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
------- |