EPA-AA-ECTD-87
Summary and Analysis of Comments  on the
      Recommended Practice for the
   Measurement of Refueling Emissions
              March, 1987
Standards Development and  Support Branch
  Emission Control Technology Division
       Office of Mobile Sources
      Office of  Air  and Radiation
 U. S. Environmental Protection Agency

-------
            Summary and Analysis of Comments on the
                  Recommended Practice for the
               Measurement of Refueling Emissions
I.    Introduction

     As a  result  of concerns  about the  emissions which  occur
when gasoline vapors  are displaced  from  fuel tanks during  the
refueling of motor  vehicles,  EPA  has  been  examining  the  need
for the control of  these refueling emissions and the methods to
do so.  One  such  method  involves  the collection on  the  vehicle
of  the  displaced  hydrocarbons  and  the  measurement  of  the
effectiveness of the refueling  vapor control system.  This  type
of  control  is  referred to  as  onboard  control  of  refueling
emissions.   On  August  22, 1985, EPA transmitted  to  interested
parties two technical reports concerned  with the measurement of
refueling  emissions.   One  report,  "Refueling  Emissions  from
Uncontrolled Vehicles,"[l ]  detailed EPA's  baseline  emissions
measurements of  refueling  emissions  and  the  second  report,
"Draft  Recommended  Test  Procedure for   the  Measurement  of
Refueling  Emissions", [2] presented  a test  procedure  for  the
determination  of  the   effectiveness  of   onboard  control  of
refueling emissions.  These reports  were accompanied by  a draft
recommended  practice,  "Subpart C  - Refueling  Emissions  Test
Procedure."[3]

     Recipients  of  the  reports and draft  test procedure  were
requested  to  review and provide  comments on  EPA's  recommended
test procedure,   including comments  on the test parameters  and
the test  equipment.  As a result  of  on-going EPA  analyses of
the test  procedure  issues and the comments provided  by  the
reviewers,  EPA  revised  the test procedure.   On April  10,  1986
EPA convened  a  technical  meeting  to  present  and discuss  the
revised  refueling test  procedures.  In   addition  to  the  oral
comments provided  during the  meeting,  EPA  requested  that  the
participants  provide  written  comments   on  the  revised  test
procedure.    Comments  on both  the  original  and  revised  test
procedures  were  received  from  the following organizations:

           American Petroleum  Institute  (API)
           California Air Resources Board (GARB)
           Chrysler Corporation
           Ford  Motor Company
           General Motors Corporation (GM)
           Motor Vehicle  Manufacturers Association (MVMA)
           Nissan Research and Development, Inc.
           Radian Corporation
           Toyota Motor Corporation
                              -2-

-------
     This  document  presents a  summary of  the comments  on the
recommended  refueling  test procedure,  EPA's  analysis  of the
issues raised by the commenters,  and  the  resulting changes made
to the recommended test procedure.

     The  remainder  of  this  document  is  subdivided  into two
major  sections.   Section II presents  the summary  and  analysis
of test procedure  issues.  The  comments received on a particular
issue  are  first  identified and  then followed  by  EPA's  analysis
and  response.   Section  II' is  subdivided  into six subsections.
These  subsections address:  test parameters, fuel  tank  heating,
facility  requirements,  canister  loading,  preconditioning,  and
miscellaneous  issues.   The final section,  Section III,   is  an
overall  description  of  the  test  procedure  which  has  been
developed as a result of the comments  and  EPA's  analysis  of the
comments.    The  Appendix  following  Section  III  describes  the
canister testing program carried  out  in support  of  the  analyses
in this document.

II.   Summary and Analysis of the Comments

     A.    Primary Parameters Affecting Refueling Emissions

     In the draft  recommended procedure,  five  key  parameters
affecting   refueling    emissions  were    identified.     These
parameters  were:   dispensed   fuel   temperature,   differential
temperature between  dispensed  fuel  temperature  and  fuel  tank
liquid temperature,  fuel volatility,   fuel  dispensing rate, and
fuel  level prior  to  refueling.   The  values  for  these  key
parameters directly  affect refueling  emissions  and were  chosen
with  the   goal   of  insuring  emissions  control  for most  all
expected  in-use  conditions.   To do  this,  the  values  of  each
parameter   were  chosen  at  approximately  the  90th  percentile
point from distributions of  in-use  survey data.   Test parameter
values  as  originally  proposed   in   the  draft   recommended
procedure  are  listed  in Table  1.  Revisions  have  been made to
three of the five test values for the parameters  as a result of
comments  received  and further  EPA  analyses.  While the reasons
for   these  changes  are  discussed  in  the  remainder  of  this
section and subsequent  sub-sections,  the  revised values  are
listed here in Table 2 for ease of comparison.

     1.    Temperature Specifications  for  Dispensed  Fuel  and
           Liquid Fuel In The Vehicle Tank

     In commenting  on  the  stringency  of  the  refueling  test
parameters, a  number  of  motor  vehicle  manufacturers  took the
position   that   the   values   were   overly  stringent.     The
manufacturers  stated that  the selection  of the  90th  percentile
of both the dispensed  fuel  temperature and the tank temperature
would result in  greater  than the 90th percentile  of refueling
events being  represented by the  test procedure.   According to
these'manufacturers,  the test values selected  in  the  draft test
procedure  would  represent  approximately the 99th percentile of
refueling  events.

                              -3-

-------
                                                  Table  1
                                        Draft Recommended Procedure,
                                          Critical Test Parameters
    Parameter

    1.  Dispensed Temperature,  TD

    2.  Temperature Differential,AT


    3.Volatility,  RVP



    4.  Dispensing Rate
i    5.Fuel Level
i
Meaning

Temperature of dispensed fuel

Tank temperature minus
dispensed fuel temperature

Test fuel volatility
expressed in Reid
Vapor Pressure

Flow rate of fuel as it is
dispensed

Level of fuel in vehicle
prior to refueling.
Percent of capacity
to nearest 0.1 U.S. gal
Value

88 + 2°F

+2 to +5 °F


11.5 + 0.5 psi



8-10 gal/min


10%

-------
                                                   Table 2
                                      Revised Critical Test Parameters
i
Ln
I
    Parameter

    1. Dispensed Temperature, TD

    2.Tank temperature TT

    3.Volatility, RVP
    4. Dispensing Rate
    5.Fuel Level
Meaning

Temperature of dispensed fuel

Soak area temperature

Test fuel volatility expressed
in Reid Vapor Pressure
Plow rate of fuel as it
is dispensed
Level of fuel in vehicle
prior to refueling.
Percent of capacity to
nearest 0.1 U.S. gal
Value

81-84°F

80 + 3°F

In range of 8 to
11.5 psi (Final
determination to
be made on
results of
Volatility Study)

Refueling
measurement:
9.8 ± 0.3
gal/min.
Canister loading:
3-4 gal/min.

10%

-------
     The commenters  are  correct in their basic  contention  that
the  selection  of the  90th percentile  for  the  dispensed  fuel
temperature  and  fuel  tank  temperature will  result  in   the
combined percentile  being  higher  the  individual  percentiles.
EPA disagrees, however, with  the assertion  that  the test values
would represent the 99th percentile of  refueling events.  First
of  all,  the  parameters  are  not  fully independent  variables,
making  it   difficult  to  assess  the  combined  probability  of
occurrence  of  extreme values.  Second,  both  parameters  do  not
have  to  be at  their  90th  percentile values  to generate  high
emissions.   As  the value  of one of the  parameters  rises beyond
that  point,  the other  can fall  correspondingly below   its  90
percent  value   and   still  produce  overall  high  emissions.
Therefore,   to  analyze  the total effect  of  these parameters  on
refueling  emissions,  EPA went  back   to  the basic  field survey
data  and  constructed  an  estimated emission  rate  distribution
from the fuel and tank temperature  data.

     The  dispensed  temperature   and  AT   data   used  in  this
distribution were taken from a  1975 gasoline  temperature survey
conducted  for  the  American  Petroleum  Institute  (API)   by  the
Radian Corporation,  the  same data that  was used  in  the draft
recommended procedure  report.[4]   The temperatures  are from the
four ozone-prone regions in the country (shown in Figure 1)  for
the. critical  months  of   May  through  September.   The  fuel
volatility  was  assumed   to  equal  the   ASTM  upper   limit.
Refueling  emission  rates were  calculated  from  the  survey  data
using the  following  emission factor  equation  developed  by  EPA
from refueling test data from uncontrolled vehicles:[l]

Emissions (g/gal) = -5.909 -0.0949(AT) +0.0884(TD) +0.485(RVP)

The  distribution of  the  estimated refueling emission rates  is
presented in Figure 2.  Assuming the  individual  90th percentile
temperatures,   TD=88°F,  AT=+2°F,   and  the  fuel  RVP=11.5,   the
resulting  emission factor  is 7.26 g/gal.  As  can  be  seen  from
Figure 2,  this  value  (7.26 g/gal) represents  approximately the
93rd  percentile  of   the  calculated  distribution  for  summer
refueling events.

     Radian  Corporation  performed  a  similar  analysis  in  a
report submitted to  EPA  at  the April 10  workshop.[5]   Radian's
analysis,,  which included  consideration  of relative  refueling
amounts, indicated that  the  specified  test parameters   require
systems  that  control  over 99  percent  of  the  refueling cases
during   ozone-prone   seasons.   While   not   disagreeing  with
Radian's  basic  approach,   EPA believes  that there  are other
factors  which  must  be  considered   in  an  overall   stringency
evaluation.  Perhaps chief among  these factors  is  the   assumed
driving pattern used to evaluate system purge.   As  will  be seen
later  in  the discussions  of  canister preconditioning,  EPA has
used a driving sequence of three trips per  day as the basis for
system  evaluation.   This  pattern  allows  a  fairly  generous
amount  of  canister  purge  and  represents  typical   conditions
                               -6-

-------
Figure l

-------
                                                   Figure  2

                           SUMMER  REFUELING EMISSION  FACTOR  DISTRIBUTION
                                          FOR SELECTED CITIES*
MIDPOINT
           COUNT











/•t
—
n
Ol
o>



K
O
O

u.

z
o
M
V)

2
















o.
. 20000
.40000
.60000
.80000
.0000
.2000
.4000
.6000
.8000
2.0000
2. 2000
2.4000
2.6000
2.8000
3.0000
3. 2000
3. 4000
n.fiooo
3.800O
• 4.OOUO
4 : 2000
4 . 4000
4 . 6000
4.flOl)0
S.OOOO
5.2000
5.400O
5.6000
5.8000
6.0000
6.2000
6 . 400O
6.6000
6.8000
7.0000
7 .2000
7.4000
7.6000
7.8000
8.0000
8 . 2000
8.4000
8.6000
8.8000
9.0000
MISSING
TOTAL
0
0
0
1
1
0
4
5
8
1 1
13
10
19
26
35
37
41
56
65
108
134
1 16
107
129
128
139
15 1
159
139
1 10
124
90
76
55
62
72
67
7O
40
31
16
a
5
4
1
0
528
3001
*
+
*
»x
«x
+
+ x
»x
»x
»x
+ x
• X
»x
»x
»x
«x
«x
• X
«x
«x
* X
»x
»x
*x
* X
•» x
«x
•* X
* X
»x
* X
»x
»x
*x
«x
« X
«x
»x
«x
+ x
»x
»x
»x
»x
*x
+


                                                            Atlanta,  Boston,   Chicago,  Cleveland,   Detroit,,  Houston,
                                                            Los Angeles,  Louisville,   Miami,  Midland,  Oklahoma  City,
                                                            Philadelphia, Pittsburgh,  San Francisco
                tXXXXXXXXXX
                •xxxxxxxxxxxxx
                tXXXXXXXXXXXXXXXXXX
                «xxxxxxxxxxxxxxxxxxx
                4XXXXXXXXXXXXXXXXXXXXX
                •xxxxxxxxxxxxxxxxxxxxxxxxxxxx
                «XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
                «xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
                »xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
                »xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
                ^XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
                »XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
                +XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX)1
                txxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
                »XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
                +XXXXXXXXXXXXXXXXXXXXXXXXXXXX
                «xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
                «xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
                «xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
                +XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
                •> XXXXXXXXXXXXXXXXXXXX
                +XXXXXXXXXXXXXXXX
                tXXXXXXXX
                        EF - -5.9O9 -O.O949(AT)  * 0.0884CT.) * 0.485(RVP)

                        RVP = ASTM Maximum allowable  for the months  involved.

-------
rather than an. upper limit.  Its use has  the  effect  of lowering
the   overall   stringency  of   the  test   procedure   from  that
represented by the fuel parameters.

     A  second stringency  consideration  is  the  fact that,  as
described  in  the technical  report  accompanying the  original
draft  procedure[2],   the   temperature  data   used  represented
smoothed average  values.   Thus,  for example,  the dispensed fuel
temperatures  were five day  averages,  and did  not  contain  the
highs that daily  values  would  give.  Nor  did they represent the
daily  maximun   values,   which  the   report   indicated   would
typically be 4 to 7°F above the average.

     Another  factor  described  in the original  report which has
not  been  directly included  in  the  procedure  is  the  effect  of
fuel  weathering  on   refueling emissions.    The  presence  of
weathered  residual   fuel  in the  fuel  tank  at  the  time  of
refueling,  rather than  unweathered fuel  as used  in the  test
procedure, would be expected to  increase  refueling emissions  by
perhaps 0.5 g/gal.  EPA's  decision  not  to use weathered fuel  in
the  tank  at  this time  is  another   factor  reducing  the  overall
stringency level  of  the refueling  test.   It would  be possible
to   use  weathered  fuel   and   make  some  adjustment   in   the
temperature   parameters,   but   this    change   would  have   no
beneficial effect on the test   and  would  add the  complexity  of
having to handle two different  test  fuels.

     Given all these factors,  the precise overall stringency of
the  test  procedure  is difficult  to  determine.   EPA  believes
that  it  is  sufficient  to insure control  at nearly all expected
conditions,  as was its  original goal.   The chief  impact  of  more
demanding  test  parameters  is   to  increase  required  canister
sizes to hold the increased amounts of  generated vapors.   This
result is not undesirable,  so  long as  no other  system changes
are  required which might markedly increase the  marginal  cost  of
compliance.   If this were  the  case,  then  more detailed analysis
of  test  condition stringency  might be  justified to determine
whether some relaxation might be appropriate.

     2.     Fuel Volatility

     The  draft  recommended  procedure  specified  that the  test
fuel have an  RVP  of 11.5  psi.   This  volatility  represents  the
RVP  of  summer commercial fuel  as  used  in current EPA emission
factors  test  programs,  as  well  as  being  the  ASTM  class  C
volatility upper  limit  for summer  months in  the  ozone  prone
areas of  the  country.    MVMA,  in its  comments,  advocated  that
the  fuel used in  the refueling test procedure have the same RVP
as that of EPA's  current certification  test  fuel, i.e.,  9  psi.
MVMA  understood  and   agreed   with  the  Agency's   desire  to
eliminate  the present  discrepancy  between  summer   commercial
fuel volatility  and  the current certification  fuel  volatility.
MVMA's solution is,  however, to  limit commercial  summer  fuel  to
                              -9-

-------
9.0 RVP as  opposed  to specifying that  the refueling test  fuel
volatility  equal present  commercial  fuel, volatility,  as  the
recommended test procedure specified.

     The  issue  of  test fuel versus  commercial fuel  volatility
is currently  being  examined by  the  Agency and EPA  is  studying
fuel  volatility to  establish  the  best  overall  approach  to
dealing  with  this   issue.   Whatever  the  resolution  of  that
process,  EPA  intends to  adopt  those  results  for  refueling
testing  as  well.   The draft  procedure used  11.5 RVP  simply
because  it  approximated  the  current  in-use  situation.   That
choice  was  not  intended  to represent  resolution  of  the  RVP
issue.   The  procedure  should  more  properly   be   viewed  as
potentially using a fuel with volatility anywhere  in the range
of possible options  being considered by EPA at  this  time,  i.e.,
anywhere from 8 to  11.5 psi.

     An additional volatility concern was  raised  by Nissan.  In
its comments, Nissan  expressed  concern  about  variations  in the
RVP  of test  fuels   and  questioned  how it  can  be  controlled.
This  concern,  i.e.,  the need  to limit the  weathering  of  test
fuel  in the  fuel cart so as  to minimize  test variability,  is
shared  by EPA.   The  revised  procedure contains the  requirement
for the collection  of a  fuel sample and measurement  of  the RVP
immediately prior  to  the  measurement  of  refueling  emissions.
EPA  recognizes  that   this   is  a  worst case  requirement  with
respect   to   its  effects   on   test   facility   and   personnel
resources.  EPA  is  open to  all  suggestions on equipment design
or test data  on the  rate  of fuel  weathering  in the  fuel  cart
which  would  allow  less frequent  measurement  of  the  RVP of the
test fuel.

     3.    Fuel Dispensing Rate

     The  final  area   of  comments  with  respect  to  the  test
parameters  concerned the  recommended  value   for  the  fuel
dispensing  rate.   In  the  draft  recommended  procedure,   the
specified  range  of  8  to  10  gallons  per  minute  (gpm)  was
identified  as covering  the  majority  of  the  refueling  events
while  minimizing nuisance  shutoff of the  fuel nozzle.   Several
commenters  took  issue  with  this  range  for   a  variety  of
reasons.  API  stated  that  some vehicles can not  be  fueled  at a
rate as high  as  8 to  10  gpm and that premature nozzle shut off
at  high  fueling rates  is  associated  with  some  filler  neck
designs.  MVMA commented that spit-back is highly probable  at a
fueling rate  of  10  gpm when the tank approaches  the 95 percent
full  level,  especially with a  liquid  seal.   MVMA  stated  that
GARB  limits  fueling to the  90  percent tank level  at a fueling
rate of 10 gpm, applicable with the 1987 model  year.   MVMA also
stated  that the  specified  fueling rate range of  8  to 10 gpm is
too  broad  and  recommended  a  fueling  rate  specification  of
9.0+0.2   gallons/minute   to   improve  test  repeatability   and
test-to-test and lab.-to-lab. correlation.
                              -10-

-------
     EPA  agrees that  the occurrence  of  nuisance  shutoffs  and
fuel  spillage may  be a  function  of  filler  neck design  when
vehicles  are fueled at  high  flow rates.   However,  it is EPA's
belief that- the design of a  refueling control system which is
capable  of  controlling  premature  nozzle  shutoff  and avoiding
spit-back  at  expected  in-use  fuel  dispensing  rates  is  the
responsibility  of  the  motor vehicle  manufacturers.   Lacking
spit-back  control,  fuel  spillage  from this  source could be  a
major  source  of  refueling  emissions.    Thus,  a  fuel  filler
system  that  prevents spit-back  at  the  upper  limit  of  the
dispensing   rate   is   integral  to  the  effective  control   of
refueling emissions.

     At  the  same  time,   EPA  recognizes  the  fact that,  given
vehicles   designed   to  operate   at   current  maximum  values,
gasoline  marketing  pressures would  be  expected  to  lead   to
increased  in-use  despensing  rates  in  the  future.   In order to
prevent  such a  situation,  it  is  likely that  some  control,
voluntary   or  otherwise,   would  be   required   over   in-use
dispensing rates.

     EPA believes that a maximum dispensed fuel rate  of  10  gpm
is  reasonable  based upon  current  in-use conditions.   The draft
procedure reported that most refuelings take  place  at  10  gpm or
less;  also  GARB has  already  specified  10 gpm  for testing to
demonstrate  compliance  with  its  refueling  control  program.
Thus,  10  gpm  will  continue  to  be  used  as the  approximate
maximum flow rate.

     Turning  to  the  question of  variability  in  test  results
between  tests  and/or between  laboratories with respect  to  the
rate  at  which  fuel  is  dispensed,   EPA  believes  that  some
variability  in  results can be attributable to this factor;  i.e.
fuel dispensing rate.   EPA also  believes  that,  in a  refueling
emissions  test, the  upper   limit  of  the  dispensing rate  is
normally the important criteria.    In selecting a  tolerance band
for  the  fuel dispensing rate to  address  the test variability
concern,  EPA also  recognized  the  need for the use of a  value
which  would  be  achievable  at   a relatively low  cost.   EPA
believes that a tolerance  band of approximately 3 percent at  a
flow  rate  of   approximately   10  gpm,   i.e.   + 0.3   gpm,  will
achieve both objectives.    Combining the selected  tolerance band
with the  objective  of holding the lower   limit close  to  10  gpm
with a minimal exceedance of 10 gpm at the upper  limit resulted
in  the  flow rate  specification  of  9.8  +  0.3  gpm.   Testing
conducted at EPA will normally dispense fuel  as close  to  10  gpm
without  exceeding  the 10.1  gpm  limit  as  possible,  since  this
value would be expected to be the most  difficult test  condition.

     B.    Fuel Tank Heating

     Heating of the fuel  in  the  fuel  tank was required  in  the
draft  recommended procedure  to bring  the  liquid fuel  and  fuel
vapors into  equilibrium  at the required test  temperature.   The
test  procedure  included  a method for heating the   fuel  tank

                              -11-

-------
using  a  single heat  blanket which  allowed  the fuel vapor  and
liquid fuel  to reach an  equilibrium condition before  testing.
Concerns raised  in  the  comments covered a wide range of  areas.
These concerns are summarized below.

     In  its   comments,  MVMA   stated  that   the  fuel   tank
configuration  greatly  influences the  ability  to  heat  the  fuel
and to  achieve the 3°F vapor  to liquid temperature  difference
required in  the  procedure.   MVMA questioned the  feasibility of
the recommended procedure .to  achieve the  required heating  on  a
variety  of  tank   configurations.    MVMA   requested  Iftiat   EPA
demonstrate  the  feasibility of  the  procedure  on several  tank
configurations.  Data submitted by  Ford showed an inability to
achieve  the  required vapor  temperature with  a single  heating
blanket on a Mercury Lynx.

     The  question  of  how  temperature measurements  on  in-use
vehicles (use  of  an external  thermocouple  was proposed  in  the
draft  procedure)  were  to  be made with plastic fuel tanks  was
also raised  by MVMA.    Another  aspect  of  the  fuel  temperature
measurement  issue  which  was  raised  by  commenters   was   the
capability to  read the  true  fuel  temperature when  heating  a
nearly empty fuel tank.   The  thermocouple  would have  to be  very
close to the heat blanket  when the  10 percent fuel volume  was
being heated  and  MVMA stated  that thermocouple  readings  could,
as a result,  be influenced by the heat blanket.

     Comments  on   the  subject   of   fuel  tank   heating  also
identified  test-to-test  variability   as   a   concern.    Toyota
stated that the initial  boiling point of 11.5 RVP  fuel  is under
88°F  and  that this  fuel  property,  in  combination  with  the
specified  temperatures   of  the  fuel  tank  and  of   the  fuel
dispensing  system,   would  result  in  high vapor  losses   and
resultant test to test  variability.   Specifying a heating  rate
was recommended  as  a means  of limiting the rate  of  boiling of
the fuel  and  to  avoid  variability  in test  results  caused by
variability in the heating rate.

     Concerns  about  the  fuel  tank  heating   procedure   and
temperature measurement  requirements  such as those expressed by
the  motor  vehicle  manufacturers  were  shared  by EPA.   As  a
result of  its  experience, EPA  set  out to  revise  the  tank  and
dispensed fuel temperatures  in an effort to eliminate  the  need
for external tank heating and tank fuel temperature measurement.

     Using   the   emission   factor   equation   given   earlier,
alternative dispensed fuel and tank  temperatures  can be defined
which would yield approximately the  same emission  conditions as
the test parameters  otherwise  selected  on the  basis  of  test
stringency.    The approach  used  was  to  select  a  fuel  tank
temperature  equal  to   ambient  laboratory   conditions   (thus
eliminating  the  need  for  tank  heating)   and  to  determine  a
dispensed fuel  temperature yielding the  same  emission  rate as
                              -12-

-------
did  the  initial test  conditions.   Temperatures developed  from
the  equation  were  80°F  +  2°F  for  the  fuel  tank  temperature
(80°F was  selected because it is the  temperature  maintained  in
the EPA's Motor Vehicle Emission Laboratory vehicle  soak  area),
83°F  +   2°F   for  the  dispensed  fuel  temperature   and   the
requirement that the temperature of  the dispensed  fuel  be  1°  to
3°F  higher  than the soak  area  temperature.  These temperatures
resulted  in a  mean refueling  emissions  value of  7.29  g/gal
which compared  very favorably  to  the  value of 7.26 g/gal  for
the  original test  conditions.   The 80°F  fuel  tank  temperature
can be readily  achieved without  the  need to  heat or measure the
temperature of  the fuel in the vehicle  tank through  the process
of  soaking  the  vehicle  at  the  required  temperature  for  a
pre-specified  soak period.   EPA chose a soak period of  six
hours as sufficient to  accomplish this  task.

     Following   the  April  10,   1986  meeting,   manufacturers
provided  comments  on  the  revised  temperature  specifications.
In  their  comments,  they  expressed   strong  support   for   the
in  rneir  comments,  tney  expressed  strong  support  ror  tne
concept  of  selecting a fuel  tank temperature equal  to  ambient
laboratory  conditions.   However,  a  number  of  manufacturers
commented   that   the   +2°F  soak   area   tolerance   was   too
restrictive.   GM  reported that  maintaining  tight  control  of
room ambient temperature in its larger  laboratories can  be very
difficult.   EPA's  main   concern  in   designating  the  +  2°F
tolerance was to limit adverse impacts  on  test variability.   In
response  to  the  comments,   EPA performed additional  analyses on
the  effects  of   test  temperature  tolerances   on   refueling
emissions   variability.     The   conclusion   reached   is   that
expansion of  the soak  area temperature tolerance band can be
accommodated  if  accompanied by an  adjustment in  the dispensed
fuel tolerance  band  to  retain approximate  equivalency in  the
refueling   emissions   tolerance  band   attributable  to   test
variability in these  temperatures.   As  a result,  the  fuel tank
temperature  is   specified   as  80°+3°F  and  the  dispensed fuel
temperature  is  specified  as  81°  to  84°F.   EPA  believes that
maintaining the  reduced dispensed fuel tolerance  band will  not
be excessively burdensome.  Under this approach,  the dispensed
fuel would  only  need to be heated to  81° -  84°F, which  would
substantially reduce  any  problems with  respect  to the  boiling
point of  the fuel  in  the  fuel cart and the associated changes
in  the   fuel  RVP.    EPA   believes   that  . the  use   of  these
temperature   specifications   will   alleviate   the   concerns
expressed by the  manufacturers   without  any  reduction  in  the
required control of refueling emissions.

     C.     Facility Requirements

     The  recommended  refueling  test procedure  requires  the  use
of  a  sealed  housing  for  evaporative  determination  (SHED),
similar  to  what  is now used  for  evaporative  emissions  testing
with .minor   alterations  to accommodate fuel  dispensing.    The
SHED is  required for the actual  refueling  test  and  for  loading
                              -13-

-------
of  the  canister  to  breakthrough.   Comments on  the  facility
requirements of the test  procedure addressed: 1)  the use of  a
SHED  to  determine  canister  loading  to breakthrough;  2)  the
impact  of .the  test   on  facility  requirements;   and  3)  the
location of the refueling hose and nozzle.

     1.    SHED Use for Breakthrough Determination

     Commenters suggested  the use  of procedures  other than  a
SHED  to  determine  canister  loading  to  breakthrough.   It  was
pointed  out  that  some  contract  laboratories  which  measure
exhaust  emissions  do  not  have  SHED  equipment   and  would,
therefore, be  unable  to perform refueling tests because  of  the
lack of  a  SHED.   The  facility requirement  impact  (as  discussed
below) was  also  a concern for  those facilities with  SHEDs.   A
procedure  involving  repeated refuel ings  to   load  the  canister
without the need for a SHED was  suggested as  an alternative.

     In  addition  to  comments recommending  the  elimination  of
the use of a SHED when loading the canister,  comments  were made
recommending  changes   to  the  SHED  loading  procedure  itself.
MVMA  stated that  the  procedure  should  be   written so  as  to
prevent  the continued  forcing  of  vapors   through  a  canister
which  is  loaded  to   breakthrough.    MVMA   believes  that  the
procedure   as    proposed  would   load   the   canisters   past
breakthrough,   and  as   a solution  recommended using a  reduced
fueling  rate,  e.g.,  3  gallons/minute,  during canister loading
to breakthrough.   MVMA also recommended that  the sample pick-up
point  for  detecting   breakthrough  be  close to  the  canister
rather than remotely  mounted in  the SHED as specified  in  the
recommended  procedure.    MVMA  believes   that    reducing  the
response time for breakthrough detection will prevent continued
forcing  of  vapors   through  a  canister   already   loaded  to
breakthrough.

     Responding first to  the  basic  issue of needing a  SHED to
detect   breakthrough,   EPA  agrees  that  a  canister  loading
approach which  would not require the use of  a SHED to determine
canister breakthrough is  desirable  so as  to simplify testing
and reduce resource requirements.   Use of the SHED was proposed
by EPA  so  as  to address the following concerns  associated with
the use  of a  sample  pick-up  located at  the  canister.   First,
that  a  small  transient puff  of  vapor from  the  canister,  prior
to  breakthrough,   could  be   interpreted  as  breakthrough  and
thereby  result  in incomplete loading of the canister.   Second,
that relatively small  air currents around the vehicle,  as could
occur  in a large room, could dissipate  breakthrough vapors and
lead to delayed detection of  breakthrough.

     A small quantity of data  recently  collected  by  EPA using
current  evaporative  emissions  canisters  suggests  that  small
premature  puffs  of  vapor may  not  be  a  significant  concern.
There  is,   however,   no  way   of  telling  whether  this data  is
                              -14-

-------
applicable  to  the  larger  and possibly  reconfigured  canisters
which  are  anticipated for  use with onboard  refueling  systems.
There  is  also  no  information  on  the  effects  of  air  currents
around the vehicle on breakthrough  detection.   EPA continues to
believe,  therefore,   that  the  SHED  needs   to   be   used  in
determining  canister  breakthrough  loading.   At the same  time,
the Agency would welcome  the submission of further data on this
area which might lead to a non-SHED based approach.

     MVMA's   concern   with   the   SHED   procedure   is   that
breakthrough  will  occur significantly  before  detection because
of the  sample pick-up location.  As  a  result  of  the  detection
delay, MVMA  is concerned  that  a  fueling  rate  of  10 g/min will
cause   a   significant  amount  of   additional  vapor  to   be
transmitted  to  the  canister   beyond  the  actual  breakthrough
point.  EPA  agrees  that  detection  lag   will  result  in  some
degree  of  canister  loading beyond  breakthrough  and   that  the
degree  of  loading  beyond  breakthrough  will  depend  on  the
refueling rate.   However,  since the  objective of  the  canister
loading  procedure   is   to  achieve   loadings  to  at   least
breakthrough, this fact of itself  is  not troubling. What  is of
concern is the  increased  amount of variability in breakthrough
measurements  at  high fuel flow rates.   For this  reason,  some
reduction  in  the  fueling  rate would  be  acceptable  provided
loading  to   at  least  breakthrough  was  achieved.    Testing
conducted by EPA at a fueling rate of 3 to 4  gallons per minute
has shown that  repeatable  loading  conditions should result.  The
fueling flow  rate  during  the  canister  loading procedure  will,
therefore,  be specified as 3-4  gallons/minute.

     2.    Testing Capacity

     Commenting  on   the   impact   of  the  test   on   facility
requirements,  Toyota  stated that  adoption  of  the  recommended
refueling test  procedure  would result  in either  a significant
reduction in  the  testing  capacity  of existing   facilities  or
would  require  significant  facility expansion  to  retain present
testing capacity.   The costs related  to the  modification and/or
construction  of  expanded  test facilities  was  a  significant
issue to a number of commenters.

     EPA recognizes  that  incorporation of  the refueling  test
procedure,    or  for   that  matter   any   other   new   testing
requirement,  into the existing  emissions  testing  procedure will
impact  test   facilities   to   some   extent.   EPA,   like  the
manufacturers,  is  desirous of  holding  to  a minimum the impact
of the procedure on  facility requirements.   EPA is making every
effort to minimize  the impact  whenever possible  in developing
the test procedure.   In fact,  the  revised procedure, which will
be described  further  below, has  a much  lower facility  impact
than did the  previous draft.   All  comments on how the  impact on
facility requirements can  be  further minimized are  encouraged
and welcomed.
                              -15-

-------
     In   the   area   of  costs,   EPA   recognizes  that   some
expenditures will  be necessary  to  expand test  facilities  to
accommodate  the  demands  of  incorporating the  refueling  test
procedure. . However,  it  appears  that,  as  a part  of  overall
cost,  these impacts  will  be  relatively  small.   For  example,
values  used by  the  Motor  Vehicle  Manufacturers  Association,
when viewed as  a  cost  per  production vehicle,  represent  only
approximately 30  cents per vehicle.  Even these values  would be
expected  to decline  in  the  face of  the procedural  revisions
being described in this  document.

     3.    Refueling Hose Location

     The  draft procedure  specified that  the  fuel  dispensing
hose  and nozzle  be  located  inside  the  SHED.   One  commenter
questioned whether non-permeable  fuel  hoses would  be  required;
fuel hose permeability,  nozzle  leakage,  and nozzle-to-fuel  hose
joint leakage may cause  a SHED contamination problem.

     During  SHED  background  and  retention  validation  tests
conducted   at    EPA's  Motor   Vehicle   Emission   Laboratory,
contamination  problems  were  experienced  as  a  result  of  the
location  of the  fuel  dispensing hose  and  nozzle inside  the
SHED.  This problem  was  resolved by moving the hose  and nozzle
outside  of  the SHED and providing access  to  the  vehicle's  fill
neck by  a boot  so  that  only the  nozzle  tip  enters the  SHED.
The  specific   criteria  developed  for  the  boot  are:  that  the
aperture through which the  nozzle tip  passes seals  against  the
tip  when the  nozzle is  inserted and closes to  form  a  vapor
tight  seal  when  the  nozzle  is  not in place;  that the  boot be
flexible  and  relatively long   so  as  to  avoid   the need  for
precise  locating  of the  vehicle  in  the SHED;  and the  boot be
large enough to  facilitate free  passage of the  nozzle through
the boot  and full  operation of the  nozzle inside of the  boot.
Location  of the  nozzle  and fuel hose outside  of the  SHED  has
solved  the  contamination  problems.   The  procedure  has  been
modified to require  the  use of equipment for refueling with the
refueling hose and nozzle located outside the SHED.

     D.    Requirement for Loading Canister to Breakthrough

     Commenting  on the   requirement  for  canister  loading,  two
commenters  took  issue with the need to  fully load the canister
to breakthrough.   MVMA and Toyota  stated  that  forced loading of
the  canister  to  breakthrough  is  not  representative of in-use
vehicle  operation  and  should,  therefore,  not  be  part of  the
test procedure.   These   comments  also claimed  that  loading of
the  canister  in  this manner  will have  a  negative impact on
exhaust  emissions,  on fuel  economy  and on  driveability.   MVMA
stated  that full  canister  loading  followed  by  one prep  LA-4
will  significantly  add  to the  difficulty  of  complying  with
exhaust  emissions   standards   and  in  meeting   fuel   economy
objectives  and  will   result   in  the  collection  of  exhaust
                              -16-

-------
emissions  and  fuel  economy  values  under  non-representative
operating conditions.  MVMA  also believes that  full  loading of
the  canister  removes any  incentive  to provide  a  safety margin
in  canister, sizing  because  excessive  hydrocarbons  have  to be
processed  during  purging  and  this  will  cause  driveability
problems.   One   commenter  pointed   out  that   the   proposed
procedure  did  not  require  loading  to  breakthrough  of  the
evaporative canisters in non-integrated onboard systems.

     EPA believes  that  loading  canisters  to breakthrough  is an
important  requirement  of  the  procedure  so  as to  demonstrate
that the system will adequately purge  the canister from a fully
loaded  condition.   The need  to demonstrate this  capability in
the  test  procedure stems  from  the wide  variations  which exist
in  the  method of  operation  of  in-use vehicles.   Since  the
degree  to  which  a canister  is purged  prior  to  refueling  is
dependent  on  vehicle   operations  preceding  refueling,  it  is
reasonable  to  expect  that  wide  variations  in the degree  of
canister  purge can  also  exist in  in-use vehicles.   Vehicles
used infrequently  and  in  short  trip operations  will experience
reduced canister  purging  while  accumulating  hot soak  emissions
after  each  trip  and  repeated  diurnal  loadings  because  of
infrequent  operation.   As a  result,  these in-use vehicles  can
be  expected  to experience forced loading  of  the  canister  to
breakthrough   or   saturation.   Forced   canister    loading   to
breakthrough   in   the  test   procedure   is,    therefore,   not
unrepresentative  of an   event  which  can  occur  on an  in-use
vehicle.  Data available  to EPA  indicate  that  the  canister
system  does not   undergo  permanent  adverse effects   by  being
highly  loaded  and  quickly recovers  its  capacity when vehicle
operating conditions provide  additional purge.   Loading of  a
canister to  breakthrough  results  in  a  readily achievable  and
repeatable  canister   loading  condition.   Retention   of   the
loading  to  breakthrough  requirement   in  the  procedure  thus
provides  a  useful,  readily  identifiable point  for  beginning
testing.

     It  is  important  to  note   that   loading  the canister  to
breakthrough is regarded  by EPA as a  minimum loading  condition
before testing.  If, because of  its in-use operating factors,  a
vehicle  comes  in  for  testing   loaded  beyond breakthrough,  it
will be tested as  received.   If systems  are  properly  designed,
such  occurrences   should  be  rare;   but  if  systems  are  not
properly   designed   and    frequently   operate   beyond   the
breakthrough point,  then  this  is a  consequence  of the design
and the systems still ought to be tested in that condition.

     Although  some of  the  commenters felt  that  loading  the
canister  to  breakthrough  would  have  an  adverse  impact  on
driveability,   exhaust   emissions,   and  fuel    economy,   and
therefore should   not  be  included in the test procedure,  EPA
does not agree.  First, since canister loading  to breakthrough
                              -17-

-------
will  occur  on  in-use  vehicles,  manufacturers  will  have  to
accommodate  this  condition in  their system  designs  regardless
of test requirements.  Manufacturers will  have to  design their
systems to .operate satisfactorily  with  respect to both canister
purge  and  driveability because  of  in-use considerations.   As
for   exhaust  interactions,   EPA   has   always   expected  that
evaporative  systems  should be  able  to begin the  evaporative
test procedure from  a  loaded  condition  and expects to introduce
this requirement apart from any  onboard actions.   The presence
of  an  onboard  canister  could  increase  the  amount of  purge
vapors  under  loaded  conditions,   but   not to  an  unmanageable
degree.  As  for  fuel economy, EPA agrees  that  impacts  on fuel
economy measurements  should  be  avoided.   The simplest  option
would  be  to  allow those manufacturers  who believe that loading
the  canister to breakthrough will  have  a negative   impact  on
fuel  economy to omit  the canister  loading  step for the fuel
economy test.   If  this  approach  were unsatisfactory,   then  a
CAFE adjustment might have to be considered.

     One revision was made  to the canister loading procedure as
a  result  of  the comment  which pointed out  that there  was  no
loading    procedure    for    the    evaporative    canister   in
non-integrated systems.  .Omission  of this  step in the procedure
was  an  oversight  since  the  intent of  the  procedure  was  to
include  a   loading   step   for   the  evaporative  canister  in
non-integrated systems.   A step  will,  therefore,  be  added  to
the  procedure requiring  the  loading   to  breakthrough   of  the
evaporative  canister  in  non-integrated  systems  prior  to the
vehicle  preconditioning.    As with  refueling canisters,  this
step  will   require  the  use  of  the  SHED  to  determine  the
breakthrough point.

E.   Vehicle   (Canister)   Conditioning   for  Performance  of
     Refueling Emissions Control Test

     Background

     The  test sequence proposed   for refueling emissions  added
two  new  tests  designed  to check the   capacity  and   purge
capability of  the refueling  control  system,   respectively.  Both
of these tests depended upon  canister  preconditioning steps for
their proper functioning.

     The  refueling  capacity  test  was  designed  to  ensure that
the   overall   vapor   control   capacity  of   the   canister  was
sufficient for a complete  fill-up, i.e. from 10  percent  of tank
volume  to  at  least  95  percent  of tank volume.   Certification
test  vehicles were  expected  to  arrive at the  test  site with
canisters  purged to a  level  commensurate with  a  nearly  empty
tank.   Prior to  testing,  the  fuel  tank  would  be  drained and
filled  to  10 percent of  capacity  with  test fuel.  Since  in-use
vehicles  could arrive in  any condition,  preconditioning by 50
miles' of   driving  using  test  fuel  on either  the   durability
                              -18-

-------
driving  schedule  or  equivalent  urban  driving  was  proposed.
Following  the  50 miles  of  driving,  the  fuel  tank would  be
drained  and  fueled to  10  percent  of capacity.   Following this
preconditioning,  the   actual   refueling   test  would  then  be
performed  to  verify  that  the  refueling  system  indeed  had
adequate capacity to handle essentially a full refueling.

     The  second,  or   purge,   test   began  with  a  drive-down
sequence  on  the  dynamometer,  consisting  of  sequential  Urban
Dynamometer Driving  Schedules (UDDS or LA-4),  alternating with
one hour hot soaks.  This sequence was intended  to  use fuel and
allow  refueling  canister  purge,   in   order  to  subsequently
perform a partial refueling with an amount of  fuel  large enough
to  adequately  test  the system's purge capability.   To do this,
the UDDS soak sequence would  be repeated until approximately 30
percent  of the tank fuel  capacity  had been  used.   A refueling
test would then be  conducted  as with the capacity  test,  except
that the refueling  amount  would approximately correspond to the
amount of fuel consumed in the drive-down.  The  purpose  of this
stage was  to  demonstrate that the refueling  control system had
adequate purge capacity to purge accumulated refueling vapors.

     Comments

     Since the condition of the refueling canister  prior to any
refueling  test is  very  important,  it is  not  surprising that
considerable comment was  directed at  the various  conditioning
steps  in the  draft procedure.   Commenters generally believed
that the 50 mile  drive for in-use vehicles was  inadequate,  and
they  opposed   the  use  of  a  conditioning  procedure  for  in-use
vehicles different  from that  used  on  certification  vehicles.
Commenters  also   expressed  concerns   with  respect   to  the
capability of  the 30  percent  drive  down  to prove  the  purge
capability of the system.

     Commenters  suggested  several  alternative  procedures  for
conditioning  of  the  canisters  prior  to  performance  of  the
refueling  capacity  test.    For   certification  vehicles  MVMA
suggested  actual*   vehicle  driving  while   Toyota  suggested
starting with  a full  fuel  tank  and driving either  14 hours on
the durability mileage accumulation  procedure  or  80 hours  of
UDDS/hot  soak  operation.    The  American  Petroleum  Institute
(API)  suggested the  use of a 50  mile  drive  for  certification
vehicles as had been proposed for  in-use vehicles.
     "Actual",  while not defined,  seemed  to imply  operation of
     the vehicle either on  a  test  track or on the road using an
     operating  schedule which  would  reflect  actual  consumer
     driving patterns.
                              -19-

-------
     For  capacity  testing  of  in-use  vehicles,  both  MVMA  and
Toyota  suggested driving  out the  fuel contained  in the  fuel
tank  at  the time  that the  vehicle entered  the test  program.
MVMA  suggested  driving  75  miles   for each  1/4  tank  volume
contained  'in  the   fuel  tank.   Toyota  suggested  driving  the
vehicle until 10 percent fuel volume remained in the tank.

     Commenters  also  questioned   the   30   percent  drive-down
associated with  the canister purge  test.   They suggested  that
actually  driving out  a whole tankful of fuel might be  the only
reliable  way to  verify  proper  system purge  characteristics.
EPA  itself  had indicated concern with  respect  to  the  adequacy
of the  30 percent  drive-down because  of the non-linear  nature
of canister  purging with  time.   The draft  procedure  contained
EPA's suggestion that a full drive-down might be required.

     In order  to effectively respond to all the concerns  over
canister  preconditioning  which  have   been raised,   EPA  has
undertaken   an    extended   analysis    of   canister    purge
characteristics  and  vehicle  operating  patterns.    From  this
analysis  the   Agency  has  derived  a  revised   approach   to
preconditioning  and testing refueling  control  canisters.   This
approach is greatly simplified compared to the draft procedure,
and  provides  a more  accurate  way to assess   the ability  of
refueling control  systems  to  perform  properly  in actual  use.
The  results  of  EPA's  analysis  are presented in the  following
sections.

     Analysis

     1.     Canister HC Purge Characteristics

     To develop  an  understanding  of how  canisters purge,  EPA
performed  a series  of  tests  on  evaporative  emission  control
canisters.   Some of  the  canisters  had been in use  (aged)  on
durability data  vehicles  and were  furnished by Chrysler,  Ford,
GM  and Nissan  while  others  were  new  units   purchased  from
dealers.  One  relatively  large canister,  constructed by EPA,
was  also  tested.   The  details  of  the  testing and  the  test
results are shown  in the  Appendix.   The  overall results  are
summarized below.

     Two  basic  steps  were  used  in  testing   the  desorption
characteristics  of   carbon  canisters.    The   first   involved
loading the  canister  to  an  appropriate  level with  refueling
vapors.    The second  was  to  draw  air  over  the carbon bed  to
purge it  of  its  hydrocarbon load.   Purge curves  were  developed
by monitoring  the  change  in  hydrocarbon load as  a function of
the volume of purge air pulled over the carbon bed.

     The  key results  of  the  testing   are  shown  in  Figure  3.
Shown are characteristic  purge  curves  for the various canisters
                              -20-

-------
   v>
   E
i   3
   CL
   3
   E
   3
   O
                    REPRESENTATIVE  CURVES
                           Normalized by Canister Volume
                                              NISSAN(COCONUT)


                                                 GMCWMO
                          10              20

                          Purge Volume (cubic feet/liter)
                                                             FIGURE 3

-------
expressed  as  the  weight  of  hydrocarbons   removed  from  the
canister versus  the volume of  air  drawn through the  canister.
The test results  have  been normalized to  a  canister volume  of
one liter to provide a standardized basis for comparisons.

     In reviewing the  results  of  this  testing,  EPA  decided that
the results of the tests on the Chrysler canister should not  be
used   for    subsequent   analysis.    This  canister  showed   a
substantially lower storage capacity  than the  other  canisters,
for  reasons  which  were  never  identified.    In any  event,  a
canister with such  a  small  storage capacity  per unit volume  of
charcoal would  not be expected to be  a reasonable choice  for
use in refueling control systems.   To characterize  the  range  of
characteristics exhibited by the  other three canisters,  EPA has
used the Nissan and Ford curves in its analysis.  At  this  time,
EPA does  not know  how representative  of all  canister  designs
these  results  are,  nor  how  much   improvement   in   canister
performance  could be  gained  by attempts  to optimize  charcoal
performance.   However, these   questions  are  not  critical  in
relation  to  the  primary  goal of  describing  general  system
characteristics and designing appropriate test techniques.

     2.    Vehicle Operation

     Evaluation   of   in-use   vehicle   operational   patterns
important   to   an  onboard  refueling   test  program  requires
consideration of  typical  daily events which contribute to  the
loading  and  unloading of  the canister.   Hydrocarbon  vapors
generated during  evaporative diurnals  and hot  soaks  along with
vehicle refuelings  constitute  canister loading  events.   Vehicle
drive events cause canister unloading.

     On the  basis of  typical  driving patterns,  in-use  vehicles
are employed under  widely varying  conditions.   As  a  result  of
this variability in daily  operational  trips,  the   loading,  at
any selected time,  of  a  HC vapor control canister,  whether it
be  a  refueling  control  canister  or  an  evaporative  control
canister, will  also vary.  At  one extreme  is  the  condition of
multiple days wherein  the vehicle is  not driven at  all.   Under
this   non-driven  condition,    the   canister   will   experience
repeated daily diurnal loadings of HC  vapor  and will eventually
reach   a   fully  loaded   (saturated)   condition;    i.e.,   the
canister's  capacity to adsorb  and  retain HC will   be  reached.
At  the  other   extreme  in  the   range  of   daily  operational
characteristics is continuous  long  trip  operation.   Under  these
conditions,  the  canister  will undergo  continuous   purging and
the amount of HC stored in the canister would approach zero.

     Between  these  limits lie  a  wide  variety  of daily vehicle
usage  patterns.    Typically,   vehicle   usage   patterns  might
                              -22-

-------
 include  two employment  related trips per  day and one  or more
 trips  for  other  purposes.   Under multiple vehicle trip  per day
 operations, the  canister  will  undergo purging while the vehicle
 is  being driven  and loading  due  to  hot soaks  and  the  daily
 diurnal  while  the vehicle is parked.  To analyze overall system
 performance,   EPA  constructed  a   simple  model   of  canister
 behavior.   Using  the canister  purge curves described above, the
 model  was  able to track canister performance for both  Ford and
 Nissan type canisters.

     In  the   model,  each  daily   trip  is  considered  to  be
 equivalent  to  one LA-4;  i.e.,  7.5 miles of  vehicle  operation.
 The  purge  rate   is  expressed  as  the volume  of  purge  air,  in
 cubic  feet, per  LA-4.   The reference information stored  in the
 model  is the   characteristic canister purge curve  for one liter
 Ford  and Nissan  canisters.   The  input  variables employed are
 desired  canister  size  and  type,  purge  air  volume  per  LA-4,
 uncontrolled hot  soak and  diurnal loadings  in  grams,   and the
 number   of  trips  per  day.  The  outputs  from  the  model  are
 tabulations of the  running tally  of the  canister   purges  and
 loadings  relative to miles  driven plus  other parameters  which
 can be derived from  these  figures (e.g.,  amount  of  HC  purged
 per  mile).    Running  losses,   if   any,   are  treated as  going
 directly to the  engine  and not impacting loading or  purging of
 the  canister.    This  assumption  is not appropriate  for  all
 current  evaporative control system  designs,   but  EPA  believes
 that such  designs will not  be found on  future  systems because
 of  their  adverse  impact  on  canister  purge.    In  addition,
 diurnal  loadings  are  treated as a  constant, neglecting the fact
 that,  for example,  immediately following a refueling,  the fuel
 tank would  be  full  and  essentially no diurnal  emissions  would
 be generated.   This  means that the  results from the model are
 representative  of   conditions   after  part  of   the  fuel  has
 actually been  used up  and not  to be  interpreted  as  the full
 time history of events beginning with a full tank.

     The results  from a  typical run of  the  computer model are
 shown  in  Figure  4.   In  this   case,   a   simple  pattern  is
 illustrated consisting  of  a single  daily  drive followed  by a
hot soak and a diurnal.   The model indicates that after  only a
 few  repetitions  of  this  pattern,  an  equilibrium   is  reached
between  purging  and  loading.   This  equilibrium indicates  the
vapor  storage  capacity  available  for refueling  control.   Note
that continuing to  operate on  this pattern produces  no further
progress toward the fully purged capacity of the canister.

     One  of the  key effects  on  system  performance   in  this
example is the daily driving pattern  which  is assumed.   Figures
 5 through  7 illustrate the  effect of using  two,  three or four
assumed  trips  per  day  instead  of one.   As  can  be  seen  from
                              -23-

-------
i
NJ
TJ
0)
cr>
i_
3
Q.
O
I
    O
          180
170 -
160 -
150 -
140 -
130 -
120 -
110 -
100 -
90 -
80 -
70 -
60 -
50 -
40 -
30 -
20 -
10 -
 0
                CANISTER  PURGE VS  CYCLIC  OPERATION
                                      FIGURE <4
                                                    CANISTER : 4.3 L NISSAN
                                                    HOT SOAK :  10 g
                                                    DIURNAL : 22 g
                                                    PURGE : 12  CU FT / LA-4
                                                    DAILY TRIPS  : 1
                                —r~
                                20
                                                40
                                                               60
                                     Miles Driven

-------
I
to
Ln
    X>
    0>
    3
    Q.
    (A
    E
    o
    o
         180
170 -
160 -
150 -
140
130 -
120 -
110 -
100 -
 90 -
 80 -
 70 -
 60 -
 50 -
 40 -
 30 -
 20 -
 10 -
 0
               CANISTER  PURGE VS  CYCLIC OPERATION
                                      FIGURE 5
                                                   CANISTER : 4.3 L NISSAN
                                                   HOT SOAK :  10 g
                                                   DIURNAL : 22 g
                                                   PURGE  : 12  CU FT / LA-4
                                                   DAILY TRIPS  : 2
                      20
                       40
60
80
100
120
                                     Miles Driven

-------
I
NJ
Q>
O>
3
0.
o
I
in
        180
              CANISTER PURGE VS  CYCLIC  OPERATION
                                   FIGURE 6
                                               CANISTER : 4.3 L NISSAN
                                               HOT SOAK : 10 g
                                               DIURNAL : 22 g
                                               PURGE : 12 CU  FT / LA-4
                                               DAILY TRIPS : 3
                                     80

                                  Miles Driven
                                         100
120
140
160

-------
i
N>
   x>
   Q>
0.

o
X

o>

E
o
        180
              CANISTER  PURGE  VS  CYCLIC OPERATION

                                  FIGURE 7
                                              CANISTER : 4.3 L NISSAN

                                              HOT SOAK : 10 g

                                              DIURNAL : 22 q

                                              PURGE : 12 CU FT / LA-4

                                              DAILY TRIPS : 4
           0
                                      120
160
200
                                  Miles Driven

-------
these  figures,  vehicles operated  on either a  one or two  trip
per  day cycle  will  possess  lower  refueling  capacities  than
vehicles operated under a  three  or four trip per day cycle.   On
the other hand,  it  is the  case  for all of the  driving patterns
that the canister reaches  an  equilibrium condition after only a
few repetitions of the daily  operating  pattern.   These results,
incidentally,  have   been   derived  for  a Nissan type  canister
because the  Nissan  canister  shows the  greatest  sensitivity  to
driving  patterns and  makes  the clearest  example.    Ford  type
canisters respond to driving patterns,  but to a lesser degree.

     Following  initial evaluation of  the  effect  of  driving
patterns, EPA  chose to do  its subsequent modeling based upon .a
three  trip  per  day  sequence.   As  will be seen  below,  this
pattern   has   also   been   used   in   the   test    procedure
preconditioning  sequence   development.    Three  trips  per  day
closely  resembles  the value  of  3.05 trips  per day  in  the  EPA
MOBILE3  model  for  determining   the  effects  of  mobile  source
emission  standards   on  pollutant  inventories.   From  the  above
modeling results, however,  it is  clear that this represents a
less demanding  requirement than that of  a one  or two trip per
day  sequence.    The  overall  effect  of  this  choice upon  test
procedure stringency has not been quantified.

     3.    Effect of Purge  Rate

     For  a   given   vehicle,   the  other   key  operating  variable
which  affects  refueling system  performance  is  the  purge  rate.
The refueling model  shows  that,  holding canister size constant,
the equilibrium  level  (which  represents the  available refueling
capacity) can  be  increased  or  decreased by  changing  the  air
purge rate.   These effects  are illustrated in Figures 8 and 9.

     The effect of increasing the purge air  rate in  the example
used above by  50 percent is shown in Figure 8.  The increase in
purge  air  rate  results   in  an  increase  in  the  amount  of
hydrocarbons which  are  purged from the  canister at equilibrium;
i.e.,  an  increase  in  refueling  capacity.   Conversely,  it  is
shown  in Figure  9 that a  50 percent decrease  in purge air rate
results  in   a   reduction   in  the  hydrocarbon  purge  level  at
equilibrium;  i.e.,  a reduction in refueling capacity.

     4.    Canister  Sizing

     Having developed  a basic model  of  refueling system loading
and purging, required  canister  sizing  for  refueling operations
was  analyzed.    The  required refueling  vapor  capacity for  a
given  vehicle  was  determined using the uncontrolled emission
factor equation  developed  in EPA's  refueling  emission baseline
study   (Refueling   Emissions   from   Uncontrolled   Vehicles;
EPA-AA-SDSB-85-6).    An  entertainment   factor   of  20  percent
(based upon early test results with a liquid seal system)  and a
safety margin  of 10  percent  were added  to  this basic  rate to
estimate overall required  design capacity.
                              -28-

-------
V£>
I
    0>
   0.
    o>
    E
    o
        180
CANISTER PURGE VS CYCLIC OPERATION
                    FIGURE 8


                                 CANISTER : 4.3 L NISSAN
                                 HOT SOAK :  10 g
                                 DIURNAL : 22 g
                                 PURGE : 18 CU FT / LA-4
                                 DAILY TRIPS  : 3
                                            100
                                    120
140
160
                                  Miles Driven

-------
o
i
           CANISTER PURGE VS  CYCLIC OPERATION

                            FIGURE 9
IOU ~
170 -
160 -
150 -
140 -
130 -
120 -
I 100 -
^ 90 -
- 80-
o 70 -
O
60 -
50 -
40 -
30 -
20 -
10 -
0 -
C
CANISTER : 4.3 L NISSAN
HOT SOAK : 1 0 g
DIURNAL : 22 g
PURGE : 6 CU FT / LA-4
DAILY TRIPS : 3




1/1/1/1/1
]/ \A A/^7 A/ V A/^ A/^ /
/V A /V fV /v ^v ^/l ^
ly /
1
1
/
) 20 40 60 80 100 120 140 160
                           Miles Driven

-------
     The required capacity was  then  related to  the equilibrium
level  of the  canister  in  the EPA  model.   More  specifically,
required   canister   size   was   determined  based   upon   the
requirement,  that  the  canister  have  the  necessary  refueling
vapor capacity at the end of the  first trip following  the daily
diurnal  loading  of  the day  wherein  the canister  first  reached
equilibrium.   This  means  that,  at equilibrium,  the vehicle  is
expected to  be  able  to handle a full  refueling  after  having
experienced a  daily diurnal and  then  driving one  trip  to  the
gas station.

     Because of  the tradeoff between  purge rate  and  effective
canister capacity  described above,  equal   canister  equilibrium
levels  and, therefore,  refueling capacity  can be  achieved from
a  relatively wide  range of  canister  sizes  and  a  corresponding
range  of  purge  air  flow   rates.   Canister  size,  for  equal
refueling capacity,  is inversely  proportional to purge  air flow
rate.   This relationship  is  shown in Figures 10 through  13  for
four different vehicle  types:   a  small  car, an  average  car,  a
full-size dual-tank  light-duty  truck,  and  a  typical heavy-duty
gasoline truck.  The  specific  characteristics assumed for  each
vehicle are given in Table 3.

     A couple of common  characteristics  are apparent from these
figures.  First,  when the  purge rate  is   relatively  high  the
Ford type canisters generally require  somewhat  greater canister
volume  than  do  the Nissan  type  canisters.   However,   as  the
purge  rate  is  decreased to the  low end  of  the  purge rates
investigated,  the Nissan type canisters  tend to be  larger  than
the Ford type.    Second,  the  curves  are fairly flat over a broad
range of purge  rates,  followed by a rapid upturning in canister
size at low purge rates.

     Since  both  diurnal  and   refueling  loads,  which  are  the
dominant vapor  sources, are  proportional to fuel tank  size,  it
is possible to normalize the results for all four  vehicles  and
produce  a   single  family  of  curves.    Figure  14  shows  the
relationship between  canister  volume  per   gallon  of fuel  tank
capacity and LA-4 purge rate per liter of canister volume.

     5.    Refueling System Effects  on Engine Operation

     Recognizing that  both  canister  size  and  purge  air  flow
rate can vary  widely,  it is  appropriate   to investigate those
factors which  could establish  boundaries  on these  parameters.
Since  a small  canister is  desirable from both  a  cost  and  a
packaging perspective,  designers can  be  expected  to  use  the
smallest canister  possible.   Since  canister size  decreases  as
purge  air  flow  rate  increases  an   investigation  of  potential
upper limits  for  purge air flow rate  is warranted.
                              -31-

-------
i
CO
ts>
u
N
(ft

£


£
z
<
o
             CANISTER SIZE VS  PURGE  AIR FLOW  RATE
        4.5
        0.5 -
                            SMALL CAR
                                         FIGURE 10
                          8
                                      16
20
24
28
      FORD Fl
                          PURGE AIR (CU FT / LA-4)

                 -I-   FORD CARB           O   MIS Fl
             MIS CARB

-------
i
CO
CO
   UJ

   N
o:
u
   z
   <
   O
            CANISTER  SIZE VS  PURGE AIR  FLOW  RATE

                           AVERAGE CAR        FIGURE 11
  D   FORD Fl
                         PURGE AIR (CU FT / LA-4)

                    FORD CARB           O   NIS Fl
NIS CARB

-------
OJ

r
   LJ

   N
o:
u
             CANISTER SIZE VS  PURGE AIR FLOW RATE

                        LIGHT DUTY TRUCK/DUAL TANKS  FIGURE 12
         10 -i
                                                                  100
  D  FORD Fl
                          PURGE AIR (CU FT / LA-4)

                    FORD CARB           O   MIS Fl
NIS CARB

-------
i
u>
tjl
I
   U

   N
UJ


g
Z
<
O
             CANISTER SIZE VS PURGE AIR FLOW  RATE
      50 -
40 -
         30 -
         20 H
         10 -
         0
                         HEAVY DUTY VEHICLE
                                         FIGURE 13
      FORD Fl
                          PURGE AIR (CU FT / LA-4)

                     FORD CARB           O   NIS Fl
                                                     NIS CARB

-------
                                       Table 3
               Vehicle Parameters Used in Canister Sizing Calculations
Vehicle
 Type

Small Car
Small Car
Average Car
Average Car
LOT (Dual tanks)
LOT (Dual tanks)
HDV
HDV
                                                      Required
Fuel Delivery  Fuel Tank     Hot Soak    Diurnal      Refueling
   System      Volume (gal)   Loading (g) Loading (g)   Capacity (g)*
Fuel Injection     8.2
Carburetion        8.2
Fuel Injection    13.0
Carburetion       13.0
Fuel Injection    38.0
Carburetion       38.0
Fuel Injection    50.0
Carburetion       50.0
6
9
10
15
29
43
38
57
14
14
22
22
64
64
84
84
 65
 65
104
104
303
303
399
399
     Refueling  capacity  required   calculated   from  refueling
     emissions at test  conditions  (i.e.  7.15 gram/gallon)  x  85
     percent of  tank  volume  x  1.2  (to  account for  20  percent
     entertainment with  liquid  seal)  x  1.1  (to  provide  a  10
     percent safety margin).
                              -36-

-------
   O
   >
   .y.
   z
i
CO
   u
   2
   D
   -J
   O
   o:
   u
   <
   o
             CANISTER SIZE  VS. PURGE AIR  FLOW  RATE
         1.8
         1.7 H
          .o -
          -
1.6 -
1.5 -
1
1.3 -
1.2 -
1.1 -
  1 -
0.9 -
0.8 -
0.7 -
0.6 -
0.5 -
0.4 -
0.3 -
0.2 -
0.1 -
  0
            0
                           NORMALIZED VALUES
                                        FIGURE
             T
             4
~r
8
12
16
20
—T~
24
              F-FI
                   PURGE AIR (CU FT/LA-4) / CAN. VOL.(L)
                     F-CARB         O   N-FI
                                N-CARB

-------
     5.1   Basic Considerations

     Control of  the  power output  from  a gasoline  engine  is
accomplished by  limiting  the amount  of air  available to  the
engine,  by  means of  a throttle  placed  in the  engine  intake
system.   The fuel  metering system  is  designed to provide  fuel
in  proportion   to  the  amount  of  air  allowed  to   enter   the
engine.   Throttling of the  intake  air  supply  causes  a reduction
in the  pressure  of the air  (or  air  and fuel  mixture) in  the
intake  manifold  downstream  of   the  throttle.    This  reduced
pressure  in the  intake manifold  downstream  of  the  throttle
provides  an essentially  zero cost  method for  moving the  air
necessary for  purging of  stored hydrocarbons  from a  canister.
Activation  of  the canister  purge system  however,  provides  an
additional source of  air and  fuel to  the  engine.   This air  is
not under  the  control of  the driver of the vehicle  and the HC
vapor (fuel) entrained in  the air is  not  under the  control  of
the engine's fuel metering  system.   Purging of the hydrocarbons
stored  in  a canister  can, therefore,  impact engine  operation
through perturbations in  the amount  of air  available to  the
engine and in the ratio of  fuel to air  supplied to the engine.

     The purpose of this segment  of  the analysis is  to develop
an understanding  of  limits which may  be  applicable  to canister
purge air  flow  and  to the  fuel supplied  by  the  canister  if
unacceptable negative impacts  on  engine  operation  are to  be
avoided.  A stepwise  presentation of  the effects  of  activating
the canister purge system  will facilitate the  desired analysis.

     5.2   Purge Air

     As was stated previously, activation  of  the canister  purge
system will allow more  air and a variable amount  of  additional
fuel  to  reach  the   engine.   The  resulting  effect  on  engine
operation  will depend  on  the range  of  control  and  rate  of
response of the  engine's fuel metering system.  If the range of
control were to  be exceeded,  the anticipated result  could  be
either  a  substantial  loss in  power  or stalling  of  the engine.
Power loss would be associated either  with  an  extremely rich or
lean but  ignitable  mixture.   Stalling would  be associated with
either  a  richening  or  leaning  of  the  mixture  beyond  the
ignition limit.

     If the  range of control  was  not  exceeded,  the  effects on
engine operation would depend on the speed with which  the  fuel
metering system could compensate  for  the perturbation caused by
the  air and fuel  coming   from the canister.   If the  response
rate was very  rapid  the effect would be for a rapid increase in
the  engine's  power output  because both  the  air  and  the  fuel
available  to  the engine  increased and increased approximately
                              -38-

-------
in the  correct  relative ratio.   As perceived by the driver, the
effect  would  be for  the vehicle to accelerate without  a driver
initiated  action  for  acceleration.   If  the acceleration was
small,  it  could go unnoticed by  the  driver.  If,  however, the
acceleration   was   large,   the   driver   could   perceive   the
acceleration  as a loss of control of the vehicle.

     One  straightforward  approach to  a  large  induced  purge
change  is to  simply  use a  damper  or  slowly  operating  purge
control valve.  Such  a valve,  by introducing the  change  in air
flow   over   a  lengthened   period,   would   allow   for  driver
compensation  as a  part of the normal  driving process.   In  this
way,  a fairly  large  change could  be made with  no perceptible
impact.  Even so,  it  is worth evaluating  reasonable  limits for
the purge  perturbation to determine  if such  a  control  strategy
is even needed.

     On the   assumption,  then,  that  the  vehicle  could rapidly
adjust  to  the sudden  onset  of  purge, one limit  for  purge  rate
would  be  the maximum  acceptable  power  perturbation  it  would
produce.   The  size  of  this  limit is  estimated  below (limits
from the purge related fuel flow will  be treated later).

     Since vehicles presently do  not  incorporate  systems  which
could  cause   relatively  large,  non-driver  induced,  changes  in
the power level at which the engine is  operating and consequent
vehicle accelerations  or  decelerations,  it was  necessary that a
surrogate   be    identified.     Vehicle    accelerations    and
decelerations associated  with  the disengagement  and  engagement
of  air-conditioning  compressors  were selected  as  a  guide  to
driver  acceptable  performance  perturbations  attributable  to
power changes at the driving wheels.  This information  was  used
in estimating a driver acceptable  limit  for purge  air induced
increases in engine power.

     Figure 15  shows  manufacturer  supplied  nominal  values for
the power  required  to  drive  air  conditioning  compressors  on
typical vehicles  (values furnished by  Honda  are  for city  type
operations  and  are,  therefore,   not  expressed  in  terms  of
vehicle speed).   Figures  16,  17,  and  18 show  nominal  engine
brake horsepower (BHP)* curves with and without air-conditioner
     Engine  brake  horsepower   (useful  external  power)  values
     were  derived   from  typical  chassis   dynamometer   power
     absorption curves with  allowances  for  power losses  at the
     tire  to  dyno  interface,  times  allowances  for drive  axle
     and transmission efficiency  plus allowances for  the power
     requirements of  the alternator,  water pump, fan,  air  pump
     and power steering pump.
                              -39-

-------
            AIR  COND  COMPRESSOR  POWER  REQUIREMENT
o
i
    Q_
    I
U
2
U
9:
D
O
LJ
or

(£.
LJ

o
d

a:
o
    u
    o:
    a.
    2
    O
    O
                                                       FIGURE 15
            8
            7 -
            6 -
            5 -
            4 -
            3 -
        2 -
            1 -
            A.

            B.

            C.

            D.

            E.

            F.
Ford. Small Car; Direct Drive

Ford. Average Car,  LOT; Over Drive

Ford. Average Car,  LDT; Direct Drive

GM. Large Car; Durability Loads

Honda Accord; City  Average

Honda Civic; City Average
                                I

                               20
                                              40
                                                60
80
                              B
                                   VEHICLE SPEED (MPH)
                                 O    C      AD

-------
               ENGINE  BMP VS VEHICLE SPEED
0.
I
CD
U

O
Q_

U
Z
O
Z
ui
                    SMALL CAR
                 FIGURE 16
         VEH
VEH
 VEHICLE SPEED (MPH)
AUX          O
VEH + AUX + AC

-------
                   ENGINE BMP VS VEHICLE SPEED
i
-c-
10

-------
               ENGINE  BMP  VS  VEHICLE SPEED
0.


CO
LJ



°
CL

Id

Z

o


Ul
     45
     40 -
     35 -
     30 -
     25 -
20 -
15 -
      10 -
      5 -
        0




       D   VEH
                     LIGHT DUTY TRUCK
                                   FIGURE 18
                      VEHICLE SPEED (MPH)

                VEH + AUX           O
VEH + AUX + AC

-------
compressor  loading for  three  sizes  of  vehicles  (small  car,
average  car  and full-size  light-duty  truck).   At  any  selected
vehicle  speed,   the  difference between  the with  and  without
air-conditioning compressor curves  represents  the  incremental
change in engine horsepower available to  accelerate the vehicle
when  the  compressor  disengages.   Incremental   increases  in
engine  power  available  to  propel the   vehicle  when  the  air
conditioning compressor' turns  off  were  extracted  from Figures
14, 15 and 16 at 20, 35  and 50  mph and are shown in  Table 4 as
percentages of the BHP required to operate the vehicle.

     The  power  consumption  figures  in Table  4 cannot  be used
directly  to  evaluate  purge rates,  because  they are  applicable
to output power, while  any purge  perturbation will  impact total
engine power.  Total, or indicated,  power includes  both output
(brake)  power  and internal  motoring  power.   However,  at  the
relatively low output (brake) power levels involved  in  the data
being  used  here,  it  appears reasonable  to assume  that losses
within  the  engine  could   approximate   the  brake  horsepower
output.   Using  this  approximation,  the percentage change  in
total  engine power  for  an  average  car when  the  compressor
cycles off at  35 mph  would be  approximately one half of the 31
percent  change  in  brake  power  shown in  Table 4,  or  about  15
percent.  Similarly, average  percentage  changes  in  total engine
power  for the  three vehicle  types  are about  8,   15  and  17
percent  for  LDTs,  average  cars   and  small cars  respectively.
For  the two car sizes  only,  the  average  change  is  about  16
percent  and  this value  was selected  as  a  representative upper
limit for the impact of an  increase in air  flow attributable to
canister  purging.    Since  the  incremental  increase   in  total
engine  power  is  directly  proportional  to   the   incremental
increase  in  air  flow,  the  16 percent  value can be applied
directly to the purge air flow rate.

     Conversion of the 16 percent of engine air flow value to a
volume of air  purged  through the  canister was derived from fuel
economy  values  for  the  vehicles  on  the  LA-4  employing  the
assumption   that  stoichiometic   air/fuel   ratio   would  be
maintained throughout.   The fuel  economy values employed  for a
small  car,  an average car  and  a  full-size LDT were 52, 25 and
14.5 mpg  respectively.   A value of 7.5  mpg was assigned  for a
heavy-duty gasoline  vehicle.   The  corresponding  volumes of air
used  by  the  vehicles  on  an  LA-4  are   178,  369  and   637  and
1231 ft3.  The canister  purge  air flow  rates  corresponding to
16 percent  of  the  engine  total  air  consumption so calculated
are  28,  59,  102 and  197 ft3 of  air  per LA-4  for  a small car,
an average car,  a  full-size  light  duty  truck and  a heavy-duty
gasoline  vehicle  respectively.   These  values,  rounded  to 30,
60,  100  and  200  ft3  of   air  per LA-4  were  used  as initial
estimates of  upper  limits  for canister  purge air  volumes for
systems characterized by the sudden onset of purge flow.
                              -44-

-------
                            Table 4


    Air Conditioner Compressor Power Requirements Expressed
     As Percentages of Power Required To Power The Vehicle
                        At Three Speeds


           Air Conditioner Compressor  Power as  Percent of
                      Vehicle  Motive Power

Vehicle
 Speed     Small Car   Average Car         Light Duty Truck

  20          43          40                  24

  35          36          31                  16

  50          25          20                  10
                              -45-

-------
     Referring back to Figures  10,  11,  12 and 13, these  values
can be  seen  to  approximately  correspond  to the maximum purge
rates evaluated.  They occur  in the region of the curves where
there  is  little  sensitivity of  canister size  to  purge rate.
They,   therefore,   do  not  appear  to   represent  any  serious
constraint on  system design or the tradeoff  between  purge rate
and canister  size.   However,  as noted at the onset,   if  it were
desirable to  operate at  higher  purge rates  than these  limits,
the power  perturbation  should  not present  a serious  limiting
factor because  of the ability to use such techniques as damped
purge control valves.

     5.3   Canister Supplied Fuel

     To this  point,  the  analysis  has  looked  at only  one  of  the
two canister  purge  factors  which  can  impact engine operation
(i.e., the amount of air  coming from  the  canister).   The second
factor, canister  supplied hydrocarbons  which become  part of the
total  volume of  fuel  supplied  to  the  engine,  is  evaluated
here.   This  analysis is  performed by  first  examining existing
evaporative  systems,  followed  by an extrapolation   to  onboard
systems.

     5.3.1 Present Evaporative Control Systems

     Purging  of  hydrocarbons  stored  in  evaporative  emission
canisters is  presently  being performed without  an  excessively
negative effect on  engine operation.   Test data reported by API
(Test  Protocol   for  Automotive   Evaporative   Emissions,  API
Publication  No.  4393) shows  a  range for evaporative canister
purge air rates from a  low  of  approximately  3  ft3  per  LA-4 up
to  approximately  11 ft1  per  LA-4  for the  six GM and Ford
vehicles tested.   Combining  the  purge  rates  for each  vehicle
with  the  purge  curve  for  Ford  type canisters  provides  an
estimate of the maximum mass of hydrocarbon  purged from  a fully
loaded  evaporative  canister  during  an  LA-4   (the  Ford  type
canister characteristic  was  selected because  it exhibits  the
highest initial desorption  rates).   For the  six vehicles tested
by  API,  the  measured volume of  air purged per LA-4  and  the
estimated  mass   of  hydrocarbon   purged  from  the  canister,
starting with a loaded canister, is shown in Table 5.

     Using the  fuel  economy  values  measured for each  vehicle
and the  assumption used previously that the air/fuel ratio is
maintained  at  14.7:1,  the  mass  of  fuel  and  volume   of  air
consumed by  each  vehicle during  an LA-4 were  calculated.  The
measured  volumes   of air  coming   from  the  canister  and  the
estimated maximum mass  of hydrocarbon  purged from the canister
during an LA-4  were  then expressed as  percentages  of  air  and
fuel used.   These values  are shown in Table 6.
                              -46-

-------
                                     Table  5
              Evaporative Canister Purge Rates and Corresponding  HC
                       Removal  for Six Production Vehicles
Test Vehicle

1983 Malibu (carb)
1983 Escort (carb)
1981 Omega  (carb)
1983 Fairmont (carb)
1984 Omega (FI)
1984 Escort (FI)
Measured Fuel        Measured
Economy on the-  Canister Purge
LA-4 (mpg)      Air per LA-4 (ft3)
   17.5
   24.1
   21.8
   17.3
   23.4
   27.0
 8.5
 6.5
11.0
 7.5
 3.0
10.0
     Estimated
     Maximum HC
Purged per LA-4 (q)*

         27.0
         25.5
         28.0
         26.5
         21.5
         27.8
     Because the HC purge  of  a canister is  very high when  the
     canister  is  fully  loaded and  decreases  as  the  canister
     loading decreases,  maximum  HC purged  during  a  LA-4  drive
     occurs when  the drive  is  initiated  with  a fully  loaded
     canister.   Depending  on  the  level  of HC  stored  in  the
     canister  at  the start  of  an LA-4  drive,  the  HC  purged
     would vary from  this maximum  down to almost zero  for  a
     drive which was initiated with a  nearly empty canister.
                              -47-

-------
                                        Table  6

                    - Air and  Fuel Coming  From  Evaporative  Canister
                       During an LA-4 for  Six Production Vehicles
   Vehicle

1983 Malibu
1983 Escort
1981 Omega
1983 Fairmont
1984 Omega
1984 Escort
               Total Fuel
               used per
               LA-4 (q)*

                  1217
                   884
                   997
                  1231
                   910
                   789
Total Air
used per
LA-4 (ft3)**

   528
   384
   424
   535
   395
   343
Percent*** of
Total Fuel From
Canister per LA-4

    2.2
    2.9
    2.9
    5.0
    5.4
    3.5
Percent Total
Air From Canister
per LA-4	

    1.6
    1.7
    2.6
    1.4
    0.8
    2.9
**
     Fuel  used   in  grams  =   (7.5  miles)   E  (MPG)  x  (3785.4
     cc/gal)  x (0.75 g/cc gasoline).
     Total air  used  in  ft1  =  (grams  fuel  used)  x  (14.7)  x
     (1/453.6 g/lb)  x 13.4  ftJ/lb.
***  Assuming LA-4  operation  starts  with  a canister  loaded to
     breakthrough.
                              -48-

-------
     Because  of the  non-linear  shape  of  the  canister  purge
curve,  the  average percentage of  total  fuel supplied  from the
evaporative  canister  over  an  LA-4  does  not  represent  the
greatest  percentage  of fuel  contributed by the  canister.   The
largest fuel contribution occurs just  after  canister  purging is
initiated  i.e.,  during the  first mile  or  fraction  of a  mile
following  initiation  of  purging.  To  investigate the  maximum*
percentage  of  fuel contributed  by the  evaporative canister  on
current vehicles,  the  computer model was used to  calculate the
percentage  of  total  fuel coming  from the canister for each of
the first five  miles  of  LA-4 operation,  based upon the average
fuel  consumption  of  the   vehicle.    The  results   from  this
analysis  are  shown  in Table 7  for   each  of the six  vehicles
analyzed.

     As can be seen in Table 7 the percentage of total engine
fuel supplied  by the  evaporative  canisters  starts at  highs  of
between  6  and  16  percent  for  the  first  mile  of   vehicle
operation and  diminishes  as the  canister  purges.  The percent
of  fuel coming  from  the  canister  would reach  zero  when the
canister is fully  purged.    The  values  of  6 to  16  percent  of
total engine  fuel  supplied by the evaporative canister can  be
used as  representative values for fuel  supplied  by  a  canister
which  do  not  presently  produce  adverse  effects   on  engine
performance.
     Because of  the  transient speed characteristic  of  the LA-4
     and, therefore,  transient engine  loading  and corresponding
     transient  air  and  fuel  flow to  the  engine,  the  term
     maximum here means the maximum averaged over a  part  of the
     LA-4 and  not  a  maximum  which may occur  during short term
     transients.   A transient maximum  would  tend to  occur when
     the air  and  fuel  flow  rates  from  the canister were high
     and the engine was working  at a  light  load, e.g.  ,  in the
     transition  period from  a  cruise  to  a  deceleration  but
     prior  to  a  reduction  in canister supplied  air and  fuel.
     Actual   determination  of  such  a  maximum  would   require
     continuous  measurements  of both  air  and  fuel   flows from
     the canister and through the  engine's primary air  and fuel
     supply  systems.   For  this analysis the use of mile by mile
     maximum values on the  LA-4  are considered to be acceptably
     accurate   values    for    comparisons   between   present
     evaporative control  systems  and onboard systems  because
     the onboard systems  are projected  to  operate  relatively
     similarly to present  evaporative systems.
                              -49-

-------
                            Table 7
       Percent  Total  Engine Fuel Coming From Evaporative
        Canisters  During  the First Five Miles of Purging
  Vehicle
1983 Malibu
1983 Escort
1981 Omega
1983 Fairmont
1984 Omega
1984 Escort
                  Percent  Total Fuel Purged from Canister for
                  each of  the first five miles of the LA-4
1st
Mile

10
12
14
 9
 6
16
2nd
Mile

3
4
3
3
4
3
3rd
Mile

1
2
2
1
3
2
4th
Mile

1
1
1
1
2
1
5th
Mile

1
1
1
1
1
1
                              -50-

-------
     5.3.2 Onboard Control Canisters

     Having  identified the first  mile fuel  contributions from
current  evaporative  emission canisters,  the previously  sized
onboard   refueling   canisters  were   similarly   reanalyzed  to
determine  their  first  mile  fuel  contributions.   The  results
from  this reanalysis  are shown  in Figures  19  through  22  as
percent  first  mile  fuel  from  the  canister  plotted  against
percent  engine  air  coming   from the  canister.   The  largest
values  for  percent  first mile  fuel  (16 percent, Table  7)  and
percent  air  (2.9 percent. Table 6)  from  the canister  for  the
vehicles  reported in the API study are also shown  to indicate
present practice.

     As can  be  seen in Figures  19  through 22,  first  mile fuel
contributions by onboard  canisters  at  higher  purge  rates  can
substantially  exceed  current evaporative  canister  first  mile
fuel  contributions.   While  there  is  presently  no  data  to
indicate  the degree  to  which  first  mile   fuel  contributions
could increase  beyond present evaporative  canister practice, it
appears  reasonable  to  assume  that  the  largest  values  shown
(e.g., 80 to  90 percent of engine full) could cause operational
problems.  Therefore,  canister  supplied fuel appears to be  a
bigger constraint than does purge air flow.

     Reproduced  in  Figures  23 through 26  are the canister size
versus purge  air flow  rate  tradeoff  curves  for  fuel injected
systems as previously presented  as Figures  10 through 13, with
information  added  to  indicate  the  points  on  these  curves
corresponding   to  various  percent   first  mile  fuel  values.
Indicated  are  values  of  15  percent  (approximately  current
evaporative  system  practice),  25 percent  and  35 percent.   As
can be  seen  from these  figures, a  first  mile  fuel  constraint
would limit  the use of the smallest canister highest  purge air
flow  systems.    The  actual   impact  of  this  constraint  would
depend  on the   degree   to  which  the  vehicle's  fuel  metering
system could accommodate fuel supplied due to canister purging.

     One possible strategy  for  dealing  with this matter  comes
from the  basic  canister purge  characteristics.   Referring back
to Table 7,  it  can  be seen that the percent  of  the  engine fuel
coming from  the  evaporative  canister  falls  off  quite  rapidly
after the first  mile  of operation.  This  trend is  similar  for
onboard  systems  (see  Table  8).   Thus,  by  modulation  of  the
purge  air  flow rate  to reduce  the  flow rate   initially  and
increase  it  later  in  the  trip  the  first mile  fuel could  be
reduced and spread out more gradually over subsequent miles.
                              -51-

-------
£
O

2
O
a:
u.
u.

*?
               %  FIRST MILE  FUEL VS  %  TOTAL AIR
                       FORD TYPE CANISTER
                 FIGURE 19
         0      2




     0   SMALL CAR
% TOTAL AIR FROM CANISTER

 +  AVG CAR          O
LOT
HDV

-------
   o:
   LJ
I
01
OJ

I
.z

o

5
O
o:
L.

_j
u
         0
                 % FIRST MILE FUEL VS  % TOTAL  AIR
                          FORD TYPE CANISTER
                                            FIGURE 20
           0
        D   SMALL CAR
                            % TOTAL AIR FROM CANISTER

                             +  AVG CAR          O
LOT
HDV

-------
   £
   u

   (/>

   z
   <
   o

   2
   O
   a:
   u_



-  y
r  2

   u
   _J

   2
                 % FIRST MILE FUEL  VS  % TOTAL  AIR

                          NISSAN TYPE CANISTER       FIGURE 21
           SMALL CAR
% TOTAL AIR FROM CANISTER

 +  AVG CAR          O
LOT
HDV

-------
   QL
   Id
   o

   e

v!n  Id
Ln  ~\
1   L.

   U
   U.


   K
         0
                 %  FIRST  MILE  FUEL VS  %  TOTAL AIR
                          NISSAN TYPE CANISTER
                 FIGURE 22
           0
            SMALL CAR
% TOTAL AIR FROM CANISTER
 +  AVG CAR          O
LOT
HDV

-------
i
Ln
   LJ
   N
   o:
   u
   z
   <
   o
             % FIRST  MILE  FUEL  FROM CANISTER  SM  CAR
                            CANISTER SIZE VS PURGE AIR   FIGURE 23
3.5
3.4
3.3
3.2
3.1
 3
2.9
2.8
2.7
2.6
2.5
2.4
2.3
2.2
2.1
 2
1.9
1.8
1.7
1-6 H
         1.5
                     15%
                                          35%
                             8
                            12
                     16
20
24
28
  	  FD CAN
       PURGE AIR PER LA-4 (CU FT)
NIS CAN            O   % IfUL FD
                                                          % FUL NIS

-------
i
Ul
   LJ
   N
   o:
   LJ
   Z
   <
   o
            %  FIRST  MILE FUEL FROM CANISTER  AV  CAR

                         CANISTER SIZE VS PURGE AIR   FIGURE 24
          8 -
          7 -
6 -
          4 -
          3 -
                15%'
                     15%
                              20
                                        40
           60
      FD CAN
                    PURGE AIR PER LA-4 (CU FT)

             NIS CAN          O   % F.IJL FD
A   % FUL NIS

-------
i
t_n
CO
I
   UJ
(ft


OH
LJ
   z
   <
   O
              % FIRST MILE  FUEL  FROM  CANISTER LOT

                         CANISTER SIZE VS PURGE AIR   FIGURE 25
           0
                                           60
80
100
     FD CAN
                          PURGE AIR PER LA-4 (CU FT)

                    NIS CAN           O   % FUL FD
    % FUL NIS

-------
i
Ln
   LJ

   N
   U
   z
   <
   o
              % FIRST MILE  FUEL  FROM  CANISTER  HDV

                          CANISTER SIZE VS PURGE AIR   FIGURE 26
         60
         50 -
         40 -
30 -
         20 -
         10 -
          0
      FD CAN
           0
       15%'
                        15%
         l


         20
 I

40
 i

60
                                       25%
                                                                  35%
                                                      I   I   I
80
100
120
140
160
180
200
                     PURGE AIR PER LA-4 (CU FT)

               NIS CAN          O   % FU4_ FD
                                         % FLIL NIS

-------
                            Table 8
         Percent Total Engine Fuel Coming From Typical
Onboard Canisters During the First
Five Miles of Driving


Vehicle
Small Car
(2.41; 9 ft3)N*
Small Car
(2.651; 9 ft3)F
Average Car
(3.01; 30 ft3)N
Average Car
(4.61; 10 ft')F
LOT
(201; 20 ft3)N
LOT
(151; 20 ft3)F
HDGV
(52.51; 50 ft3)N
HDGV
Percent
Each of'
1st
Mile .
21

40

27

23

14

26

16

31
Total
the Fi
2nd
Mile
16

28

11

17

13

24

14

28
Fuel Purged
rst Five Mi
3rd
Mile
12

17

9

15

12

2.1

14

21
From Canister
les of the LA-
4th
Mile
8

9

8

10

11

20

12

21
for
4
5th
Mile
7

5

7

7

10

15

12

19
 (32.31;  50 ft3)F
*    Size of canister in liters,  purge air flow rate in
     ft3/LA-4,  N = Nissan type,  F = Ford type.
                              -60-

-------
     6.    Summary

     Prior to proceeding with the  section  of the analysis which
addresses  test   procedure   revisions,   it  is  appropriate  to
summarize  the key  findings of  the analysis  to this  point  as
they    relate    to    vehicle    conditioning    for   refueling
measurements.  The analysis  of  canister  and vehicle operational
characteristics has shown:

0    That  the level  of hydrocarbon  stored in the canister when
     the vehicle  is operated under  repetitive cyclic drive,  hot
     soak  and diurnal  conditions reaches stabilization after at
     most a few days of operation.

0    That the canister  stabilization level  is  highly dependent
     upon  the  operating pattern  of  the  vehicle  and  on  the
     amount  of  purge  which  occurs  with   each drive  and  the
     amount  of  loading  which  occurs with each  hot  soak  and
     diurnal.

0    That, for  continuous driving with  no hot  soak  or diurnal
     emissions,  the  canister will be  rapidly purged to  a very
     low level.

0    That, for given hot  soak and diurnal  loadings, appropriate
     selection  of canister  size and purge air flow  rate will
     provide adequate storage capacity  for  refueling  vapors  at
     the stabilized canister loading.

0    That  a   range  of  canister  size  and  purge air  flow rate
     choices  are  available  for   any  given  vehicle  which  should
     not adversely affect vehicle performance.
o
     That   while   required    canister    size   is   inversely
     proportional to  purge air flow  rate,  the  relationship  is
     not linear.

     7.    Test Procedure Revision

     7.l   Evaluation of Preconditioning in Draft Procedure

     The  preceding  analysis  has  identified  how  an  onboard
canister .would  be expected  to purge and  load during  typical
in-use  operation and has  provided a method  for  modeling this
operation.   A comparison of the effects  on  canister purging due
either  to   the   50  mile  continuous  drive  or  the 30  percent
drive-down steps proposed  in  the  original draft procedure with
a representative onboard canister  performance curve is shown in
Figure 27.   As  can be seen,  both the 50  mile drive and  the  30
                              -61-

-------
Q
U
O
K
D
0.

Z
o
00
01
o

-------
percent  drive-down  purge  the  canister  well  beyond  expected
in-use  levels.   These  procedures  would,  therefore,  produce
non-representative canister purges  and  would not be appropriate
for  canister  conditioning  prior to the measurement of refueling
emissions.

     It is  also apparent  from the modeling work  that canister
purge  and refueling capacity  are  not  separable items, and that
the  approach used in the original  draft procedure of evaluating
each aspect  with a separate test  is  inappropriate.   The actual
storage capacity which  the vehicle  will  have available  upon
refueling  is  a function  of   the  canister  purge  rate  and the
vehicle  driving  pattern  as   well  as  of  the  canister  size.
Because of this,  the  performance of the entire refueling system
can  be  evaluated with  a  single test which  first  preconditions
the  canister  to  a  level  near its equilibrium level  and  then
performs  the  refueling operation.  This  greatly  reduces  the
overall  complexity  of  the  refueling  test and  its  resource
impacts and also allows it to  be separated  from the  testing for
exhaust  and   evaporative  emissions.    The  following  section
develops the  preconditioning  procedures needed  for  the revised
test.

     7.2   Revised Canister Conditioning

     Based upon  the modeling  which has been done, there are two
options for   simulating  in-use  performance.   The  first  is  a
cyclic  drive-down  of  alternating drives   and soaks  directly
simulating a  few "days" of vehicle operation  to  approximately
establish  the  canister equilibrium  level.   The  second  is  a
short,  continuous, drive-down  to the  equilibrium level  with no
soak periods.

     Reproduction in  the laboratory of  the  cyclic  in-use daily
operating  pattern  would  be  accomplished  by  the  repetitive
performance  of  a simulated  daily  pattern  consisting  of  three
LA-4s,   each  separated  by  a  one  hour  hot   soak  plus  the
performance  of  a  diurnal  heat build  following  the  last  hot
soak.   This   "daily"   canister   conditioning   sequence  would
constitute the  basic building  block for the construction of the
canister  conditioning  phase   of   the  onboard  test  procedure.
Sequential  repetitions  of   this   basic   sequence   until  the
canister  stabilization  level  was  reached  (or  approximated,  in
the  case  of  a canister system  requiring an  unusually long time
to completely stabilize) would constitute  the  primary procedure
whereby canisters would be conditioned  prior  to measurement of
refueling  emissions.

     There are  several  things  to note about  this  approach.
First,   it  should accurately  simulate  a realistic  conditioning
sequence.   Of course,  as noted  in  the earlier  discussion of the
                              -63-

-------
effects of driving patterns,  in-use patterns of  less  than  three
trips  per  day  would  purge  less  than would  this  procedure.
However,  there  are compensating  conditions of  test  condition
stringency  which  act  to  offset  this difficulty.  Secondly,  it
appears that nearly all vehicles  will reach equilibrium within
three  to  five  "days"  of  simulated  operation.   In fact,  after
only three  "days", all vehicles  appear to  be  within  at most  a
few  grams  of  equilibrium.   Therefore,  three  simulated   days
should  provide   adequate  canister  conditioning.   This,   of
course,   means   that   conditioning  can   be   performed   with
substantially  fewer  testing  resources  than   the   originally
proposed  30 percent drive-down.   Third,  because  of  the  rapid
rate  of  purge  when  the   canister  is    fully   loaded,   this
conditioning sequence  is  relatively  insensitive to the initial
starting  condition of  the  canister.   The initial  period  of
canister  purging  to  a level near  the equilibrium level occurs
within the  first dozen or so driving miles,  so the  total  time
to  reach  stabilization  is  not  significantly  affected by  the
initial loading on the canister.

     While  greatly reducing  resource impacts from the  original
draft  procedure,  multiple repetitions of the  "daily" operating
sequence  will  still be  somewhat time  and  facility  intensive.
Remembering that continuous  vehicle  driving (i.e.,  no hot  soaks
or diurnals) will  result  in  rapid purging of the  canister,  EPA
investigated this  approach  as  another  alternative for  canister
conditioning.   The computer model  was  used  to determine  the
number  of continuous  LA-4  miles required  to   achieve  canister
purging equivalent to  the  cyclic  drive   stabilization  level.
The  results  of this  evaluation,  plotted  as  continuous  LA-4
miles  versus purge  air  flow  rate,  are shown in  Figures  28
through 31 for the systems previously evaluated.

     As   can  be   seen   from  these  figures,   the   number  of
continuous   LA-4   miles   required   to   reach   the   canister
stabilization level is  under 20  miles  in  most  cases,  although
it goes as  high as 33 miles for  heavy-duty vehicles with very
low  purge   rates.   Relative  to   the   "daily"   cyclic   drive
conditioning  procedure,   the  continuous  drive procedure  would
provide significant savings in both time and facilities.

     As envisioned for the  test  procedure,  the continuous  drive
would  operate  as  follows.   Following  canister  loading,  the
vehicle  would  be  driven  continuously  over   repetitive   LA-4
cycles until the  stabilized  level  was reached.  In the case of
a  partial cycle needed  to  complete the required mileage,  the
vehicle would be stopped at the first  idle  point after  reaching
the  desired  mileage.    The  refueling  test   would  then  be
performed.   The number of  miles to be  driven would  be  based
upon previous testing to establish  equivalence with  the  cyclic
                              -64-

-------
Ui
I
z
UJ
E
Q
u
_J
2
          25
24 -
23 -
22 -
21 -
20 -
19 -
18 -
17 -
16 -
15 -
14 -
13 -
12 -
11 -
10 -
 9 -
 8 -
 7 -
 6 -
 5
                 CONTINUOUS  DRIVE MILES  FOR  PURGE
                            FORD TYPE CANISTER Fl
                                      FIGURE 28
                    20
                   40
                                 60
80
100
120
         D
   SMALL CAR
                           PURGE AIR PER LA-4 (CU FT)
                            +   AVG CAR           O
           LOT
140
                HDV

-------
UJ
>
a:
a
V)
LJ
      25
24 -
23 -
22 -
21 -
20 -
19 -
18 -
17 -
16 -
15 -
14 -
13 -
12 -
11 -
10 -
 9 -
 8 -
 7 -
 6 —
 5 -
            CONTINUOUS  DRIVE  MILES  FOR  PURGE
                      FORD TYPE CANISTER CARS.
                                     FIGURE 29
               —r~
               20
                 T~
                 40
60
~nr~
80
100
120
         SMALL CAR
                    PURGE AIR PER LA-4 (CD FT)
                     -I-   AVG CAR          O
                  LOT
—I—
140

HDV

-------
I
ON
   z
   LJ
Q


(A
LJ
           0
              CONTINUOUS  DRIVE MILES FOR PURGE
                       NISSAN TYPE CANISTER Fl
                                       FIGURE 30
                                    80
100
120
           SMALL CAR
                        PURGE AIR PER LA-4 (CD FT)

                         +  AVG CAR         O
   LOT
140




HDV

-------
Id



Q


U
_J

2
             CONTINUOUS  DRIVE  MILES  FOR  PURGE

                      NISSAN TYPE CANISTER GARB.   FIGURE 31
                                        80
                     100
     120
         SMALL CAR
PURGE AIR PER LA-4 (CU FT)
 +   AVG CAR          O
LOT
140



HDV

-------
drive.  Based, .as  it would  be,  upon the purge  level developed
by the  cyclic  drive,  this test would not be. a fully  independent
operation.   Rather,   it  would  be  an  abbreviated  approach  to
obtaining  the  same  results  as  the  cyclic  drive.  It  could be
used  by  manufacturers  in  repeated  testing  of  the  same  or
substantially  similar  vehicles,  and  by EPA in all phases of its
testing.

     The continuous drive  will  allow  canister  conditioning with
a  greatly abbreviated procedure.   Because of  this,    it  would
likely   serve   as   the    principal   approach   used   by   EPA.
Manufacturers,   once  they had  initially conducted  the  cyclic
test  in  order to  develop  the  appropriate  continuous  drive
mileage,  would also  be  able to use  the  continuous drive cycle
for subsequent repetitive  work.   The  continuous  drive procedure
should give equivalent results to the cyclic drive.

     On the  other  hand,  the continuous drive  procedure has the
limitation that it does  not  itself physically demonstrate that
the control  system has  the  ability  to actually  purge  hot soak
and diurnal  loads.   Thus,  in spite  of its advantages  in terms
of resource  impacts,  the  continuous  drive  cannot be  the only
preconditioning  sequence.   However,   so  long  as  the  cyclic
drive-down  is  also  retained,  EPA  would in  all likelihood  be
able  to  use  the  continuous drive-down  for  the  bulk  of  its
testing.   The  longer  cyclic  procedure would  be  reserved for
those cases when the  Agency  felt the need to  fully demonstrate
system  performance  via  direct  testing.   Overall  then,  the
revised  refueling  procedure will  specify both  preconditioning
sequences with the requirement  that  any vehicle be able to pass
the test regardless of which is used.

     7.3   Conditioning of Non-Integrated Control Systems

     Throughout  the  preceding  analysis,  the   use   of  fully
integrated refueling  and  evaporative  emission  controls  has been
assumed.   That is,  the  analysis has  presumed  that the  same
canister  is  used to  collect  and  store  hot  soak,  diurnal and
refueling emissions*.   Since manufacturers may elect  to use one
canister  dedicated  to the collection and storage  of refueling
vapors  alone  and  another canister   for  hot  soak  and  diurnal
emissions, conditioning  of these canisters  will  be addressed at
this  time.
     The  preceding  analysis  is  also  applicable  to  systems
     wherein the  refueling canister is  used to  collect  either
     hot soak  or  diurnal  emissions  (i.e.,  partially integrated
     syterns).
                              -69-

-------
     Since  the- refueling  canister in  a non-integrated  system
may not experience hot  soak  or  diurnal  loadings, purging  would
be  the same  whether the  vehicle  was  driven  continuously  or
under  cyclic  conditions.   Such  a system  would  also  not  be
expected to come  to an  equilibrium condition since each  drive
would continue the process  of purging the canister to  lower and
lower  levels.   Thus,  the only way  to fully  simulate  conditions
of a nearly empty  fuel  tank  would be to actually drive  out the
required amount of fuel, beginning with a loaded canister  and a
full fuel  tank.   Since  the  refueling  emissions measurement  is
initiated  from  the 10 percent tank volume level and  fueling is
continued  until  the fuel  level  in the  tank  is  at  least  95
percent  of  tank  volume,  the  continuous  drive  duration  for
non-integrated   systems   would   have   to   be   the   mileage
corresponding to the consumption  of fuel  equal  to  85  percent of
fuel tank volume.

     A  full  drive-down   for  non-integrated systems  would  be  a
time and  resource intensive  process.   In  addition,   given the
non-linear nature  of  the purge process (refer  to the canister
purge  curves  given  in  Figure  3) most  of  the  purge  would
actually   be   accomplished  in  the  initial   phases  of   the
drive-down.   An  alternative  procedure,  therefore might   be  a
partial drive-down of perhaps  30   to 40 percent of tank  volume
followed  by  a nearly  full  refueling.  Since this  procedure
would  not  directly  verify full  system performance,  it  would
have  to  remain  as  an  optional  test,  similar  to   the  short
continuous  drive   for  integrated  systems.    It  might  also  be
coupled with  supporting test  data or  engineering analysis  to
demonstrate that  satisfactory  performance  at  the  intermediate
level would be  expected  to result in full performance on a full
test.   The potential  use  of  this option  has  not  yet  been
analyzed in detail to determine  adequate  drive miles  and fill
amounts.  Such an analysis  is planned for the future.

     F.    Miscellaneous Issues

     In addition to the comments  addressed above,  comments were
provided   on  several   other   areas  of  the  recommended  test
procedure.     These  areas    included   the  baseline   refueling
emission  factor,   provisions  for  retests,  vehicle  temperature
prior  to  the  refueling  test,  specifications  for  refueling
nozzles, and numerous other minor comments.

     The California  Air Resources  Board  (GARB)  took  issue with
the baseline  refueling  emission factor.  GARB  stated  that data
collected  in  California from  refuelings of  in-use vehicle at
service  stations  showed  refueling  emission  factors  of  3.7
g/gallon with 8.0 RVP  summer  fuel  and 5.6 g/gallon  with 12.0
RVP winter fuel for  an  average of  4.5  g/gallon.  GARB  went on
to  state  that  a  national  refueling  control program  should be
based  on  the  California annual average value  of 4.5 g/gallon
                              -70-

-------
rather  than  the 5.9  g/gallon  value used  by  EPA  for  12.6  RVP
summer  refuelings.  The  purpose  of  this approach was to achieve
a  lower  refueling emission  standard  under  a  fixed  percent
reduction approach.

     EPA does  not agree with CARB's analysis.   First,  EPA does
not believe  that  it can  equate  refueling emissions  measured at
service  stations   under  unknown  measurement  conditions   to
emissions measured  under controlled laboratory  conditions such
as are  included in the  draft  procedure and  described  in EPA's
report, "Refueling Emissions from Uncontrolled Vehicles."  This
report  details  EPA's baseline  program  to  measure  refueling
emissions and  EPA will  continue to use the  baseline refueling
emission factor  resulting  from  its baseline  program.   Second,
the  GARB  comment  appears  to  focus  on  achieving  the  lowest
possible  numerical  emission  standard  associated  with  a  95
percent reduction  of  baseline emissions.   The EPA  standard is
not  intended to be a  simple  percent reduction  of  the  baseline
refueling emissions.  The  refueling standard  will  be chosen to
be a  measurable and reasonable  level as near  to zero emissions
as is possible.   Thus, a change  in the baseline emission level
would not automatically result  in a change in the standard.

     In commenting  on the need  for  retests, MVMA  stated that
the   recommended   procedure   did   not   provide   a   clearly
identifiable  route for  the  performance  of  a retest of  one
segment, e.g.,  tailpipe  emissions   measurement,  of  the overall
procedure either  because  of  a  test  void  or  a failure  in  one
segment of the  test.   A clear line of demarcation  between  the
refueling segment  and other segments  of the  overall procedure
was requested.

     The draft  test  procedure  report   discussed provisions  to
rerun  tests  if  needed.    For  the   refueling  tests,   there were
several  appropriate  places identified where  partial  testing
could be restarted to  avoid having  to  rerun the entire sequence
in case of a test void in one segment of the  test.   The revised
refueling  procedure  is   now  completely   separable  from  the
exhaust and  evaporative test  procedures,   providing the  clear
line  of demarcation  requested by   MVMA.   In  the  event  that  a
retest  of the evaporative  or exhaust test were to be required,
the retest would  be  initiated  at the first step in  the vehicle
preparation  procedure  with  the  requirement  to   re-load  the
canister(s)   prior  to  the  40 percent  fueling  for  the prep LA-4
preceding the cold soak.

     A  concern  regarding the test  vehicle's  temperature  prior
to  the  refueling  test   was  noted  by  Toyota.   To  preclude
inclusion of evaporative emissions  into the  refueling  emission
measurement,  Toyota recommended that the  test vehicle be cooled
to  soak area  temperature  prior  to  performing  the  refueling
emission  measurement   test.   This  is  a  valid point, and  in
                              -71-

-------
response  the   procedure   will  be  modified  to   include   the
stabilization  of  the  vehicle  temperature  at  the  soak  area
temperature.   This will  be  achieved  by  soaking the  vehicle,
following the preconditioning of the canister, for  a  minimum of
6 hours and a maximum of 24 hours.

     Two commenters  stated that a specification was  needed for
the  accuracy  of  the  dispensed fuel  meter.   EPA  agrees  with
these comments  and is  including a  dispensed fuel meter  accuracy
specification in the revised procedure.

     In addition to  comments  about  limiting the dispensed  fuel
flow  rate  reviewed  earlier  in  Section  II  A,  a  number  of
manufacturers commented oh the need to  control refueling nozzle
specifications.  These applied  to both  in-use and test  nozzles,
in areas of  the nozzle which could impact the effectiveness of
onboard control.   Examples of the areas  of nozzle  design which
could  be   considered   for  standardization  under   a   uniform
specification  focus on  the  nozzle  spout  and  include  length,
angle of bend in the spout and its location along the length of
the  spout,  and   position  of   the   automatic  shut-off  port.
Presently  there are no  standardized specifications  applicable
to these areas of the nozzle.

     EPA is  concerned  about  the impact  that nozzle geometry may
have on  refueling  emissions,  but has  little data  at  present
with which  to  evaluate  these claims.   If  it  were true  that
nozzle geometry could  substantially affect  the  performance of
refueling   systems,   then   a  standardized  design  might  be
considered.  If this were the case,  such  standardization would
have  to  be  applied  both  to  test  equipment  and  to  in-use
nozzles.   Otherwise, refueling system performance  would suffer
in practice.

     Lacking detailed  information,  no  decision  can  be  made on
this   issue  at   present.   The   submission   of    test   data
demonstrating  the  degree  of  sensitivity  involved  would  be
especially useful.   It would also  be necessary to  determine to
what  degree  fill   neck   designs  could  be modified to  accept
greater nozzle variability.

     Finally,  numerous  minor comments were provided.  Examples
of  these   minor   comments   include   recommendations  for  the
expansion of the tolerance bands for the hot  soak times and the
driver  trace   during  the   canister  conditioning  drive  to
facilitate testing  and to avoid unnecessary test voids.   These
types  of   comments  are   addressed   by  minor  changes,  where
appropriate, in the test procedure.
                              -72-

-------
Ill.       Test Procedure Overview Summary

     A.    Onboard Test Procedure

     The  onboard  refueling  emission  test procedure,  resulting
from  the  preceding  reanalysis  would  consist  of  four  basic
steps.  In the  first  step,  the onboard canister would be loaded
to  at least  breakthrough.   The second  step  in the  procedure
would  be  canister purging to the appropriate  level  by means of
the applicable vehicle  drive.   The  third step would  be vehicle
cool  down to ambient temperature  followed  by  the  fourth step
wherein the  refueling emissions are  measured.   The  details  of
the tasks performed  within  each of  these  steps  are  shown  in
Figure 32.

    . Briefly, the execution of the procedure as  shown in Figure
32  would  be  as   follows.    In   the  canister   loading   to
breakthrough  step,  the vehicle,  in  as-received condition,  is
brought into  the  laboratory and the  fuel tank  is  drained.  The
vehicle is soaked for six hours in the laboratory to  bring  the
temperature of the  vehicle  into  equilibrium  with the laboratory
ambient temperature.   Following  temperature  equilibration,  the
vehicle is moved  into the  SHED  and  fuel  is added to  the fuel
tank  until canister breakthrough is  detected.   In  those  cases
where the canister  loading  is already at or  beyond breakthrough
in  the  as-received  condition  (such  as might  occur  during
testing   of   in-use   vehicles),   the   analyzer  response   to
hydrocarbons  emanating from  the  canister  would closely coincide
with the  initiation of  fueling and  little fuel would have to be
added to the  fuel  tank for the purpose of loading the canister.

     Upon  completion  of  the  canister  loading  step  of  the
procedure,  the  vehicle  will  enter  into   the  canister  purge
step.   In the canister  purge step,  the procedure which  will  be
followed will depend,  first,  on  whether  the  vehicle is equipped
with an integrated  or a non-integrated emission control system
and second,  if an  integrated system is employed,  whether  the
cyclic drive  procedure or  the  continuous drive procedure  has
been  selected.   In  Figure  32,  the  blocks  headed  "Integrated
System Canister  Purge,  Cyclic  Drive"  and  "Integrated  System
Canister  Purge, Continuous  Drive"  are  applicable  to integrated
systems and specify the details  of  the  steps  in each  of  these
purge  procedures.   The  block  headed  "Non-integrated  System
Canister Purge,  Continuous Drive" specifies  the details  of  the
purge procedure for non-integrated systems.

     In each  of  the  purge  procedures,  the  first  steps are the
same,   i.e.,  to  disconnect  the  canister  vapor line to  avoid
disturbing the  canister  loading,   to drain  the fuel  tank,  to
fuel  with  the  specified  volume  of  fuel  (40   percent  for
                              -73-

-------
                    JL
        Integrated  System
   Canister  Purge,  Cyclic Drive
   a.  Disconnect Vapor Line to Canister
   b.  Drain Fuel Tank
   c.  40%  Fueling
   d.  Reconnect Vapor Line to Canister
   e.  Drive One LA-4
   f.  One  Hour Hot Soak
   g.  Repeat  (e) and (f) Twice
   h.  Disconnect Vapor Line to Canister
   i.  Drain Fuel Tank
   j.  40%  Fueling
   k.  Reconnect Vapor Line to Canister
   1.  Heat Build (60° + 2°F Initial,
      24 + 1°F Rise)
   ro.  Repeat  (e) Through (1) Twice
   n.  Drive One LA-4
•e-
I
                                             Canister Loading  to Breakthrough
                                                 0  Drain Fuel  Tank
                                                 0  Soak Vehicle for  6 Hours
                                                 0  Fuel in SHED to Breakthrough
                ±
       Integrated System
Canister Purge, Continuous Drive
 a.
 b.
 c.
 d.
 e.
Disconnect Vapor Line to Canister
Drain Fuel Tank
40% Fueling
Reconnect Vapor Line to Canister
Drive Repeated LA-4s Until Mileage
Accumulated = Mileage Reguired for
Purge Equivalent to Cyclic Drive.
Mileage Accumulation Stops at the
First Idle Past the Mileage
Requirement
                                              Non-integrated System
                                       Canister Purge,  Continuous Drive
a. Disconnect Vapor Line to Canister
b. Drain Fuel Tank
c. 95% Fueling
d. Reconnect Vapor Line to Canister
e. Drive Repeated LA-4s Until
   85% of Tank Volume Is
   Consumed
                                                        Vehicle Cool Down
                                                 0 Disconnect Line to Canister
                                                 0 Drain Fuel Tank
                                                 0 10% Fueling
                                                 0 Soak Vehicle
                    6 to 24 Hours
                                                            I
       Refueling Emissions Measurement
 0 Reconnect Vapor Line to Canister
 0 Fuel to Automatic Nozzle Shut-off
   (85% Mimimun Fueling).  Restart Fueling
   Following Any Premature Shut-offs Within
   15 Seconds)
                                                ONBOARD TEST PROCEDURE FLOW CHART
                                                                                                           FIGURE  32

-------
 integrated  systems and  95 percent  for  non-integrated systems)
 and   finally  to  reconnect  the  canister  vapor   line.    For
 non-integrated   systems  actual  purging  of  the  canister  will
 consist  of  driving the  vehicle,  using repetitions  of  the LA-4
 cycle,  until 85 percent of tank  volume  has  been consumed.  For
 integrated  systems,  actual purging  of  the  canister  will  be
 performed either by  a  short  continuous  drive  using repetitions
 of the  LA-4  until the canister is purged  to the  level equal to
 that  achieved  with  the  cyclic  drive or by  the  cyclic  drive
 procedure.    In  the   cyclic   drive  procedure,  the   LA-4  is
 performed three times with each  performance of the LA-4  being
 followed by  a one hour  hot soak.  Following  completion  of the
 third  LA-4/hot  soak,  the  canister vapor  line is  disconnected,
 the fuel tank is drained and fueled to  40 percent  with chilled
 fuel, the vapor line  is reconnected and a diurnal heat build is
 performed.   The three  LA-4,  three  hot  soaks  and   one  diurnal
 heat  build  cycle  is repeated  twice and   is  followed by  the
 performance  of  one LA-4.   At  this  point,  the canister  purge
 drives have  been completed and the vehicle enters the cool down
 step of the procedure.

     In the  cool down step, the  vapor line is  disconnected to
 ensure that  canister  loading  is not disturbed, the fuel tank is
 drained  and  fueled with 10  percent  fuel  and  the vehicle  is
 allowed to cool  to laboratory temperature for 6 to 24 hours.

     Measurement  of  refueling  emissions  is the  fin^l step  of
 the procedure and follows the cool down  step.   In  the refueling
 emissions measurement step, the  vapor line is  reconnected and
 the vehicle  is  placed in the  SHED.  The  SHED  is sealed  and .an
 initial  measurement  of  the hydrocarbon  level  in  the SHED is
made.    The  vehicle  is  fueled  in  the SHED with  at   leas:t  85
percent  of   the tank  volume  of  fuel.   The  final  hydrocarbon
 level in the  SHED is  measured.   The 85 percent fueling and the
 final  hydrocarbon measurement  are  the  last  two  steps in the
 refueling test procedure.

     B.     Associated Changes  to Present Test Procedures

     Existing   test   procedures   for   the   measurement   of
evaporative  and exhaust emissions  were  developed prior  to any
consideration   of   onboard  refueling   controls.    Since   these
procedures (evaporative  and exhaust  tests for  LDVs and LDTs and
evaporative  tests  for  HDGVs)   include  two  forty  percent  tank
volume fuelings,  changes to account for  the effects of onboard
controls are necessary  to  achieve continuity in the  results of
these tests.   The  necessary  changes  are the  disconnecting  of
the fuel tank  to canister  vapor  lines  prior to each  fuel tank
drain and forty percent fueling  event and  the  reconnecting of
the  lines   following  each  forty   percent   fueling.    These
                              -75-

-------
disconnecting and reconnecting events will  ensure that new  and
non-representative canister  loadings  are not  incorporated  into
existing  test   procedures.    Specifically,   within   the   test
procedure, disconnecting the vapor  lines  would occur first  when
the test vehicle  enters  a  test  program and residual fuel is  to
be  drained  and  the first  forty percent  fueling is  performed
prior  to  the preconditioning  drive and  cold  soak  and  second
immediately  prior to  the second  fuel tank  drain  and  fueling
with chilled fuel in preparation for the heat build.

     In  addition to   the  previously  indicated  changes   three
other  changes  to existing  test procedures  are proposed.   The
first  of  these  changes  is  the  addition of  two  steps  at  the
start  of  all  testing  on vehicles undergoing  evaporative  and/or
exhaust emissions testing.   These  two steps  are a  fuel  tank
drain  and a six-hour  vehicle temperature equilibration soak  at
a  room temperature of between  68°F  and 86°F.   The  second  of
these  changes  is the  requirement that  all  canisters  be  loaded
to  at  least  breakthrough  prior  to  the  performance  of   the
preconditioning  drive  for LDV  and  LDT evaporative  and  exhaust
emissions tests  and prior to the preconditioning drive for  HDV
evaporative   emissions  tests.    The  third   change  is   the
requirement   that all  applicable  canisters  be  installed  and
operational   during   the  performance  of  HDGE  exhaust  emission
testing.   Prior  to  being  instated  on  the  HDGE  undergoing
testing,  the   canisters  are   required   to  be   loaded   with
hydrocarbons to  a level  equal  to  that existing  at  the end  of
the diurnal  heat build  in  the  HDGV evaporative  emission  test
procedure;  i.e.  the  loading   resulting  from   a  loading  to
breakthrough,   followed  by  a  vehicle  preconditioning  drive,
followed  by a  vehicle  cold soak  and  finally followed  by  a
diurnal heat build.
                              -76-

-------
                    Appendix
Evaluation of the Purge Response Characteristics
          of  Activated  Carbon Canisters
                      -77-

-------
I.  Introduction

     EPA is currently in the process  of  developing a  procedure
to test  the  performance of refueling emission control  systems.
Regardless of the specifics of  the procedure, it must  evaluate;
1)  the  hydrocarbon storage  capacity  of the  system and 2)  the
ability  of  the  system   to   restore  that  capacity   between
refuelings.  Testing  the  hydrocarbon storage capacity of  the
system is  relatively  straightforward,  but  testing the  ability
of  a system  to  restore  -that  capacity is  significantly  more
complex.

     The  standard  hydrocarbon  storage  medium used  in  today's
evaporative  emission  control systems  is activated carbon,  and
it  appears  likely  that  activated carbon  would  be  used  for
refueling emission control as well.   The Draft Recommended Test
Procedure for the Measurement  of Refueling Emissions  published
in July  1985 was  developed with  limited detailed  information
about the  purge characteristics  of  activated carbon  beds.[2]
The  test  of purge  capacity  was  developed around  a  general
knowledge of the  stripping characteristics  of activated carbon
beds, i.e. that for a  given  purge air flow, the  rate  at  which
hydrocarbons  are stripped  from the carbon  bed is high  when  the
bed  is heavily  loaded with hydrocarbons  and this rate decreases
as the hydrocarbon load is reduced.  Since  its publication,  the
proposed  procedure  has  been further  analyzed.    This  analysis
has lead EPA to the conclusion that the  procedure  as  originally
proposed  would   not   adequately  test  control   system  purge
capability.   In  order  to  develop  a  procedure which  does
evaluate the purge  capability of  the  control system,  a better
understanding  of  the  desorption  characteristics  of  activated
carbon was needed.  The test program  described in this appendix
was undertaken for this purpose.

     The  rate  at  which  hydrocarbons  are  stripped  from  an
activated  carbon  canister  is  influenced by  several  variables.
Some  of  these  are  associated  with  the  canister  design  and
include;   1)  size,  2)  shape, 3)  carbon base  (the  material from
which the  carbon  is  produced)  and  4)  internal  configuration
(how  the  vapors  are  routed through  the  carbon  bed).   Other
variables, such  as  purge  air  flow rate  and purge  temperature
are  related  to the purge  process. The  main body of  this test
program  addressed  the  canister-related variables  by  evaluating
the  purge characteristics of  several  canisters  of  different
sizes, designs,  etc.   Although no  attempt  was made  to isolate
the   impact    that   individual   variables   had  on   canister
performance,  the data were used  to estimate the variability in
purge   response  that   could  be  expected  among   different
canisters.   In  addition  to  the  evaluation  of   the  purge
characteristics  of several  canister  designs,   the effects  of
temperature,   purge  air   flow   rate   and  canister   aging  on
hydrocarbon  stripping  were  also  investigated  to  a  limited
extent.
                              -78-

-------
 II.  Test Procedure

     The  basic  objective of this  test  program was  to evaluate
 the hydrocarbon desorption  characteristics  of  various activated
 carbon  canisters  when  loaded  with refueling  emissions.   There
 were   two  basic   steps   used   in  testing   the   desorption
 characteristics of  carbon  canisters.   The first  step involved
 loading the  canisters to an  appropriate level with  the  chosen
 hydrocarbons -  in this case refueling vapors.  The  second step
 was  to  draw  air over  the  carbon  bed  to   purge   it  of  its
 hydrocarbon  load.  During  purging,  the change  in  hydrocarbon
 load  was  measured as a function  of the  volume  of   purge  air
 pulled  over  the carbon bed.   Each of these steps is described
 in greater detail below.

     A.  Canister Loading

     In order  to  evaluate  the  stripping characteristics   of  an
 activated carbon  canister,  the  canister must  first  be  loaded
 with  hydrocarbons.    Because   adsorption   and  desorption  are
 mechanical  processes   they  are  affected  by  the size  of  the
 molecules being  adsorbed  or   desorbed.   Therefore,  the  purge
 characteristics of a  carbon bed can be  affected  by  the type of
 hydrocarbons   used    to   load   the  canister.    Because   the
 information  gathered   in this  program  is  being used  in  the
 development  of a procedure to test  the performance of refueling
 emission  control  systems,  canisters were  loaded  with refueling
 emissions.

     Refueling emissions were generated  by  dispensing fuel with
 a volatility of approximately  11.5 psi  RVP into a fuel tank for
 a  1983  Cutlass  Supreme.  The  fillneck  for this  fuel tank  was
modified  so  that  a  tight seal  was formed  between the fillneck
 and the fuel dispensing  nozzle when  the  nozzle was  inserted
 into the  fillneck.  The  fuel  sender unit for this tank was also
modified  by  adding  an  orifice  and nipple   to  which  a  s/«"
vapor  line   could be attached.   This  vapor   line   routed  the
vapors  displaced  during the  refueling  event  to  the  carbon
 canister.

     The  performance  of   a   canister   during .purge  is  also
 affected  by  the extent to  which  the canister  is loaded.   The
more fully loaded the canister  is, the  higher  the rate at which
hydrocarbons will be  removed  by  a given volume  of  purge air.
Therefore,  to  compare   the results  of  various  tests,   it  is
desirable to load  all  canisters  to  the  same  extent.    This
program was  designed to evaluate  the performance of activated
carbon   canisters  over  their  normal   range  of  hydrocarbon
 loading.  Therefore,   it  seemed  logical  to load the canisters to
 approximately the "breakthrough" point.   The breakthrough point
 is that point at  which  the  canister can no longer adsorb  all of
the hydrocarbon being put into  it, and   some hydrocarbon  passes
through the  canister.   Although breakthrough  is  easy to  define
 in theory,  there  are several  methods of  defining a practical
measure  of   breakthrough,   each  of  which  could result  in  a

                              -79-

-------
somewhat different canister load for  a  given canister.   In this
program,  the  breakthrough  point  was  detected  using  a  flame
ionization detector  from an  exhaust  gas analyzer.   Initially,
the analyzer probe was placed  near  the  canister outlet  and  the
hydrocarbon concentration  of  the gas  leaving the canister  was
monitored during the loading process.   When  a  sharp,  persistent
rise in  hydrocarbon  content was observed, canister  loading  was
discontinued.

     During  the  early  stages  of   the  test  program,   some
variability was observed  between  canister  loadings for  tests
performed on the same  canister  loaded to the  same  breakthrough
point.    It  was hypothesized  that  this  variability was due  to
the technique used to determine breakthrough,  i.e.  that  because
the FID pickup was  located so  near  the  canister  outlet  that
intermittent "spikes"  of  HC coming through  the canister  prior
to  breakthrough might have  been mistaken  for breakthrough  in
some instances.  In  an attempt  to  get more  repeatable  canister
loadings, the  test  procedure  was changed somewhat.   Instead of
measuring a breakthrough point  for  each test  on a  canister,  a
breakthrough point was only measured  for the  first  test on  the
given canister.  For each subsequent test on that canister,  the
canister was loaded  with the  vapors  displaced by dispensing the
same number  of  gallons  of  fuel  that   were dispensed  in  the
original test.

     Bl.  Canister Purge

     Hydrocarbons can  be  stripped  from  a carbon bed by passing
hydrocarbon-poor gas over the  bed.   In  this program, purge  was
accomplished by using  a vacuum  source  to  pull  air over  the
carbon bed.  In the  purge  characterization  portion of  the test
program,  both  the canister ambient  and purge  air  temperatures
were maintained at  95°  F.   Purge  air flowrate  was  measured
using  a  rotometer  downstream from the  canister.  The rotometer
read  in  standard  cubic  feet  per  minute  and  was  monitored
throughout the canister  purge.  A valve  was  installed in  the
air supply line downstream  from the rotometer  and  was  adjusted
throughout the  purge process  to maintain the  desired purge air
flowrate.  Most  of  the testing was  performed  using  a  flowrate
of  approximately  one  cfm,  although  flowrates  of one  half  and
two cfm  were used  in the investigation of the  effects  of  purge
air flow rate on the rate of hydrocarbon removal.

     B2.  Measurement of  Canister Loading

     The  performance of  an  activated   carbon  canister  can be
defined  in  terms  of  the  change  in  canister  loading  as   a
function  of the volume of  purge air that is pulled through the
canister.   Changes  in  canister   loading  were   measured   by
weighing  the canister  before and after loading and  at  several
points  during  the  purge  process.    The   difference   in  the
                              -80-

-------
canister  mass  between  weighings  is  equal  to  the  change  in
canister  load.   Because  hydrocarbons  are  more easily stripped
from the  carbon bed when the  bed  is heavily  loaded,  data were
collected  more  frequently  during  the  initial portion  of  the
purge process.  Canisters were weighed at the following times*:

     0,1,2,3,4,5,7,9,11,13,15,20,25,35,50,70, and every twenty
     minutes thereafter as needed.

     After  completion of .the  test  program a  procedural  error
was discovered.   Specifically, the  time  clock was not  stopped
when  the  canisters  were  disconnected  from the  purge  line  for
weighing,   the clock  was  allowed  to continue  running for  the
5-10   seconds   that  elapsed  during   each  weighing.    The
consequence  of  this  error  is  that  slightly less  purge  air
actually  passed through  the  canisters  than is  represented  in
the data  tables accompanying  this  report.   Although  this  data
recording  practice skews the  results,  it  should skew  all  the
results in the same direction  and  approximately the  same amount
for  tests  in  which  the  purge  air  flow rate is similar.   To
compare tests  done at different flow  rates,  the data must  be
adjusted  by  shifting the  data to  account  for  the  time  lost
during each canister weighing.

     The only time this issue  becomes  important in this program
is   in   the    evaluation   of '  purge   rate   on   stripping
characteristics.   One canister  was  purged  at three  different
purge  rates  in  order to compare  the  effect  of purge  rate  on
hydrocarbon stripping characteristics.   A ten  second  gap in the
purge at  2  cfm represents four times as much  purge  air as does
a ten second gap  in the  purge  at 1/2 cfm.   In order  to compare
the results of  tests  done at  different purge rates,  the results
were corrected  by  shifting  each data  point to account  for  the
gaps in purge  air  flow  corresponding to canister weighings.   It
was estimated that each weighing took  approximately  ten seconds
and that  the  canister was  first disconnected  after  55 seconds
of purge.   The correction procedure  is  more throughly described
in  the discussion of the  results  of  the tests dealing  with
purge rate.
     Because   flowrate   is   constant,   time   elapsed   between
     canister  weighings  is proportional  to the amount   of air
     passing through the canister in that time interval.
                              -81-

-------
III.  Canisters

     Ideally, the  results  of this test program would  provide a
characterization  of   the   performance  of   typical   refueling
emission control canisters that had  been  well aged on refueling
vapors  (i.e.  they would  previously  have  been  subjected  to
multiple  refueling vapor  loadings  and  subsequent purges)  and
had been well maintained.   Since vehicles do  not  presently have
refueling    control    systems,    refueling   canisters    were
unavailable.   However, evaporative emission  control  canisters,
which perform essentially the  same function,  have been used for
more than  a  decade.   The question then became  one of  finding
several  canisters  that  could   be   expected  to  be  in  good
condition - that is well  maintained in use.   One  source  of such
canisters  is   the  fleet   of  vehicles   used   by  automobile
manufacturers to gather emission  control  system  durability and
deterioration  information  -  durability  data vehicles.   These
vehicles are operated  for  50,000  miles and are well  maintained
and  serviced   during   this   mileage   accumulation.    However,
although the canisters on the  durability  vehicle  were subjected
to  a great  deal  of  mileage  accumulation,  the  canisters  were
probably  not  as  well  aged  as  a  typical  in-use  canister.
Durability   vehicles   typically   accumulate   mileage   while
operating  on the  AMA  driving cycle.   This  cycle consists  of
essentially  continuous  operation  with infrequent  stops.   This
kind  of  operation  results   in' infrequent  loading  and  more
extensive purge than would normally occur.

     Although the  durability canisters may not  have  been fully
aged, they were the best  canisters  readily  available  for use.
Three domestic  automakers  (Chrysler,  Ford,  and  General Motors)
and  one foreign  maker  (Nissan)   were contacted  and asked  to
supply EPA with  a  canister  from a durability data vehicle.  All
of these manufacturers obliged.

     A   description   of  each  of   the   canisters   from  the
durability-data vehicles  (durability canisters)  is provided in
Table Al.   Most of  the  information provided  in the  table  is
self-explanatory,   but   one   item   deserves   some   further
attention.   That  is   the  item   labeled  "Treatment  after  50K
Testing".  As  discussed above,  these canisters  were  taken off
of durability data vehicles  that  had  accumulated 50,000 miles.
Upon.completion of durability  mileage accumulation and testing,
manufacturers  typically  store   these vehicles   in   case  the
vehicles  are needed  for  any  further  testing.    It can  be seen
that  at  least  three  of the vehicles  had  been  stored  outside
between  the  time  they  finished   mileage  accumulation  and the
time their canisters were  removed.  Also, one  manufacturer ran
the  durability  vehicle on the test  track prior  to removing the
canister.  The significance of this  information  is discussed in
the analysis of the test results.
                              -82-

-------
00
U)
Canister
Size (ml)

Activated
Carbon Base

Design
Butane(l)
Working
Capacity(gm)

Estimated
Design(2)
HC Working
Capacity(gm)

Approximate
Observed HC
Working
Capacity (gin)

Vehicle
Type
           Canister
           Shape

           Open/Closed
           Bottom

           Treatment
           After 50K
           Testing
                             Chrysler

                             1320


                             Wood


                             50




                             30-35




                             31
                             S-Body
                             (Caravan,
                             Voyager)
                                 Ford
925
Coal
50
30-35
33
Taurus(3)
                  Cylindrical    Rectangular
                  Closed
                  No
                  Information
                  Available
Closed
        Table Al

Description of Canisters

   GM

   850


   Wood


   35




   20-25




   35
  J-Car
   (Sunbird)
                        Cylindrical
                                                         Open
Completed 50K Jan 85
Stored outdoors on
vehicle until 8/28/85
Vents covered with
tape when removed
from vehicle. No
special treatment
after removal.
   Completed  50K 7/19/85
   Stored outdoors on
   vehicle  until 9/5/85
   4  hrs AMA  mileage
   accumulation  (92 miles)
   prior to canister
   removal  and delivery
Nissan

1230


Coconut


57




34-40




57




Maxima



Cylindrical


Open(4)
Completed 50K Jan 84
Stored outdoors until
10/85.  No special
treatment after
removal
EPA

5000


Wood


260




160-180




190
                                                        Cylindrical


                                                        Open
           1.  These are "Design Working Capacities" as given in CERT application.
               Those that specify, specify Butane W.C.
           2.  60-70% of "Design Butane Working Capacity"
           3.  Vehicle representing a Taurus.
           4.  Open bottom with a cover over bottom with a 5/8" opening for air to enter.

-------
     Two  other  canisters  were  tested  in   addition   to   the
durability  canisters..   One  was  a  new  925  milliliter  Ford
canister of  the  same design  as the  Ford durability  canister.
The  results -of  the  tests  on  the new  canister  are  used  in
comparison  with  results  of  the Ford  durability  canister  to
evaluate the effects  of  aging.   The sixth canister evaluated in
this program was a  canister built  by  EPA for  refueling tests.
This  canister  was  not  actually tested  as  part  of this  test
program, nor  was the purge procedure  used  identical  to  that
used  in  testing  the  other  canisters.   It was,  however,  loaded
with  refueling vapors and purged at  approximately 2 cfm.   The
canister is cylindrical  with  a volume  of about five liters  (h =
16  cm,  r  = 10  cm) .    The  canister  was  loaded with  Westvaco
extruded  activated  carbon.    Although  this  canister  was   not
fully  aged prior to  the start  of  this  program,  it  had  been
exposed to  several  refueling  vapor  loading/purge cycles as  part
of another project.   The results of tests on  this canister  were
used to compare  results  from  a large canister to those from the
smaller evap canisters.

IV.  Results

     The data  obtained  in  the  test  program  are presented  in
Tables  A2-1 through  A2-20.   Within  the tables,  the  data  are
organized by canister.  For each test,  information is presented
on;  1)  refueling  conditions,  2)   purge  conditions,   and  3)
canister mass  as a function  of purge  volume.   For  each test,
the following information is presented:
     o

     o

     o

     o

     o

     o

     o
           TTI,  fuel tank temperature prior to refueling (°F)
           TTr,  fuel tank temperature following refueling (°F)
           TD ,  dispensed fuel temperature (°F)
           Gallons of fuel dispensed (gallons)
           TP,  purge air temperature (°F)
           f, purge air flowrate (cfm)
           t, cumulative  purge time  elapsed  prior to  canister
           mass measurement
     0     Canister mass  1)  prior  to loading with  refueling
           vapors,  2)   following  HC  loading,  and 3)  following
           each purge interval.
     0     Cumulative decrease  in canister mass  at  each purge
           interval.

V.  Analysis

     The main purpose  of  this test  program was to compare  the
performance  of  activated  carbon canisters  of various  designs
under  various  conditions  of  purge  temperature  and  flowrate.
The  primary  information  of   interest   was   how  readily  the
                              -84-

-------
I
00
en
TEST NUMBER                      Cl

FUEL TflNK TEMP. PRIOR TO         72
REFUELING (deg. F)

FUEL TflNK TEMP. FOLLOWING        73
REFUELING (deg. F)

DISPENSED FUEL                   73
TEMPERflTURE (deg. F)

GflLLONS OF                        5
FUEL DISPENSED

PURGE RIR                        95
TEMPERflTURE 
-------
                                                           TRBLE  R2-2

                                                      CHRYSLER DURHBILITY  TESTS
I
oo
CT\
I
TEST NUMBER                      C3

FUEL, THNK TEMP. PRIOR TO         71
REFUELING Cdeg. F)

FUEL TflNK TEMP. FOLLOWING        65
REFUELING (deg. F)

DISPENSED FUEL                   65
TEMPERRTURE 
-------
                                                        TflBLE R2-3
                                                    FORD OURRBILITY TESTS
I
oo
-j
TEST NUMBER                      El

FUEL TflNK TEMP. PRIOR TO         78
REFUELING (deg. F)

FUEL TRNK TEMP. FOLLOWING        75
REFUELING (deg. F)

DISPENSED FUEL                   75
TEMPERRTURE (deg. F)

GRLLONS OF                        8
FUEL DISPENSED

PURGE RIR                        95
TEMPERRTURE 
-------
                                                          TRBLE R2-4
                                                     FORD OURHBILITY TESTS
I
oo
TEST NUMBER                      E3

FUEL TflNK TEMP. PRIOR TO         76
REFUELING (deg. F)

FUEL TRNK TEMP. FOLLOWING        75
REFUELING 

     0.0
    14.0
    17.8
    19.6
    20.9
    22.0
    23.4
    24.5
    25.0
    25.6
    25.9
    26.4
    26.9
    27.9
    29.2
    29.9
Cansister weight
 (grams)

   715.1
   701.4
   697.7
   695.5
   694.1
   692.9
   691.3
   69O.O
   689.0
   688.5
   688.0
   686.7
   686.2
   685.3
   684.2
   683.1
Cumulative HC
purged (grams)

     0.0
    13.7
    17.4
    19.6
    21.0
    22.2
    23.8
    25.1
    26.1
    26.6
    27.1
    28.4
    28.9
    29.8
    30.9
    32.0

-------
I
oo
                                                           TflBLE  R2-5

                                                      FORD  DURRBILITY TESTS
TEST NUMBER

FUEL TflNK TEMP. PRIOR TO
REFUELING 
-------
                                                          TflBLE R2-6

                                                     GM DURflBILITY TESTS
o
I
TEST NUMBER

FUEL TflNK TEMP. PRIOR TO
REFUELING (deg. F)

FUEL TflNK TEMP. FOLLOWING
REFUELING (deg. F)

DISPENSED FUEL
TEMPERflTURE (deg. F)

GflLLONS OF
FUEL DISPENSED

PURGE flIR
TEMPERflTURE (deg. F)

PURGE flIR
FLOW RflTE (cf,n. )

CflNISTER WEIGHT
PRIOR TO LOflDING (gms)
                                         BI

                                         69


                                         73


                                         73
                                         95
                                        1.0
                                     445. 1
                                          B2

                                          70


                                          73


                                          74


                                           8


                                          95


                                         1.0


                                       424. 1
       Volume of purge
       air  (ftA3)

              0
              1
              2
              3
              4
              5
              7
              9
             11
             13
             15
             20
             25
             35
             50
             70
             90
                  Canister weight
                   (grams)

                     471. 1
                     455.8
                     451.5
                     449.4
                     447.7
                     447.0
                     445.6
                     444.6
                     444.0
                     443.2
                     442.5
                     441.2
                     439.9
                     437.5
                     433.8
                     429.5
                     424.7
Cumulative HC
purged (grams)

     0.0
    15.3
    19.6
    21.7
    23.4
    24. 1
    25.5
    26.5
    27. 1
    27.9
    28.6
    29.9
    31.2
    33.6
    37.3
    41.6
    46.4
Canister ueight
 (grams)

   451.8
   438.4
   433.2
   431.1
   429.5
   428.7
   427.3
   426.2
   425.4
   424.6
   424.0
   422.6
   421.4
   419.7
   417. 1
   415.0
   414. 1
Cumulative HC
purged (grams)

     0.0
    13.4
    18.6
    20.7
    22.3
    23.1
    24.5
   - 25.6
    26.4
    27.8
    29.2
    30.4
    32.1
    34.7
    36.8
    37.7

-------
                                                  TflBLE H2-7

                                             GM DURflBILITY TESTS
TEST NUMBER                      B3

FUEL TflNK TEliP. PRIOR TO         71
REFUELING (deg. F)

FUEL TflNK TEMP. FOLLOWING        65
REFUELING (deg. F>

DISPENSED FUEL                   65
TEMPERflTURE (deg. F>

GRLLONS OF                       10
FUEL DISPENSED

PURGE RIR                        95
TEMPERflTURE (deg. F)

PURGE flIR                        1.0
FLOW RflTE (cfm.)

CflNISTER WEIGHT               417.1
PRIOR TO LORDING  (gms)
                                                            B4

                                                            70


                                                            65


                                                            65


                                                            10


                                                            95


                                                           1.0


                                                         417.2
Volume of purge
air  (ftA3>

       0
       1
       2
       3
       4
       5
       7
       9
      11
      13
      15
      20
      25
      35
      50
      70
      90
Canister weight
 (grams)
   456.1
   446. 1
   442.3
   439.7
   437.5
   435.9
   433.4
   431.2
   429.7
   428.5
   427.4
   425. 1
   423.6
   421.0
   418.9
   417. 1
Cumulative HC
purged (grams)

     0.0
    10.0
    13.8
    16.4
    18.6
    20.2
    22.7
    24.9
    26.4
    27.6
    28.7
    31.0
    32.5
    35. 1
    37.2
    39.0
Canister ueight
 (grams)

   450.2
   443.3
   439. 1
   436.4
   434. 1
   432.3
   429.7
   428.0
   426.6
   425.5
   424.6
   422.9
   421.2
   419.3
   417.8
   416.6
Cumulative HC
purged (grams)

     0.0
     6.9
    11.1
    13.8
    16.1
    17.9
    20.5
    22.2
    23.6
    24.7
    25.6
    27.3
    29.0
    30.9
    32.4
    33.6

-------
                                                          TRBLE R2-8

                                                      GM OIJRHBILITY TESTS
NJ
I
TEST NUMBER                      85

FUEL TftNK TEMP. PRIOR TO         72
REFUELING (deg. F)

FUEL TRNK TEMP. FOLLOUING        65
REFUELING (deg. F)

DISPENSED FUEL                   65
TEMPERflTURE (deg. F)

GflLLONS OF                       10
FUEL DISPENSED

PURGE RIR                        95
TEMPERflTURE (deg. F)

PURGE AIR                       1.0
FLOW RRTE (cfm.)

CRNI5TER WEIGHT               416.8
PRIOR TO LOflOING (gms)
                                          B6

                                          73


                                          65


                                          65


                                          10


                                          95


                                         1.0


                                       417.6
         Volume of purge
         air  (ftA3)

                0
                1
                2
                3
                4
                5
                7
                9
               11
               13
               15
               20
               25
               35
               50
               70
               90
                  Canister ueight
                   (grams)
Cumu1at i ve HC
purged (grams)
                     443.8               0.0
                     437.0               6.8
                     433.9               9.9
                     431.7              12.1
                     430.0              13.8
                     428.7              15.1
                     426.9              16.9
                     425.6              18.2
                     424.6              19.2
                     423.8              20.0
                     423.2              20.6
                     421.7              22.1
                     420.6              23.2
                     419.1              24.7
                     417.6              26.2

                  (not used in average)
Canister weight
 (grams)

   450.6
   442.6
   438.7
   435.6
   433.3
   432.1
   429.3
   427.7
   426.5
   425.7
   425.0
   423.4
   421.9
   420.2
   418.5
   417.4
Cumulative HC
purged (grams)

     0.0
     8.0
    11.9
    15.0
    17.3
    18.5
    21.3
    22.9
    24. 1
    24.9
    25.6
    27.2
    20.7
    30.4
    32. 1
    33.2

-------
                                                            TflBLE H2-9

                                                       NISSflN qURHBILITY TESTS
VD
V
TEST NUMBER                      Ql

FUEL TfiNK TEMP. PRIOR JQ         70
REFUELING (cleg. F)

FUEL TflNK TEMP. FOLLOWING        65
REFUELING (deg. F)

DISPENSED FUEL                   65
TEMPERATURE (deg. F)

G.HLLONS OF                       25
FUEL PJSPENSEP

PURGE flIR                        95
TEMPERflTLJRE (deg. F)

PURGE FUR                       1.0
FLOW RPTJE (cfm.>

CflNISTER WEIGHT               1103.9
PRIOR'TO LOflqiNG 

                 0
                 1
                 2
                 3
                 4
                 5
                 7
                 9
                11
                13
                15
                20
                25
                35
                50
                70
                  Canister ueight
                   (grams)

                    116^.3
                    1152.9
                    1148.3
                    1145.1
                    1141.7
                    1139.0
                    1134.8
                    1132.0
                    1128.9
                    1126.5
                    1124.4
                    1120.4
                    1116.8
                    1111.3
                    1104,.9
                    1100.3
Cumu1at i ve HC
purged (grams)
     O.Q
    10.4
    15.6
    18.2
    21.6
    24.3
    28.5
    31.3
    34.4
    36.8
    38.9
    42.9
    46.5
    52.
    58.
                                                  63.0
Canister ueight
 (grams)

  1162.9
  1154-3
  1150.2
  1146.7
  1143.7
  1141.3
  1136.8
  1132.5
  1129.5
  1126.9
  1124.6
  112Q.2
  1117.0
  1112.0
  1107.3
  1103.2
Cumulative HC
purged (grams)

     0.0
     8.6
    12.7
    16.2
    19.2
    21.6
    26.1
    30.4
    33.4
    3fa.Q
    38.3
    42.7
    45.9
    50.9
    55.6
    59.7

-------
                                                  THBLE R2-10

                                             NISSHN DURflBILITY TESTS
TEST NUMBER                      D3

FUEL TflNK TEMP. PRIOR TO         66
REFUELING (deg. F)

FUEL TRNK TEMP. FOLLOWING        59
REFUELING (deg. F>

DISPENSED FUEL                   59
TEMPERflTURE (deg. F)

GRLLONS OF                       16
FUEL DISPENSED

PURGE RIR                        95
TEMPERRTURE (deg. F)

PURGE RIR                       1.0
FLOW RRTE (cfm.)

CRNISTER WEIGHT              1097.4
PRIOR TO LORDING  

       0
       1
       2
       3
       4
       5
       7
       9
      11
      13
      15
      20
      25
      35
      50
      70
      90
Canister weight
 (grams)
  1150.4
   137.8
   131.2
   126.2
   121.
   118.
   114.
   110.6
   108.
   106.
   105.
   102.
   100.0
  1097.2
  1094.4
.5
,7
.3
 1
Cumulative HC
purged (grams)

     0.0
    12.6
    19.2
    24.2
    28.7
    32.0
    36.3
    39.6
    41.9
    43.7
    45.1
    48.3
    50.4
    53.2
    56.0
Canister weight
 (grams)

  1145.9
  1138.2
  1133.5
  1130.0
  1126.9
  1124.5
  1120.7
  1117.9
  1115.9
  1114.1
  1112.3
  1109.2
  1107. 1
  1103.3
  1098.5
  1093.5
  1089.7
Cumulative HC
purged (grams)

     0.0
     7.7
    12.4
    15.9
    19.0
    21.4
    25.2
    28.0
    30.0
    31.8
    33.6
    36.7
    38.8
    42.6
    47.4
    52.4
    56.2

-------
                                                         TflBLE R2-H

                                                    NISSRN DURflBILITY  TESTS
ui
TEST NUMBER                      05

FUEL TRNK TEMP. PRIOR TO         64
REFUELING (deg. F)

FUEL TRNK TEMP. FOLLOWING        59
REFUELING (deg. F)

DISPENSED FUEL                   59
TEMPERRTURE (deg. F)

GRLLONS OF                       16
FUEL DISPENSED

PURGE RIR                        95
TEMPERRTURE (deg. F)

PURGE RIR                        1.0
FLOW RHTE (cfm.>

CRNISTER WEIGHT               1086.8
PRIOR TO LORDING  (gms)
                                          06

                                          65


                                          59


                                          58


                                          26


                                          95


                                         1.0


                                      1082.1
        Volume  of  purge
        air  (ftA3)

               0
               1
               2
               3
               4
               5
               7
               9
              11
              13
              15
              20
              25
              35
              50
              70
              90
                  Canister ueight
                   (grams)
Cumulative HC
pur ged (grams)
                    1127.4               0.0
                    1120.2               7.2
                    1115.8              11.6
                    1112.8              14.6
                    1110.3              17.1
                    1108.4              19.0
                    1104.9              22.5
                    1102.5              24.9
                    1101.1              26.3
                    1099.6              27.8
                    1098.0              29.4
                    1095.4              32.0
                    1093.8              33.6
                    1091.1              36.3
                    1088.1              39.3
                    1085.1              42.3
                    1082.7              44.7
                 (not uied irt average)
Canister ueight
 < grams)

  1146.4
  1138.4
  1133.6
  1129.4
  1126.6
  1124.4
  1120.2
  1117.3
  1114.9
  1113.0
  1111. 1
  1107.9
  1105.4
  1101.2
  1096.6
  1090.8
  1086.4
Cumulative HC
purged (grams)

     0.0
     8.0
    12.8
    17.0
    19.8
    22.0
    26.2
    29.1
    31.5
    33.4
    35.3
    38.5
    41.0
    45.2
    49.8
    55.6
    bO.G

-------
                                                  TRBLE 02-12

                                             NISSRN OURHBILITY TESTS
TEST NUMBER                      07

FUEL TflNK TEMP. PRIOR TO         59
REFUELING (deg. F)

FUEL TRNK TEMP. FOLLOWING        58
REFUELING (deg. F)

DISPENSED FUEL                   53
TEMPERRTURE 
-------
TEST NUMBER                      09

FUEL TRNK TEMP. PRIOR TO         67
REFUELING (deg. F)

FUEL TflNK TEMP. FOLLOWING        59
REFUELING (deg. F)

DISPENSED FUEL                   59
TEMPERflTURE (deg. F)

GflLLONS OF                       16
FUEL DISPENSED

PURGE flIR                       115
TEMPERflTURE (deg. F)

PURGE flIR                       1.0
FLOW RflTE (cfm.)

CflNISTER WEIGHT              1G38.B
PRIOR TO LORDING
                                                  THBLE R2-13

                                             NISSRN OURRBILITY TESTS
Volume of purge
air (ftA3)

       0
       1
       2
       3
       4
       5
       7
       9
      11
      13
      15
      20
      25
      35
      50
      70
      90
Canister weight
 (grams)

  1093.1
  1088.2
  1085.5
  1083.4
  1081.7
  1080.3
  1078.0
  1076.3
  1075.0
  1073.9
  1072.7
  1071.0
  1069.6
  1067.4
  1064.8
  1061.8
  1O58.8
Cumulative HC
purged (grams)

     0.0
     4.9
     7.6
     9.7
    11.4
    12.8
    15.1
    16.8
    18.1
    19.2
    20.4
    22.1
    23.5
    25.7
    28.3
    31.3
    34.3

-------
                                                        TflBLE R2-14
                                                      NEH FORD TESTS
CO
         TEST  NUMBER

         FUEL  TRNK  TEMP.  PRIOR  TO
         REFUELING  
-------
                                              TflBLE R2-15
                                             NEW FORD TE5T5
TEST NUMBER

FUEL TflNK TEMP. PRIOR TO
REFUELING (deg. F)

FUEL TflNK TEMP. FOLLOWING
REFUELING (deg. F)

DISPENSED FUEL
TEMPERRTURE (deg. F)

GflLLONS OF
FUEL DISPENSED

PURGE flIR
TEMPERRTURE (deg. F)
PURGE flIR
FLOW RRTE
(cfm.)
CRNISTER WEIGHT
PRIOR TO LORDING 

       0
       1
       2
       3
       4
       5
       7
       9
      11
      13
      15
      20
      25
      35
      50
      70
      90
        Canister ueight
         (grams)

           634.2
           617.4
           613.0
           610.4
           609.0
           608.0
           606.5
           605.3
           604.6
           603.9
           603.3
           602.0
           600.8
           599.1
           597.  1
           595.8
           594.7
      Cumulative HC
      purged (grams)

           0.0
          16.8
          21.2
          23.8
          25.2
          26.2
          27.7
          28.9
          29.6
          30.3
          30.9
          32.2
          33.4
          35.1
          37. 1
          38.4
          39.5
Canister weight
 (grams )

   630.8
   616.8
   614.0
   611.1
   609.9
   609.0
   607.9
   607.2
   606.6
   605.9
   605.4
   604.4
   603.3
   601.4
   599.7
   598.3
   597.3
Cumulative HC
purged (grams)

     0.0
    14.0
    16.8
    19.7
    20.9
    21.8
    22.9
    23.6
    24.2
    24.9
    25.4
    26.4
    27.5
    29.4
    31. 1
    32.5
    33.5

-------
                                                      TflBLE H2-16

                                                     NEW  FORD TESTS
o
o
TEST NUMBER                      fl5

FUEL TflNK TEMP. PRIOR TO         70
REFUELING (deg. F)

FUEL TflNK TEMP. FOLLOWING        74
REFUELING (deg. F)

DISPENSED FUEL                   74
TEMPERflTURE (deg. F)

GflLLONS OF                       10
FUEL DISPENSED

PURGE flIR                        95
TEMPERHTURE (deg. F)

PURGE flIR                       1.0
FLOW RflTE (cfm.)

CflNISTER WEIGHT               602.4
PRIOR TO LORDING (gms)
                                                            R6

                                                            69


                                                            74


                                                            74


                                                            10


                                                            95


                                                           1.0


                                                         602.8
       Vo1ume of purge
       air  (ft"3)

              0
              1
              2
              3
              4
              5
              7
              9
              11
              13
              15
             20
             25
             35
             50
             70
             90
Canister ueight
 (grams)

   636.3
   622.0
   616.8
   616.6
   615.3
   614.3
   613.0
   612.2
   611.4
   610.7
       1
                     610
                     608.8
                     607.6
                     606. 1
                     604.6
                     603.3
   602.8
Cumulative HC
purged (grams)

     0.0
    14.3
    17.5
    19.7
    21.0
    22.0
    23.3
    24. 1
    24.9
    25.6
    26.2
    27.5
    28.7
    30.2
    31.7
    33.0
    33.5
Canister weight
 (grams)

   640.8
   628.3
   622.0
   620.1
   616.2
   614.3
   612.8
   611.8
   610.7
   610.0
   609.3
   608.0
   606.8
   604.7
   602.7
Cumulative HC
purged (grams)

     0.0
    12.5
    18.8
    20.7
    24.6'
    26.5
    28.0
    29.0
    30. 1
    30.8
    31.5
    32.8
    34.0
    36.1
    38. 1

-------
                                                          TRBLE R2-17
                                                         NEW FORD  TESTS
I
M
O
V
TEST NUMBER                        H7

FUEL THNK TEMP.  PRIOR TO          72
REFUELING Cd«g.  F)

FUEL TRNK TEMP.  FOLLOWING         61
REFUELING Cdog.  F)

DISPENSED FUEL                     61
TEMPERRTURE  Cdeg. F)

GHLLONS OF                         13
FUEL DISPENSED

PURGE flIR                          95
TEMPERHTURE  Cdeg. F)

PURGE flIR                         l.O
FLOW RHTE (cfm.)

CRNISTER WEIGHT                 627.1
PRIOR TO LORDING Cgms)
                                                                                            R8

                                                                                            71


                                                                                            62


                                                                                            62


                                                                                            13


                                                                                           115


                                                                                           1.0


                                                                                        628. 1
Volume of  purge
air 
-------
                                              TRBLE R2-18

                                             NEW FORD TESTS
I
M
O
I
TEST NUMBER                      fl9

FUEL TflNK TEMP. PRIOR TO         67
REFUELING (deg. F>

FUEL TflNK TEMP. FOLLOWING        61
REFUELING 
-------
                                                        TRBLE  R2-19
                                                       NEW FORD TESTS
o
to
TEST NUMBER                     fill

FUEL TRNK TEMP. PRIOR TO         71
REFUELING (deg. F)

FUEL TRNK TEMP. FOLLOWING        61
REFUELING (deg. F)

DISPENSED FUEL                   61
TEMPERRTURE (deg. F)

GRLLONS OF                       13
FUEL DISPENSED

PURGE RIR                        75
TEMPERRTURE (deg. F)

PURGE RIR                       1.0
FLOW RRTE (cfm.)

CflNISTER WEIGHT               636.5
PRIOR TO LORDING (gms)
                                         fl!2

                                          67


                                          61


                                          61


                                          13


                                          75


                                         1.0


                                       643.4
          Volume of purge
          air (ftA3)

                 0
                 1
                 2
                 3
                 4
                 5
                 7
                 9
                11
                13
                15
                20
                25
                35
                50
                70
                90
                  Canister weight
                   (grams)

                     673.9
                     668.3
                     665.7
                     663.7
                     662. 1
                     661.0
                     659.0
                     657.8
                     656.8
                     656.0
                     655.5
                     653.6
                     652.9
                     650.9
                     648.6
                     645.9
                     643.4
Cumulative HC
purged (grams)

     0.0
     5.6
     8.2
    10.2
    11.8
    12.9
    14.9
    16.1
    17.1
    17.9
    18.4
    20.3
    21.0
    23.0
    25.3
    28.0
    30.5
Canister weight
 (grams)

   678.2
   671.0
   667.2
   664.7
   663.0
   661.7
   659.5
   657.8
   656.9
   655.9
   655.1
   653.3
   651.8
   649.7
   646.8
   643.7
Cumulative HC
purged (grams.)

     0.0
     7.2
    11.0
    13.5
    15.2
    16.5
    18.7
    20.4
    21.3
    22.3
    23. 1
    24.9
    26.4
    28.5
    31.4
    34.5

-------
                                              TflBLE H2-20

                                             NEW FORD TESTS
I
I—'
o
I
TEST NUMBER                     fl!3

FUEL TflNK TEMP. PRIOR TO         67
REFUELING (deg. F)

FUEL TflNK TEMP. FOLLOWING        61
REFUELING (deg. F)

DISPENSED FUEL     •              61
TEMPERRTURE (deg. F)

GflLLONS OF                       13
FUEL DISPENSED

PURGE flIR                        75
TEMPERflTURE (deg. F)

PURGE flIR                       l.O
FLOW RflTE (cfm.)

CflNISTER WEIGHT               650.4
PRIOR TO LORDING Cgms)
                                                           R14

                                                            69


                                                            61


                                                            61


                                                            13


                                                            75


                                                           1.0


                                                         643.4
Volume of purge
air (ft*3)

       0
       1
       2
       3
       4
       5
       7
       9
      11
      13
      15
      20
      25
      35
      50
      70
      90
Canister" ueight
 (grams)

   681.4
   672.4
   668.3
   665.8
   663.8
   662.4
   660.4
   658.8
   658.0
   657. 1
   656.7
   655.2
   654. 3
   652.5
   650.4
   649.5
                                               Cumulative HC
                                               purged (grams)

                                                    0.0
                                                    9.0
                                                   13.1
                                                   15.6
                                                   17.6
                                                   19.0
                                                   21.0
                                                   22.6
                                                   23.4
                                                   24.3
                                                   24.7
                                                   26.2
                                                   27.1
                                                   28.9
                                                   31.0
                                                   31.9
Canister ueight
 (grams)
   683.1
   676.1
   672.3
   670.3
   668.4
   667.3
   665.0
   663.
   662.
   661.9
   661.4
   660.1
   659.2
   656.8
   654.7
   652.6
   650.4
.7
.7
Cumulative HC
purged (grams)

     0.0
     7.0
    10.8
    12.8
    14.7
    15.8
    18. 1
    19.4
    20.4
    21.2
    21.7
    23.0
    23.9
    26.3
    28.4
    30.5
    32.7

-------
 canister releases hydrocarbon  in response to a given volume  of
 purge air  when  purged at  a  given flowrate  and  temperature.
 Because  the amount of  HC purged  does not  vary  linearly with the
 purge air volume, canister performance  should  be compared over
 a continuum  rather  than  at  any  given time  during  the purge
 sequence.   The  simplest  way to  do  this   is  with  a  graphical
 representation   of   the results  which  were  presented   in  the
 previous section.   Throughout  the  rest of the  data  analysis,
 canisters   are  compared  by  comparing   plots  of  cumulative   HC
 purged  from  the canister   (ordinate)  versus  the  cumulative
 volume of  purge  air  pulled  through the  canister  prior  to the
 given mass  measurement   (abscissa).    The  data  points  were
 connected  with  a smooth curve to approximate the performance  of
 the  canister  at  all  times in  the  purge sequence.

      Because  there  is  some variability  in canister performance
 from one test to another the results from a  series of  tests  on
 a  given  canister were  averaged to  find  a  "typical"  purge curve
 for   each   series.    The  averages  were  found  by   finding  the
 average  amount  of hydrocarbon purged from the  canister  at each
 data point.  The average values  were found  and  plotted, and a
 smooth curve  drawn  between  the  average  values.   Most   of  the
 analysis   in  this   report   is   based   on  the  average  curves
 developed as described  above.

      The  remainder  of  the  analysis of   results  is  divided into
 five  sections.    The  first  section addresses  some differences
 that  were observed between the  first test  or  tests performed  on
 a  given  durability  canister  and  later  tests  performed  on the
 same  canister.   The  next  section  touches on  the  effects    of
 canister  aging  by evaluating the results  from sequential tests
 on a new canister and  by  comparing results of  tests  performed
 an  aged  (durability)  Ford  canister of  identical  design.   The
 effect of  purge rate  is  then briefly discussed  followed  by  a
 description of  the  results  of the tests  designed  to  evaluate
 the  effect of  temperature  on  purge  rate.   The  next   section
 describes the most  substantial  portion  of  the work in this test
 program.   That  is the  development  of  a "representative" purge
 curve  for  each  canister type  examined in this  program.   The
 final  section  discusses   the   internal   temperature   of  the
 refueling canister during the purge process.

A.    Initial Tests versus Average Curves

     As  part  of  the analysis of  the canister  test  results,  the
 results  from  tests   performed  on  individual  canisters  under
 similar  conditions  of  purge  were  plotted individually  on  the
 same  set of axes.   When plotted  in this way,  the  results from
 three  of the  four  durability  canisters  tested show a pattern  in
the test results.   The pattern observed was  that the  shape   of
                              -105-

-------
 the  curve generated from  the  results of  the  first one  or  two
 tests  on  a given canister was markedly different from the shape
 of  the curve generated  in subsequent tests.   In two  of these
 three  cases- the  canister  working capacity  appeared  to  improve
 after  the first  few tests, and  in  the third  case  the canister-
 working   capacity  appeared  to  decrease  after  the   first  few
 cycles.

     The  durability canisters  supplied  by Ford  and  Chrysler
 both  showed a lower working capacity  in  the initial  tests than
 in subsequent tests.  Figure Al  shows  two curves generated from
 the  results of the  tests  on the Chrysler  durability canister.
 The curve  labeled "Initial" was  generated by averaging the data
 taken  in  the  first  and second tests on the canister.   The curve
 labeled   "Average"   is  the  average  of  the  other  two  tests
 performed  on  that canister.   The results of the  first two tests
 were nearly identical  to each other as were the  results  of  the
 third  and fourth  tests.   An examination of Figure Al shows that
 the  purge  curve  representing the  initial  tests  is  different
 from  that  representing  later tests  in  terms of  both working
 capacity   (lower  for  initial  tests)  and  shape.    A  similar
 pattern is  seen  in  Figure A2.   Figure A2 shows results of tests
 performed  on  the durability canister  provided  by Ford Motor
 Company.   An  examination  of the plots  shows a lower working
 capacity  and  a distinctly different  pattern of  purge response
 in the initial test  as compared with later tests.

     Results of  the tests performed  on  the durability canister
 supplied  by General  Motors are  shown  in  Figure A3.   As  in  the
 tests on  the  Ford and  Chrysler  durability canisters,  there is a
 noticeable  difference  between  the  initial  tests  and  later
 tests.  In  this  case,  however,  the working capacity observed in
 the initial test  on the canister  is  greater  than  the capacity
 observed  in succeeding tests.

     The   results  of  the  initial   tests  on  the  durability-
 canisters raised two questions:  1) Why did  the initial test  (or
 tests) on these  durability canisters  produce  different results
 than  subsequent  tests?  and  2)   Why  did  the capacities  of  the
Ford  and   Chrysler  canisters   apparently  increase  over  the
capacity observed in their initial tests,  while the capacity of
the General Motors  canister  appeared  to decrease?  One possible
explanation  of these  results  is described  in  the  following
paragraph.

     An   increase   in   canister  working  capacity  with  time
suggests that the canister is  being purged  somewhat  more fully
than  it  recently  had  been,   and that   some hard  to  remove
hydrocarbons are being stripped  from  the  carbon.  Conversely,  a
decrease   in   canister   working  capacity   suggests   that  the
canister   was   initially  fairly  well   stripped,    but   that
                             -106-

-------
I
h-1
O
~J
(ft
E
Q>
0>
k_
3

OL
E
3
O
30


28


26


24


22


20


18


16


14


12


10


 8


 6


 4


 2


 0
                                 FIGURE A-1


                INITIAL CURVES  VS.  AVERAGE  CURVE

                                Chrysler Durability
                                                         AVERAGE
                                                        INITIAL
         0
                          20
                                                          40
                              Purge Volume (cubic ft)

-------
o
oo
i
w.

3

Q.


O

I


Q>
        32



        30
        26
        22
20



18



16



14



12



10



 8



 6



 4



 2



 0
                                    FIGURE A-2



                   INITIAL  CURVE  VS. AVERAGE CURVE

                                   Ford Durability
             7
                         AVERAGE
                        INITIAL
           0
                               20
                                                 40
                                 Purge Volume (cubic ft)

-------
I
H-1
O

I
3
Q_

O
I

0)
E
3
O
     40
     25
20
     15
     10
      0
        0
                               FIGURE A-3

                INITIAL  CURVE vs. AVERAGE CURVE
                               GM Durability
           /   /
                                             40
                            Purge Volume (cubic ft)

-------
 load/purge   cycles   are   increasing   the  canister   "heel."
 Therefore,  the  difference  in  the  results  on  the  Ford  and
 Chrysler  canisters  versus  the  General Motors  canister  could
 have  been . due  to  the  condition  of  the  canister  prior  to
 testing.  The  results  suggest  that the  General Motors canister
 was  fairly  well purged and the  Ford  and Chrysler  canisters had
 more  of  a residual load.   This  was  exactly  the case.   Each of
 these durability canisters  (Ford,  Chrysler,  G.M.)  was taken off
 of  a  vehicle that had completed  its  durability testing months
 earlier.  As can be  seen in Table Al,  the Ford and GM vehicles
 were  stored  outdoors between  completion of durability  testing
 and  the  time of canister  removal.   (Though no information was
 available, it  is  safe to assume that the Chrysler  vehicle was
 treated   similarly).   During   that   time   the  canisters  were
 subjected to  multiple diurnals  and  could be  expected  to  have
 been  thoroughly saturated.  The difference between the Ford and
 the General Motors  canisters  is that  General Motors  attempted
 to  "stabilize"  the canister prior to  delivering it  to EPA.  In
 this  case,  stabilization was  achieved  by driving  the  vehicle
 for  four  hours   of  AMA mileage accumulation  with  the  evap
 canister  onboard.   This  type  of operation probably purged the
 canister  quite thoroughly,  and led  to the difference  between
 the plots of the initial tests on these canisters.

     It should be pointed out  here that  the  Nissan canister was
 also  subjected  to  multiple diurnal  loadings before delivery to
 EPA.  There was, however, no significant difference between the
 plot of the initial canister tests and later tests.

 B.   Canister Aging

     Canister aging  refers  to  the process by which an activated
 carbon  bed  loses  working  capacity  with  repeated   load/purge
 cycles  until   a  stabilized level  is  reached.   On  a  molecular
 level,  aging   is   the  process   by  which  certain  molecules  are
 adsorbed  onto  the  carbon bed  in such a way that  they are very
 difficult to  remove.   Although  it might be possible  to remove
 them with an extensive amount  of purging,  the effective working
 capacity of the carbon bed  under normal, in-use purge modes is
 reduced.   The  aged condition  appears  to develop  gradually over
 repetitive load/purge cycles.

     One of the secondary goals  of this  program was to evaluate
 the magnitude  of  the  aging effect.   Because of  the  extensive
 amount of testing   that  would  be  required,  it was  outside the
 scope of  this  project to  age  a  new canister from  its virgin
 state to its stabilized level.    It is  unlikely that aging could
be observed  after  the limited  number  of cycles possible during
this  test  program.    Figure  A4  shows  the  first  six  tests
performed on  the  new  Ford canister.   Although the  first  five
                              -110-

-------
                                    FIGURE A-4
                             FORD  DURABILITY
0>
o>
t_
3
Q_

U
Z

o>
3
£
3
o
      40
      35
      30
      25
20
      15
      10
       0
                                      AGING
         0
                           20
40
                                Purge Volume (cubic ft)

-------
 tests  tend to show a decrease  in  capacity  with time,  the sixth
 test  shows the second  highest working  capacity in the  group.
 This  suggests  that  the difference between tests is being masked
 by  test-to^test variability  and no  trend  in  working  capacity
 can be established from this data.

     Because  of the time  involved  in aging  a  virgin  canister,
 an  alternative method  of  evaluating  the  effect  of  canister
 aging was  needed.   One logical approach was to perform a series
 of tests  on a new canister  (as outlined above) and to compare
 the  results  of these  tests with  the  results  of  a series  of
 tests done on  an  aged canister of  identical  design.  Figure  A5
 shows four plots.   The two upper curves were generated from the
 results of  the tests on the new canister.  The  top  curve  is the
 purge record  from  the initial test on the virgin canister.  The
 next curve is  an  average of  the  six plots shown  previously  in
 Figure A4.  The  lower  two  curves  in Figure A5 were  generated
 from the  results  of  the tests on the Ford  durability  canister.
 The lowest  shows  the  results of the first test performed on the
 durability  canister after  it was received by EPA and the final
 plot  (labeled  "durability  average")  shows the average  of  all
 the tests performed on the durability canister.

     Figure  A5  illustrates  a few  significant  points.   First,
 the  original test  on  the  virgin  canister  shows  the  highest
 working capacity of all  the tests performed on the new and the
 durability  canisters.   Second,  the  average  working  capacity
 observed  in the tests on  the new  canister  is  higher   than the
 average  working  capacity  of  the  durability  canister.   This
 suggests  that  some  aging has  taken place.    Finally, the first
 test  on  the  durability  canister   showed  the  lowest  working
 capacity  of all the  tests,  suggesting  that the  canister  may
 have  "aged" more  than  the average  durability plot shows  and
 that the  repeated bench  purge  performed  in this  program  has
 restored  some  capacity.   Although  the  results  of these tests
 are not   a definitive measure of  the  effects of  aging,  the
 results do suggest  that the  durability'canister has been aged
 to some extent.  In this  case, the durability  canister appears
 to have lost about twenty percent of its original capacity.

 C.   Purge Rate

     Another  secondary  goal   of   this   test  program  was  to
 evaluate  the  effect  that  the  rate of  purge  air   flow  has  on
 canister  stripping characteristics.   If  the  rate of hydrocarbon
 stripping  is  independent   of  purge  air  flow  rate  then  the
 cumulative  amount of  hydrocarbon  purged  from  the canister would
 be a  constant  function of the  volume of  purge  gas  passing
 through the canister.   Traces of  cumulative  hydrocarbon  purged
versus volume of purge air  pulled through the canister  would  be
                             -112-

-------
X)
0>
en
t.
3
0.

o
I
E
3'
O
       40
       30
>0
       15
       10
        •5 -
        0
                                        FIGURE A-S


                                  CANISTER  AGING

                                   New Ford vs. Ford Durability
                                                     JRABILITY AVERAGE
                                                    DURABILITY FIRST
         £
       //.
          0
                              20
40
                                    Purge Volume (cubic ft)

-------
similar regardless of the purge  air  flowrate.   However,  because
desorption  is  a  mechanical  process  and  the   molecules   in
refueling  vapors range  widely  in  size,  a dependence  between
purge air  flow  rate  and the  rate  of hydrocarbon  removal  could
be possible.  In  this program,  an  attempt was  made to determine
whether  the  amount  of   HC  stripped  from  an  activated  carbon
canister  is  a  constant  function  of  the volume  of purge  gas
pulled over the bed,  independent of purge air flowrate.

     The effect of purge air flow rate on hydrocarbon stripping
was investigated  by performing  three  sets of purge tests on the
Ford  durability  canister.   Each  set  of  tests  was  performed
under identical  conditions  of purge, except that  the purge  air
flowrate was  different  for  each  set of  tests.    The flowrates
chosen were nominally 1/2,  l,  and  2 cfm.   Although these values
may be  somewhat high for current  evaporative emission  control
system  purge   flowrates,   it   seems   that  flowrates  of  this
magnitude  may  be necessary  for  some  evap/refueling  control
systems.

     As  noted  in  the   discussion  of  the  canister  weighing
procedures, the  results  of  this  series  of  tests   had  to  be
corrected  to  account for  time spent  weighing the canister.   For
each test,  it was assumed that the  canister was  purged  for 55
seconds,  and  then the  canister- was  disconnected and  weighed.
The weighing  procedure was  estimated  to  take  approximately  ten
seconds.   Therefore,  ten  seconds  were   substracted from  the
cumulative  total  for each  canister weighing.   Table A3  shows
the corrected results of  the tests used in the  evaluation of
the effect  of  purge  rate on  hydrocarbon  stripping  (tests  El  -
E6).  Each column represents  the average  of the  two tests  done
at the purge rate shown  at the top of the column.

     The results  of the  tests designed to evaluate  the  effects
of purge  rate are plotted in Figure A6.   Each curve represents
the average of  the tests done at one flowrate as  marked on the
figure.     If    the   rate  of   hydrocarbon   desorption   were
proportional to purge air flow rate,  one would expect to  see  a
pattern  in the  purge  curves  in  Figure  A6.   Specifically,  one
would expect  to  see a  steeper  purge  curve,  and   possibly  a
greater working capacity in the tests performed  at  the highest
flowrate.   An  examination  of  Figure A6  shows   that  no  such
pattern  is apparent.   The curve generated from results  of  the
tests performed using  the  lowest  flowrate  falls  between  the
curves  of  the  tests  done  using the  higher  flowrates.   In
addition, all  of the tests are very  similar  and  the differences
between  them  are certainly  within the  range of  test  to  test
variability.   Therefore,  within  the   range  of  purge  rates
examined  here,   the  amount  of  hydrocarbon  stripped from  the
carbon bed is  a function of  the volume of  purge  air  pulled over
the  carbon  bed  and  is basically  independent   of purge  air
flowrate.  What will  happen  at higher purge rates  is unclear.
                             -114-

-------
                              TRBLE fl3

                  CORRECTED flUERRGE PURGE HISTORIES
                      FORD DURflBILITY CflNISTER
Molum* of Purge                           Purge Rir Flow Rate
   flir                    1.0               0.5               2.0

       Q                          000
       1                       14.5              15.9              17.0
       2                       18.2              19.5              20.4
       3                       20.3              22.2              21.8
       4                       21.7              23.6              23.3
       5                       22.8              24.2              24.6
       6                       23.6              25.0              25.4
       7                       24.3              25.4              26.0
       B                       24.8              25.6              26.5
       9                       25.3              25.9              26.9
      10                       25.7              26.1              27.3
      15                       27.2              27.4              29.0
      20                       27.6              27.8              29.8
      25                       28.2              28.4              30.4
      30                       28.6              29.0              30.8
      35                       29.0              29.6              31.2
      40                       29.4              30.1              31.5
      50                       30.2              31.0              32.3
                              -115-

-------
en
a>
01
i_
3
0.

O
I

a>
    D

    E
    3
    O
            0
                                      FIGURE A-6


                           EFFECT  OF PURGE  RATE

                                     Ford Durability
                               20
40
                                  Purge Volume (cubic ft)

-------
D.   Temperature of Purge

     Canister  temperature  can  effect  both  the  loading  and
stripping o.f carbon beds.   Increased temperature is equivalent
to  an  increase in the  kinetic energy of  the molecules  in  the
gas.  An  increase  in  the kinetic energy of air and hydrocarbons
associated with  a  temperature increase  in an  activated  carbon
bed  should  cause  a  decrease  in  the   amount of  hydrocarbon
adsorbed but should also aid the desorption  process.   This test
program was  focused on  the process of purging hydrocarbons from
a   carbon   bed,   and   therefore   the   temperature/loading
interactions   are  not   addressed.   The  effect   of  canister
temperature on purge was tangentially investigated, however.

     Fourteen  tests  were  performed on  the  new Ford canister.
In the first six tests, both the canister ambient and purge air
temperatures were  maintained  at  95°F.    In  the thirteenth  and
fourteenth   tests,   the  canister   ambient   and   purge   air
temperature were held  at 75°.   Figure A7 shows two  plots;  one
representing the average of the results of   the  tests  in which
canisters were purged at 95°,  the  other  representing the tests
using a 75° purge.

     From the  graph  it  appears that  temperature  has  a distinct
effect  on  purge.   There  are  circumstances  of  the  testing,
however,  which  suggest  that the results  may be confounded.   As
mentioned above, the  tests  in  which the canister was purged at
95° were  the  first  six tests  done  on  the  new  canister.   The
tests in which the  canister was purged at  75°  were  the 13th and
14th tests performed  on that canister.   Although it  is expected
that  extensive aging  would not  be  observed  after  a  limited
number of  cycles,  any  aging effects would  bias  the  results of
these  tests  toward  the   pattern   observed  in  Figure  A7.
Although there is  a possibility  that the results of these tests
may be somewhat  confounded  by aging effects,  it  appears  that
the  increased  purge  temperature  did have  some  effect  on  the
amount of hydrocarbon that  was stripped from the canister  by a
given volume of purge air.

E.   Representative Curves

     As  was stated   previously,  the  main  goal  of  this  test
program was  to evaluate the purge response  characteristics  of
several activated  carbon  canisters when  loaded  with refueling
vapors.  This was  done  by  loading  each canister to or  near  the
breakthrough  point  with  refueling  vapors,   and  then  pulling
hydrocarbon-poor   air   over   the   activated  carbon   bed   and
monitoring the  canister mass change as  a  function  of purge air
flow.   The  purge  air flow  rate  and  temperature  of  purge  were
generally held  at  1.0  cfm  and  95°F, respectively,  during the
                             -117-

-------
                                  FIGURE A-7
CD
Q>
O>
k,
3
Q_

O
I

Q>
   3
   E
   3
   o
                       EFFECTS  OF  TEMPERATURE
                                 75 deg vs. 95 deg
           0
                             20
40
                               Purge VOLUME (cubic ft)

-------
canister purge  sequences.*   Average curves were  then  generated
for  each  canister.   The  average  curves, for  the  four  evap
canisters  from  durability  vehicles  are  shown  in Figure  A8.
Figure  A9  shows  the curve  for  the EPA  canister,  which  was
designed and  built  as  a refueling  canister  and has  a  working
capacity much larger than the evaporative canister capacities.

     The plots of  the average  curves are valuable  in  that  some
understanding of  the differences in the  purge response  of  the
various canisters (as measured by the differences  in the shapes
of the  various  curves)  can be  gained.   It is  difficult  to  use
the  curves  to  fully evaluate the  performance of  the  various
canisters  however,  because  some of the differences are simply
due  to  the  fact that the  canisters are  not  all  of   one  size.
Although there  are several  variables other  than size that  may
affect  the  performance  of  the  canisters  (carbon  type,  shape,
and  interior configuration,  to mention  a  few),  this program  was
not  designed  to  investigate the effects  of  the  differences  in
canister designs.   The  differences  in  the  curves due  only  to
differences in size .can  be effectively eliminated, however.

     The  differences  in  canister   sizes  were  eliminated   by
normalizing the  average  curves presented  in  Figures  A8  and  A9
by  canister  volume.   In  scaling   the  canister  curves,   the
characteristic shape of each curve  was  preserved, but  the purge
curves were scaled to represent  the results  expected  for a  one
liter canister.   The normalization  was  done  by dividing the  X
and  Y  components  of the  points  on  the average  curve  by  the
canister volume (in  liters).  The use of  this scaling  technique
implicitly assumes that a small  canister  designed  exactly  like
a  large  canister  (in   terms   of   length to  diameter   ratio,
interior configurations,  carbon  base,  etc.)  would demonstrate
stripping characteristics (for  equivalent  amounts  of  activated
carbon) identical to those of the large canister.   For example,
a  half-sized  canister purged  half   as  much  would  release  half
the hydrocarbons that a  full  sized canister would.

     The volume-normalized  purge curves   are  shown   in  Figure
AID.   The curves  are labeled with  the  source  (or  supplier)  of
the  canister  as well  as with  the  base  material  used  in  the
production of the  activated carbon.  Several  features  of Figure
A10  are  worthy   of  discussion.   The   first  thing  that   is
noticeable in Figure A10 is that the  curve  generated from  the
     Three  purge  rates  were used  in  the tests  on  the  Ford
     durability canister,  but as  discussed in  the section  on
     the effects of purge  rate,  this had  little impact  on the
     results.
                             -119-

-------
                                     FIGURE A-8
                           DURABILITY CANISTERS
K
O
3

Q.


O

I


0>
    O

    3

    E
    3

    o
         10
                                   AVERAGE CURVES
                                  20
                                                    40
                                 Purge Volume (cubic ft)

-------
                                           FIGURE A-9
i
(-•
to
v
    lf>
    E
•Q
0>
O>
L.
3
Q.

O
    >
    •*•<
    o
    3
    E
    3
    O
      170

      160

      150

      140

      130

      120

      110
           70
           40

           30

           20

           10
                                      EPA  CANISTER
                                           Average Curve
                                        20
                                                               40
                                       Purge Volume (cubic ft)

-------
                                         FIGURE A-10
K
to
     t>
    0.

    0>
    3
    E
    3
    o
                             REPRESENTATIVE  CURVES
                                    Normalized by Canister Volume
                                                    NISSAN (COCONUT)

                                                          -*~~~~
                                                          GM (WOOD)
EPA (WOOD)

       FORD (COAL)
                                                                CHRYSLER (WOOD)
                               10                 20


                                   Purge Volume (cubic feet/liter)
        30

-------
results  of  the tests on the Chrysler canister  is  isolated from
the rest of  the curves.  An examination  of the  information on
canisters presented  in Table  Al reveals  nothing  extraordinary
about the Chrysler canister.   The activated carbon  used  in this
canister and the canister construction  appear very  similar to
the material and  construction  of the other canisters  tested in
this  program.   Chrysler was  unable to  provide information on
the  history of  the  canister,   and  there  may have  been  some
unusual  treatment   of  the   canister   which  lead   to   these
unexpected   results.    Although   no   explanation   for   the
performance  of  the  Chrysler  canister  is  apparent,  it is clear
that  the results  of  the tests on this  canister  fall  well
outside  of   the  range  predicted by  the  tests  of  the  other
canisters.    Because  the results of  the  tests  on  the  Chrysler
canister cannot  be  explained  by the  information  available  to
EPA, the curve  for  the Chrysler canister  will not be included
in any further  analysis of the  results.

     The main  feature  of interest  in  Figure A10  is that  it
shows the differences  in  the purge  response characteristics of
similarly  sized  canisters  of  several   designs.    The  Nissan
canister apparently  releases hydrocarbons  relatively  grudgingly
during  the  initial   stages  of  purge, but  shows a  less  drastic
decay of hydrocarbon  stripping  as the purge process  continues.
The  Ford  canister   lies   at   the  opposite end  of  the  purge
response spectrum.  This  canister type  apparently  gives  little
resistance  to  hydrocarbon  removal  during  the  initial  stages of
purge, but  the hydrocarbon stripping rate drops  quite  rapidly
thereafter.   The  other two  curves  (the G.  M.  and EPA  curves)
fall within  these extremes.

     Also shown  in  Figure A10  is  the type of carbon used in
each  of  the canisters  tested.   Looking  at the  right side of
Figure A10  it appears there might be a distinct purge  curve for
each carbon  type.  Upon examining the left side of the figure,
however, it  can  be  seen  that  the  Ford  (coal base)  and  EPA
(wood) curves  are almost  identical  through the early  stages of
purge.   The Nissan  curve  is  distinct   throughout  its  purge
history, however,  and  there  may be some  differences   in  the
fundamental   absorption/desorption  characteristics   of  coconut
based carbons.    The  comparison of purge  curves . by carbon type
should not  be  emphasized, however,  because  there  are  several
other variables  in  canister  design  that could  not  be separated
from carbon  type by this experimental design.

F.   Canister Temperature  During Purge

     The canister listed  in  Table Al under "EPA"  was designed
and  built   for   refueling  emission  control tests.   When  this
canister was loaded  with  activated  carbon,  a  thermocouple  was
installed   in    the    canister   so   that   internal   canister
                             -123-

-------
temperatures could be measured  as  needed.   The thermocouple was
used to monitor internal canister temperature during two  of the
purge sequences done on the refueling canister.

     The  trace of  canister temperature  as a  function  of  the
volume  of  purge  air  pulled through the  canister for   one  of
these tests  is shown in  Figure All.  As  can  be seen  from the
trace, the desorption process absorbs heat.  The  temperature of
the canister drops  from its peak (measured at  135°F immediately
after  loading) down  to   its   lowpoint  (70°F,   6°F  below  the
ambient)  in  under  10  minutes.   The canister  temperature  then
climbs  back  to the  ambient in about ten  minutes and  remains
near the ambient throughout the remainder of the purge.

     This  information  is  significant  for two  reasons.   First,
as hydrocarbon is stripped  from the canister,  the  temperature
of  the canister  falls  rapidly.   The  decrease  in  temperature
could  tend  to   inhibit   the   removal   of   more  hydrocarbons.
Second, as noted  above,  the canister temperature can fall below
the  ambient  during  the  purge  and  in  certain   situations  the
internal  canister temperature   could fall  below the  dewpoint
resulting  in  condensation inside  the  canister.   Although  this
situation  would   probably  not   arise   with   any  regularity,
condensation inside the canister could occasionally occur.

G. Conclusions

     As stated in  the  introduction to  this  paper,  this  test
program had  one primary goal and  several  secondary goals.   The
primary goal of the program was to  evaluate the  purge  response
characteristics  of  several  canisters  of  different  designs.
Part  of  this evaluation was the development of  "purge curves"
which  could  be  used  in  the   development  of   a procedure for
evaluating the purge  capability of  onboard refueling  emission
control systems.   _The  secondary goals  of  the program  were to
investigate the effects of aging, purge  air  flow rate  and purge
air  temperature on hydrocarbon stripping  characteristics.   In
the  course of gathering  data  to  address  the  topics  mentioned
above,  information  was  also  obtained  on  internal  canister
temperatures  during  the  purge process.   Although  this  test
program  had  a limited  scope  several  useful  conclusions can
still be drawn from the data.

     The   conclusions   that  can   be   drawn   concerning   the
secondary   goals   of  the  program  were  stated  as  part  of the
analysis  of   results.   The  results  of   the tests  to  develop
representative  curves  for  the various canisters  merit  some
further  discussion,  however.   The  remainder  of this   section
describes  the  manner  in which  the  representative curves can be
                              -124-

-------
                          FIGURE A-11
to
en
               CANISTER TEMPERATURES

                          DURING PURGE


X"X
U.
w
9
rtt
V
t>
»
TJ
>_x
UJ
fV
UL
D
1-
TEMPERA

140 -
130 -
120 -

1 10 -
100 -

90 -
80 -
70 -
60 -
SO
"







\ AMBIENT TEMPERATURE


au II 1 1 1 1 1 1 1
                20
40       60



  TIME (mfn)
80
                                                 100

-------
used  in the development of a refueling test procedure, and some
recommendations  for  improvements  in  the  experimental  design
used  in this program.

      A  test procedure designed  to evaluate the effectiveness of
onboard control  systems must  test the ability of the  system to
provide capacity for  the storage  of refueling  emissions.   In
the case of a system that uses activated carbon  as  the storage
medium   (which   is  expected  to  be  the  case),  this  involves
stripping hydrocarbons  from the activated  carbon  bed by pulling
hydrocarbon poor air across  it.   A basic understanding  of the
relationship  between  hydrocarbon  stripping  and  purge  air  flow
is  needed  in order  to  develop  a  procedure which  adequately
tests  the  purge  capability  of the control  system.    The  main
purpose of  this test program  was to  develop  a  series of purge
curves  which could  be  used  to  represent  the  range  of  purge
response  patterns  that could  be  expected  of onboard control
system  canisters.

      The  representative purge  curves developed in  this program
are shown in Figure  AlO.   These  curves represent  the expected
performance of  one  liter canisters of the same design  as  those
used  in  the  test  program.   As  discussed  in the  analysis  of:
data,  the curve  for the Chrysler  canister  falls well  outside of
the  range  of  curves  generated  from the data  from  the  other
canisters.   Since there  is  no evidence   to  suggest  that  the
Chrysler  canister   is   radically   different   from  the  other
canisters in  terms  of  material or design, the  curve generated
for   this   canister   is  probably  not  representative  of  this
canister's  typical performance.

     The  remaining  curves  fall  within a relatively  narrow band
in  Figure   AlO.  Although  the   curves   are  closely  grouped
spatially,  there are significant   functional  differences  across
the   curves  in  that   range.    The   curve   representing  the
performance of  the  Ford canister  rises quite  sharply  and shows
a  relatively  clear  breakpoint early  in the  purge process after
which the  curve flattens  out.  The curve generated  from tests
on the  Nissan canister  is less steep  in  the  initial stages of
purge and   tends to break over  more gradually.   The  curves for
the Ford and Nissan canisters are  the  most  dissimilar of those
in Figure AlO  (excluding the  Chrysler  curve).   Because  these
.curves  are  the  most  dissimilar,  they can be  used  to represent
the range  of response  patterns expected  from activated  carbon
canisters.   Therefore,   these  two  curves  were  used  in  the
analysis  performed  in  the development  of  the   refueling  test
procedure.
                              -126-

-------
                           References

     1.     "Refueling  Emissions  from  Uncontrolled  Vehicles,"
D.  Rothman. and R.  Johnson,  EPA-AA-SDSB-85-6,  U.S.  EPA,  OAR,
QMS, ECTD,  July, 1985.

     2.     "Draft    Recommended   Test    Procedure   for    the
Measurement     of     Refueling    Emissions,"    L.D.     Snapp,
EPA-AA-SDSB-85-5, U.S. EPA, OAR, QMS, ECTD, July  1985.

     3.     "Subpart C -  Emission  Regulations  for 19XX and Later
Model  Year  New or In-Use  Light-Duty  Vehicles and New or  In-Use
Light-Duty  Trucks;  Refueling Emissions  Test  Procedure,"  Draft,
U.S. EPA, OAR, QMS, ECTD,  July  1985.

     4.     "Summary   and  Analysis   of   Data   From   Gasoline
Temperature  Survey  Conducted by  American Petroleum  Institute,"
Radian Corporation, May  1976.

     5.     "An  Assessment  of EPA's Certification Procedure  for
Onboard  Refueling  Emission  Control  Systems,"  Technical  note
prepared  for American  Petroleum  Institute by Robert Kausmiez,
Radian Corporation, February  19, 1986.

     6.     "Fuel Volatility   Trend,"  Final Report,  Letter  from
Bruce  B.Bykowski,  Southwest   Research Institute to Craig  Harvey
and Amy Brochu, U.S.  EPA,  September 28, 1984.
                              -127-
                                            U.S. GOVERNMENT PRINTING OFFICE: 1987 - 744-622

-------