EPA-AA-SDSB-88-05

                   Technical  Report
  Analysis of Motor Vehicle Fuel Tank-Related Fires
                          By

                   Kathleen Steilen


                    February 1988
                        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.
       Standards Development and Support Branch
         Emission Control Technology Division
               Office of Mobile Sources
             Office  of  Air  and  Radiation
         U.  S.  Environmental Protection Agency

-------
                               -2-





                        Table  of  Contents



Section                                                    Page



I.    Introduction 	 	  3



II.   Summary of Available Data 	  4



     A.    Motor Vehicle Fire Data Bases 	  5



     B.    Previous Motor Vehicle Fire Analyses 	  7 .



III. Analysis of Data 	 12



IV.   Development of 1990 Vehicle Fire Data 	 14



V.    Summary and Conclusions  	 17



Appendix	 A-l

-------
                               -3-
I.    Introduction

     Recently several  parties  have expressed  concern regarding
the  safety  of  onboard  refueling  emission  controls.   These
parties  have  commented   that   adding  canister-based  onboard
controls to current motor  vehicles would add some complexity to
the  fuel  system which could  increase  the  risk  of  crash  and
non-crash  fires.    In   a   separate  analysis,  EPA conducted  an
evaluation of the safety  implications  of canister-based onboard
systems  to  address   the   concerns  raised.   This  evaluation
concluded    that    straightforward,     reliable,    relatively
inexpensive  engineering   solutions  exist  for   each  of  the
potential problems identified and that no  increase in risk need
occur  or  be  accepted  because  of  the  presence  of  an onboard
system.[1]

     Since  onboard  refueling  controls  will modify  the  fuel/
evaporative systems of motor vehicles and  thus  could have some
impact on the risk  of  fire, it is  important to  have a baseline
assessment  of  the  risk for current  vehicles.   The purpose of
this  technical   report is  to  provide  an  assessment  of  the
current  risk of  fuel tank-related  fires  for   motor  vehicles
which  meet  current  safety standards,  and  therefore  provide  a
baseline  from  which  to  assess  any  potential  change  in  risk
(increase or  decrease) which  may  occur due  to   onboard  vapor
recovery systems.

     A  preliminary  assessment  of  fuel  tank-related  fires  was
drafted and  released  for   comment  and review  to  agencies  with
expertise  in  this  area.[2]  (This  preliminary  assessment  was
also placed in EPA Docket A-87-11  for  public review.)  Agencies
which were  asked  to review  this  analysis  include  the National
Highway Traffic  Safety Administration   (NHTSA),   the  Office  of
Motor   Carrier    Safety   (OMCS)    of    the   Federal   Highway
Administration,  the Federal Emergency  Management  Agency (FEMA),
and  the  Insurance  Institute for  Highway Safety  (IIHS).   Based
on comments  and  further data received from these  parties,  the
preliminary analysis has been refined  and is summarized in this
technical report.[3,4,5]

     Before presenting  this analysis,  it  is important  to  note
that the potential  hazard  of fuel  tank-related fires has been a
concern in  the  past.    To  address  such  concerns,  Federal  Motor
Vehicle  Safety  Standard   (FMVSS)  301  was  promulgated by  the
Department  of  Transportation  in  1967  (effective  January  1,
1968)  to  decrease   the  potential  of  crash  fires.[6]   This
standard improved the  fuel  system crashworthiness  of passenger
cars by  limiting  the  amount of allowable  fuel   leakage  in  the
event  of  an accident.   Since  that  time,   there  have  been  two
additional  FMVSS  301   rulemakings  to  upgrade  the  standard  and
further   improve    the   fuel    system   integrity   of   motor
vehicles.[7,8] '   In  1975,   this   Standard  was   substantially
enhanced  by  extending  the  coverage   of  impact   types   and
applicable  vehicles.    According   to   this  upgraded  standard,
FMVSS 301-75,  which  was effective for  1976  model  year passenger

-------
                               -4-


cars  and  1977   and   1978  model  year  light-duty  trucks,  all
vehicles with  a  Gross  Vehicle  Weight  Rating  (GVWR)  of 10,000
pounds  or  less  must  restrict  fuel  leakage  to less  than five
ounces  per  five  minutes  when  subjected  to  a  rollover  test
following front and rear collisions at 30 miles per hour (mph),
and side collision(s) at 20  mph.[7]   Furthermore, beginning in
April 19-77, school buses with  a  GVWR  greater  than 10,000 pounds
must  meet  the  same  requirements  when  subjected to  an impact
with  a  contoured  moving  barrier  at  any  speed  up  to  and
including  30 mph, at any  point and  angle.[8]   Therefore,  past
concerns regarding crash  fires  have  lead  to  three  FMVSS  301
rulemakings with  increasing stringency.

     As  a  result  of  these variations of  FMVSS  301,  the motor
vehicle  fire data base  is  more complex.   The  fire  data which
exists  today  includes  a mixture  of  vehicles  of  varying model
years which comply with different  versions  of  FMVSS  301 and in
some  cases  meet  no  standard.   These  data  must  be  analyzed
closely  to  assess the  fire  risk associated  with  current motor
vehicles, to which FMVSS 301-75 is applicable.

     As  mentioned  above,   the  purpose  of  this  report  is  to
analyze the motor vehicle fire data and  to  estimate the current
risk  of fuel  tank-related  fires.    The  report  begins  with  a
summary  of  currently available  data  on the  annual  number  of
total  motor  vehicle  fires,   post-collision   fires,  and' fuel
tank-related fires and  the  subsequent  associated  fatalities,
injuries  and  property  damage  with  these   fires.    This  is
followed by  a  discussion of  how these  data  were  analyzed  and
adjusted to  project  the number  of  fuel  tank fires, fatalities,
injuries, and property damage  in 1990  and beyond when refueling
controls could  be required.[9]  The  report  closes with a brief
summary of  the  analysis and draws some conclusions.

II.  Summary of Available Data

     The available data on  fuel  tank-related  fires  and their
subsequent  effects are  limited;  no  single body  of  data exist
which accurately  assesses  the  overall  hazards.   Virtually  all
of  the   available data  arise  initially  from  police  and  fire
marshall reports  from different states.   While  police reports
from  such   states  as  Michigan,   Illinois,   North  Carolina,
Maryland,  and   Pennsylvania   are quite  useful,  these  records
alone are  not  adequate  to characterize  the problem.   Data from
these  records  can be  extrapolated   to  obtain  an  estimate  of
national values.

     Further,   the nature  and  extent of  the  fire  crash  data
varies  in  different  state  reports.   For  example,   some  state
accident reports  are  very explicit  in  evaluating  fires  from
fuel  leakage,   the type of   accident  (front  end,   rear  end,
rollover)  and  the accident severity.  Others  are less  complete
perhaps  only  reporting information  on  fires  and   fatalities
without  much supporting detail on the cause  or location of  the
vehicle fire.

-------
                               -5-
     The  available   fire   information   from  police  and  fire
reports has been used by several  different  motor vehicle safety
experts  to  develop  data  bases  and  other  analyses which  can
estimate  the  number  of  'fuel  tank-related  fires  and  their
subsequent  effects.    This  analysis  of  fuel  tank  fires  uses
information .from  two data  bases,  the Fatal  Accident Reporting
System  (FARS)  and the  National Fire Incident  Reporting System
(NFIRS),  and  four past  analyses  on motor  vehicle  fires.   Each
of  these  data bases  and analyses  will  be described below,
together with the important information drawn from the source.

     A.    Motor Vehicle Fire Data Bases

     Fatal Accident Reporting System (FARS)[10]

     The  Fatal Accident  Reporting  System  (FARS)   is   a  motor
vehicle  accident  data  base  operated  and  maintained  by  NHTSA
which gathers  data  on all  nationwide  police-reported accidents
(for   all  types   of   vehicles,   including   passenger  cars,
light-duty trucks,  buses,  and  heavy-duty  vehicles)   in  which a
fatality occurs.  FARS  specifies  the occurrence of  fire in such
motor vehicle  accidents,  but  does  not  define  the  severity  of
the  fire,  the  cause of the  fire,  which  vehicle the  fire was
initiated  in.  (when  two  or  more  vehicles  are involved),  or
whether a fatality was  caused directly by fire.  Since  the FARS
data  represent only accidents  with fatalities,  the   reported
motor  vehicle  accidents   contained   in  this  data  base  are
generally  more severe.   Therefore,  use  of   these   data  alone
could misrepresent  or overestimate  the  overall  fire hazard for
current vehicles.

     Nevertheless, FARS does  offer  some useful  information with
respect  to  the  number  of fatalities  from  all motor  vehicle
accidents .and  those  in which  a  post-collision  fire occurred.
The data  cover  vehicles of all model years (i.e., both  pre- and
post-standard  vehicles).   The  key data  extracted from  this
source were  averaged over  the  time period  of 1980  through 1984
and are summarized below:

           1,605  annual post-collision  motor vehicle  fires  in
           which a fatality was reported.
           46,237  annual  fatalities   resulted  from all  motor
           vehicle accidents.
           1,655  (3.6   percent)   of   these   total   fatalities
           resulted  in  accidents  where  a post-collision  fire
           occurred.

     National Fire Incident Reporting System  (NFIRS)[11]

     The  National Fire Incident  Reporting  System (NFIRS)   is  a
fire  data base operated and  maintained by  FEMA which gathers
data  on  all  types of reported  fires,  including motor  vehicle
fires.   (This  motor  vehicle  data  are  based  on  all  types  of

-------
                               -6-
vehicles,   including  passenger cars,  light-duty  trucks,  buses,
and  heavy-duty trucks.)   The system  receives   a  sampling  of
state  fire reports  and extrapolates  national  statistics  from
these.  (For  example,  NFIRS  received  about 40  percent of  the
total nationwide  fires  for 1985.)   NFIRS  supplies data on  the
number  of  fires   and  the  fatalities,  injuries  and  property
damage  in  these  fires.   Furthermore,   this  data  base  can
characterize  the   fire   information  by  type  of  vehicle,   the
vehicle area  where the fire  originated  (i.e.,  fuel  tank area),
and whether the  fire was  a result of  a  collision or  had  some
other non-collision cause.

     For   the  purpose   of   this  analysis,   NFIRS   data   were
extrapolated  to  represent   nationwide  estimates  and  were  then
averaged  over  the  period of   1980  through  1985.   (Detailed
information obtained from NFIRS is presented in  the  appendix to
this report.)   Since NFIRS  statistics  are  compiled on an annual
basis,  the data   cover  both  pre- and  post-standard  vehicles.
The important motor  vehicle fire estimates (expressed as annual
average values for both collision  and  non-collision  fires)  from
the NFIRS data include:

           488,000  total  motor  vehicle fires  (of all  vehicle
           types).
           10,800   motor  vehicle  fires   (2.2  percent)  which
           originated in the fuel tank area.
           237  fatalities    in   motor  vehicle  fires   which
           originated in the fuel tank area.
           631 injuries in  motor  vehicle  fires which originated
           in  the  fuel tank area.
           $32 million  in  property damage from  motor  vehicle
           fires  which originated in the fuel tank area.

The NFIRS  annual  average nationwide  fire  estimates using  data
from 1981 to 1985  for post-collision fires  exclusively include:

           1,871   motor  vehicle  fires  which  originated in  the
           fuel tank area.
           185  fatalities    in   motor  vehicle  fires   which
           originated in the fuel tank area.
           336 injuries in  motor  vehicle  fires which originated
           in  the  fuel tank area.
           $14 million  in  property damage from  motor  vehicle
           fires  which originated in the fuel tank area.

     Another accident data  base,  the MCS  250  Accident  Reports,
which   is  operated  and   maintained  by  the  Federal  Highway
Administration/OMCS,   also  reports   the   occurrence   of  motor
vehicle fires  in   accidents.   As  is  prescribed  in the  Code of
Federal  Regulations   (49  CFR 394.3),  all  heavy-duty  vehicle
accidents  in  which death,   bodily  injury,  or property  damage of
$4400 or  greater  have occurred  must be reported to this  data
base.   Since the  data  from this  source only covers one class of

-------
                               -7-
vehicles and its nationwide  representativeness of all accidents
can not  be  determined,  the  MCS 250 Accident  Reports were  not
used  for  this  analysis.   Other sources of  helpful  information
for this assessment include  previous analyses  by vehicle safety
experts.  These analyses will be presented and discussed below.

     B.    Previous Motor Vehicle Fire Analyses

     "Evaluation  of  Federal   Motor   Vehicle   Safety  Standard
301-75, Fuel System Integrity:   Passenger  Cars," Glenn Parsons,
NHTSA (1983)[12]

     This analysis  was  published by NHTSA in  1983  to evaluate
the effectiveness  of  the improved  FMVSS  301-75.   The  report
characterizes  the  in-use  accident  and fire experience  of  both
pre-  and  post-standard  vehicles.   The data  contained  in  this
report  are   obtained  primarily from  Michigan  police accident
reports  for  the  calendar  years  1978,   1979,  and  1980  for
accidents   involving   passenger  vehicles  only.   From  data
presented in this  report,  Tables 1  and 2  show the  distribution
of  impact  type   and   vehicle   damage  severity  in  accidents
involving both  pre- and post-standard  passenger cars.   Table  1,
which  presents  the frequency   of  accidents  from   the  various
impact   types,   shows   that   rollover   accidents    are   rare
occurrences   (about 1.4  percent  frequency)  and  most  accidents
are frontal  impacts (about  60.9 percent  frequency).    Table  2
presents the damage severity sustained by each vehicle involved
in an accident based on a Vehicle Damage  Severity  (VDS)  scale.
A VDS  of  1  represents  very minor vehicle  damage, whereas a VDS
of 8  represents  extreme damage.  As expected,   this  table shows
that accidents  with Lo-Moderate severity  damage (VDS  = 2, 3,  4,
5) are more  frequent in comparison  to  major  severity  accidents
(VDS =6,7,8).

     Furthermore, this   report supplies data which characterizes
the  occurrence  of  fires   in   accidents   involving  pre-  and
post-standard  vehicles.   Table  3  presents   the   fire  rates
(number of  fires per number  of  vehicle crashes)  for  both types
of vehicles  by impact  type and  damage severity.  The  data show
that  rollover  and  rear end  collisions,  especially those  with
major damage severity,   have  the highest fire rates.    This table
also shows  that  FMVSS   301-75 gave  a  greater  reduction  in  the
fire rates  for major severity accidents (50.6 percent) than for
Lo-Moderate  severity accidents  (22.4  percent)  of these  impact
types.

     Other data  provided  by  this NHTSA technical report include
the number  and  degree  of injuries  which  result from  vehicle
accidents   and   post-collision   fires.    It   categorizes   the
injuries  into    four  types:    fatal   injuries,  incapacitating
injuries, non-incapacitating injuries,  and possible  injuries.
In addition to the  key  data  with regard to  fire rates shown  in
Table  3,  some  further  information  presented  in   this  report
include:

-------
                         -8-





                       Table 1



      Distribution of Accidents by Impact Type*
Impact
Type .
Frontal
Rollover
Rear end
Other (Side,
Pre-Standard
Freo^iency (%)
62.9
1.5
22.9
etc.) 12.8
Post-Standard
Frequency (%)
59.7
1.4
24.5
14.4
Overall
. Freojiency (%)
60.9
1.4
23.9
13.8
Based on Michigan data for calendar year 1980.

-------
                              -9-


                            Table 2

     Distribution of Accidents by Vehicle Damage Severity*


Vehicle
Damage       Pre-Standard       Post-Standard      Overall
Severity     Frequency (%)      Frequency (%)'      Frequency (%)

   1             3.858              2.892              3.26
   2            25.595             26.108             25.92
   3            26.134             27.162             26.78
   4            20.220             20.532             20.42
   5            12.447             12.398             12.42
   6             6.552              6.350              6.43
   7             3. 112              2.821              2.93
   8             2.076              1.707              1.85
     Based on Michigan data for calendar year 1980.

-------
                              -10-
                            Table 3

             Fire Rates for Pre- and Post-Standard
          Vehicles by Impact Type and Damage Severity*

       tt of fires/tt of vehicle crashes = fire rate(xlO~3)
Pre-Standard
  Front end
  Rear end
  Rollover
   TOTAL

Post-standard

  Front end
  Rear end
  Rollover
   TOTAL
                     Lo-Moderate Severity
                     (YDS = 2,3,4,5)
360/229,433 = 1.569
139/99,663 = 1.395
14/2946 = 4.752
513/332,042 = 1.545
311/261,002 = 1.192
139/119,141 = 1.167
10/3541 = 2.824
460/383,684 = 1.199
                        Majority Severity
                        (YDS = 6,7,8)
91/9465 = 9.614
90/6304 = 14.28
49/3389 = 14.46
230/19,158 = 12.01
79/14,425 = 5.477
40/6944 = 5.76
29/3599 =8.05
148/24,968 = 5.93
Reduction from Standard
             22.4%
            50.6%
     Based on Michigan  data  for calendar years  1978,  1979,  and
     1980.   Information for  other  types of  impacts  were  not
     reported.

-------
                              -11-
           20,600  annual  motor  vehicle  fires  resulting  from
           passenger car accidents.
           1,099 annual  fatalities  in  passenger car  accidents
           with fire (average from FARS for  1978-1981).
           Total injuries  in all  passenger car accidents:
                   385,892 incapacitating  injuries
                   865,642 non-incapacitating injuries
                 1,095,180 possible  injuries
           Injuries   in    passenger    car    accidents    with
           post-collision  fire:
                 3,867 incapacitating  injuries
                 3,836 non-incapacitating  injuries
                 2,486 possible  injuries

     "Fires in Motor Vehicle Accident," Peter Cooley (1974)[13]

     This analysis  done  by Peter Cooley  of the  Highway Safety
Research  Institute   (HSRI,  now  the   University  of  Michigan
Transportation  Research  Institute or  UMTRI)  provides  data  on
the   total   number   of  vehicle   fires   and   the   resulting
fatalities.    In this  analysis,  Cooley  studies  accidents  of
passenger cars  to  estimate the number  of post-collision fires.
Furthermore,    he   analyzes   the   fatalities    involved   in
post-collision  fires  and  determines  whether  the  fatality  was
"accompanied by"  or  "directly  caused   by"   the  post-collision
fire.  Unfortunately, his data do not  specify the source of the
fire (i.e.,  fuel tank).   Since this  report was written in 1974,
the  data  are  based entirely on pre-standard  vehicles.   The key
data presented in this analysis  include the  following:

           17,000  annual   fires   result   from   motor  vehicle
           accidents.
           720-1,250 fatalities  are  accompanied by these fires.
           450-650   fatalities  are  directly  caused   by  the
           vehicle  fires.

     "General  Motors and Fuel System Collision Fires"  (1974)[14]

     This study  provides  General  Motors'- point  of  view on the
fire  hazards  of motor  vehicles.   It  offers statistics  on the
number   of   passenger  car   fatalities  caused   directly   by
post-collision   fires   and  more   specifically,    fuel   system
collision- fires.   Since  this study was performed  in  1974,  it
only relates to pre-standard vehicles.   Key  data  include:

           560-1,870 annual  fatalities  due to collision fires.

           250  annual  fatalities  due  to  vehicle  fuel  system
           collision fires.

-------
                              -12-
     "A  Perspective  on  Automobile  Crash  Fires,"  SAE  850092,
     Warner,  James and Woolley,  (1985)[15]

     This SAE  publication  analyzes  previous  studies  regarding
automobile  crash   fires.    The  major   conclusion  from  this
analysis   is   that   post-collision   fuel-fed   fires   cause
approximately 1-1.5  percent of  all  vehicle occupant  fatalities
in spite of  the FMVSS 301 improvements.

     This concludes  the  data  available  on  fuel  tank-related
fires which  were  used in  this  analysis.    Based  on information
gathered from these  motor  vehicle  fire  data bases  and previous
studies, the fire hazard of current vehicles can be estimated.

III. Analysis of Data

     As  shown   in  the previous section,  Summary  of  Available
Data,  the  data  bases and  previous  analyses  regarding  motor
vehicle  fires  provide different  types  of  information covering
different time  periods.  With careful analysis,  the information
obtained from  these sources can be  used to  generate  estimates
of fuel  tank-related  fires,  fatalities,   injuries,  and property
damage  for vehicles operating  in the post-1990  time  frame.   In
the  post-1990  time  frame,   essentially  all  light-duty vehicles
and  light-duty  trucks would be in compliance  with  FMVSS 301-75
and  onboard   refueling   controls  would  be  required  on  these
gasoline-powered motor vehicles.[9]  Therefore,  for purposes of
this  analysis,  the  year  1990  is  used  to  adjust  the  data  by
vehicle miles travelled (VMT) to account  for such a time frame.

     Before   these   data  can  be used  to  generate  the  desired
estimates,  two  adjustments  are  necessary to  get  all  data  on a
common  basis.   First, all  data must be  put  in  terms  of  post-
standard fire rates, since  refueling controls  would be required
on  post-standard  vehicles  only.   To put  the  data in  terms of
post-standard  rates,  the  fraction  of   fires  contributed  by
pre-standard  vehicles   was  determined   using   the   estimated
percentage  of  VMT  by  these  vehicles  when   the  data  were
taken.[16,17]   These  pre-standard  fractions were  then- adjusted
to  post-standard  rates  based  on  the  effectiveness  of  FMVSS
301-75  given  in the  aforementioned NHTSA report.[12]   Since it
is  reasonable  to  assume   that  most   rear-end   and  rollover
accident fires  occur  in  the fuel  tank area, a fuel tank-related
fire reduction factor for FMVSS 301-75 was  calculated  using the
pre-  and post-standard  fire rates  for  rear-end  and rollover
collisions shown  in Table  1.   For purposes  of  this  analysis,
these   overall   post-collision   and   fuel  tank-related   fire
reduction factors  for FMVSS 301-75  were estimated to  be  32.7
and 36.9 percent,  respectively.

     The  estimated  fire  reduction  factors  for  FMVSS  301-75
discussed above are based on fire data involving  only passenger
cars.   It  should be  noted  that  these  adjustment  factors  were

-------
                              -13-


applied to  all data  sources,  including  PARS and  NFIRS,  which
include fires from other vehicles besides  passenger cars (e.g.,
light-duty trucks, buses, and heavy-duty  vehicles).  Absent any
other data,  it  is  reasonable  to assume that FMVSS  301-75  would
reduce fires involving  light-duty  trucks  and  school buses  as
effectively  as the  NHTSA  report  shows  for  passenger  cars.
Using  the  fire reduction  factors  on  heavy-duty vehicle  (over
10,000  pound  gross  weight)   data,   however,   could  possibly
underestimate the number of  fires and their consequences,  since
these vehicles do not have to comply with  FMVSS  301-75.  To put
such  a possible  underestimation  in proper perspective,  it  is
important  to note that heavy-duty vehicles  account for  less
than five percent  of  the in-use vehicles and VMT each year, and
information  in  the appendix  to this report  suggests  that  these
vehicles  contribute  only  a  few  percent  of  the  annual  motor
vehicle  fires. [ 18,19]   Nevertheless,  it  is  important  to  note
that  by   1973  all   fuel   tanks   manufactured  for  heavy-duty
vehicles which  are involved  in interstate  commerce must comply
with  OMCS  performance  requirements.[20]    Therefore,  applying
the  fire  reduction  factors   estimated from  NHTSA's  report  to
data sources which include  fires  from heavy-duty vehicles (FARS
and  NFIRS)   should not introduce  an   appreciable  error  to  the
fire estimate projections presented in this  analysis.

     Second, after  the data  are  adjusted to post-standard fire
rates  it  is also necessary  to  consider  that  increasing  VMT
between  the time  when the data  were generated  and 1990  would
increase  the absolute  number of  fuel tank  fires  and  related
after  effects.    It  is  logical  to project  that  absent  other
measures, fires and  subsequent fatalities,  etc., would  increase
as VMT  increases.   It  should be noted that  fatality  rates based
on   VMT   have  generally   declined   over  the   last  several
years.[21]   However,  absent  any other method to account for the
effects  of  the increase in  vehicles  and  VMT between the time
when  the  data were  collected  and  1990,   using  VMT for  this
adjustment  seems  reasonable.   Thus,  all  data were  adjusted to
1990 using  the ratio of the  1990 total VMT to the total VMT of
the year(s)  in which the data were  generated.[18]

     One   other  adjustment  which  was    considered but  not
incorporated into this analysis  is the  effect of vehicle aging
on  fire  rates.   If  vehicle  aging  does  affect  fire  rates,  the
data  adjustments  for pre- to post-standard vehicles would have
to  be adjusted further.  This concern arises  because the pre-
and  post-standard fire rate data used to develop the adjustment
factor  did not use  all vehicles  of  the  same  age.   The  post-
standard  fire  rate was based on vehicles  0-4 years  old, and the
pre-standard rate was based on  vehicles  3-8  years  old.[12]
Thus,  there is the possibility that the  adjustment  factor could
be too large since it  incorporates  the effect of different fire
rates   on   different   age   vehicles,   in  addition  to  the
improvements brought by the safety  standard.

-------
                              -14-
     Fire  rates   (fires  per  crash)   may  vary  by  vehicle  age
because  changes  in driving  patterns  (less  high  speed freeway,
more lower speed  urban  area),  ownership,  fuel  system component
deterioration,  and lack  of  maintenance  with  vehicle  age  may
have an effect on  the  possibility of  a fire in the  event  of an
accident.  The  net effect of these factors was  evaluated  using
data from  the NHTSA technical  report.[12]   Using this  report,
preliminary analysis of  the data  concluded  that the  effect of
aging  is  probably  small.   However,  an  accurate  and   more
conclusive    determination     (either    quantitatively     or
directionally) cannot  be  made  until   more  information  becomes
available  with  greater  range  in  vehicle  age.    NHTSA plans to
update  their  evaluation  of  FMVSS 301-75  in  the   near  future
based  on  more  recent  data.   Regarding the  effect  of  vehicle
aging,   it  is  interesting  to  note,  that  the  1985   PARS  data
suggest  that  the  percent  of vehicle  accidents  (not  fires) by
model year  are  directly proportional  to  registrations  of  that
model  year's   vehicles,  not   VMT.[21]    This   supports   the
hypothesis  that   vehicle  aging   has   no  effect   on  vehicle
accidents.

IV.  Development of 1990 Vehicle Fire  Data

     The  data on  motor vehicle  fires extracted from  the  six
sources   were   collectively   analyzed,    according   to    the
methodology described  above, to  estimate  the annual  number of
post-collision and fuel tank  related  fires  and the resulting
fatalities,   injuries   and  property   damage.    As   previously
mentioned,  to  be  useful  the characteristics  of all  available
data must  be put  in  terms  of  a  common basis.   Therefore,  the
fire  data  were  scaled  to  represent  post-standard  vehicles
operating  in  the  year  1990 when essentially all vehicles in-use
will meet FMVSS 301.

     Very  few  sources   directly   provide  statistics  for   fuel
tank-related fires.  In  fact, only three  of the sources studied
provide such  statistics:   NFIRS,  the  General  Motors  study,  and
SAE  paper   850092.[11,14,15]   Another  method  to estimate  the
statistics for fuel tank  fires  in particular  is  to  approximate
the  percentage  of  total  post-collision  fires  which  are  fuel
tank-related.   Since it  is logical to  assume  that  most of  the
fuel  tank-related  fires  result   from rollover  and  rear  end
collisions, this  percentage  can   be  estimated  from  the  total
fraction of  these  collision  types.   Using  this  assumption  and
the  NHTSA  report,  it  can be  calculated  that  fuel  tank  fires
account  for   approximately  31  percent of  the total  collision
fires.[12]   With   this  percentage  factor,  post-collision  fire
data from  the remaining  sources   can  be  used  to estimate  the
number of fuel tank-related fires  and  their  subsequent effects.

     Table 4 presents  the  1990 post-standard motor  vehicle  fire
data generated from the  six  sources used  in this analysis.   The
data presented include  the number  of vehicle  fires  and/or  their
aftereffects for post-collision fires  and particularly for fuel

-------
                                    -15-


                                  Table 4

                    1990  Post-Standard Vehicle Fire Data
Source/Statistic

PARS
Fires with Fatalities
Fatalities

NFIRS
Vehicle Fires
Fatalities
Injuries
Property Damage

NHTSA Report
Vehicle Fires
Fatalities
Injuries:
  Incapacitating
  Non-Incapacitating
  Possible

Cooley's Report
Vehicle Fires
Fatalities

GM Study
Fatalities

SAE 850092
Fatalities
Post Collision
    Fires
    1,629
    1,679
                                             Fuel Tank-Related Fires
    19,550
    1,043

    3,669
    3,640
    2,350
    15,313
    405-1,126
     505-1,685
Collision
    505
    520
 Collision &
Non-Collision
                       1,871
                       185
                       336
                       $14 million
    6,060
    323

    1,137
    1,128
    728
    4,747
    126-349
    211
                        560-839
                   10,730
                   236
                   625
                   $32 million

-------
                              -16-
tank  fires  (for  both  post-collision  fires  only  and  post-
collision  and  non-collision  fires).   As previously  mentioned,
only the  NFIRS,  GM Study,  and SAE 850092  directly report data
for  fuel  tank-related  fires.   Fuel  tank  fire statistics  for
collisions  from  the   remaining   sources  (PARS,  NHTSA  report,
Cooley's  report)  were estimated  by  scaling  the  post-collision
fire data  with the percentage of fuel tank  fires  approximated
from  Parsons'  report  (31  percent).   Therefore,   the  data  in
Table  4  encompasses  all  of the  available  data and  presents  a
reasonable range  of the  fire hazards  for motor vehicles  in  the
year 1990.

     The  data  from the six sources presented in Table  4  can be
summarized for both  the  post-collision  and  fuel  tank-related
fires as presented below:

     •     Post-Collision Motor Vehicle Fires  (annual
           projections)

                 15,300-19,600 annual  fires
                 400-1,700 fatalities
                 3,700 serious injuries
                 3,600 moderate injuries

     •     Fuel Tank-Related Fires (annual projections)

               .  4,750-10,700  fuel tank-related fires (for both
                 collision and non-collision fires)*
                 1,870-6,060   fuel   tank-related   fires   (for
                 collision fires  only)*
                 126-839 fatalities
                 625-1,140 serious injuries (for both collision
                 and non-collision fires)
                 336-1,140 serious injuries (for collision fire
                 only)
                 1,130 moderate injuries
                 $32   million   property   damage    (for   both
                 collision and non-collision fires)
                 $14  million  property  damage  (for  collision
                 fires only)

     These  statistics,   based   on   extrapolations  to   1990,
represent  a  reasonable  range for the  fire   hazards of  motor
vehicles for the  post-1990  time  frame, when  refueling  controls
may be required.   As  can  be seen from these statistics,  a large
percentage of  fuel tank-related fires  are non-collision  fires.
     Statistics for non-collision fires  only  cannot be obtained
     directly  by  subtracting these  ranges.    These ranges  are
     based on  several sources which  report  collision  fires  or
     collision and non-collision  fires.

-------
                              -17-
V.   Summary and Conclusions

     In response to the  recent  concerns regarding the potential
safety   implications   of   onboard   refueling   controls,   this
analysis  estimated the  potential hazard  of  fuel  tank-related
fires  for both  current  motor  vehicles  and  in the  post-1990
timeframe  when  essentially  all   in-use   passenger  cars  and
light-duty  trucks  would  meet  FMVSS 301-75.   This  assessment
provides  a  baseline  from which  to  assess  any change  in risk
which could occur  as  a result of the addition of  onboard vapor
recovery systems.

     As shown  in this report, very few sources of motor vehicle
fire  data exist which accurately  assess  and  characterize  the
fire  hazard  for   current  vehicles.   This  analysis uses  fire
information from two data bases and  four past analyses  of motor
vehicle  fires  to  perform a vehicle  fire  hazard  assessment.
With careful analysis, the  fire  information obtained from these
sources  were  used to estimate  a range  for the number  of fuel
tank-related  fires  and their consequences  for vehicles  in  the
post-1990 time frame, when onboard  refueling controls  could be
required  on gasoline-fueled  motor  vehicles.   In  order  to  put
all data  on a common basis,  this analysis accounts for both the
effectiveness of FMVSS 301-75 and the increase in  VMT from when
the  data were generated  and  1990.   From  this  analysis,  the
following statistics  summarize  the  hazards  of fuel tank-related
fires  for motor vehicles on  an  annual  basis, projected for the
post-1990 timeframe:

     -     1,870-10,700 annual fuel tank-related fires

           126-839 fatalities

     -     336-1,140  serious  injuries

           1,130 moderate  injuries

     -     $14-$32 million  property damage

-------
                              -18-
                           References
     1.    "Safety  Implications   of   Onboard  Refueling  Vapor
Recovery Systems," U.S. EPA-AA-SDSB-87-05, June 1987.

     2.    "Analysis   of   Fuel    Tank-Related   Fires,"    EPA
Memorandum, Kathleen  A.  Steilen to  Charles L. Gray,  U.S.  EPA,
OAR, QMS, OMS, ECTD, SDSB, June 10, 1987.

     3.    National  Fire   Incident  Reporting   System,   Motor
Vehicle Fire Tallys, FEMA, .U.S. Fire Administration, 1980-1985.

     4.    Letter,  Ralph  J.   Hitchcock,  NHTSA  to  Chester  J.
France, U.S. EPA, September 21, 1987.

     5.    Letter,    John   Glenn    Hart   III,    U.S.    Fire
Administration to  Kathleen A.  Steilen,  U.S.  EPA, September 16,
1987.

     6.    Motor Vehicle  Safety Standard  No.  301,  Fuel  Tanks,
Fuel Tank  Filler Pipes,   and  Fuel  Tank  Connections  - Passenger
Cars:  32 FR 2416, February 3, 1967, Part 571: S  301-1.

     7.    Motor  Vehicle   Safety   Standard  No.   301-75,   Fuel
System  Integrity:  39  FR  10588,   March  21,   1974,   Part  571:
S 301-75-5.1,  5.2, 5.3.

     8.    Motor  Vehicle   Safety   Standard  No.   301-75,   Fuel
System  Integrity:   40  FR  48352,  October  15,  1975,  Part  571:
S 301-75-5.1,  5.2, 5.3, 5.4.

     9.    "Control of  Air Pollution From  New  Motor Vehicles
and  New Motor Vehicle  Engines; Refueling  Emission Regulations
for   Gasoline-Fueled   Light-Duty   Vehicles   and   Trucks   and
Heavy-Duty  Vehicles,"  Notice  of  Proposed  Rulemaking,  FRN:  pp.
31162-31271, August 19, 1987.

     10.   Fatal   Accident   Reporting    System,   NHTSA,   DOT,
1,980-1984.

     11.   National Fire  Incident  Reporting System,  FEMA,  U.S.
Fire Administration, 1980-1985.

     12.   "Evaluation of  Federal  Motor  Vehicle Safety Standard
301-75,   Fuel   System   Integrity:    Passenger   Cars,"    DOT
HS-806-335, January 1983.

     13.   "Fires In Motor  Vehicle Accidents," UM-HSRI-SA-74-3,
Peter Cooley,  April 1974.

-------
                              -19-


     14.   "General  Motors  and  Fuel  System Collision  Fires,"
Environmental Activities  Publication  No.  A-3177, December  16,
1974.

     15.   "A  Perspective  on  Automobile   Crash  Fires,"   SAE
850092, Warner,  James and Woolley, 1985.

     16.   "Revised  Diesel  Sales Fraction  Projections,"  U.S.
EPA  Memorandum,  John W.  Mueller  to  Charles  C.  Gray,  January 9,
1987.

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

     18.   "MOBILES  Fuel  Consumption  Model,"   U.S.  EPA-AA-TEB-
EF-85-2, Mark A. Wolcott and Dennis F. Kahlbaum, February 1985.

     19.   "MVMA  Motor  Vehicle Facts and Figures  1987," Motor
Vehicle  Manufacturers  Association of  the United  States,  Inc.,
1987.

     20.   "Federal  Motor  Carrier Safety Regulations and Noise'
Emission Requirements," Bureau  of Motor Carrier Safety, Federal
Highway Administration, U.S. DOT, October 31, 1983.

     21.   "Fatal  Accident  Reporting  System 1985, A Review of
Information  on  Fatal Traffic  Accidents  in  the U.S.  in 1985,"
NHTSA, DOT,  1985.

-------
     Appendix





NFIRS Detailed Data

-------
                                  A 2
                               Table A-l

                  NFIRS Tally by Mobile Property  Type

           - Annual  Nationwide Projection Averages 1980 through 1985
           - Collision and Non-Collision Fires
Mobile Property Type
Automobile
Bus, Trackless Trolley
Terrain Vehicles
Pass. Road Trans, Other
Pass. Road Trans, Unc.
     Pass. Road Trans

Truck - Over 1 ton
Truck - Under 1 ton
Semi-Trailer Truck
Tank Truck - Non-Flam.
Tank Truck - Flam Lqd.
Tank Truck - Comp. Gas
Trash Truck
Freight Road Trans, Other
Freight Road Trans, Unc.
     Freight Road Trans
 Fires
Injuries   Deaths   Loss (x!06$)
443,958
3,622
8,212
1,458
3,463
460,713
19,382
35,994
7,821
289
1,131
372
3989
530
4961
74,469
3,344
41
246
29
23
3,683
260
455
114
10
110
14
15
8
31
1016
726
4
24
1
2
758
53
113
46
2
20
1
0
0
10
245
689.92
10.31
11.30
2.87
1.96
716.36
60.86
49.78
59.65
1.43
10.50
1.91
8.05
2.15
16.99
211.33
Total
535,183
  4,699
1,003
927.68

-------
                                 A 3
                              Table  A-2

                 NFIRS Tally by Area of  Fire Origin*

           - Annual Nationwide  Projection Averages 1980 through 1985
           - Collision and Non-Collision Fires
Vehicle Area         Fires      Injuries    Deaths     Loss (x!06$)

Passenger Area      100,616         868      202        250.52
Trunk Area           18,659         274       30         36.79
Engine Area         332,681       1,964      286        388.92
Fuel Tank Area       10,828         631      237         32.18
Control Area          7,121          58        8         17.70
Exterior Surface      6,931          80       14         10.97
Not Classified       11,179         136       43         21.66
Total               488,015       4,010      821        758.74
     This tally  includes  only vehicle  fires  in which  the fire
     originated  in  the  vehicle.   The total is  smaller  than the
     total  of Table  A-l  because  it  excludes  apparent  NFIRS
     reporting errors  and vehicles that were  involved  in fires
     which did not originate within the vehicle.

-------
                                  A 4
                              Table A-3

                       1990 Post - FMVSS  301-75
              Annual Projections for Motor Vehicle Fires

                 - Collision and Non-Collision Fires
                 - Based on 1980-1985 data
Area of Fire Origin     Fires     Injuries   Deaths    Loss (xlp6$)

Passenger Area         101,300       874      204       253.04
Trunk Area              18,782       275     -  30        37.12
Engine Area            335,358     1,978      287       393.43
Fuel Tank Area          10,731       625      236        31.97
Control Area             7,175        58        9        17.88
Exterior Surface         7,005        80       14        11.09
Not Classified          11,255       137       43        21.86
Total                  491,605     4,028      822       766.39

-------
Year
                                       A 5
                                          Table A-4

                              NFIRS Annual Fuel Tank Fire Tally

                                    -  Collision Fires Only
        Nationwide  Projections
                                  1990  Post-FMVSS  301-75  Projections

1981
1982
1983
1984
1985
Fires
2418
1945
1771
1725
1494
Injuries
366
353
329
387
245
Deaths
228
159
160
222
154
Loss(M$)
17.66
10.76
12.08
16.51
12.75
Fires
2363
1942
1786
1747
1515
Injuries
357
352
332
392
248
Deaths
223
159
162
225
156
Loss (x!05$)
17.26
10.74
12.18
16.71
12.92
Avg.
1871
336
185
                                     13.95
1871
336
185
13.96

-------