year  averages
are  88  ug/m3  for conventional  homes  built  using  significant
amounts  of  ureaformaldehyde   pressed wood  and  125 ug/m3  for
mobile homes.[45] These exposures  are significantly higher  than
current outdoor urban exposures.
                              -42-

-------
 4.0  BENZENE

     The  Office  of  Mobile  Sources has  recently  completed a
 thorough   analysis   on  the   carcinogenic   impact  of  benzene
 emissions.[47]   This   section will  summarize  that  analysis.
 Also  discussed is  an  alternative  approach  to  estimating  the
 risk   using   ambient   monitoring   data   in  conjunction  with
 emissions  data.  A  risk  analysis  performed by  the California
 Air Resources  Board (CARB) is then  summarized and compared with
 the risk estimates presented  in this report.

 4.1  Formation, Composition,  and Gontrol Technology

     Benzene   is  an   aromatic  hydrocarbon  with  the  formula
 C6HS.    It  is   present   in  both  exhaust   and  evaporative
 emissions.   Current  data  show  the  benzene  level  of  current
 gasoline   to   be  about   1.3%,   with  diesel  fuel  containing
 insignificant  levels  of  benzene.   Very little  exhaust  benzene
 is unburned fuel  benzene.  Some work indicates that non-benzene
 aromatics   in  the  fuels  cause   about  70-80%  of  the  exhaust
 benzene  formed.  Benzene  also forms from  engine  combustion of
 non-aromatic fuel hydrocarbons.  The fraction of  benzene in  the
 exhaust  varies  depending   on   control   technology  and  fuel
 composition but  is  generally  about   3-5%.   The  fraction  of
 benzene  in the  evaporative  emissions  also depends  on  control
 technology  (e.g.,  whether  the vehicle  has fuel  injection  or a
 carburetor)  and fuel composition  (e.g., benzene  level and RVP)
 and is  generally about  1%.  These  data also show  that diesel
 vehicles account  for only  about  3% of the total mobile source
 benzene emitted.

 4.2  Emissions

 4.2.1  Emission Factors for Calendar Years  1985 and 1995

     An  approach   similar   to   that   described   earlier    for
 formaldehyde was  employed  for benzene.  Benzene emissions were
 expressed   as   a   percentage   of  exhaust   and  evaporative
 hydrocarbons.   These  percentages  were  then  applied  to   the
 hydrocarbon emissions  data   in  MOBILES  to  obtain  composite
 emission  factors  for  calendar  years  1985  and  1995.   For  the
 purpose of  this report,  it will  be assumed  that  calendar  year
 emission    factors   for   1985   and  1986   will  not   differ
 significantly.

     Benzene emissions,  expressed  as  a percentage of  exhaust
 and evaporative emissions  for the  various  vehicle classes,  are
 given in Table 4-1.

     The   test  vehicles   used   to   determine   the   benzene
percentages were  low  mileage, well maintained  vehicles.   Like
 formaldehyde,   it   is   important   to   address   whether    the
percentages  chosen   are   adequate   to  use  for  the   excess
hydrocarbons  that  come  from  deterioration,  malfunction   and
tampering/misfueling.

                              -43-

-------
                           Table 4-1

        Benzene  Emissions Expressed  as Percentages of HC

- Vehicle Class          % of Exhaust HC      % of Evap.  HC

LDGV                                            0.35-1.53*
  3 Way Cat                   5.12
  3 Way + Ox Cat              2.78
  Non-Cat or Ox Cat           3.95
LDGT1,2                       3.24                 1.1
LDDV                          2.40
LDDT                          2.40
HDGV                          3.48                 1.1
HDDV                          1.10
   l.ll-l.53% for carbureted LDGV.
   0.35-0.46% for fuel injected vehicles.
                              -44-

-------
     For  non-malfunction conditions,  there did not appear to be
 any  consistent  differences  in  benzene  percentages   among  the
 various  control technologies (non-catalyst, oxidation catalyst,
 3-way  catalyst,  and 3-way  plus  oxidation catalyst),   regardless
 of  whether the vehicles were tested  new or in-use.   There were
 some  differences in benzene  percentages  found  between certain
 malfunction  modes  and  the unmodified test  conditions.  The most
 significant  of  these were 1) the 12  percent misfire  mode which
 consistently  decreased  the benzene  percentage  (while  greatly
 increasing total  HO   and  2)  the  rich  best  idle  mode  which
 consistently  increased  the  benzene  percentage  as well  as  the
 total  HC  (but not as much  as the total  HC increase  in  the 12%
 misfire  mode).   Due  to the  lack  of  differences among control
 technologies, the  offsetting nature  of  these  two malfunctions,
 and   the  lesser   effects  of   the   other   malfunctions,   no
 adjustments   were   made   to   the    benzene   percentages   for
 malmaintenance/tampering.

     Resulting  composite  FTP  benzene emissions  for  1985  and
 1995   are   0.128-0.135   g/mile   and   0.055-0.057   g/mile,
 respectively.   The ranges  result from the ranges given for the
 percentage of evaporative hydrocarbons  for LDGV.  The  MOBILES
 runs  assumed the pr5sence  of  a standard,  minimum I/M program.
 As  stated  previously,  diesel   vehicles   account  for  a  small
 percentage (3%) of the  total mobile source benzene emissions.

     Based on  the QMS analysis,  RVP  control,  which would be
 accompanied  by  a  small increase in  both  benzene content  and
 total  aromatic   content of  gasoline, would  have little  or no
 effect on overall fleet  emissions  or on  the  number  of  cancer
 incidences.[47]

 4.2.2   Contribution  of Mobile  Sources  to Nationwide  Benzene
        Emissions

     In  1982,  total  benzene  emissions   were  roughly  293,000
 metric  tons.[47]   Mobile  sources  account  for  250,000  metric
 tons,  or  85 percent   of   the  total.   Of  the  mobile  source
 contribution, 70%  comes  from exhaust  of motor  vehicles  while
 14%  of the   benzene  emissions   are  motor vehicle  evaporative
 emissions.   About  1%  of  the   total benzene  emissions  occur
 during   motor  vehicle  refueling.    The  remaining  benzene
 emissions  come  from stationary sources  with  coke ovens  being
 responsible   for  10%   of  the  total.   Obviously,  in  regions
without coke  ovens virtually all benzene is from mobile sources.

 4.3  Health Effects of  Benzene and the Unit Risk Estimate

     Several  epidemiology  studies on  workers  exposed to benzene
have  identified benzene  as a  carcinogen  causing leukemia  in
humans.  The  upper  confidence  limit unit risk estimate has been
determined from these studies to be 8.0 x  10"6.[47,48]
                              -45-

-------
4.4  Current and Projected Health Risk

     Nationwide   exposure   levels    from   both   exhaust   and
evaporative  emissions  were estimated  using  the  modified  NEM
exposure model.  The  speed  correction factors at 10 mph  and 35
mph   required   for   the   NEM   inputs   were  those  used  for
formaldehyde.  The nationwide exposure  levels  predicted  by NEM,
adjusted  upward to  account  for increased  VMT, are  3.09-3.25
ug/mj  and  1.68-1.77  ug/m3   for   1985  and  1995,  respectively.
The range  is due  to  consideration of both a  low and  high range
evaporative  emissions  estimate  for   light-duty  gasoline-fueled
vehicles.    Annual   cancer   incidences   from   exhaust   and
evaporative  emissions  are  estimated  to be  84-89   in 1985  and
50-52  in  1995.   The  reason for  this  marked decrease  is  the
decrease in projected HC in 1995 and,  thus,  benzene emissions.

     Exposure to benzene  during  refueling  includes  self-service
refueling,  occupational exposure  (service station  attendants)
and   community   exposure   in  an  urban  area.    Exposure   to
self-service refueling  and  occupational exposure was  determined
by  measuring benzene  levels  in  the  region  of  the  face of  a
person  refueling  a  vehicle  tank.  The exposure  in  a  typical
urban area  was  estimated  using  a dispersion  model.[48]   Annual
cancer  incidences  from benzene refueling are  estimated  to be 8
in 1985 and 7 in 1995.

     The total  estimated cancer  incidences due to  mobile source
benzene  and  the  contribution   of   evaporative,   exhaust  and
refueling emissions are given below.

      Annual Cancer Incidences due to  Mobile Source Benzene

Year      Evaporative*     Exhaust      Refueling      TOTAL

1985        17-22            67             8          92-97
1995         8-10            42             7          57-59
*  Includes low and high range evaporative emissions estimate.

     These  numbers   indicate  that,  as   vehicle  hydrocarbon
evaporative and exhaust  emissions  are controlled,  the estimated
carcinogenic impact due to motor vehicle  benzene emissions will
decrease.  They  also indicate that  the impact  is  somewhat less
than that  predicted  for  diesel particulates or  formaldehyde.
However,  it  should  be  pointed  out  that   the potency  on which
these estimates are  based  is  derived  from several epidemiology
studies and is thus  a much firmer  potency than that derived for
diesel  particulates  or  f or.naldehyde.   The  potency for  diesel
particulates  is  based  on  several  animal  studies  rather  than
human data.  For  formaldehyde,  it  is  based on  a  single animal
study.
                              -46-

-------
     An  alternative  approach for  estimating the  exposure and
 resulting  risk of  mobile  source  benzene emissions  is  to use
 available  ambient monitoring  data and  assign a  mobile source
 fraction.  This  approach was used for formaldehyde  in  order to
 account for photochemistry.

     An  urban population  weighted average  of 10.24 ug/m3 was
 calculated based  on  data from six  metropolitan  areas.   Ambient
 data   for  Baltimore,   Los   Angeles,   Northern   New   Jersey,
 Philadelphia and Chicago were  taken from reference 49.   Ambient
 data  for  Houston  were  obtained  from  reference  46.    Their
 estimated metropolitan area  populations  in  1983  were taken from
 reference  15  and  used  to   calculate  a  population  weighted
 average.  Since these cities  are  in ozone non-attainment areas,
 the  resulting  urban  concentration  may represent  more of  an
 upper-bound.    A concentration  of  7.52  ug/m3  was  selected  to
 represent rural areas.[46]

     Since mobile  sources  appear  to  be  responsible for roughly
 85  percent  of  the   total   benzene   emitted,  a  mobile  source
 fraction  of   0.85  was  selected  and  applied  to  the  estimated
 urban  and  rural  concentrations   above.   Resulting  urban  and
 rural benzene  concentrations due  to mobile sources are 8.70 and
 6.39 ug/m3, respectively.

     These exposure estimates  are  then combined  with the annual
 risk  estimates  to  obtain  estimates of  cancer  incidence  from
 current mobile sources.   In  1986,  the  urban and  rural  risks,
 using this alternative  approach,  are estimated to be 179 and 44
 cancer  incidences,   respectively,  for  a total   of  223  cancer
 incidences.   Based  on  the NEM modeling,  emissions of  benzene
 from  mobile  sources  are  projected  to  decrease  roughly  40
 percent from  1985 to 1995.  The  mobile  source urban,  rural and
 total  risk  in  1995,   accounting for  this  decrease  and  the
 projected  population  increase,  is  116,   29,   and  145  cancer
 incidences, respectively.

     These  risk  estimates   are   somewhat   higher  than  those
 calculated  with  the  NEM  approach.   Both   approaches  contain
 uncertainties.   For  example,  temperature   effects  are  only
 partially  accounted  for.   With the NEM  approach,  it is assumed
 that  CO  and  benzene have  similar dispersion  characteristics.
 This may not  be  true.   With the ambient apportionment approach,
the  ambient   data   may   be  from  fixed  site   monitors  that
 overrepresent  24-hour  exposures  of the  population.   It  is  also
not certain whether the cities chosen are  representative of the
 entire urban  population.   For the purposes  of  this report,  a
 range of risk estimates for benzene will be  reported, using the
 results  of both approaches.   The  resulting  range of  cancer
 incidences is 92-223 in 1985 and 57-145 in 1995.
                              -47-

-------
4.5  GARB Analysis of Current and Projected Health Risk

     In  January,  1985,  the  California  Air  Resources  Board
(GARB)  identified benzene  as  a  toxic  air  contaminant.  As  a
result, CARB released a draft  report  which includes information
regarding  present (1984)  and  future (2000) benzene  emissions
and levels of benzene in California as well  as  the magnitude of
risk  posed by  benzene  emissions  from various  sources.[50]   An
addendum   to   the  report  was  recently   prepared.[51]    The
information in  these  reports  will  be briefly summarized in this
section.   The  following  section  provides a comparison of  the
EPA   and   CARB  risk   estimates.    For   reference,  California
contains  11  percent  of the U.S.  population  and  11 percent  of
the motor vehicles.[19,52]

     Total benzene emissions  in California  were  roughly 18,400
tons/year in 1984 and projected to  decrease  to  13,000  tons/year
in  the year 2000.   The vehicular  contribution,  which includes
both  on-road  and  off-road vehicles  as  well  as  other  mobile
sources,  constitute  roughly  91 percent  of  the total  emissions
in  1984  and  84  percent of the total emissions  in 2000.   The
statewide  population  in 1984 was 25.8 million  and is  projected
to increase to 31.4 million in 2000.

     The population exposure  to benzene  in  California  for  1984
and  2000  was   estimated  using   benzene  emissions  data  and
population and monitoring data.  The  estimated  statewide annual
average  exposure to  benzene  in  1984,   8.4  x  107  ppb-persons,
is  equal   to  25.8 million,   the  number  of   people exposed  to
benzene, times  3.3  ppb,  the  population  weighted  annual average
benzene  concentration.    The  vehicular  contribution  (68.1xl06
ppb-persons)  was  calculated based  on the  emissions data.   This
is  similar  to  the  alternative   approach   described  in  the
previous  section.   Based on  exposure of  25.8 million  people,
the  annual  average  benzene  concentration  due  to  vehicular
sources  in 1984  is  2.64 ppb (8.5 ug/m1) .   This  accounts  for
both urban and rural  exposure.

     The estimated statewide  annual average  exposure to benzene
in  2000  is  74  x  10s  ppb-persons.    This   was   calculated  by
adjusting the 1984 value to account for  the  projected  reduction
in  total  benzene  emissions  and  the  projected  increase  in
population.    The   vehicular   contribution   of   63.5  x   106
ppb-persons  is  again  based  on  the  emissions  contribution.
Based  on  exposure of 31.4  million people,  the  annual  average
benzene concentration due to vehicular  sources  in  2000  is  2.0
ppb (6.4  ug/m3).   This is  a  25 percent  reduction  relative  to
1984.
                              -48-

-------
     CARB  used  a  range  of  risk  estimates  of  22-170  excess
 cancers  per  million people  exposed per ppb per  70 years.   The
 range  is based  on  an examination of both mouse  and human data.
 Using  this range  together  with the estimated  statewide annual
 average benzene exposures in 1984 and  2000,  the resulting range
 of  risks  in  1984  and  2000  are 1,900-14,500  and  1,630-12,600
 excess cancer cases  in  California  per  70  years.   On  an annual
 basis,  this  translates  to 27-207 cancer  cases  in  1984  and
 23-180 cancer  cases  in  2000.   The  vehicular   contribution  is
 21-166  cancer   cases  in  California in  1984  and 20-154 cancer
 cases  in 2000.   Expressed  as  individual  risk,  the  vehicular
 contribution  is 8.11 - 64.3 x  10"7 in 1984  and 6.4  - 49.0  x
 10"7 in 2000.

 4.6  Comparison of EPA and CARB Health Risks

     The range  of  cancer incidences  given  in  this  report  is
 92-223 in  1986  and  57-145  in  1995.   In order  to compare these
 risk  estimates  to  those   of   CARB,   they   are  expressed  as
 individual  risks.   The  resulting individual  risk estimates are
 3.8 -  9.3  x 10"7  in 1986  and  2.2  -  5.6 x  10~7 in  1995.   The
 upper end  of these risk estimates are roughly equivalent to the
 lower end of the CARB estimates.

     CARB  used   a  range  of  risk  estimates  of  22-170  excess
 cancers  per  million people exposed per ppb  per 70  years.   In
 terms of unit  risks, this  translates  to  6.81  - 52.6  x 10"6.
 The upper  confidence limit  unit risk used in this  study is 8  x
 10"s;   this is  near the   low  end  of   the  range of  CARB  unit
 risks.   CARB's  high  end  risk  estimate  is  based   on  animal
 studies,  whereas EPA's risk estimate is based on human studies.

     The CARB   estimate  of  the  benzene  concentration  due  to
vehicular  sources   in  1984  is  8.5   ug/m3 .   In  comparison,  the
 range of estimated exposures in  this  study for  1985  is 3.2  -
 8.1 ug/m3  (accounting  for  both urban  and  rural  exposure,  as
CARB  has  done).   Clearly,  the   CARB estimate  exceeds   the
 exposure estimates  used in this  study.  This  is not surprising
 since monitoring data  indicate  that benzene concentrations  in
urban areas  of  California  are  higher  than most  other areas  of
the  country.    (The high benzene   level  in  Los  Angeles  was
factored into the national estimate  used in this study.)

     The  CARB   projections  do  not   show   benzene  emissions
dropping as rapidly as  predicted  in  this report.  CARB predicts
that vehicular  benzene  emissions will  decrease  35  percent  from
 1984 to  2000.   This can  be compared to a  40 percent decrease
from  1985  to 1995  given  in this report.   Also, when the  CARB
risk is expressed in terms of total   cancer cases, the  number  of
cancer cases  only  decreases  by  13  percent  from 1984  to 2000.
                              -49-

-------
This is because the decrease in emissions is  slightly  offset by
the  projected  increase  in  population.    The   population  in
California  is  projected to  increase 22  percent  from 1984  to
2000.   In  comparison,   the U.S.  population  is  projected  to
increase 8 percent from  1986 to 1995.

     When reviewing the GARB risk  estimates,  it  should also be
remembered  that  the  vehicular  contribution  includes  on-road
vehicles,  off-road vehicles,  trains,  ships,  aircraft,  mobile
equipment  and  utility  equipment.   The   EPA  estimates  only
include   emissions   from  on-road   vehicles.    Most   of   the
difference seen in the  two  estimates,  however, is attributed to
the wide range of  unit risks used by CARB.

4.7  Current Activities

     The California Air  Resources Board  (CARB)  is  considering
implementing  regulations  requiring  control   of  motor  vehicle
benzene  emissions.   This may  include  a  limit  on the  benzene
content  of  gasoline  and/or  stricter   light-duty  exhaust  HC
standards.   The first  measures being  considered are  a  more
stringent exhaust  HC  standard  by the end of  1987,  and possible
changes  in  the  evaporative  test  procedure,  e.g.,   multiple
diurnal tests,  longer  soak  times,  and higher soak temperatures.

     The  EPA  Office  of  Air  Quality  Planning  and  Standards
(OAQPS)  has  designated benzene  as  a  hazardous  air  pollutant
under  Section   112  of  the  Clean  Air  Act  and  is  implementing
necessary controls for  stationary sources.   The EPA  Office of
Mobile  Sources  (QMS)  is   determining   if  any  motor  vehicle
controls  specific  to  benzene   are  needed.    Also,  QMS  is
determining  how  other  regulations  that  may be  proposed for
additional hydrocarbon control  would affect benzene.
                              -50-

-------
 5.0  GASOLINE VAPORS

     Several  years   ago,  the   American   Petroleum  Institute
 released  the  results  of  a  lengthy  animal  inhalation  study
 showing that totally vaporized  gasoline resulted in an increase
 in  kidney  tumors  in   male  rats  and  liver  tumors  in  female
 mice.[53]   EPA  evaluated these  data  in depth to  determine the
 potential carcinogenic  impact.[48,54]  From these  studies,  EPA
 calculated  the  following plausible  upper  limit  carcinogenic
 potencies  for  exposure  to  1   ppm  for  a  year.    The  maximum
 likelihood  potency is  lower  by almost  a factor  of  two for  both
 data sets.

       Gasoline Carcinogenic Potencies  Based on API Study

           Rat data          4.9 x 10~s/ppm
           Mice data         2.9 x 10"5/ppm

     Exposure  to  gasoline  vapors  and benzene  from  gasoline
 during  refueling was estimated  based  on  an American Petroleum
 Institute  study  that   involved  measuring gasoline  and  vapor
 levels  in  the   region   of  the  face  of   a  person  refueling  a
 vehicle tank.  The exposure  in a typical  urban area  for  these
 refueling emissions  was also estimated by using the Industrial
 Source  Complex   (ISC)   dispersion  model  to  calculate  annual
 concentrations.[48]

     The exposure numbers  determined  were  combined with the rat
 potency data to calculate the potential carcinogenic impact due
 to  refueling emissions.   These  numbers  for  both  benzene  and
 gasoline vapors are as below.

    Potential  Carcinogenic  Impact of Refueling  Loss  Emissions

                                       Annual Incidences
     Exposure                        Benzene    Gasoline Vapors

     Self-service refueling             5             35
     Occupational exposure
       (service station attendants)      2             17
     Community exposure in
       an urban area                    0.5           13

     TOTAL EXPOSURE                     7.5           65

     There  are  several  limitations   to  the  use   of  the  API
bioassay,  however.  First, totally  vaporized gasoline  was  used
which  includes  higher   molecular weight components as well  as
the lighter  components.   Some  short  term  animal tests done  by
API indicate that the branched  chain  paraffins (e.g., C-6,  C-7,
C-8) are  the  ones that  would  cause kidney  damage.  It is  not
certain  what relation   the  kidney  damage  has to the  kidney
                              -51-

-------
tumors  in the  rats.   However,  it  is  known  that  most  (i.e.,
80-90%) of the  gasoline  vapors in a realistic refueling episode
are below C-6 while only about 23% of the vaporized  gasoline is
below C-6.[55,56]

     Another  limitation  to   the  API   work  is  the  apparent
involvement   of  a   low   molecular   weight   protein   (alpha
2-globulin) that  is  present only  in  rats.   It  is  thought that
this protein, which is synthesized in the liver  of  mature,  male
rats,   accumulates    in   the   kidney   following   hydrocarbon
exposure.   The  effect results  in  kidney damage  (nephropathy,  a
chronic inflammation and vascular  thickening of  portions of the
kidney) similar  to  that  known  as  "old rat  nephropathy".   This
condition  frequently occurs  spontaneously  in  older rats;  the
presence  of  hydrocarbons  accelerates  this  change.   There  is
some concern that this kidney  condition is  related to the small
cancerous growths  found.   Alpha  2-globulin is  not  present  in
humans which are not  affected  by this kidney condition.  Female
mice  have  a  large  and  variable  rate  of  spontaneous  liver
tumors.   Like  the case  of  the  rats,  some  questions have been
raised  about  extrapolating  the  results  with   the  mice  to
humans.   These  issues have been discussed  in  a report  by the
Health Effects Institute  (HEI)  which concluded the following.

     ...the usefulness  of  available  animal and  human  data  in
     helping  to  determine  health   risks   is   quite  limited.
     Unburnt gasoline  vapors  may,  upon further  investigation,
     prove  to   present   significant   carcinogenic   risks  for
     humans.    The  evidence  is  not   available  to  make  that
     statement  today.    Significant   additional  research  would
     have to be undertaken  to  understand  important  mechanisms
     of action,  physiological  differences between  test animals
     and people and the extent and nature of exposures.[57]

It should be  noted  that  HEI  has  since decided  to  perform  no
research  in  the  gasoline  vapor  area.   Also,  aside from  some
limited work being done  by  API, EPA  is  aware of  no  additional
health research being done on gasoline vapors.

     In an  internal  EPA  memo,  EPA  has concluded  the following
about the potential  carcinogenicity of  gasoline vapors  and the
EPA risk assessment  discussed above.

     All issues  raised by HEI  were known to EPA and thoroughly
     assessed in EPA's  analyses of  the  same  issues.   EPA's
     quantitative    risk   assessment   for   carcinogenicity  of
     unleaded gasoline  vapors  was prepared according  to  EPA
     Guidelines  and  expresses  its uncertainties,  which are not
     greater than  EPA  routinely  addresses.   Further  research
     would   not  change   the   underlying   evidence   of   the
     carcinogenicity of  gasoline  vapors but may shed  light  on
     which  components  are   responsible  for  the  carcinogenic
     response and this could alter the risk calculation.[58]

                              -52-

-------
     On an overall basis  then,  it  appears that gasoline  vapors
must  be  regarded  as  a  potential  human  carcinogen.    Also,
gasoline  vapors represent a hydrocarbon  source which may  need
to be  controlled  to  help attain the ozone National  Ambient Air
Quality Standard.   EPA  has  not yet made  a decision to  propose
controls for  gasoline vapors or a  decision  on the  form  of the
control.
                             -53-

-------
 6.0   GAS  PHASE ORGANICS

 6.1   Formation and Control Technology

      Gas  phase  organics, or  volatile organic  compounds  (VOC),
 are  present  in  both  exhaust  and evaporative  emissions.   Over
 300  gas  phase organics  have been  identified.[59]   The majority
 of  VOC are formed from  incomplete combustion of the fuel while
 others are  simply evaporated fuel components.   The  emphasis of
 this   section  will  be  on  exhaust  compounds;   evaporative
 compounds are covered in Section 5 discussing gasoline vapors.

      The  total  mass  of  VOC  has  decreased  since  1975 due  to
 introduction   of  the   oxidation  catalyst   (caused   by  more
 stringent  HC  standards).   More  advanced  control  technology
 (i.e., 3-way  and 3-way  plus oxidation catalysts) introduced in
 the  early 1980's to  allow  further simultaneous  control  of NOx
 and HC has reduced the total mass even further.

 6.2   Composition

      The  majority of  gas phase  organics  consist of unsaturated
 and  saturated hydrocarbons  along with benzene,  alkyl  benzenes,
 aliphatic  aldehydes  and   a  variety  of  polycyclic  aromatic
 hydrocarbons  (PAH)  including  nitro-PAH.    A  list   of  the  VOC
 measured  in  ECTD  contractor studies  (that  include unregulated
 emissions characterization)  is given  in  Table 6-1.   It  should
 be noted  that  this  is not meant to be a complete list; instead,
 it represents those compounds for which the  most emissions data
 are available.

      Studies have been  conducted to determine  the  mass and the
 detailed  hydrocarbon  composition  of  vehicle  exhaust.[30,60]
 Gas  chromatographic  (GO analysis was used  in these studies to
 identify individual hydrocarbons of carbon numbers  1 through 10
 (Ci  - Cio).   In  a  recent  study,  82 individual  hydrocarbons
 and  10  aldehydes were  measured  in  the  exhaust  of 46  in-use
 1975-1982     gasoline-fueled     vehicles.[30]      The     gas
 chromatographic    conditions    employed    did    not    permit
 identification   of   each   individual   compound    above   Cio-
 Fortunately,  the  hydrocarbon composition  of  gasoline  engine
 exhaust consists  primarily  of components with carbon  numbers 1
 through 10  so a  fairly  complete  description of VOC  emissions
 from  gasoline-fueled  vehicles   is  possible  with GC  analysis.
 The  exception is  the  small  quantity of  PAH  which  is  also
 present in gasoline exhaust.   A  more detailed discussion of gas
phase PAH will be presented later in this  section.

     Unlike  gasoline-fueled  vehicles,  the  VOC  emitted  from
diesel-fueled vehicles   range  from Ci  to about  C40,   with  the
majority  being  below  C2S.    The  C^Cio  hydrocarbons  result
 almost  entirely  from the  combustion process,  which  involves
                              -54-

-------
                      Table 6-1

VOC Unregulated Emissions Most Commonly Characterized

                  Total  Hydrocarbons

               Individual Hydrocarbons

                       Methane
                      Ethylene
                        Ethane
                      Acetylene
                       Propane
                      Propylene
                       Benzene
                       Toluene

                Aldehydes and Ketones

                     Formaldehyde
                     Acetaldehyde
                       Acetone
                   Isobutyraldehyde
                 Methyl Ethyl Ketone
                   Crotonaldehyde
                    Hexanaldehyde
                     Benzaldehyde
                         -55-

-------
 cracking,  and  possibly  subsequent  polymerization,  of  higher
 molecular   weight  materials.    It  is   postulated   that  the
 Cio-Czs   hydrocarbons   result,   to   a   large   extent,   from
 uncombusted    fuel,    and   the   Cis-C40   hydrocarbons   from
 lubricants.[4]

     Gas  phase PAH have  been characterized  in  the  exhaust  of
 both  gasoline- and diesel-fueled  vehicles.[61,6]  Some  of  the
 PAH    identified   included   anthracene    and    phenanthrene,
 fluoranthene,  pyrene,  benzo(a)anthracene and isomers, chrysene,
 indenopyrene,         benzo(ghi)perylene,         benzo(e)pyrene,
 benzo(a)pyrene  and coronene.   These  compounds,  which  contain
 two to  six benzene rings,  have  also been  identified in diesel
 particulate   extracts.    With  the   exception   of    anthracene,
 phenanthrene  and possibly  pyrene   (the  data  for pyrene appear
 conflicting),  the  majority of  these  PAH are  emitted  in  the
 particulate    phase.     This    is    particularly    true   for
 benzo(a)pyrene.    Particulate/gas   phase   mass   ratios   for
 benzo(a)pyrene range from roughly 4:1 [61] to 15-27:1.[6]

 6.3  Mutagenicity of VOC

     Two   studies  were  conducted  to  determine  the  relative
 mutagenicity  of  the gas  and  particulate phases  in  the exhaust
 of  gasoline  and  diesel-fueled vehicles.   A  XAD-2  trapping
 system  was  used  in   both  studies  to  collect   the  gas  phase
 compounds  (greater  than  C7).   The  highly volatile  compounds,
 i.e.,  less  than C7,   can  not  be  collected on  a XAD-2 trap.
 Mutagenicity  of  the trapped  compounds  was determined  with  the
 Ames test  (Salmonella typhimurium).

     In   the  first   study,   conducted   with   gasoline-fueled
 stratified   charge  engine,  only   3   percent   of   the  total
 direct-acting  (i.e.,  mutagenicity  decreases  with  the  addition
 of S9  activation) and  5 percent of the  total  indirect-acting
 mutagenicity was  found  in the gas phase sample. [6]   Most of the
 gas  phase   indirect-acting  mutagenicity  was   found   in  the
 non-polar fraction which may be due to the presence of PAH.[6]

     In the  second study,  similar  results were  found.[62]  The
 mutagenic  activity  of the  gas  phase emissions  from the three
 gasoline-fueled  vehicles  tested were   at or  near  background
 levels.   For  the  single  diesel-fueled  vehicle   tested,  the
 mutagenic  activity of  the  gas phase  emissions,  expressed  as
 revertants   per   mile,   was  less   than  11  percent   of  the
 mutagenicity of the particle-bound organics.

 6.4  Risk Associated with Individual VOC

     For the purpose  of  this  report, available data on detailed
 hydrocarbon  emissions  were reviewed,  and  individual compounds
were selected  which have unit  risks associated with them.   Of
                              -56-

-------
the  gas  phase  organics  emitted  in  motor   vehicle  exhaust,
benzene,     formaldehyde,     1,3-butadiene,     ethylene     and
benzo(a)pyrene have unit  risks.   Benzene and  formaldehyde  have
been discussed in previous sections.

1,3-Butadiene

     It  is  difficult  due  to  the  limited  data  available  to
develop accurate  NEM  inputs for  1,3-butadiene.   Only data  for
light-duty gasoline-fueled vehicles  exist.  1,3-Butadiene poses
an  additional   problem   because  1,3-butadiene  and  n-butane
coelute  and  thus have  the same retention  point  on  the  gas
chromatograph.  Emission  characterization  studies to date  have
not attempted to determine  the percentage of  the  peak  due  to
1,3-butadiene.  Therefore,  assumptions  must  be  made  about  the
percentage  each  compound  contributes   to this peak.    In  a
previous study, it was assumed  that  20  percent of the peak  was
due to  1,3-butadiene,  for the purpose of  grouping the  detailed
HC  data  into various  compound  classes.[63]    Based  on  data
collected in  the  Lincoln  Tunnel,  1,3-butadiene constitutes 13.9
percent of  the total  peak,  although unknown  dilution with  air
containing    n-butane   and   1,3-butadiene    complicates   the
analysis.[64]    Morning    ambient   samples,    probably   daily
maximums,    collected   in   1984   and  1985   show  much   lower
percentages,  with overall  averages  of  5.08  and 4.21  percent,
respectively.[64]  For this  report,  fifteen percent  was  chosen
as an upper limit.

     1,3-Butadiene emissions were  expressed as a percentage  of
the total  exhaust HC  predicted  by MOBILES.   Based on  the data
from 46 in-use gasoline-fueled  vehicles  provided in reference
30  (and  the  fifteen percent   assumption),   1,3-butadiene  is
roughly 0.94  percent of the total FID exhaust  HC.    Due  to  the
lack of data  for  the  other vehicle classes, this percentage was
simply  applied to the  MOBILES  composite exhaust  HC  emission
factor.   It  was  further  assumed  that  the  percentage  would
remain  the  same  from  1986  to   1995.   Composite  1,3-butadiene
emission factors  for 1986  and  1995 are  0.0238-0.0263  g/mile and
0.0121-0.0143  g/mile,  respectively.   The  range  accounts  for
both  the  presence  and   absence  of  an  Inspection/Maintenance
program.   The speed  correction  factors  were  those used  for
formaldehyde.

     The  modified NEM model was used  to estimate  exposures.
Nationwide urban  and rural  exposure  in  1986  is estimated  to  be
0.69-0.76    ug/m3   and   0.32-0.35   ug/m1,   respectively.    In
1995,  nationwide  urban  and  rural  exposure is  estimated to  be
0.42-0.49  ug/mj and 0.19-0.23 ug/mj,  respectively.

     These  exposure  estimates   are  for  direct  emissions  of
1,3-butadiene and do not  account  for  reactions  of 1,3-butadiene
in  the  atmosphere.   Available   ambient  monitoring  data  were
                              -57-

-------
reviewed  and compared to  the exposure estimates.   A condensed
summary  of  1,3-butadiene  monitoring  data  is  given  in  Table
6-2. [64]   As seen in  the  table, average  mean values  in  urban
settings  range from  0.77-24.23 ug/m3.   The  1986  NEM estimate
of  urban  exposure  from motor  vehicles  (0.69-0.76  ug/m3)  lies
near  the  low  end of  the  range.  The  monitoring data indicate
that  1,3-butadiene  is,  in fact, emitted  from  motor  vehicles as
evidenced  by the elevated  levels  in tunnels.   The average mean
value  in  a rural  setting  (0.67  ug/m3)  is also within the same
order  of   magnitude   as  the  NEM  rural  estimate   (0.32-0.35
ug/m3).

     More  recent  1986   1,3-butadiene  monitoring  data are  now
available  for  18 cities.[65]   These  data  represent  6-9  a.m.
averages  of  about 10-15 samples for each  city,  measured during
the  summer  months.   The  average  concentrations  range  from
0.24-1.98  ug/m3  (Data  for  Los Angeles  were  not  obtained).
Again,  the  1986  NEM  estimate of  urban  exposure  from  motor
vehicles  lies within this range.

     The  EPA estimate  for  the upper confidence limit  unit  risk
for   1,3-butadiene  has  changed   substantially.    Based   on
inhalation studies of 1,3-butadiene  in  mice,  a 95 percent  upper
confidence  limit  unit  risk for 1,3-butaciiene  is estimated to be
2.8  x  10"4.[66]   This  is  much  greater  than  the  previous
estimated unit risk of 4.6 x  10~7.

     Estimates  of cancer incidence  for 1986  and  1995 are  given
in Table 6-3.  The total risk  in  1986  (with  the fifteen percent
assumption)  ranges  from 593-656 cancer incidences and drops to
391-460   cancer   incidences  in   1995.    Preliminary  emission
characterization    results     indicate    the    presence    of
1,3-butadiene,  but the  amount  has  not   yet  been  quantified.
Therefore,  a lower risk estimate  of  zero  will  also  be  used.
The resulting ranges  of cancer  incidences  for  1986  and  1995
given in Table 6-3 are 0-656  and 0-460, respectively.

Ethylene

     The  upper  confidence  limit unit  risk  of 2.7  x  10"6  for
ethylene  was provided in  the Six Month  Study although it  was
not developed by  EPA.   Unlike 1,3-butadiene, more extensive and
reliable  emissions data  exist to  construct  NEM  inputs  and
predict  resulting exposure  and risk  from direct  emissions  of
ethylene.   Ethylene  was  handled  in  much  the  same  way  as
formaldehyde.  Emissions were expressed  as  a  percentage of the
total exhaust HC  predicted by MOBILES.  Ethylene  in evaporative
emissions is negligible and was not considered.

     Ethylene  emissions,  expressed as  a percentage  of  total
hydrocarbons,  for the  vehicle  classes  in  1986  and  1995  are
given below:
                              -58-

-------
                           Table 6-2

       1,3-Butadiene Monitoring Data;  Condensed Summary
ug/m^
Average Mean Maximum
Location Value Value
Houston, TX area*
Los Angeles, CA area*
Riverside, CA
Atlanta, GA
Lincoln Tunnel, NY
Columbus, OH
Denver , CO
5.22
2.81
24.23
3.98
5.57
20.11
1.64
0.77
1.7
88.61
33.76
88.39
6.19

24.09
2.71
5.47
7.6
Comments
incl . tunnels
excl. tunnels



in tunnel
outside air

only one
Jones State Forest
0.67
2.41
sample set

non-urban
setting
     All  locations   within   approximately  15  miles   of  the
     metropolitan area.
                             -59-

-------
                           Table 6-3

            Annual  Cancer  Risk  from  Direct  Emissions
              of  1,3-Butadiene  from  Mobile  Sources

                         Urban      Rural      Total

1986
     With I/M            0-514       0-79      0-593
     Without I/M         0-568       0-88      0-656

1995
     With I/M            0-339       0-52      0-391
     Without I/M         0-399       0-61      0-460
                            -60-

-------
                                   % of Total Exhaust HC
Vehicle Class                      1986             1995

LDGV                                9.0              6.4
LDGT 1,2                            9.0              6.4
LDDV                               11.0             11.0
LDDT                               11.0             11.0
HDGV                               13.0             13.0
HDDV                                9.0              9.0

     Data  were taken  from references  30,   32-34,  36,  38,  and
40.  The percentages of  exhaust  HC for the  light-duty  gasoline
classes were projected to  decrease from 1986 to 1995 due to the
increasing  penetration  of  3-way  and  3-way   plus  oxidation
catalyst-equipped  vehicles and  trucks.   The  speed  correction
factors used to compute 10 mph and 35 mph emission  factors were
those  used for  formaldehyde.   Composite FTP emission  factors
for  1986   and  1995,  with  and  without  I/M,  are  0.2367-0.2607
g/mile and 0.0946-0.1092  g/mile,  respectively.

     Estimates of  cancer incidence for 1986 and  1995  are given
in Table  6-4.   As seen  in this table, the  total risk  in 1986
ranges from  55-60  cancer  incidences  and  drops  to  29-31  cancer
incidences in 1995.

     Two  important  limitations  need  to   be   mentioned  when
discussing the  risk  estimates for  ethylene.   The first  is the
fact  that  ethylene  is   photochemically  reactive.   The  risk
estimates  are  for direct emissions  of  ethylene  and do  not
account for reactions of  ethylene in the atmosphere.

     The  second,  and  most important,  limitation  is  the  unit
risk estimate.  The unit  risk  quoted  in the Six Month Study was
estimated by Clement Associates,  Inc.  for EPA but has  not been
endorsed  by  EPA.[67]    There  is  no  available  evidence  that
ethylene is carcinogenic  although  ethylene  oxide,  a metabolite
of ethylene,  has  been  shown to be  an animal  carcinogen.   EPA
has calculated a unit risk for ethylene  oxide.   Clement assumed
a  downward difference in  potency of  100 to  calculate  a  unit
risk for  ethylene.   The  basis  for this  assumed difference  in
potency was  not given.   Since there  is no  available  evidence
that  ethylene  is  carcinogenic,  the  risk  estimates  must  be
regarded as extremely tentative.   For  this  reason,  a lower risk
estimate of zero will also be used.

Benzo (a)  Pyrene (B(a)P)

     Gas phase B(a)P is  emitted  in small quantities.   The risk
posed  by  gas  phase  B(a)P  will  be  considered together  with
particle-associated  B(a)P.   It   will   be   assumed   that  the
particle-associated B(a)P  emission factor used  will adequately
represent what little gas phase B(a)P may also  be present.  The
risk from particle-associated B(a)P is discussed in Section 7.

                              -61-

-------
                           Table 6-4

            Annual Cancer  Risk  from  Direct  Emissions
                Of Ethylene  from  Mobile  Sources*

                          Urban         Rural          Total

1986
  With I/M                 0-47          0-8            0-55
  Without I/M              0-52          0-8            0-60

1985
  With I/M                 0-25          0-4            0-29
  Without I/M              0-27          0-4            0-31
   Lower limit of zero used due to the uncertainty of the unit
   risk estimate.
                             -62-

-------
6.5  Reactivity of VQC

     The  atmospheric  photochemical reaction products  of  mobile
source  volatile organic  compounds (VOC)  are  largely  unknown.
It  is  likely that  carcinogens  may  be  formed after  emissions
leave  the  vehicle  as  well  as  some  carcinogens  degrading  to
non-carcinogenic  compounds.   A  study was conducted by Calspan
Corporation   for  the   Coordinating  Research   Council,   the
objective  being to  investigate  the  fate  of diesel exhaust  in
the  atmosphere.[68]   Smog  chamber  experiments   with  diluted
diesel  exhaust  were  conducted to investigate parameters such as
the  presence  or  absence  of  UV  irradiation,  dilution  ratio,
aging  and exposure  to  ozone  and  N02.  The  presence   of  ozone
increased    the   mutagenic   response    of   the    collected
particle-bound  organics,  in some  cases  by as much  as an order
of  magnitude.   Similar  results  were  obtained  when NO2  and/or
irradiation  is  present  in  the chamber.   Thus,  it  appears  that
the mutagenicity  of  diesel particle-bound organics  is affected
by ambient conditions.

     It also appears likely  that  irradiation of  even  innocuous
VOC compounds  may lead to the formation  of  mutagens.   In  one
study,    irradiation    of   propylene,    S02   and   NOX   did
demonstrate  a mutagenic response.  [69]

     Mobile  source  VOC  is  known to  contain  photochemically
reactive   compounds.    The   percent   composition   of   exhaust
hydrocarbons  for  pre-1975, 1975-1980  and  1981-1982 model  year
gasoline-fueled  vehicles  is  given  in  Table  6-5.[30,63]   The
table  provides a general indication  of  changes   in reactivity
with  advancing control technology.   Pre-1975 vehicles  are  not
catalyst-equipped.  Model  year 1975-1980  vehicles  are  primarily
equipped  with  oxidation  catalysts  whereas  1981-1982  vehicles
are  primarily  equipped  with  3-way  and   3-way  plus  oxidation
catalysts.

     Olefins  and aromatics  are  all  fairly  reactive;  paraffins
and  acetylene are  less  reactive.   In  particular,  methane  is
essentially  non-reactive.   From  Table 6-5,  it can be  seen that
vehicles equipped with  catalysts,  particularly 3-way  and 3-way
plus  oxidation catalysts,  emit  a  higher  percentage of- methane
in their  exhaust  than do  non-catalyst-equipped vehicles.   Most
catalytic  converter  systems  preferentially  oxidize non-methane
hydrocarbons  because  methane  is  harder  for   the  catalytic
converter    to    oxidize.     Since   methane   is   essentially
non-reactive,  the  total   photochemical  reactivity  of  the  HC
mixture tends  to be reduced by  the  catalyst.   In addition,  the
catalyst  reduces  the  total  hydrocarbon  mass  and  generally
oxidizes  the unsaturated HC compounds to  a  greater extent than
the saturated compounds.   This  is evidenced by  the decreasing
percentages   of  olefins  and  aromatics  coupled   with   the
increasing percentage of paraffins in the exhaust.


                               -63-

-------
                           Table 6-5

          Percent Composition of Exhaust Hydrocarbons
Paraffins
(Methane)
Olefins
Acetylene
Aromatics
% of Total
Pre-1975*
41.0
(9.2)
30.0
8.8
20.4
Exhaust HC
1975-1980**
49.2
(9.3)
23.2
3.4
25.1
by Model Year
1981-1982**
66.5
(21.8)
15.2
0.9
17.4
 * Reference 63
** Reference 30
                             -64-

-------
     Aldehydes  were  not   included  .in  Table  6-5.    Aldehydes,
particularly  formaldehyde,  are  photochemically reactive.   The
catalyst has  also been found  to  reduce formaldehyde  and total
aldehyde    emissions.     Thus,     it    appears    that,    as
non-catalyst-equipped vehicles are phased out of the  fleet,  the
reactivity  of  vehicle exhaust  will  decrease.   The  reaction
products  of  mobile  source  VOC   to which  people  are  exposed
remains to be determined.

6.6  Current Activities

     EPA-ORD   is   conducting  a   long-term  research   project
referred to  as the Integrated Air  Cancer Project  (IACP).   The
goal  of  the   IACP   is   to  identify  the  principal  airborne
carcinogens and their sources.  Initially, airsheds with cne or
two  emission  sources  will be  examined,  followed  by areas  of
increasing complexity.

     In Phase  I,  initial  field measurements  were  conducted  in
Albuquerque,   New  Mexico   and  Raleigh,   North  Carolina.   The
purpose of  Phase  I  was  to  evaluate and  select the  sampling,
analytical   and   bioassay  methodologies,  as   well   as  other
approaches for an integrated field study.

     In Phase  II,  which will  be  conducted in FY86 and FY87,  an
integrated field  study will be conducted in  Boise,  Idaho.   The
two major emission sources will be  mobile sources  and emissions
from  wood combustion for  residential  heating.  Some   of  the
objectives  are  to:    1)   identify  and  quantify   classes   of
compounds  in the  ambient air resulting  from  residential  wood
combustion  and  motor  vehicles,   2)   quantify  the   relative
contributions  of  these  sources   to  the  mutagenic  activity,
organic   and  fine   particulate   mass   of   ambient   airborne
pollutants,  and  3) characterize the chemical changes  which  may
occur to the source  emissions  in  the atmosphere and  assess  the
resultant changes in mutagenic response.

     In conjunction  with  the  integrated  field study  in Boise,
separate smog  chamber studies will  be  conducted under  IACP to
assess  the mutagenicity  of  vehicle exhaust  before  and after
irradiation.    These  results  could provide  at least an  initial
indication of  the  photochemical transformation  of  mobile source
emissions, both  gas   phase  and  particle-associated,  and  the
resulting effect on mutagenicity.
                              -65-

-------
7.0  ORGANICS ASSOCIATED WITH NON-DIESEL PARTICIPATE

7.1  Emission Rates and Composition

     This   section  will   deal  with  the   organic  compounds
associated with gasoline particulate emissions.

     Table   7-1   presents  total  particulate   and  associated
soluble  organic  fraction  (SOF)  emissions   for  the  different
vehicle  classes.[2,70]   Gasoline-fueled vehicles emit  far less
particulate  than  their  diesel  counterparts.   As  discussed  in
Section  2.1,  it  is  thought  that  nitro-polycyclic  aromatic
hydrocarbons  (PAH)  such  as  nitropyrenes,  dinitropyrenes  and
nitrohydroxypyrenes   together   account   for   much   of   the
mutagenicity  of   diesel   particulate  emissions.    Particulate
emissions  from  gasoline-fueled  vehicles  contain  significantly
less of  these nitro-PAH's; however,  as seen  in  Table  7—1,  the
mutagenicity of gasoline SOF, expressed as  reverents/ug SOF,  is
greater  than   diesel   SOF.    Also,  unlike   diesel  SOF,  the
mutagenic activity  of  gasoline  SOF increases with  the  addition
of  S9  activation,  indicating  indirect-acting activity.   This
suggests that  the classical  PAH's  may be  responsible  for  the
mutagenicity of gasoline SOF,  rather than the nitro-PAH's.

7.2  Risk from Non-Diesel Particle-Associated Organics

     For  gasoline-fueled vehicles,  three different approaches
were   taken    to   estimate    the    risk    from    non-diesel
particle-associated  organics,  referred  to  here  as  gasoline
particle-bound  PIC  (products  of  incomplete  combustion).   The
first  approach  estimates  the   risk  of  B(a)P  emissions  from
gasoline-fueled vehicles and assumes no risk  from  the remaining
gasoline  particle-bound  PIC  emissions.   The  second  approach
uses   B(a)P   emissions    as    a   surrograte   for   gasoline
particle-bound PIC  emissions.   Unlike the  first approach, which
uses the unit risk  for B(a)P,  the second approach  uses  the  PIC
unit risk presented  in the Six  Month  Study.   The third approach
uses estimated  gasoline  particle-associated  organic  emission
rates    together    with   a   unit    risk    for    gasoline
particle-associated  organics.    All   three   approaches   and
resulting risk estimates will  be described in this section.

     The  first   approach   uses B(a)P  emission  factors  from
gasoline-fueled vehicles  together  with the  B(a)P  unit  risk.
The  annual  cancer  risk  of  B(a)P from  gasoline-fueled  vehicles
was  determined  by multiplying  the  B(a)P  risk  obtained  in  the
Six  Month  Study  by the  ratio  of the emission factors  (this
                              -66-

-------
                                Table 7-1

                 Particulate Emissions and Mutaqenicity*
Total particulate,
  mg/mile

SOF, mg/mile

SOF, as % of total
  particulate
Nitropyrene, ug/mi
TA98, +S9, rev/ug SOF
Leaded
LDGV
102.5
21.1
1
20.6
g/mile 14.6
i 0.20
SOF** 7.3
SOF 12.5
(x!03) 152
(xlQ3) 258
Unleaded
LDGV
31.7
14.4
45.4
3.2
0.24
7.6
13.4
42.1
79.3
Leaded
HDG
735
27.6
3.8
39.5

5.3
16.1
110.5
428.3
LDP
606.8
124.1
20.5
4.5
7.4
4.1

509

HDD
1948
385.2
19.8
2.3

0.88
0.84
287.4
279.4
**
For the light-duty vehicles, emissions were collected during the
FTP; for heavy-duty vehicles, the transient test procedure was used,
Denotes revertants per microgram of the soluble organic fraction
(SOF), using Salmonella typhimurium strain TA98 with or without
metabolic activation  (S9) .
                                   -67-

-------
study/Six  Month Study).*  This  approach  assumes that  there  is
no risk from the remaining gasoline particle-bound PIC.

     The  B(a)P emission  factors were  taken from  references  2
and 70 and summarized in  Table  7-1.   B(a)P emission factors can
vary substantially  from one  reference to another.  The emission
factors used  represent the    higher  end  of the  range.   Since
gas  phase B(a)P  is  emitted  in such  small quantities,  it  is
assumed  that   the  emission  factors  used  adequately  represent
what little  gas phase B(a)P may also be  present.   The unleaded
and leaded LDGV weightings  were 0.88 and  0.12,  respectively  to
compute a  composite  LDGV emission factor.   These weightings are
based  on  MOBILES  estimates  for 1986.   The resulting  overall
composite  emission  factor   is  5.56  x   10 ~6   g/mile.    The
exclusion  of  diesel-fueled vehicles  does  not have much impact
on  the composite  emission  factor  since  most  of  the  B(a)P  is
emitted from gasoline-fueled vehicles.

     The  composite emission factor  of  5.56 x  10~6   g/mile  is
roughly one-third  of the emission  factor used  in  the  Six Month
Study  (1.66  x  10"sg/mile).    The  primary reason   for  this
difference is  that  the  Six Month  Study assumed  a  50/50 split
for leaded and unleaded  gasoline-fueled vehicles.   A  B(a)P unit
risk of  3.3  x  10~J  was  used  for  both the  Six  Month  Study and
this approach.

     The B(a)P risk from motor vehicles  estimated  in  the  Six
Month  Study  is 0.02 cancer  incidences per  urban  million.   The
current  (1986)  cancer   risk   of   B(a)P   from   gasoline-fueled
vehicles  using the first   approach  was  determined   by  simply
multiplying the risk  obtained  in  the Six  Month  Study  by  the
ratio  of   the  composite  emission  factors  (5.56  x 10"6/1.66  x
10~s = 0.33).   The  resulting  annual  cancer  risk  is  0.007  per
urban  million, or  1.3  cancer incidences,  assuming  an  urban
population of  180  million in 1986.

     In  1995,  virtually  all  LDGV  and  HDGV  will   be  using
unleaded  fuel.   The   composite  B(a)P  emission  factor  will
decrease  as  a result  to roughly  2.95 x  10~6  g/mile.   When
this emission  factor  is  used, the  resulting   cancer risk  is
0.004 per urban million,  or  0.78 cancer  incidences,  assuming  an
urban population of 195 million  in 1995.
*    This  is  the  first place in the report where  the  Six Month
     Study  exposures  are  used   in   place   of   NEM-predicted
     exposures.   The  Six Month  Study  used the  GAMS  dispersion
     model  developed  by  EPA's  Office  of Toxic Substances  to
     calculate exposures.   Area  sources were  generally  assumed
     to  be distributed  equally  throughout each county.   As  a
     very  crude  comparison,   assuming  the  same  FTP  emission
     factor, the NEM-predicted exposure  is roughly  1.6 times as
     high as that predicted by GAMS.
                              -68-

-------
     The  second  approach  uses  B(a)P  emission  factors  from
gasoline-fueled vehicles together with the  PIC unit risk (which
is  expressed  per unit  of exposure  of B(a)P)  to  estimate  the
annual  cancer  risk of  PIC  from gasoline-fueled  vehicles.   The
PIC unit  risk  is 4.2 x 10"l.   The  Six  Month  Study also  used
B(a)P  emission  factors together  with  the  PIC  unit  risk  to
estimate the PIC risk  from motor  vehicles.  The  annual  cancer
risk  of PIC  from  gasoline-fueled  vehicles  was  determined  by
multiplying the PIC risk obtained in the  Six Month Study  by the
ratio of the B(a)P emission factors.

     The  PIC  risk  from motor  vehicles   estimated  in  the  Six
Month Study is  2.07 per  urban million.  When  this is multiplied
by  the  ratio  of the  B(a)P  emission factors  (5.56 x 10"s  /1.66
x 10"s  =  0.33), the  resulting  annual cancer  risk is 0.68  per
urban million,  or  122   cancer  incidences  in  1986.   When  the
estimated  1995  B(a)P emission  factor is  used  (2.95  x  10"6),
the  resulting  cancer  risk is  0.37  per  urban  million,  or  72
cancer incidences in 1995.

     The  third  approach  uses  estimated  emission rates   of
gasoline particle-associated organics as an unspeciated  mixture
together with  a unit risk for  these organics.   Exposures  were
estimated using the modified NEM model.

     Emission rates of  gasoline particle-associated organics  in
1986 and 1995 were estimated using the methodology in reference
71.    Unfortunately,  the organic  emission  factors  contained  in
reference  71  include  both  soluble  organics   and  elemental
carbon.    The  organic   emission  factors   in  reference  71  were
revised to  include only the  soluble organic  fraction  (SOF).
The SOF emission factors used for  the various vehicle  classes
are given  in Table 7-2.   SOF  emission  factors  for  light-duty
gasoline-fueled vehicles and  trucks were taken from references
70  and   72.     SOF  emission   factors   for   the   heavy-duty
gasoline-fueled trucks  were  taken  from  references  73,  74  and
75.    The  pre-1987  heavy-duty  data  were  obtained using  EPA's
transient  chassis test  procedure.   It is assumed  that the  more
stringent   emission  standards  for  1987   and  later  heavy-duty
gasoline vehicles in the 8,501-14,000  Ibs  range (classes  2B and
3) will  require the use  of catalysts.  The  SOF  emission-factors
in Table 7-2 differ somewhat from  the emission factors  in Table
7-1.  The  reasons  for   this  difference   are:   1)  the  emission
factors  in Table 7-2  are  specific  by model  year unlike  those in
Table 7-1, and 2) more  references  were used in compiling  Table
7-2.

     MOBILES VMT  fractions  for  1986 and  1995 for  the  various
vehicle   classes  were  used  to   calculate   composite  emission
factors.   Composite emission  factors were estimated  assuming
both  the  presence  and  absence  of  an   Inspection/Maintenance
(I/M)  program.   (The effect of the  I/M program is to reduce the
rates  of  misfueling   and   catalyst  removal,    which   affect
particle-bound  organic  emission  rates.)   An  FTP  average  speed
                             -69-

-------
                             Table 7-2

       Gasoline Particle-Associated Organic  Emission Factors
              Light-Duty Gasoline Vehicles  and Trucks
Model Year

Pre-1970
1970-1974
1975+
1975+
1975+
1975+
Control
System

NOCAT
NOCAT
NOCAT
CAT
CAT
NOCAT
 Fuel
 Used

Leaded
Leaded
Leaded
Unleaded
Leaded
Unleaded
Emission Factor
grams per mile

   0.0873
   0.023"
   0.019b
   0.005b
   0.023C
   0.019d
                    Heavy-Duty Gasoline Trucks
Model Year

Pre-1987
1987+
1987+
1987+
Control
System

NOCAT
CAT
CAT
NOCAT
 Fuel
 Used

Leaded
Unleaded
Leaded
Leaded
Emission Factor
grams per mile

   0.072"
   0.013f
   0.0509
   0.105h
a   From reference 72
b   From reference 70
c   Same as 1970-1974 no catalyst leaded value.
d   Same as 1975+ no catalyst leaded value.
e   From references  73,  74  and  75.   The  average  class  2B truck
    value was assigned a weighting of 60% based on VMT.
f   Average  class  2B no  catalyst  leaded  truck  value multiplied
    by ratio of  light-duty 1975+ catalyst unleaded and  1975+ no
    catalyst leaded values (0.050 x 0.005/0.019).
g   Average class 2B no catalyst leaded truck value.
h   Average no catalyst leaded truck value for classes 5 and 6.
                               -70-

-------
of  19.6  mph   was   assumed.    The  estimated  fleet  composite
emission factors are given below.

    Gasoline Particle-Associated Organic Emissions  (q/mile)

                       With I/M                Without I/M

     1986               0.0075                   0.0082
     1995               0.0048                   0.0058

     Based on  data  obtained over a number  of  different driving
cycles,  no   trends   are  apparent  with  regard  to  speed.[70]
Therefore,  the  g/mile  emission  factors   above  were  used  to
calculate g/min emission  factors  at  10  mph,   19.6  mph  and  35
mph, as  required by the  model.   Idle emission factors were also
determined   based    on   data   for   non-catalyst-equipped   and
oxidation   catalyst-equipped   vehicles.[40]     Idle   emission
factors  for  1986  and 1995  are  0.0196 g/min  and 0.0138  g/min,
respectively.

     A  unit  risk  estimate  for  gasoline  particle-associated
organics  was estimated  by  EPA scientists  in  1983. [7]   Like
diesel  particulate,  the   unit  risk   estimate  has  not  been
reviewed by  the Science Advisory  Board (SAB)  and  an  official
EPA risk assessment  has  not been done.  It is  based on data for
only one catalyst-equipped vehicle.   Furthermore,   the  vehicle
had exceptionally  high  exhaust  emissions, comparable  to those
from a non-catalyst-equipped vehicle.  It  was  originally chosen
on this  basis  since it  was easier  to collect  enough extractable
organics  for   analysis.   The   mutagenic  activity  of   the
particle-associated organics from this vehicle,  as  indicated by
the Ames Salmonella strain TA-98 bioassay,  is on the low end of
the   range,    when   compared   with   other   catalyst-equipped
vehicles.  As a result,  the vehicle should be considered to  be
of  uncertain  representativeness.   An upper  confidence  limit
unit risk estimate  based on  this  vehicle is  2.5  x  10"4.   The
bioassays used  to estimate  the  unit risk were the same as those
used to  estimate the unit  risk  for diesel particulate,  and the
same approach was  used.

     Nationwide urban  and  rural  exposure in  1986, using the
modified  NEM,   is   estimated   to  be   0.20-0.21    ug/m3   and
0.17-0.19 ug/m3,  respectively.   The  range accounts for  both
the presence and absence of an  Inspection/Maintenance  program.
In 1995, nationwide  urban  and  rural exposure  is estimated to be
0.13-0.15 ug/m3 and 0.11-0.13 ug/m3, respectively.

     Estimates   of lung  cancer incidence for  1986 and  1995 are
given in Table  7-3.  The total  risk in 1986 ranges from 163-176
cancer  incidences  and drops  to  115-136  cancer  incidences  in
1995.

     For this  report,  a  range  of  risk estimates  for  gasoline
PIC will be  reported,   which  encompasses  the  results   of  all
three approaches.    The  resulting range of  cancer  incidences is
1.3-176 in 1986 and 0.78-136 in 1995.
                              -71-

-------
                            Table  7-3

        Annual  Lung Cancer  Risk from Gasoline Particle-
                       Associated  Orqanics

                          Urban     Rural     Total

1986
     With I/M              1.3-127   0-36      1.3-163
     Without I/M           1.3-136   0-40      1.3-176

1995
     With I/M              0.78-90   0-25      0.78-115
     Without I/M           0.78-106  0-30      0.78-136
                              -72-

-------
8.0  DIOXINS

8.1  Composition

     Over  75  different  chlorinated  dioxin  isomers  have  been
identified.  One  of  the 22 isomers with four chlorine  atoms is
2,3,7,8-tetrachlorodibenzo-p-dioxin  (2378-TCDD).   This  dioxin
compound  has  a  high  molecular  weight   and   exists   in  the
particulate  state or is  adsorbed  onto  particulates.   It is the
dioxin compound of most interest since it  is thought to be the
most  toxic  of  the  chlorinated  dioxins  and  is  most  often
associated with  exposure and potential  health  risks to humans
based on available data.

8.2  Emissions

     Some qualitative analytical measurements have found dioxin
to be present in the muffler  scrapings  of  vehicles using either
leaded  or  unleaded  gasoline.   Also,  some  measurements  have
shown this compound  to  be present  in diesel  particulate.   Some
work  has   been   done  attempting to   measure  this  compound
quantitatively  in vehicle  exhaust.   Since  it  is  apparently
emitted  in  only  trace quantities (e.g.,   10"9  g/mile),  it  is
very  difficult   to   collect  enough  particulate   sample  for
analysis.[76]  Also, a  radioactive tracer  compound must be used
to correct for losses in sample work-up.

8.3  Concentrations of Dioxins

     Most  reports of environmental  contamination from dioxins
have  concerned  non-air releases  of dioxin from  industrial  or
chemical processes  (e.g., pesticide production)  or  measurement
of  dioxin  levels  in  fish  in waters  that  could  have  been
contaminated by non-air routes.   At  this  point,  there  are  no
accurate  measurements  of  ambient levels  of  dioxin  since  it
would be present  in  such  low levels in the ambient  air.   Thus,
it is not  possible to  say for certain whether the known sources
of dioxin such as municipal  and industrial incinerators account
for most of  the dioxin  found in the ambient air  although it is
thought that such sources account  for a  significant  fraction of
dioxin emitted.   A  recent EPA OAQPS report states that,  due to
the low chlorine level (0-100 ppm)  in gasoline  and diesel fuel,
mobile sources may not be a significant source of dioxin.[76]

8.4  Current Activities

     OAQPS plans  to  make a decision within the   coming year on
whether  to  list  dioxin  as  a  hazardous   air  pollutant  under
Section 112  of  the  Clean Air Act.   Also,  OAQPS  has  recommended
that  QMS  consider a  program to  quantitatively measure dioxin
emissions  from  mobile  sources.   QMS  is  presently  considering
whether work  in  this area  would  be  useful and  has asked  for
some additional OAQPS input  on  the relative priority of work on
dioxins from mobile sources versus  other sources.
                              -73-

-------
9.0  ASBESTOS

9.1  Emissions  and Ambient Concentrations

     Asbestos,  due  to  its  high  friction and  heat  resistance
characteristics,  is  used  in brake  linings,  clutch facings  and
automatic  transmissions.  About 22%  of  the total asbestos  used
in 1984  was used in  motor vehicles.  Previous EPA work  involved
measurement  of  asbestos  emissions  from vehicles during  typical
vehicle  operations  for  a  vehicle  with  front  disc  brakes  and
rear  drum brakes. [77,78]   This work  showed  that  the  asbestos
emissions  were  4  ug/mile and  possibly  as high  as 28  ug/mile.
It is  estimated that the 4 ug/mile  emission figure would result
in a  maximum annual average asbestos  level  in  a  central  city
area  of  about  0.25  ng/m3.[79]   This is  in agreement with  what
would  be predicted  in  a typical  street  canyon,  using a method
developed    by    Southwest    Research   Institute   under    EPA
contract.[80]   Urban asbestos levels from  all sources in  large
cities   show average   readings  of  2.6-5  ng/m3  but   average
levels   in   New York  range  from   8  to   30  ng/m3;  localized
asbestos  levels in dense traffic can be somewhat higher.   Thus,
asbestos  from  mobile  sources  is  responsible  for  1-10%  of  the
total  asbestos  although a higher  fraction  (7-70%)  could result
with the high emission factor.

9.2  Cancer  Risk

     The  National   Academy  of  Sciences   (NAS)  has  estimated
lifetime  risks  for  persons  in  urban areas.[81]  Based  on  the
data  in  the NAS  report, unit  risk estimates range  from 6.6 x
10~7  to  2.6 x 10~5  per  ng/m3  exposure   (6.6  x  10~4 -  2.6
x  10~2  per  ug/m3) .    On an  annual  basis,  this  translates to
9  x  10~9  to 3.6  x  10~7 per  ng/m3 exposure  (NAS  assumed an
average lifetime of  73 years).

     Maximum annual  average  asbestos levels in  urban  areas due
to motor  vehicle  emissions are  estimated  to  range  from  0.25 to
1.75  ng/m3,  based  on  estimated   emission  rates  of 4 to 28
ug/mile.   Using these  exposure estimates  in  conjunction  with
the range  of unit  risks, and  assuming  an  urban  population of
180,000,000,  the  resulting  cancer  risk  is estimated to  range
from 0.405-113.4 cancer  incidences per year.

9.3  Current Activities

     Due  to  the risk  to the  general population as  well as the
risk   to  workers  exposed   to   asbestos,   the  EPA  Office  of
Pesticides and  Toxic Substances, in January 1986,  has proposed
regulations  under  Section  6 of TSCA  to ban  certain  uses of
asbestos  (i.e.,  asbestos-cement  pipes,  flooring  tiles,   and
asbestos  clothing) and  to allocate  permits to mine  and import
asbestos  which   would  restrict its remaining  uses.[79]   The
                              -74-

-------
National  Resources  Defense Council  petitioned EPA  in 1984  to
ban the  use of asbestos in motor vehicles.  However,  EPA feels
that  effective substitutes are  still  not available  for  many
applications   of  asbestos  in  brakes.    EPA   feels  that  the
restrictions on use of  asbestos  will encourage the  development
of  suitable  replacements  for   brake  linings.    EPA  is  also
considering a  ban on asbestos friction  products  about 5  years
after    the    final    rules   are    promulgated.     Presently,
semi-metallic disc brake pads  made without asbestos  are used in
about 85%  of  new domestic  cars  with front wheel  drive.   Also,
some  manufacturers  are  introducing  aramid  fiber  instead  of
asbestos   for   disc   brakes.    Development   of   non-asbestos
substitutes for drum  brakes has  not  yet  been  as  successful  as
for  disc  brakes  due  to  concerns  about durability  and  heat
resistance.    Still,  one  automobile  manufacturer   is   using
semi-metallic  drum brake  linings  in  its  new minivans.  Another
has reported progress  in developing  a drum brake  lining  with
aramid fiber.

     The proposed regulations would  involve EPA  issuing permits
for mining of  asbestos  in the U.S.  and importation  of  asbestos
and its  products.  The  total amount  of asbestos  to be permitted
under these permits for the first  year after  the  regulation  is
promulgated would be  30%  of  the  average amount  of  asbestos
mined or  imported yearly  during the  base period of  1981,  1982,
and  1983.   The amount  of the  asbestos used  in  the  following
years would decrease annually by  3% from the 30% level  until  it
reached  a 3%  level in  the tenth year  of the regulation  after
which no  asbestos could be  mined  or   imported.   Recent  findings
that  there are not  yet  available  good substitutes to  replace
asbestos  in certain  automobile  and  truck  brakes,  however,  may
push  back  EPA's  goal  for banning  and phasing out  asbestos for
such uses.
                              -75-

-------
10.0 VEHICLE  INTERIOR EMISSIONS

10.1 Composition and Concentrations

     Two   different  EPA  projects   were   conducted  measuring
different  compounds in vehicle interiors.[82,83]

     One project  showed  vinyl chloride to be present  at levels
ranging  from below 2  ppb to  7  ppb  in  several  closed vehicles
under  high temperatures  as  might exist  on a  summer  day.[82]
Under  similar  conditions  other  carcinogenic  compounds  were
identified  qualitatively.   The carcinogenic  compounds detected
are listed below.

                             aniline
                            biphenyl
                        1,2-dibromoethane
                         dichlorobenzene
                         dimethylphenol
                         isobutyl alcohol
                        maleic anhydride
                           naphthalene
                             benzene
                      carbon tetrachloride
                           chloroform
                             phenol

A total of 147 compounds were identified.

     In another EPA project,  58  vehicle interiors  were sampled
for  nitrosamine which   is  a potent  carcinogen.[83]   The  main
compound measured  is  N-nitrosodimethylamine which was  found at
levels  of   about   0.05   ug/m3.    This  exposure  level  for  3
hours/day  which would represent  a  long commute  is  less  than
that from a can of beer or a strip of bacon a day.

     Also,   from  time  to time, complaints have  been received by
EPA  on  the  formation   of  a  white  semi-opaque  film  on  the
interior  of  a vehicle  windshield.   This  film very  slightly
obstructs  visibility  and can be  somewhat  difficult though  not
impossible  to  remove.   EPA  ORD  has  analyzed  samples  of  this
film  and  finds  it to   consist  of  a phthlate  ester;  dioctyl
phthlate   is  used as   a  plastisizer  inside  vehicles.    The
exposure level to  and the health  effects  of this  compound  are
not known.

10.2 Cancer Risk and Current Activities

     At this  point, no  risk assessment calculations  have  been
made for these vehicle  interior  emissions  (with the  exception
of  some  simple calculations  for  the nitrosamines).   Since  the
exposure level is  low,  it is  thought that no  significant  risk
could  be   occurring  due  to  exposure  to these  substances.   No
further work is planned to identify  other  substances present in
vehicle interiors at this time.

                              -76-

-------
11.0 METALS

     This  section  discusses  metals  which are emitted in vehicle
exhaust.    Lead,   manganese  ,   platinum   and   cadmium   are
considered.   For  each  metal,  available  information  regarding
the source of  the  metal,  emission rates and  health  effects are
given.  In addition, for cadmium, a risk estimate is presented.

11.1 Lead

11.1.1  Source and Emission Factors

     Lead,  in  the  form  of tetraethyl lead and tetramethyl lead,
is  used  in gasoline  fuel  to  increase  the octane  number  of
gasoline   and   suppress   knock.    Knock   is   the   premature
autoignition and very rapid  combustion  of the fuel-air  charge
in the  engine  combustion  chambers.   Knock is frequently audible
as  a  sharp metallic  rap,  and it may  cause  damage  to  engines.
Lead  alkyl compounds  were first added to  gasoline  in  1923  as a
means of  suppressing  engine  knock by promoting uniform burning
of the fuel-air mixture in the engine combustion chambers.

     When  lead  alkyl  compounds  are  burned,   lead  oxide  is
formed.   To reduce the tendency of  lead  oxide to  build deposits
in  automobile  engines,  halogen  compounds are  included  in  the
fuel  to  scavenge the lead  deposits from  the engine.   Ethylene
dichloride  and ethylene dibromide  are  commonly used.   Analysis
of  lead  particulate indicates that  most of  the  exhausted  lead
appears as PbClBr.[71]

     Vehicles  equipped  with  catalytic  converters  are  required
to use unleaded gasoline since the presence  of  lead poisons the
catalytic    converter.      Catalyst-equipped   vehicles    were
introduced  in  1975.    Unleaded  gasoline  is  subject  to   the
requirement that it not contain any lead  additives  and that it
not include more than 0.05 grams per gallon (gpg)  lead.

     Due  to the  widespread  and persistent  misuse  of  leaded
gasoline  in vehicles designed  for unleaded gasoline  (termed
misfueling or fuel  switching) and the adverse health effects of
lead  in  gasoline,  EPA promulgated regulations  to reduce  the
lead content of gasoline.

     In 1973,  EPA  required  refiners to  meet  a 0.5  gpg  standard
for  the  average  lead content  of all   gasoline.   EPA  later
replaced this  standard with  a standard  for the lead content  of
leaded  gasoline   only.   Effective  November  1,   1982,   large
refineries were  required  to  meet a  standard of 1.10 grams  per
leaded gallon  (gplg).   Certain smaller  refineries  were subject
to a  1.90  gplg standard until July 1,  1983,  at which  time  they
would  also be  subject  to  the  1.10   gplg   standard. [84]   EPA
                              -77-

-------
further  reduced  the   standard   to   0.10  gplg   effective   on
January 1, 1986, with  an  interim standard of 0.5 gplg effective
on July 1,  1985. [85]   EPA is now investigating  the  possibility
of  regulatory  action  banning the  use of  leaded gasoline.   A
range of alternatives  is  being considered,  ranging from  no  ban
to  a  ban  in  1995  to  a  ban effective as  early  as  January 1,
1988.[86]

     Lead emissions from mobile  sources are  calculated based on
the percentage  of  burned lead  exhausted  at different  speeds,
the  lead   content  of  gasoline,  vehicle  fuel economy and  the
model year mix  of  vehicles on  the  road.   A report  containing
guidelines    for    predicting    lead   emission   factors    is
available.[87]   Based  on   the   information  in  this  report,
simplified  lead  emission   factors  have  been   estimated   for
light-duty  and  heavy-duty  model year  1986   (MY1986)  vehicles.
These emission factors are summarized in Table 11-1.   Note  that
these estimates  are  for the mass of lead alone.   To  account for
the  halogens,   these  lead  emission   estimates   should   be
multiplied by 1.557.[71]

     Composite  calendar  year  1986   emission  factors  will  vary
from area  to  area  depending on  the  model  year  mix  of vehicles
on  the  road,  the  driving  conditions   (which  will affect  fuel
economy) and  the misfueling rates  in   a  particular   area.  Lead
emission  factors for  heavy-duty gasoline-fueled vehicles  are
projected to  decrease  in  later  years  since  it  is assumed  that
emission standards effective in  1987 will  require virtually all
new heavy-duty  gasoline-powered vehicles under   14,001  pounds
gross vehicle weight  to  use  catalytic  converters and  thereby
burn unleaded fuel.

11.1.2  Health Effects

     A  strong   correlation   has   been  demonstrated  between
gasoline  lead  usage  and blood  lead  levels.   The Centers  for
Disease Control  (CDC) has  defined 25 ug/dl as an  elevated blood
lead level.   The list of demonstrated health effects  at blood
lead levels exceeding  30  ug/dl  is  well established.[88]   Such
effects include:   l)  death  at  blood  levels of  80+  ug/dl;  2)
frank  anemia,  anorexia,  abdominal   pain,  and  vomiting  at  70
ug/dl;   3)  reduced  hemoglobin  at 40 ug/dl and  interference in
heme synthesis  at  levels  down  to 15-20 ug/dl; and 4)  vitamin D
metabolism interference at PbBs possibly as low as 12 ug/dl.

     In  addition,  there   is   a great  deal  of  evidence  on
neurological  effects   in  children.   At  levels   of  30+  ug/dl,
studies have  found significantly slowed  nerve  conduction,  fine
motor dysfunctions,  impaired concept  formation,   lower  IQ,  and
altered behavior  among pre-school children.   At  PbBs as  low as
15-30 ug/dl, a number of  studies suggest  possible small effects
on  IQ,   behavioral  dysfunctions  (attentional   deficits,  poor
classroom  behavior),  and  changes   in  electrical  brain  wave
activity and  hearing function among children.   Children appear
                              -78-

-------
Vehicle

LDGV catalyst
LDGV cat (misfueled)
HDGV
LDGV non-cat**
                                       Table 11-1

                        Lead Emission Factors  for MY1986  Vehicles
 1986 Fuel
Economy mpg*

    23.8
    23.8
     9
    13
   1986 Lead
Fuel Content qpg

      0.014
      0.10
      0.10
      0.10
Fraction of Lead
 Burned That is
   Exhausted

      0.75
      0.44
      0.75
      0.75
 MY1986 Lead
Emissions gpm

    0.0004
    0.002
    0.008
    0.006
     Combined city/highway fuel economy.

     This estimate is  for  a non-catalyst-equipped vehicle for  comparison to  the  1986
     fleet.   Fuel  economy  was  based  on data  for pre-1975  light-duty  gasoline-fueled
     vehicles.
                                          -7Q-

-------
to be  particularly susceptible to  adverse  health effects  from
lead  exposure.    Recently  emerging  data from  several  ongoing
longitudinal studies consistently indicate  that  fetal  exposures
to  lead,  at maternal  or umbilical  cord PbB  levels as  low as
10-15 ug/dl are  associated with  reduced birth weight  and early
growth,  and  delays  in  early  mental,  motor,   and  emotional
development.

     In addition, more  recent  data  provide  convincing evidence
for  strong associations  between blood pressure increases  and
blood lead levels, even at  blood lead levels below 30 ug/dl.

     It is difficult to conclude what role lead may  play in  the
induction  of   human   tumors.    Epidemiological    studies   of
lead-exposed  workers  provide  no  definitive  findings.   Lead
acetate has produced renal  tumors  in  some  experimental animals
but does not seem to be a potent carcinogen.[89]

11.2 Manganese

11.2.1  Source and Emission  Factors

     A nonleaded organometallic  additive  which was  used widely
in the middle 1970's as  an  octane improver  in unleaded gasoline
is methycyclopentadienyl manganese tricarbonyl (MMT).

     In  September,  1978,   EPA  banned   its  use   in   unleaded
gasoline because  of evidence  which showed  that MMT  increased
hydrocarbon emissions and plugged catalysts.[90]   This  ban  was
temporarily suspended  for  four  months  (June  through September)
in 1979 because of  concerns  over potentially short  supplies of
unleaded gasoline.[91]   MMT has  always been allowed  in leaded
gasoline,  but has been  used  at  only low  levels due  to  its  high
cost compared to lead additives.

     More  recently,  a  particulate  trap  regeneration  system
based on use of  a manganese  (Mn) containing fuel additive  has
been  selected  by  Volkswagen as  a  means of  meeting  the  1986
California  and  1987  Federal   light-duty   diesel   particulate
standard of  0.2  g/mile.   EPA  has  given Volkswagen  temporary
approval to use  this additive for the 1986-1988 model years.

     Volkswagen   estimates  that  for highway  driving,  emission
rates will range from  an expected value of  1.38 mg Mn/mile to a
worst-case value of 4 mg Mn/mile.  For  street  driving,  expected
emissions are 3.13  mg  Mn/mile  with  a worst-case value  of 10 mg
Mn/mile.[92]    The   worst-case   numbers   are  based    on   the
assumption that  all  of the  manganese is  emitted  in  the exhaust
while the expected numbers reflect  expected  manganese  retention
in the engine and  particle  trap.  Data are  presently lacking on
the size distribution  of particles  emitted  in the  exhaust  and
the amount and species  of manganese  they contain.
                              -80-

-------
     The  California  Air Resources  Board  (GARB)  estimates that,
by 1990,  between  2.6 percent and 6.6 percent of  the California
vehicle  fleet will  have  traps.   Using  this  estimate  of trap
usage,  ambient   Mn   concentrations  corresponding  to  highway
emissions of  1.3  - 4 mg Mn/mile are estimated to be 0.03 - 0.24
ug  Mn/m3.    Ambient  concentrations   corresponding  to  street
canyon  emissions  of  3.13  - 10 mg' Mn/mile are estimated  to be
0.06 - 0.5 ug Mn/m3. [92]

11.2.2  Health Effects

     In  June,  1984,   Volkswagen   requested   that  the  Health
Effects  Institute  (HEI)  undertake  an  evaluation  of  potential
health  issues  related  to  emissions  from  diesel  automobiles
using  manganese-containing  fuel   additives  for  particle  trap
regeneration.  This  section briefly  summarizes  the highlights
of HEI's October,  1985  response to Volkswagen.[92]

     The  effects   of  greatest  concern   are   neurotoxic  and
respiratory.  The neurotoxic effects in humans  reguire at least
several   months   of   exposure   and  progress  from  an  early
reversible  stage  to  a more advanced  and irreversible  stage.
The human clinical  and epidemiologic literature  suggests that
neurotoxicity is  not strongly  indicated until  exposure exceeds
5  mg  Mn/m3,  but  that  neurological  symptoms  may  occur  at
levels as low as 0.3 mg Mn/m3.

     The  respiratory  effects   of  manganese  are  identical  to
those   associated   with   exposure   to   fine   or   respirable
particulate  matter  in  general,  and  involve  an  inflammatory
effect  or pneumonitis, which  may  lead to  diminished  pulmonary
function,  bronchitis,  or  altered  susceptibility  to infection.
EPA considers respiratory  symptoms  to be  the critical  effect
because respiratory  effects are reported  at  levels lower  than
those   reported   for  neurotoxicity.    EPA  derived  human  No
Observed Effect Levels  (NOEL)  of  5 ug Mn/m3  based  on  rat data
and 8.7 ug Mn/m3  based on monkey data.  [93]

     Occupational   studies   generally  show  that   respiratory
effects follow  exposures  in excess of  5 mg  Mn/m3.   However,
data  from  a Japanese  study  of  junior   high  school  students
exposed  to   manganese   emissions  from  a  ferromanganese  plant
apparently   associates  increased   respiratory   symptoms   and
diminished  pulmonary function  with exposures  below  5  mg/m3.
Exposure  was determined by the  amount  of  manganese in  the
dustfall.   EPA estimated, on the basis of  analyses  of  dustfalls
near  a ferromanganese plant  in  the  Kanawha Valley in  West
Virginia,  that the dustfall in the Japanese study  may  translate
to  between  3  and 11  ug Mn/m3 .   These are  the  lowest  levels
associated   with    adverse   health   effects;    however,    the
correlation   of  settled dust  with  suspended particulate  matter
is subject  to broad variability,  even in  the same geographic


                              -81-

-------
locations, as a  result/of  air turbulence,  humidity,  topography,
windspeed,  etc.    As  a  result,   there  is  a  great  deal  of
uncertainty  regarding  the   level   of   exposure  to  manganese
concentrations  associated  with  the  health  effects  in  this
study.  HEI is presently evaluating the study in detail.

     There  is   some  evidence  of   carcinogenic   activity  of
manganese  in  laboratory animals,  although the  value  of  these
studies is questionable.   There is  no epidemiologic information
relating manganese exposure to cancer occurrence in humans.

11.3 Platinum

11.3.1  Source and Emission Factors

     Platinum is  a catalyst  attrition  product.   It  is emitted
in  small  quantities.    Most  studies  that  have  attempted  to
characterize  platinum  emissions  report no  detectable levels.
The measurement  methods  used are often not  sensitive enough to
detect the small  quantities of platinum emitted.

     Limited  quantitative  measurements  of  platinum have  been
made.     In  1978,   Ford   measured   FTP   particulate  platinum
emissions from two vehicles  (1976 LTD with monolithic three-way
catalyst, oxidation  catalyst  and  air pump  and a 1978 Pinto with
monolithic  three-way  catalyst,  oxidation   catalyst  and  air
pump).   Platinum emissions  ranged  from 0.4  to 1.4  micrograms
per mile  (ug/mile).[94]  In  1977,  General  Motors  reported that
platinum emissions from pelletized  catalysts  ranged from 1 to 3
ug/mile.[95]

     It  has   been thought  for  some  time  that  much  of  the
platinum  emitted may  accumulate  alongside roadways.   This was
confirmed  in  a  recent  study.[96]   Dust samples collected from
the leaves of roadside plants  contained  as high  as  0.7  ppm of
platinum  and  0.3 ppm  of palladium  (another  catalyst  attrition
product).  The highest  concentrations of both metals were found
in  dust  collected from plants  growing  at  the edge  of heavily
trafficked  streets  and  highways  and  the   lowest   in samples
collected from plants growing in the  yards of  houses located on
lightly trafficked streets.   The concentrations of  both metals
in  the dust  samples  are much higher than the reported natural
abundances of these elements.

11-3.2  Health Effects

     In 1977, the  National Academy  of Sciences prepared a study
which  examined  the  uses,   sources   of  supply,  and   potential
health  effects  of noble  metals.    With regard  to  platinum and
palladium  emissions  from  catalyst-equipped  vehicles,  the HAS
study concluded:
                              -82-

-------
     "Minute  quantities  of  platinum  and  palladium  (about  1-3
     ug/mile)   are  emitted   from   the   exhaust  systems   of
     automobiles  equipped with  catalytic  converters;  much  of
     this material may  accumulate  alongside roadways.  However,
     this   material   is   in   a   chemical   form   that   is
     physiologically  innocuous (no  detectable soluble  salts),
     and  it  is concluded that such  emission  poses  no threat to
     the  environment.    Because  there   is  no  evidence  that
     platinum  metal  can be  methylated  by microorganisms  and
     solubilized  in  the  same  way that  mercury  is methylated,
     this deposited  material  should not have an  adverse effect
     on the environment."[97]

     Recently, Dr.  Hans A.  Nieper of  the  Paracelsus  Clinic  in
West   Germany  has   claimed  that   platinum  emissions   from
catalyst-equipped   vehicles   are   associated  with   rising
incidences of AIDS,  cancer  and leukemia.[98]   He also  states
that  slight  traces  of  platinum   are   extraordinarily  toxic
against  the  genetic defense  systems.   These claims  are  not
currently scientifically supportable.

11.4 Cadmium

11.4.1  Source and Emission Factors

     It is postulated that cadmium is present  in  fuel  as  a lead
contaminant  and  therefore  emitted  by vehicles  burning  leaded
fuel.   Cadmium emissions from  non-catalyst-equipped  vehicles
are  roughly   1.6  x  10~s  g/mile.[32]   Assuming  12 percent  of
the   current   light-duty   fleet    are   non-catalyst-equipped
vehicles, the resulting  light-duty fleet  emission factor is 1.9
x 10~6 .

11.4.2  Health Effects and Risk Estimate

     A unit risk estimate for  cadmium is estimated  by EPA to be
1.8  x   10"3.    The  risk  from   cadmium  was   determined   by
multiplying  the  risk  determined  in  the   Six Month  Study  by
ratios  of  the updated to the  Six  Month  Study emission factors
and unit  risks  to obtain a crude  estimate  of  the  risk.   The
resulting risk  of 0.001  per  urban million  appears negligible.
This risk is  further  projected to  decrease  to zero by  1995  as
non-catalyst-equipped vehicles are  phased out of the fleet.
                              -83-

-------
 12.0 SIX MONTH  STUDY: SUMMARY AND COMPARISON OF RESULTS

 12.1 Purpose  of Six Month Study  and Summary of Results

     The  final  study   is   formally  titled,  "The  Air  Toxics
 Problem  in  the  United States:  An  Analysis of Cancer  Risks  for
 Selected Pollutants."[1]  The  EPA  study attempted to assess  the
 magnitude and nature of  the air  toxics problem  by developing
 estimates of  the cancer risks  posed  by selected air pollutants
 and their sources.

     Three  major  analyses were  used  to estimate  excess cancer
 risk from  exposure to 15-45 toxic  air  pollutants.   The Ambient
 Air Quality study used ambient data for five  metals, 11 organic
 compounds,  and  benzo  (a)   pyrene  (B(a)P)  together with  unit
 risks  for  these  compounds   to  estimate national  excess cancer
 incidence.  The other two  analyses  (the NESHAPS  study  and  the
 35 County study)  used exposure models  together with unit risks
 to  estimate   excess  cancer   incidence.    The   NESHAPS  study
'provides  national estimates  for about 40  compounds.    The  35
 County  study  was  limited  to  22 compounds and 35  counties  but
 was  designed   to  allow  more   detailed  assessment  of  source
 contributions.  Since the 35 County  study  was  used to estimate
 the mobile  source contribution to  the  air toxics  risk,  it  will
 be  discussed  in  more  detail.   The  other two  studies  do  not
 contradict any  relevant findings from the 35 County  study.

     The 35 counties were chosen from counties with the highest
 expected ambient  exposures,  based  on  high populations and large
 aggregate  emissions  from  all  source  categories.   Also,  some
 counties  were  included simply  because  they  contain  large
 industrial  point   sources   of  potential  interest.    The  35
 counties  contain  roughly  20  percent  of  the   total  U.S.
 population (45  million people),  20 percent  of  total releases of
 VOC, and  10 percent  of  total  particulate  matter  loading, based
 on  1982 data.   They  represent  a  variety of  industrial   and
 population    distributions,    but    are   not    considered   a
 statistically representative sample  of the country.[99]   When
 comparing the results of the  35 County study with the results
 of this  study,  cancer incidences will be expressed  per  million
 people exposed.

     The  aggregate  cancer   risk from   all  sources  in   the  35
 County study is 207, or 4.6  cancer  incidences  per  million,  when
 products of  incomplete  combustion  (PIC) are  included.   Mobile
 sources were  estimated to  account for  60  percent  of  the total
 incidence when  PIC  are  included and 23 percent  of the total
 incidence when  PIC are  excluded.   A  breakdown of  the  specific
 mobile  source  pollutants  responsible  and  the risk associated
 with each pollutant is given in Table 12-1.
                              -84-

-------
                           Table 12-1

           Six-Month  Study Results  for Mobile Sources*
Pollutant
PIC
Benzene
Formaldehyde
Ethylene
B(a)P
Cadmium
Total
93.2
14.0
le 3.1
ibromide 0 . 9
0.8
0.4
Annual Cancer Incidence
Per Million** %
2.07
0.31
0.07
0.02
0.02
0.01
Contribution
82.9
12.5
2.7
0.8
0.7
0.4
        TOTAL:     112.4         2.50               100.0
*    Gasoline  vapors  were considered  separately.   The  total
     risk was 6.8 or 0.15 per million.

**   Based on 45 million people in 35 County study.
                              -85-

-------
     PIC  refers  to  a  large  number   of  hydrocarbon  compounds
(mainly  polynuclear  organics).    The  PIC  unit  risk value  is
derived   from  various  older   epidemiological   data   and  is
controversial.    Epidemiological   studies   of   the   general
population  exposed  to  ambient  air  and  studies  of  workers
occupationally  exposed to  PIC (e.g., coke oven  emissions, hot
pitch  fumes)  were used to  estimate the  PIC  unit  risk.   B(a)P
was used  as  an index of exposure to  PIC.   As  a result,  the PIC
unit risk  is  expressed per  unit of exposure of B(a)P.   The PIC
unit  risk  used  in  the  Six  Month   Study is  4.2X10'1.   The
uncertainties  associated  with  this unit  risk  are  detailed  in
the Six Month Study.

     The  B(a)P  emission factor  for mobile sources  was  used  to
calculate  an  annual average  exposure.   The B(a)P  exposure was
then multiplied by the PIC unit risk to  estimate  annual cancer
incidence due to PIC from mobile sources.   The  same process was
performed  for  all  other  sources   emitting B(a)P  to  determine
cumulative annual cancer incidence  from PIC.

     In the  35 County  study,  roughly 75  percent  of the  total
B(a)P  emissions was said to  be  attributable to mobile  sources.
Sources  emitting  the  remaining   25  percent   are  residential
fireplaces,  woodstoves and waste  oil  burning,  all considered
area sources.   Point  sources were assumed to emit no  B(a)P,
since  coke oven  emissions were dealt with  separately  in the
report.   As   a  result,  75  percent  of  the  total  PIC  cancer
incidence  is  due  to  mobile  sources.   PIC  cancer  incidence
accounts for  a  large  portion  of the  aggregate  cancer  incidence
because the  unit  risk for PIC  is orders  of magnitude  greater
than the unit  risks for the  other toxic compounds  examined  in
this study.   This  explains  why,   when  PIC are  included,  the
fraction  of  total  cancer  incidence  related  to mobile  sources
increases from 23 to 60 percent.

     From Table 12-1,  the  pollutants  responsible  for the total
estimated cancer incidence from motor vehicles in  the 35 County
study,  in  order of importance,  are PIC,  benzene,  formaldehyde,
ethylene dibromide  (EDB),  B(a)P and  cadmium.   Gasoline  vapors
from motor vehicles  were  not  included  in the  motor  vehicle
category.    Instead,  gasoline  vapors  from  service  stations,
which includes  vehicle  refueling and  delivery  of  gasoline, were
considered separately.

12.2 Comparison of Six Month Study  Results  with Results  of This
     Study

     For   benzene,     formaldehyde,    gasoline    vapors    and
1,3-butadiene,   the    annual   cancer    incidence   per   million
estimated  in  both studies  can  be  directly compared.  For PIC,
the PIC risk  estimated in the Six  Month Study will  be  compared
to  the  combined  risk from   diesel  particulate  and  gasoline
POM/PIC estimated  in  this  study.   For EDB and cadmium,  risks
have not  been  estimated  in this report  due to the  rather low
                              -86-

-------
emissions and the  lack  of data on  emissions  at various  speeds
required  as inputs  to the  NEM model.   For  these  pollutants,
rough  estimates  of  the  risk  will  be made  by  comparing  the
emission factors used  in the Six  Month Study with  those which
will be developed in the subsequent sections, and  adjusting  the
risks  obtained  in  the  Six Month  Study  accordingly.   This
approach  assumes  that  the  emissions  are  directly  associated
with ambient concentrations.  This appears to have  some  merit,
considering  that    EDB  and   cadmium   are   not   considered
photochemically reactive.

     In  addition  to emission factors, the  unit risks used in
both studies should be compared.   In some cases, the  unit risks
have  changed  substantially  since  release  of  the  Six  Month
Study.   1,3-Butadiene is one notable example.

     In the  following  sections,  the risks  from each  pollutant
will be  compared.   The  emission factors and unit  risks used to
calculate the annual cancer  incidences will also be  compared in
order  to pinpoint  the discrepancies.   Finally,  the  aggregate
risks will be compared.  Table  12-2 presents  a  summary of this
comparison   that   will   be   referred  to   in   the   following
subsections.

12.2.1   Formaldehyde

     Before  comparing  the   composite  emission  factors   for
formaldehyde,  the   vehicle  class  VMT  fractions  used  will   be
briefly discussed.   For  this study,  the VMT fractions  are those
used in  MOBILES  for calendar  year  1986.   For  the Six  Month
Study,  the  supporting  document  prepared  by Versar  for  the  35
County  study  included  VMT  data   for   only  one  of  the   35
counties. [99]  Also,  the  emissions  and  VMT  data  in  the   35
County study were  grouped  into three major vehicle  categories
(LDG,  HDG  and  HDD).   The MOBILES  VMT fractions  were  grouped
into these   three  vehicle categories  and compared  to the  VMT
fractions  for   the  single   county.   The  VMT   fractions were
generally  similar;  the  MOBILES  heavy-duty  VMT fractions were
somewhat  higher.    As  a   result,   the  MOBILES  grouped   VMT
fractions were applied  to the emissions  data for LDG, HDG  and
HDD given in the Versar report  to  calculate  composite  emission
factors for the 35  County study.

     As seen in  Table 12-2, both the  composite  emission  factor
and  the  unit risk  for  formaldehyde  used in this report are
higher  than  those  used in  the  35  County study  (referred to as
the Six Month Study  in  Table 12-2).  The emission factors used
in the 35  County study were  those derived from studies  of  low
mileage vehicles.   In  this  study,  the  emission  factors were
percentages  of  total exhaust HC,   as  predicted by MOBILES  for
1986,  and  were designed  to account  for  in-use  deterioration.
Also,  since  the  release of  the Six Month Study,  the  unit risk
                              -87-

-------
 0)
 O

 10
O

 0)
 u
 u
 3
 O
01

 0)
rH  tl
•H  3
            — T3
            C  3
            O 4-»
            i-H  (Q
            •H -H
            E x:

            CH
            (o
                       o
                        i
               r—
               vo

               ro
                 I
               vO
                                                             vO
                                                             t-
                                                             O
                                                              1
                                                                    O
                                                                    O
                              VD
                              ro
                                                                           vO
                                                                                   ro
                                                                                    I
                                                                                   o
                                      Cn
                                      CNJ


                                      O
                                       I
                                      O
                                                                                          ro
                                                                                                  o
                                                                                                   i
                                                            O
                                                            o
o
o
                                                                                                                        CO
                                                                                                                        in
         I
        o
        oo
(M
i-H


 0)

X)
 01
 01
 0)
M.
        u

        a

        o
            Oi  C
        rH *-  O

        3*S

        C  01 vO
               •o

               4->
               00

                01
                              r-
                              o
                                      CO

                                      o
                                      CM
                                      O
                              in
                              i-H

                              o     o
                                                                                   p
                                                                                   2
                                                     o

                                                     o
                                                            (M
                                                            o
                                                                                  in
                                                                                  vO
    u

    4->
    •rl
    c
                               X
                              o
                              CO
o
rH
 X


VO
                                                      X
                                                     CO
                                               I
                                             o
                                             (M
                                                                            X
                                                                            CO
                                                                                          T       o     7
                                                                                          O     vO     O
                                      X
                                      CO
                                                                           rg
                                              I
                                             vO
                                                                                          vO
                                                                    X
                                                                    00
                                      X
                                      CM
                                                      X
                                                     ro
                                                             X
                                                            in
                    0)  3
                   rH  4J
                   •H  ryj

                   ^  w

                   *-  !c
                       H
                    u
                    o
                   4-1
                    o
                0)   IT]

               •rl      XI
                M   C  4-1
                o   o  c
                Ql-H  O
                E   en  S
                O   CO
               U  -H  VO
                       o
                        I
               o

               o
                       CO
                       CO
                       o
                              o
                              o
                                      X
                                      VO
                                                                    vO
                                                                    eg
                                                                    o

                                                                    o
                                                                     I
                                                                                   p
                                                                                   2
                                                                                   VO
                                                                                   rg
                                                                                   o
                                                                                   CO
                                                                                   X
                                                                                   oo
                                                     X
                                                    ro
                                                                                                  (Nl
                                                            o
                                                            rH
                                                             X
        o
        i-H
        X


        in
        o
        rH
        X
        rH

        in
                                                            o
                                                            i-H
                                                            x
                                                                           3
                      CO
                      rH


                      O
                                     U

                                      P
O
i-H
 X

VO
                                                            T)
 X
VO
vO
               4J
                C
                (0
                o
               O4
                       0)
                              x;
                               0)
                              T3
                               U
                               O
                       0)
                       C
                       0)
                       N

                       0)
                       CQ
U
M
a,
                                              m
                                                     p
                                                     0)
                                                     4-1
                                                     (0
                                                     i-H

                                                     3
                                                     U
                                                     m
                                                     [0
                                                     0)
                                              X
                                             O
                                                                                   p
                                                                                          p
 X
in
                                                                                                  00
                                                                            01

                                                                            a
                                                                            (0
                                                             0)
                       o
                       01
                       (0
                      o
                                      0)


                                      0)
                                     •H
                                     TJ
                                      (0
                                     4J


                                     CQ
                                      I
                                     CO
                                      O)
                                      C
                                      o>
                                                            01
                                                            O

                                                            01
                                                            0)
                                                            XI
                                                            0}
                                                                                                          (0
                                                                                                         U
 x
CO
                                                                                                                 O)
                                                                                                                 -O
        O

       XI
       •H
       P

        0)
        C
        0)
       rH

       x;

       u


•o
vV
01
01
g
c
0
•H
1— 1
1— 1
•H
S
o
00
rH
MH
O
C
o
•H
) 1
m
rH
3
O4
O
O4

c
1C

^J
3

fj
<
«
anics from gasoline-fueled vehicles were considered to
risk was not determined in this study.
?0
*O —
0) «
4-1 „
H
1 X
U

if
•s w
* .H
X
(U i)
4-)
(0 c

y o
•rl r-l
4J p,
U
(0 4->
ft C


0)
a>
GO
•3
                                                                                                                                                               I
                                                                                                                                                              oo
                                                                                                                                                              OO
                                                                                                                                                               I

-------
for  formaldehyde  has  increased  substantially.   The  unit  risk
used in this  report  is an upper 95 percent confidence limit and
includes only malignant tumors.

     The annual cancer  risk  for  both studies assume  an  average
lifetime of  70  years.   The  risk per million for the 35 County
study is based  on  a  population  of 45 million.  This  population
was  assumed  to  be  urban.   The calendar year  1986 annual cancer
risks in this study presented  in Table  12-2 are  based on  an
estimated 1986 urban population  of 180 million.

     When the annual  cancer risk for formaldehyde  given in the
35 County  study is  adjusted to  account   for  the   updated  unit
risk estimate,  the  resulting annual  cancer risk  is 0.15  per
million.   This still  falls below the low end of the range given
in  this  study.   The  range   (0.22-0.70   per   urban  million)
attempts to account for formaldehyde  photochemistry.   The range
is  large  because  1)  it   is difficult  to  determine what  the
annual  average formaldehyde  concentration  is  for any particular
area,  much  less  the  entire  urban  population,   and  2)   the
relative contribution  of  mobile and  stationary  sources  is  not
known.   For  example,  the  low estimate  (0.22  per urban million)
is based  on  a  NEM-predicted concentration  of   1.21  ug/m3  for
direct   emissions  of   formaldehyde   from  mobile  sources.   The
highest estimate  (0.70 per  urban  million)   assumes  a  maximum
annual    average   concentration  of    12.7   ug/m3    based   on
monitoring  data,  with   30   percent   attributable   to   mobile
sources,    and    attempts   to    account    for    formaldehyde
photochemistry.

     From  comparing  the  two  risk  estimates  (the  35  County
estimate and the low estimate in this study) and accounting for
differences in the emission  factors  and unit risks,  it  appears
that  the  NEM  used  in  this  study  predicts  slightly  higher
exposures per g/mile emitted than  the  dispersion modeling  used
in  the  35  County study.    This  should be  kept  in mind  when
comparing the results for benzene,  diesel  particulate,  gasoline
POM/PIC and  1,3-butadiene as well  since  the modified  NEM was
used to estimate exposures for these pollutants.

12.2.2   Benzene

     As seen  in Table  12-2, the  risks  for benzene are  more
similar;  however,  the  risk obtained  in  the Six  Month  Study
still falls  below the  range of  risk  estimates  determined  in
this study.   The  low  estimate  in  this   study  is  based  on  a
NEM-predicted concentration  whereas  the high estimate  is based
on available  ambient monitoring  data,  assuming  85  percent  is
attributable  to mobile  sources.   The  35  County   study  used
greater  emission   factors  but  a   lower  unit   risk.    Like
formaldehyde, the  unit risk for  benzene  has  increased  since
release of  the Six Month Study.
                              -89-

-------
12.2.3  PIC, B(a)P, Diesel Particulate and Gasoline PIC/POM

     The   Six   Month   Study  calculated  a  risk  estimate  for
products of incomplete combustion  (PIC).   The  approach used has
been  previously   described.    Benzo  (a)   pyrene  (B(a)P)  was
treated  separately.   In  this  study,  information  existed  to
enable  separate treatment  of  diesel and  gasoline  PIC.   Diesel
particulate  was   considered   to  represent   diesel  PIC  while
particle-associated   organic   emissions   from  gasoline-fueled
vehicles were considered to represent gasoline PIC.   The sum of
these  risks will  be  compared to  the  PIC  risk obtained  in the
Six Month  Study.

     The risk  from diesel  particulate  ranges  from 0.76  - 3.67
per urban  million.  The  range is due to the range of unit risks
chosen.

     For gasoline-fueled vehicles,  three  different  approaches
were  taken  to estimate  the  risk  from  gasoline  PIC.   These
approaches  are discussed  in  detail  in  Section  7.   The  first
approach   estimates  the  risk  of  B(a)P  from  gasoline-fueled
vehicles.   B(a)P  emission  factors are used  together with  the
B(a)P  unit  risk.    The  annual  cancer   risk  of  B(a)P  from
gasoline-fueled  vehicles  was  determined  by  multiplying  the
B(a)P risk obtained in the Six Month Study by the  ratio  of the
emission factors (this study / Six Month Study).   This approach
assumes that  there is no  risk from the remaining  gasoline PIC
emissions.

     The   composite  B(a)P  emission  factor   of   5.56   x  10"6
g/mile used in this study is  roughly one-third of  the emission
factor used in the Six  Month  Study (1.66 x  10"s).   The  main
reason for this difference is that  the Six Month Study assumed
50 percent of  the gasoline-fueled  vehicles  were  leaded  while
this study used a more realistic percentage of 12  percent.

     When  the B(a)P risk estimated in the  Six Month Study (0.02
cancer incidences per urban million) is multiplied  by the ratio
of  the    composite  emission   factors   (5.56  x   10"6/1.66   x
10~s),  the resulting  annual  cancer risk  is  0.007  per  urban
million.

     The  second  approach  uses  B(a)P  emission   factors  from
gasoline-fueled vehicles  together with the PIC unit  risk (which
is expressed  per  unit  of  exposure  of B(a)P)  to estimate  the
annual cancer risk  of  PIC from  gasoline-fueled vehicles.   This
is the approach used  in  the Six Month Study.   The annual cancer
risk  of  PIC  from  gasoline-fueled  vehicles was  determined  by
multiplying the PIC risk obtained  in the  Six Month Study by the
ratio of the B(a)P emission factors.
                              -90-

-------
     When the PIC  risk  estimated in the  Six Month Study  (2.07
per urban  million)  is multiplied by the ratio of  the  composite
emission  factors  (5.56  x  10"6/l-66  x  10~5),   the   resulting
annual cancer risk is 0.68  per urban million.

     Unlike  the  second approach,  the  third  approach  does  not
use B(a)P as a surrogate for PIC emissions.  The  third approach
uses  estimated  emission rates  of  gasoline  particle-associated
organics (as an  unspeciated mixture) together  with a  unit  risk
for these  mixed organics.   Exposures  were estimated  using  the
modified NEM.

     Estimated   composite    emission   factors   for    gasoline
particle-associated organics in  1986 are  estimated to  be 0.0075
- 0.0082 g/mile,  depending  on whether a minimum I/M program  was
assumed to exist.  An upper  confidence limit unit risk is  2.5 x
io-4.

     Nationwide urban exposure in  1986,  using the modified  NEM
is  estimated to be  0.20-0.21 ug/m3.   Estimates  of urban  lung
cancer  incidence  in  1986  range  from 127-136,  or  0.71-0.76  per
urban million.

     The resulting  range  of  risk  estimates  for  gasoline  PIC,
using the  results  of all  three approaches,  is   0.007-0.76  per
urban million.

     The sum of  diesel  particulate  and gasoline  POM/PIC risks,
or  equivalent   "PIC"  risks,  range  from  0.77-4.43  per  urban
million.   The  PIC risk  obtained in the  35 County study  (2.07
per urban million)  lies  within this range.

12.2.4  Gasoline Vapors

     The  annual  cancer  risk  from gasoline vapors   reported
(based  on  other  documents)   in  this  study is  0.36   per  urban
million.   In  the  Six  Month  Study,  the  risk   from   gasoline
marketing was  considered.   Gasoline marketing in  the  Six  Month
Study refers to  emissions  from  service  stations.  It  includes
emissions  from  vehicle refueling  and  from  the  delivery  of
gasoline to  the  service station.  The resulting  risk was  0.15
per million.  This is less  than  half the risk presented in this
study.   It  is  mentioned,  however,  in  the  supporting  Versar
report  that high  exposures  near  the pump  from  self-service
refueling  were  excluded.    The  vehicle  refueling  in  the  Six
Month  Study  then  refers   to  occupational  exposure   (service
station attendants).   The  risk presented in the Six Month  Study
is, as  a  result, more  in  line with the risk  presented  in this
study, since the risk from self-service refueling was  projected
to be 54 percent of the  total.
                              -91-

-------
12.2.5  1,3-Butadiene

     The   35   County  study   found   no   risk  associated  with
emissions  of  1,3-butadiene  from mobile sources.  A  cancer risk
estimate for  1,3-butadiene  was estimated in this study using an
updated emission  factor  and  unit  risk.   The risk  estimate  is
discussed  in detail in Section 6.

     Determination  of  an emission  factor  for  1,3-butadiene  is
difficult.  This  is  because 1,3-butadiene  and  n-butane coelute
and   thus  have   the   same   retention   point   on   the   gas
chromatograph.  Emission  characterization studies to  date have
not  attempted to  determine the percentage  of  the  peak  due  to
1,3-butadiene.  Therefore,  assumptions must  be made  about  the
percentage each  compound contributes to  this peak.   It will  be
assumed   that    15   percent   of    the    peak   is   due   to
1,3-butadiene.[64]  1,3-Butadiene emissions  were  expressed as a
percentage of the total exhaust HC predicted by MOBILES.   Based
on the data from  46 in-use  gasoline-fueled vehicles provided in
reference  30,  1,3-butadiene  is  roughly 0.94  percent  of  the
total FID  exhaust HC.   Due to the  lack  of data for  the other
vehicle  classes,   this  percentage was  simply  applied  to  the
MOBILES composite exhaust HC emission factor.

     The  modified NEM  model  was  used  to  estimate exposures.
Nationwide urban  exposure in 1986  is estimated to  be  0.69-0.76
ug/m3.   The range accounts   for  both  the presence  and absence
of an Inspection/Maintenance  program.  These exposure  estimates
are  for  direct emissions  of  1,3-butadiene  and do  not account
for  reactions  of  1,3-butadiene  in  the   atmosphere.   Available
ambient  monitoring  data were  reviewed and  compared  to  the
exposure  estimates.   Average mean   values  in  urban  settings
range  from 0.77-24.23  ug/m3. [64]    The   NEM  estimate  of  urban
exposure  from motor vehicles  lies  near the  low  end of  this
range.

     The   unit    risk    for    1,3-butadiene    has    changed
substantially.  Based on inhalation  studies  of  1,3-butadiene  in
mice,  a  95  percent  upper  confidence   limit   unit  risk  for
1,3-butadiene   is   estimated to  be   2.8   x   10~4.[66]   This  is
much greater  than the  previous estimated  unit risk  of  4.6  x
10~7 used in the Six Month Study.

     Estimates of urban cancer  incidence  in  1986  range  from
514-568,    or   2.86-3.16   per   urban  million.    Due  to   the
uncertainty associated  with  the emissions  estimate,  a  lower
risk estimate  of zero will  also be used.

12.2.6  Ethylene

     Ethylene  from motor  vehicles  was not considered  in  the  35
County study.   Based on  the unit  risk for ethylene  provided  in
the  Six Month Study,  risk  estimates  for  ethylene  were computed
in  this  study  (Section  6.4).  The  urban   risk  in  1986 was
estimated to be 47-52 cancer  incidences,  or 0.26-0.29  per urban
                              -92-

-------
million.  As  mentioned in Section  6.4,  however,  the unit  risk
and  resulting  risk  estimates  must  be  regarded  as  extremely
tentative,  since  there is no  available  evidence  that  ethylene
is  carcinogenic.    The unit  risk  for  ethylene  was  estimated
based on assumptions regarding its  potency  relative  to  ethylene
oxide, a metabolite  of ethylene  and an animal carcinogen.   As a
result, a lower risk estimate of zero is also used.

12.2.7  Asbestos

     Asbestos emissions from motor  vehicles  were not considered
in  the Six Month Study.   Based  on available  emission  factors
and resulting expected ambient  concentrations,  it is  estimated
in this study that  asbestos  emissions  from motor vehicles  could
currently  be  responsible for  as  many  as  0.002-0.63   cancer
incidences per urban million.

12.2.8  Ethylene Dibromide (EPS)  and Cadmium

     EDB and  cadmium were handled  similarly.   Updated  emission
factors  and unit  risks  were  used  in  this study.    For  each
pollutant,  the  risk  determined  in  the  35 County study  was
multiplied by ratios  of the updated to  the 35 County  emission
factors and unit risks to obtain a crude estimate  of  the risk.

     EDB is used  as a  lead scavenger.    It  is  therefore  emitted
from  vehicles  using  leaded  gasoline.    £DB  data   came  from
reference 100.   The light-duty vehicle  fleet was assumed  to be
88     percent     catalyst-equipped      and    12      percent
non-catalyst-equipped   based   on   MOBILES   data.     Of    the
catalyst-equipped vehicles, a  misfueling rate  of  14.5  percent
was  used.[87]   The composite  emission  factor for  light-duty
vehicles   is    7.1x10"5   g/mile.   Heavy-duty  gasoline-fueled
vehicles were not considered.  The  impact of this vehicle  class
is not  expected to be great  due to its small contribution to
total VMT (4 percent).

     The composite  emission  factor and  resulting annual cancer
risk  is  roughly half  that calculated  in the  35  County study.
This    is    because    a    much    higher    percentage     of
non-catalyst-equipped  vehicles  was  assumed  in the  35  County
study  (50  versus  12 percent).   The resulting  risk  of  0.01 per
urban million appears negligible.

     In 1995,  with the  phase-out  of  leaded-fueled vehicles,  the
emission  factor   is  expected  to  decrease   to  1.8   x   10"5
g/mile.  The resulting risk is 0.003 per urban million.

     Data  for  cadmium were  obtained  from  references  32-35.
Cadmium does  not appear  to be  emitted  from  catalyst-equipped
vehicles or  it  is emitted below the detection limit.   Cadmium
emissions   from  non-catalyst-equipped  vehicles  are   roughly


                            .  -93-

-------
1.6xlO~s  g/mile.   It  appears  that  cadmium  may  be  a  lead
contaminant.   Assuming 12 percent  of the light-duty  fleet  are
non-catalyst-equipped  vehicles,  the resulting  light-duty  fleet
emission  factor  is  1.9xlO~6.   Again,  this  is  less  than  what
was used in the 35 County study  because  of  the  lower percentage
of non-catalyst-equipped vehicles expected.   The  resulting  risk
of 0.001 per million is negligible compared to the other risks.

12.2.9  Total Aggregate Risk

     The sum of all the individual  annual  cancer  risk estimates
in the  35  County study (including  gasoline vapors)  is 2.65  per
million.  In this study, the aggregate risk ranges  from 1.80 to
10.58 per  urban  million.   If the  formaldehyde  risk  in the 35
County study is increased to reflect  the updated  unit  risk,  the
aggregate  risk in the 35 County study  would increase  to  2.73
per million.   The range in  this study  is  due to  a  number  of
factors  including:    1)  the  uncertainty of   annual  average
formaldehyde and benzene concentrations  and the  contribution of
mobile  sources,   2)  the  range  of  unit  risks  used for  diesel
particulate, and  3) the different  approaches  used  to  determine
the contribution of gasoline POM/PIC.

     It  should be  noted when  reviewing  Table  12-2  that  the
pollutants given  do  not represent  a complete list.   This  list
does  not  include  pollutants  which are  formed  photochemically
from  mobile  source  emissions.   This  category  of  pollutants
could have  considerable impact but  not enough is known to  make
a quantitative estimate.
                              -94-

-------
13.0 SUMMARY AND LIMITATIONS

     The  aggregate  risk  from  mobile   source   pollutants  in
calendar year  1986 was estimated  to  range from  1.80-10.58 per
urban  million.   This  translates  into  roughly  325-1900  urban
cancer  incidences.   The  majority of  risk  is  attributed  to
formaldehyde,   diesel   particulate,   and  benzene.    Due   to
increasing use  of  advanced control technology,  the risk in 1995
is projected to decrease to roughly 60 percent the risk in 1986.

     When  reviewing these  estimates, the following  important
limitations should be considered.

     o     This  report only  accounts  for  a small  number  of
           mobile  source  pollutants  known to be emitted.   In
           reality, mobile  sources emit  hundreds  of  compounds.
           A  combination  of  health and/or  exposure  and/or
           emissions  data   are   lacking   for  many   of   these
           compounds.

     o     With the exception  of formaldehyde, this report does
           not consider reactions  of  mobile source pollutants
           in  the  atmosphere.    Little   data   are   available.
           Resulting   pollutants   may    be   more    or    less
           carcinogenic than what was  originally  emitted.

     o     Exposures for many  of the  pollutants  were  estimated
           using a modified version of the NAAQS Exposure Model
           (NEM)  for  CO.    In order  to  apply   this  model  to
           other mobile  source  pollutants, it was assumed that
           the  other   pollutants   have   the  same  'dispersion
           characteristics  as  CO.   This   is  probably not  true
           for every pollutant; however,  this was determined to
           be   the  best   approach  currently   available   to
           determine exposure  to mobile  source pollutants  on a
           national scale.

           An   alternative    is   to    use    ambient    data.
           Unfortunately,   ambient  data  for many mobile  source
           pollutants,  particularly VOC,  are  scarce  and  not of
           high    quality.     In   addition,    the    relative
           contributions of  mobile and  stationary sources  are
           not known.

     o     This  report  does  not  fully  take   into   account
           seasonal  variations    in   emissions.   Hydrocarbon
           emissions from mobile sources  are  known to  increase
           as the temperature  decreases.    Work is being done to
           characterize emissions  of  formaldehyde  under  cold
           temperature   conditions.   This  will  be  followed  by
           further characterization of individual HC  emissions
           from   mobile   sources   under   cold   temperature
           conditions.

                              -95-

-------
The unit risk  estimates' are  always  subject  to  some
uncertainty.   With the  exception  of  benzene,  they
are generally  based on  the  results of  animal rather
than human studies.

The  risks   are assumed  to  be additive.   It may  be
that   certain   combinations   of   exposures   have
synergistic or antagonistic effects.

The  risk  projections   for   1995  are  based   on  the
emission standards currently  in  place.  Changes  in
fuel composition are not considered.
                   -96-

-------
REFERENCES

1.   Elaine  Haemisegger,  Alan   Jones,   Bern  Steigerwald  and
     Vivian  Thomson,  "The  Air   Toxics  Problem  in the  United
     States:   An   Analysis   of   Cancer   Risks   for   Selected
     Pollutants," EPA  Report,  Office of  Air  and Radiation,  May
     1985.

2.   Roy  B.  Zweidinger,  "Emission  Factors  from  Diesel  and
     Gasoline  Powered  Vehicles:   Correlation  with  the  Ames
     Test,"   In:    Toxicological   Effects   of   Emissions  from
     Diesel  Engines,  Joellen  Lewtas,   ed.,  Elsevier  Science
     Publishing Co., Inc., 1982.

3.   Ronald   L.   Williams,   "Diesel   Particulate   Emissions:
     Composition,    Concentration,     and     Control,"      In:
     Toxicological  Effects  of  Emissions from  Diesel  Engines,
     Joellen Lewtas,ed.,  Elsevier  Science Publishing Co., Inc.,
     1982.

4.   Frank Black  and Larry  High,  "Methodology  for  Determining
     Particulate and Gaseous Diesel  Hydrocarbon  Emissions,"  SAE
     Paper 790422, February 1979.

5.   Joellen Lewtas, Marcia  G. Nishioka  and Bruce  A.  Peterson,
     "Identification   and   Comparative   Risk   Assessment   of
     Airborne Carcinogens from  Combustion  Sources," U.S.  EPA,
     Office  of  Research and   Development,   EPA/600/D-86/013,
     January 1986.

6.   Dennis  Schuetzle,   "Sampling   of  Vehicle   Emissions  for
     Chemical Analysis  and Biological Testing,"   Environmental
     Health Perspectives, Vol.  47, pp.  65-80,  1983.

7.   Roy  E.  Albert, Joellen Lewtas,  Stephen Nesnow,  Todd  W.
     Thorslund,   and Elizabeth   Anderson,   "Comparative  Potency
     Method for Cancer  Risk  Assessment:   Application to  Diesel
     Particulate Emissions,"  Risk Analysis,  Vol.  3, No.  2, 1983.

8.   "Diesel Particulate  Study," EPA  Report,  Office of  Mobile
     Sources,  October 1983.

9.   EPA  Emission  Factors  Program  Data   Base.   Report   in
     preparation.

10.   Mahlon  C.   Smith,   IV,    "Heavy-Duty  Vehicle   Emission
     Conversion  Factors,  1962-1997,"   EPA  Technical   Report
     EPA-AA-SDSB-84-1,  August 1984.

11.   "Users Guide to Mobiles (Mobile Source  Emissions  Model),"
     Office of  Mobile Sources,  EPA-460/3-84-002,  June 1984.

12.   MVMA Facts  and  Figures and Fact Sheets.

                              -97-

-------
13.  EPA Memo from Paul Laing, Standards Development  and Support
     Branch   (SDSB)    to   Chester   J.   France,   Chief,   SDSB,
     "Historical  and  Projected Light-Duty  Truck and  Heavy-Duty
     Vehicle Sales Data,"  February 1986.

14.  EPA Memo  from John  W.  Mueller,  Standards Development  and
     Support Branch to Charles L.  Gray,  Jr., Director,  Emission
     Control Technology Division,  "Revised  Diesel  Sales  Fraction
     Projections," January 9,  1987.

15.  Draft MOBILES Fuel Consumption Model,  February,  1985.

16.  "The  Motor   Fuel  Consumption   Model,   Tenth   Periodical
     Report,"  prepared by  Energy and  Environmental  Analysis,
     Inc. for U.S. EPA, Office of Mobile  Sources,  November  1983.

17.  "National Air Quality  and Emissions  Trends Report,  1984,"
     U.S.  EPA,  Office  of Air  Quality  Planning and  Standards,
     EPA-450/4-86-001,  1986.

18.  Melvin N.  Ingalls,  Southwest Research  Institute,  "Improved
     Mobile Source Exposure Estimation," EPA-460/3-85-002,  March
     1985.

19.  "Statistical  Abstract of the  United States,"  105th  Edition,
     U.S. Department  of Commerce,  Bureau  of the  Census,  1985.

20.  Charles T. Hare  and  Thomas M. Baines,  "Characterization of
     Particulate   and   Gaseous    Emissions   from   Two   Diesel
     Automobiles  as  Functions of  Fuel  and  Driving Cycle,"  SAE
     Paper 790424, February 1979.

21.  James D. Hyde, Richard E. Gibbs,  Robert A. Whitby,  Stanley
     M.   Byer,  Benjamin J.  Hill,  Thomas  E.  Hoffman,  Robert  E.
     Johnson, and  Paul  L.  Werner,  "Analysis of Particulate  and
     Gaseous Emissions Data  from  In-Use Diesel Passenger  Cars,"
     SAE Paper 820772,  June 1982.

22.  Terry L.  Ullman  and  Charles  T.  Hare,  Southwest  Research
     Institute,  "Emissions Characterization  of  Heavy-Duty  Diesel
     and Gasoline  Engines and Vehicles,"  EPA-460/3-85-001,  May
     1985.

23.  "Compilation  of Air  Pollutant Emission  Factors,  Volume  II:
     Mobile  Sources,"  AP-42,  Fourth  Edition,   September  1985,
     Appendix N.

24.  R.G. Cuddihy, W.C. Griffith,  C.R. Clark and R.O.  McClellan,
     "Potential  Health  and Environmental  Effects of  Light-Duty
     Diesel   Vehicles   II,"   Inhalation   Toxicology   Research
     Institute,  Lovelace  Biomedical and  Environmental  Research
     Institute,  prepared  for  U.S.  Department of  Energy,  LMF-89,
     UC-48, October 1981.

                              -98-

-------
25.  Roger 0.  McClellan,  "Health Effects  of  Diesel Exhaust:   A
     Case  Study in  Risk  Assessment," Am.  Ind. Hyg.  Assoc.  J.,
     47(1):  1-13, 1986.

26.  A.K.  Pollack,  A.B.  Hudischewskyj and  A.D. Thrall,  Systems
     Applications,  Inc.,  "An Examination of  1982-83 Particulate
     Matter  Ratios   and  Their  Use   in  the  Estimation  of  PM10
     NAAQS Attainment Status, "EPA-450/4-85-010, August 1985.

27.  Jeffrey  E.  Harris,   "Potential   Risk  of  Lung Cancer  from
     Diesel  Engine   Emissions,"   report  to  the  Diesel  Impacts
     Study   Committee,   National   Research  Council,   National
     Academy Press,  Washington, D.C., 1981.

28.  Todd  W.   Thorslund,   "A   Suggested   Approach   for   the
     Calculation  of  the Respiratory  Cancer Risk  Due to  Diesel
     Engine  Exhaust,"  presented at   the  EPA  Workshop  on  the
     Evaluation of  Research in  Support  of  the  Carcinogenic Risk
     Assessment for  Diesel Engine Exhaust,  February 24-25,  1981.

29.  W.E. Pepelko and W.B.  Peirano,  "Health Effects of  Exposure
     to  Diesel  Engine  Emissions,"   Journal   of   the  American
     College of Toxicology,  Volume 2, No. 4,  1983.

30.  John  E.  Sigsby,  Jr.,   Silvestre Tejada  and  William  Ray,
     Mobile   Sources  Emission   Research  Branch,   U.S.   EPA,
     "Volatile  Organic   Compound  Emissions   from   46   In-Use
     Passenger Cars," Environ.  Sci.  Technol.,  Volume 21,  1987.

31.  George J.  Nebel,  "The Effect of  Misfueling on Aldehyde  and
     Other Auto Exhaust  Emissions," APCA Journal,  Vol.  31,  No.
     8, August 1981.

32.  Charles  Urban,  Southwest  Research  Institute,  "Unregulated
     Exhaust   Emissions  from Non-Catalyst  Baseline  Cars  Under
     Malfunction Conditions," EPA-460/3-81-020,  May 1981.

33.  Charles  Urban,  Southwest Research Institute,  "Regulated  and
     Unregulated   Exhaust    Emissions    from    Malfunctioning
     Non-Catalyst  and  Oxidation Catalyst Gasoline  Automobiles,"
     EPA-460/3-80-003,  January 1980.

34.  Charles  Urban,  Southwest Research Institute,  "Regulated  and
     Unregulated Exhaust Emissions  from  Malfunctioning Three-Way
     Catalyst Gasoline  Automobiles,"  EPA-460/3-80-004,   January
     1980.

35.  Charles  Urban,  Southwest Research Institute,  "Regulated  and
     Unregulated  Exhaust   Emissions   from   a   Malfunctioning
     Three-Way  Catalyst  Gasoline  Automobile,  EPA-460/3-80-005,
     January  1980.
                              -99-

-------
36.  Lawrence R.  Smith,  "Characterization  of Exhaust  Emissions
     from    High    Mileage    Catalyst-Equipped    Automobiles,"
     EPA-460/3-81-024, September 1981.

37.  "Investigation  of  Diesel-Powered  Vehicle  Emissions  VII,"
     EPA-460/3-76-034, February 1977.

38.  Karl  J.   Springer,"Characterization  of   Sulfates,   Odor,
     Smoke,  POM  and  Particulates  from  Light   and  Heavy-Duty
     Engines  - Part IX," EPA-460/3-79-007, June 1979.

39.  M.W.  Phillips   and  G.E.   Wilkins,  Radian   Corporation,
     "Preliminary  Source  Assessment  for  Formaldehyde,   Final
     Report,"  prepared  for   U.S.   EPA,   Office  of  Air  Quality
     Planning and Standards,  September  3,  1985.

40.  Lawrence   R.    Smith,    Southwest    Research    Institute,
     "Unregulated  Emissions   for  Vehicles  Operated  Under  Low
     Speed Conditions," EPA-460/3-83-006,  May 1983.

41.  "Control  Techniques  for  VOC  Emissions   from  Stationary
     Sources," U.S. EPA, Draft Report,  March 1986,  Table 2-5.

42.  1983  NMHC  Emissions  Inventory,  Office  of  Mobile  Sources,
     Technology Evaluation Branch,  Draft,  November,  1986.

43.  Gary Z.  Whitten and Henry Hogo, Systems  Applications,  Inc.,
     "Impact   of  Methanol on Smog:   A  Preliminary  Estimate,"
     prepared for ARCO Petroleum Products Company,  February 1983.

44.  W.D.  Kerns,  et  al. , "Carcinogenicity  of  Formaldehyde  in
     Rats and Mice after  Long-Term Inhalation Exposure,"  Cancer
     Research, 43:  4382-4392.

45.  "Assessment of Health Risks to  Garment Workers  and  Certain
     Home  Residents  from  Exposure to Formaldehyde," U.S.  EPA,
     Office  of   Pesticides  and  Toxic  Substances,   Draft  Final
     Report,  March 1987.

46.  William  F.  Hunt, Jr., Robert  B. Faoro,  Thomas  C. Curran and
     Jena Muntz,  U.S.  EPA,  Office  of  Air  Quality Planning  and
     Standards,   "Estimated Cancer   Incidence  Rates  for  Selected
     Toxic Air  Pollutants Using  Ambient Air  Pollution  Data,"
     April 23, 1985.

47.  Memo from  Charles  L. Gray, Jr., Director,  Emission  Control
     Technology Division to Richard  D.  Wilson,  Director,  Office
     of  Mobile  Sources,  "Mobile Source  Benzene  Emissions  and  a
     Preliminary Estimate  of  Their  Health Impacts,"  May 15,  1986.
                             -100-

-------
48.  "Evaluation  of  Air  Pollution  Regulatory  Strategies  for
     Gasoline  Marketing  Industry,"  EPA  Report  450/3-84-012b,
     July 1984.

49.  Memo  from  William  F.  Hunt,   Jr.,  Chief,  Data  Analysis
     Section,  OAQPS  to  Stan  Meiburg,   Special  Assistant  for
     Program   Development,   OAQPS,   "Comparison   of   Additive
     Individual Lifetime Risks for Five Cities," April 3, 1986.

50.  Donald J.  Ames,  Barbara Fry and Richard  Vincent,  "Proposed
     Benzene  Control  Plan," State  of California  Air  Resources
     Board Draft Staff Report,  May 1986.

51.  "Addendum  to  Proposed  Benzene  Control   Plan,"  State  of
     California Air Resource Board Staff  Report, July 1986.

52.  Ward's Automotive Yearbook 1986, Forty-Eighth Edition.

53.  H.N.  MacFarland,   C.E.  Ulrich,   C.E.    Holdsworth,   D.N.
     Kitchen,  W.H.  Halliwell   and   S.C.  Blum,  Journal  of  the
     American College of Toxicology, 3,  4,  1984.

54.  "Level   I   Options  Package   for   a   Gasoline   Marketing
     Decision," Office of Air Quality Planning and Standards and
     Office of Mobile Sources Report, July  11,  1985.

55.  Andrew E.  Bond,  Vinson L. Thompson,  and  Gordon C.  Ortman,
     Environmental  Monitoring   Systems   Laboratory,  EPA,   and
     Francis  M.   Black   and  John  E.  Sigsby,   Jr.,   Atmospheric
     Sciences  Research  Laboratory,   EPA,  "Self Service  Station
     Vehicle Refueling Exposure Study," no  date given.

56.  Memorandum from  Charles L.  Gray,   Jr.,  Director,  Emission
     Control Technology  Division  to James B.  Weigold, Office of
     Air  Quality  Planning  and  Standards,    "Lower  Limit   on
     Gasoline Vapor Risk," April  19, 1985.

57.  "Gasoline Vapor  Exposure  and  Human Cancer:  Evaluation of
     Existing  Scientific  Information and  Recommendations  for
     Future Research," Health Effects Institute, September 1985.

58.  Memo from Donald J.  Ehreth,  Acting Assistant  Administrator
     for   Research   and   Development   to   Lee   Thomas,    EPA
     Administrator,  "Brief  Statement Concerning Health  Effects
     Institute (HEI)  Report," October 8,  1985.

59.  "Feasibility of  Assessment of Health Risks  From Vapor-Phase
     Organic Chemicals in Gasoline  and Diesel  Exhaust,"  prepared
     by the Committee on Vapor-Phase Organics,  National  Research
     Council,  National Academy  Press,  Washington, D.C.,  1983.
                             -101-

-------
60.  Penny  Carey and  Janet Cohen,  U.S.  EPA,  Office of  Mobile
     Sources,  "Comparison  of  Gas  Phase  Hydrocarbon  Emissions
     From Light-Duty  Gasoline Vehicles  and Light-Duty  Vehicles
     Equipped  with  Diesel  Engines,"   CTAB/PA/80-5,  September
     1980.

61.  J. Kraft  and  K.-H.  Lies,  "Polycyclic  Aromatic  Hydrocarbons
     in the Exhaust of Gasoline  and Diesel  Vehicles," SAE Paper
     810082, February 1981.

62.  Fred Stump, Ronald  Bradow,  William Ray, David  Dropkin,  Roy
     Zweidinger,  and John  Sigsby,  "Trapping Gaseous  Hydrocarbons
     for Mutagenic  Testing," SAE Paper 820776,  June 1982.

63.  Frank  Black   and  Larry   High,  "Automotive   Hydrocarbon
     Emission  Patterns    in   the  Measurement   of   Nonmethane
     Hydrocarbon Emission Rates," SAE 770144, February 1977.

64.  Summary  of  1,3-butadiene  monitoring  data  provided  by  Ila
     Cote,  EPA,  Office of  Air Quality  Planning and Standards,
     1986.

65.  EPA  Memo  from  Edward J.  Li 11 is,  Chief,  Air  Management
     Technology  Branch  to  List  of Addressees,  "Urban  Toxics,"
     June 12,  1987.

66.  "Mutagenicity     and    Carcinogenicity    Assessment    of
     1,3-Butadiene," final  report  prepared   by  U.S.   EPA,  Office
     of Health  and Environmental Assessment,  EPA/600/8-85/004F,
     September,  1985.

67.  Carl O.  Schulz  and  David  M.  Siegel,  Clement  Associates,
     Inc.,  "The  Relative  Carcinogenic Potential of  50 Chemicals
     That May Be Air Pollutants," final report prepared  for U.S.
     EPA,  Office  of Air  Quality Planning  and  Standards,  March
     30, 1984.

68.  T.M.  Albrechcinski,  J.G. Michalovic, B.J.  Wattle,  and E.P.
     Wilkinson,  Calspan Corp.,  "Laboratory   Investigation  of  the
     Fate of Diesel  Emissions in the  Atmosphere.  Task 2.   Final
     Report," CRC-APRAC-CAPA-13-76-05, December 1985.

69.  Larry  D.  Claxton  and H.M.  Barnes,   "The  Mutagenicity  of
     Diesel-Exhaust     Particle    Extracts     Collected    Under
     Smog-Chamber Conditions Using  the  Salmonella  typhimurium
     Test System,"  Mutation Research,  88,  255-272,  1981.

70.  John M.  Lang,  et  al. ,  "Characterization  of   Particulate
     Emissions from In-Use Gasoline-Fueled  Motor Vehicles,"  SAE
     Paper 811186,  October 1981.

71.  Energy  and  Environmental  Analysis,   Inc.,  "Size  Specific
     Total  Particulate  Emission  Factors   for  Mobile  Sources,"
     final  report   prepared  for  U.S.   EPA,  Office  of  Mobile
     Sources,  EPA 460/3-85-005,  August 1985.

                             -102-

-------
72.  Charles  T.   Hare  and  Frank   M.   Black,   "Motor  Vehicle
     Particulate  Emission  Factors,"   APCA  Paper  81-56.5,  June
     1981.

73.  James  N.  Braddock  and  Ned  K.  Perry,   Jr.,   "Gaseous  and
     Particulate  Emissions from  Gasoline  -   and  Diesel-Powered
     Heavy-Duty Trucks," SAE Paper 860617, February 1986.

74.  Frank  Black  et   al. ,   "Emission  from   In-Use  Heavy-Duty
     Gasoline Trucks," SAE Paper 841356, October 1984.

75.  H.E. Dietzman  et al. , "Emissions  from Gasoline  and  Diesel
     Delivery  Trucks  by Chassis  Transient  Cycle," ASME  Paper
     81-DGP-6, January 1981.

76.  "National   Dioxin   Study   Tier   4-Combustion   Sources,"
     EPA-450/4-84-014a, February 1985.

77.  M.G.  Jacko,   R.T.   DuCharme,  and  J.H.  Somers,  "Brake  and
     Clutch Emissions  Generated During  Vehicle Operation,"  SAE
     Paper 730548, 1973.

78.  S.  Cha,   P.   Carter,  and  R.L.   Bradow,   "Simulation  of
     Automobile   Brake   Wear   Dynamics   and   Estimation   of
     Emissions," SAE Paper 831036,  1983.

79.  "Asbestos;   Proposed  Mining  and  Import  Restrictions  and
     Proposed   Manufacturing    Importation    and    Processing
     Prohibitions,"  Federal Register,  Vol.  51,  No.   19,  3738,
     January 29, 1986.

80.  Melvin N.  Ingalls  and Robert J.  Garbe,   "Ambient  Pollutant
     Concentrations   from    Mobile    Sources   in    Microscale
     Situations,"  SAE 820787,  June 1982.

81.  National  Research Council, "Nonoccupational Health Risks  of
     Asbestiform  Fibers,"   National  Academy  Press,  Washington,
     D.C.,  1984.

82.  Ruth  A.  Zweidinger,  Joan  T.  Bursey,  Nora   C.   Castillo,
     Ronald  Keefe  and  Doris   Smith,   "Organic Emissions  from
     Automobile Interiors," SAE Paper  820784,  June  1982.

83.  Lawrence  R.  Smith  and Thomas M.  Baines, "Nitrosamines  in
     Vehicle Interiors," SAE Paper  820785, June 1982.

84.  "Regulation of Fuels and Fuel Additives,"  Federal  Register,
     Vol. 47,  No.  210,  49322,  October  29,  1982.

85.  "Regulation  of  Fuels  and  Fuel  Additives;   Gasoline  Lead
     Content," Federal Register, Vol.  50, No.  45,  9385, March  7,
     1985.

86.  Regulation of  Fuels   and  Fuel   Additives;   Gasoline  Lead
     Content," Federal Register, Vol.  50, No.  45,  9400, March  7,
     1985.
                             -103-

-------
87.  Energy  and  Environmental  Analysis,   Inc.,   "Supplementary
     Guidelines   for   Lead   Implementation  Plans   -   Updated
     Projections for Motor Vehicle  Lead  Emissions,"  final report
     prepared   for   U.S.   EPA,   Office  of   Mobile   Sources,
     EPA-460/3-85-006,  August 1985.

88.  "Regulation of Fuels  and Fuel Additives;  Lead  Phase Down,"
     Federal Register,  Vol.  49, No.  150,  31032,  August 2, 1984.

89.  "Air  Quality Criteria  for Lead,"  External  Review  Draft,
     EPA-600/8-83-028B,  September 1984.

90.  "In Re Applications for  MMT Waiver,"  Federal Register,  Vol.
     43, No. 181,  41424,  September  18,  1978.

91.  "Regulation of Fuel  and Fuel  Additives:   MMT -  Suspension
     of Enforcement,"  Federal Register,  Vol.  44,  No.  109, 32281,
     June 5, 1979.

92.  Letter from  Thomas  P.  Crumbly, Executive  Director,  Health
     Effects Institute to Wolfgang  Groth,  Manager,  Emissions and
     Fuel Economy, Volkswagen of America, Inc.,  October 7, 1985.

93.  "Health     Assessment     Document      for      Manganese,"
     EPA-600/8-83-031F,  Final report, August 1984.

94.  Memo  from  Lawrence  E.  Slimak,   Director,  Planning   and
     Industrial   Department,    Motor    Vehicle    Manufacturers
     Association to Phil Lorang, Chief,  Technical Support Staff,
     EPA, May 6, 1986.

95.  R.F. Hill  and  W.J.  Mayer,  IEEE  Trans.  Nucl.  Sci.,  NS-24,
     2549-2554, 1977.

96.  Vernon  F.  Hodge  and  Martha  O.   Stallard,  "Platinum  and
     Palladium  in Roadside  Dust,"   Environ.  Sci.  Technol.,  Vol.
     20, No. 10, 1986.

97.  "Medical  and Biologic  Effects  of  Environmental  Pollutants,
     Platinum-Group  Metals,"   National    Academy  of   Sciences
     Report, 1977.

98.  Open Letter  from Dr.  Hans A, Nieper,  Medical  Department,
     Paracelsus  Clinic   to  the   Board  of   Directors  of   a
     Technological Firm,  published  in Space & Time,  1985.

99.  Versar Inc. and American Management Systems, "Hazardous Air
     Pollutants, A  Preliminary  Exposure  and  Risk Appraisal  for
     35  U.S.   Counties,"  draft  report  prepared for  U.S.  EPA,
     Office of Policy  Analysis,  September 1984.

100.  John E. Sigsby, Jr.,  David L.  Dropkin, Ronald L.  Bradow and
     John M. Lang,  "Automotive Emissions of  Ethylene Dibromide,"
     SAE Paper 820786,  June  1982.
                             -104-

-------
GLOSSARY OF TERMS

Airshed:  A geographical area which, because of topography,
   meteorology, and climate, shares the same air mass.

Air Toxic:  A compound in the air capable of causing adverse
   health effects.  For the purpose of this report, the air
   toxics examined were limited to known or suspected
   carcinogens.

Aldehydes:  A class of fast-reacting organic compounds
   containing oxygen, hydrogen, and carbon.  They contain the
   group -CHO.

Aliphatic:  A class of hydrocarbon compounds which are open
   chained and fully saturated (e.g., no double bonds).

Ambient Air:  That portion of the atmosphere, external to
   buildings, to which the general public has access.

Ames Test:  A mutagenicity bioassay using the bacteria.
   Salmonella typhimurium.

Aromatic:  A class of hydrocarbon compounds originating from
   benzene, C6H6, or containing at least
   one benzene ring or similar unsaturated heterocyclic ring.

Benzo(a)Pyrene (B(a)P):  A polycyclic aromatic hydrocarbon with
   the molecular formula C20H12.

Bioassay:  Using living organisms to measure the effect of a
   substance, factor, or condition.

Catalyst:  Used in this report to denote a catalytic converter,
   a chamber in the exhaust system of vehicles containing a
   catalyst system which aids in oxidizing the carbon monoxide
   and unburned hydrocarbons in the exhaust gases or in reducing
   nitrogen oxides in the exhaust gases to innocuous products
   (carbon dioxide, N2, O2,  and water).

Control Technology:  A combination of measures designated to
   achieve the aggregate reduction of emissions.

Diesel Particulate Trap:   A device located in the exhaust stream
   of a diesel vehicle that filters  a certain percentage of
   exhaust particulate.  This device must include some means by
   which accumulated particulate can be burned,  thus
   regenerating the trap and making the trap available for
   continued particulate filtration.
                             -105-

-------
Dynamometer:  A device that is used to measure or to simulate
   loads, engine torque,  and driving forces on vehicles or
   engines.

Exhaust Gas  Recirculation (EGR):   A system or device (such as
   modification of the engine's carburetor or positive crankcase
   ventilation system) that results in engine operation at an
   increased air-fuel ratio so as to achieve reductions in
   exhaust emissions of nitrogen oxides.

Emission Factor:  For motor vehicles,  an emission factor is the
   amount of pollutant emitted per unit of distance.  In this
   report, emission factors are expressed in units of grams of
   pollutant per mile travelled.

Epidemiology:  The study of diseases as they affect populations
   rather than individuals, including the distribution and
   incidence of a disease;  mortality and morbidity rates;  and
   the relationship of climate, age, sex, race and other factors.

Evaporative  Emissions:  Hydrocarbons emitted into the atmosphere
   from a motor vehicle through fuel evaporation.

Exhaust Emissions:  Substances emitted to the atmosphere from
   any opening downstream from the exhaust port of a motor
   vehicle engine.

Federal Test Procedure (FTP):   A multistage (multimodal) test
   procedure for new car  certification by the Environmental
   Protection Agency.

Halogen:  Any one of the  nonmetallic elements chlorine, iodine,
   bromine,  and fluorine.

Heavy-Duty Vehicle:   Any  motor vehicle rated at more than 8,500
   pounds gross weight or that has a vehicle curb weight of more
   than 6,000 pounds or that has a basic vehicle frontal area in
   excess of 45 square feet.

Hydrocarbon:  Any of a vast family of compounds containing
   carbon and hydrogen in various combinations: found especially
   in fossil fuels.

Ketone:  An  organic compound derived by oxidation from a
   secondary alcohol; it  contains the carbonyl group (= CO).

Light-Duty Vehicle:   A passenger car or passenger car derivative
   capable of seating 12  passengers or less.
                              -106-

-------
MOBILE3:  A computer program that calculates emission factors
   for hydrocarbons (HC), carbon monoxide (CO), and oxides of
   nitrogen (NOx) from highway motor vehicles.

Mutagenic:  The property of a substance or mixture of substances
   to induce changes in the genetic structure in subsequent
   generations.

NAAQS Exposure Model (NEM):  An exposure model suitable for
   evaluating alternative ambient air standards.

Neurotoxic:  Harmful to nerve tissue.

Noble Metal:  Metal such as platinum or palladium that is
   non-reactive to most chemical substances.

Organic:  In chemistry, any compound containing carbon.

Oxidation Catalyst:  A catalytic converter used to oxidize the
   carbon monoxide and unburned hydrocarbons in the exhaust
   gases to innocuous products.

PAH:   Polynuclear aromatic hydrocarbons.

Particulate:  A particle of solid or liquid matter.

Photochemistry:   Chemical changes brought about by the radiant
   energy of the sun acting upon various polluting substances.
   The products  are known as photochemical smog.

POM:   Polycyclic organic matter.   Many POM compounds are also PAH
   compounds.

PIC:   Products of Incomplete Combustion.

Soluble Organic  Fraction (SOF):   In this report, defined as the
   organic fraction of particulate soluble with dichloromethane.

Steady-State:   Constant operating conditions with no variation
   in fuel supply or load.

Street Canyon:   A street lined with buildings.  In general, the
   streets are less than seven lanes wide and the buildings
   under 26 stories.  The minimum canyon height to width ratio
   for a building-lined street to be considered a street canyon
   is approximately 0.3.

Three-Way Catalyst:   A catalytic  converter that is capable of
   both oxidizing carbon monoxide and unburned hydrocarbons and
   reducing nitrogen oxides  in the exhaust.
                             -107-

-------
  TSCA:   Toxic Substances Control  Act.

  Unit Risk:   The individual  life  time  excess cancer risk from
     continuous exposure to 1 ug carcinogen per m3 inhaled air.

  VMT:  Vehicle miles travelled.

  Volatile Organic Compounds  (VOC):   Any compound containing
     carbon and hydrogen or containing  carbon and hydrogen in
     combination with any other element which has a vapor pressure
     of  1.5 pounds per square inch absolute or greater under
     actual storage conditions.
                                -108-
;,. Environmental Protection Agenc.
•ion 5, Library (5PL-16)
j S, Dearborn Street, Room 1670
     IL   60604

-------