United States          Air and Radiation        EPA420-P-02-003
            Environmental Protection                    August 2002
            Agency                          NR-013a
vxEPA     Refueling Emissions for
            Nonroad  Engine Modeling
                                        > Printed on Recycled Paper

-------
                                          EPA420-P-02-003
                                               August 2002
                               for
        Report No. NR-013a
  Assessment and Standards Division
Office of Transportation and Air Quality
 U.S. Environmental Protection Agency

-------
       This technical report describes the methods and assumptions used in the draft
NONROAD2002 emissions model to estimate refueling emissions from nonroad equipment.
The discussion primarily focuses on refueling emissions for gasoline fueled equipment.
Refueling emissions for diesel fueled equipment are discussed briefly at the end of this report.
For ease of reference, the previous version of this technical report has been included as an
appendix.

Background

       Refueling emissions can be divided into two components: spillage and vapor
displacement. Spillage emissions result when fuel is spilled during the refueling process.  Some
or all of the spilled fuel will subsequently vaporize, adding hydrocarbon compounds to the
atmosphere. Vapor displacement emissions result when new liquid fuel being added to a fuel
tank displaces fuel vapors already present in the tank. For example, if one gallon of gasoline is
added to a fuel tank already containing some gasoline, one gallon of gasoline vapor will be
displaced to the atmosphere by one gallon of liquid fuel.

       Although spillage and vapor displacement both occur during a single refueling event, the
draft NONROAD2002 model shows them as separate outputs.  This is to provide additional
flexibility in using NONROAD output for atmospheric modeling.  Spillage emissions and vapor
displacement emissions will typically have different chemical characteristics.  Spillage emissions
will be composed of all of the compounds found in gasoline, while vapor displacement emissions
will be primarily composed of the lighter compounds that have vaporized in the gas tank (the
heavier compounds will remain as liquids in the tank).  Separating refueling emissions into the
two components in the output allows atmospheric modelers to account for the effects of these
differences on atmospheric chemistry.

       For both spillage and vapor displacement, the model initially calculates an emission
factor in terms of grams of emissions per gallon of fuel consumed.  Fuel consumption is then
used to calculate total emissions based on the g/gal emissions factors.

Refueling modes - Gasoline Pump vs. Portable Container

       Many types of nonroad equipment are commonly refueled from a portable container
rather than  from a gasoline pump. Refueling nonroad equipment from a portable container
results in different emissions for both spillage and vapor displacement compared to refueling
from a gasoline pump. These differences are described in detail below.  In addition, the use of
portable containers also results in extra refueling events. Both  spillage and displacement
emissions will also occur when the container is filled from a gasoline pump. However, due to
lack of data, we have not attempted to quantify this extra set of refueling emissions in the draft
NONROAD2002 model.  We welcome comments and data submissions that would help us
quantify these refueling emissions in future versions  of NONROAD.

-------
       Because the different refueling modes result in different emissions, we must make
assumptions in NONROAD about which equipment will be refueled predominantly using a
gasoline pump and which will be refueled predominantly from a portable container. Table 1
shows refueling mode assumptions that have been used in the past compared to the current
version of NONROAD. The draft NONROAD2002 model allows the refueling mode to be
based on horsepower or tank volume. For some types of gasoline-powered equipment, versions
with larger horsepowers are fueled at the pump while  versions with smaller horsepowers are
fueled with a container. All equipment powered by diesel engines are assumed to be fueled at the
pump

                   Table 1:  Alternative Refueling Mode Assumptions.
 Source
Portable container
Gasoline pump
 NEVES
Lawn and garden (except
chippers/stump grinders),
recreational, light
commercial, all other
equipment with tank volumes
less than 6 gallons.
Chippers/stump grinders, all
other equipment with tank
volumes greater than 6
gallons except lawn and
garden, recreational, and light
commercial equipment.
 ARB Small Engine Model
All gasoline 2-stroke engines
and all equipment less than
15 hp.
All gasoline 4-stroke engines
greater than 15 hp
 Draft 2002 NONROAD
- All lawn and garden
equipment.
- Smaller horsepower
gasoline recreational,
industrial, commercial,
logging, recreational marine,
and railway maintenance
equipment.
- All gasoline construction
equipment.
- All gasoline agricultural
equipment.
- All gasoline aircraft ground
support equipment.
- Oil Field Equipment.
- Larger horsepower gasoline
recreational, industrial,
commercial, logging,
recreational marine, and
railway maintenance
equipment.
- All diesels.
Spillage Emissions

       EPA has received no significant new information on spillage emissions since the Nonroad
Engine and Vehicle Emission Study (NEVES) was published in 1991.  NEVES described two
refueling factors; a value of 17 grams of fuel spilled per refueling event for equipment refueled

-------
from a portable container, and a value of 3.6 grams of fuel spilled per refueling event for
equipment refueled from a gas pump. The first value was derived from an OPEI study and the
second value was derived from MOBILE4 estimates for refueling of on-highway vehicles.
NEVES gives the following reasons for the difference in these two values: (1) fuel containers are
more difficult to use than gas pumps, and (2) fuel containers do not have automatic shutoff
capability. Given the lack of new information, we have kept the NEVES values in draft
NONROAD2002 using the following equations (all gasoline spilled is assumed to evaporate into
the atmosphere):

For refueling from portable containers:
       Spillage (g/gal.) = 17.0 + tank volume

For refueling from gasoline pumps:
       Spillage (g/gal.) = 3.6 ^ tank volume

       By using tank volumes in these equations, we assume that all refueling events are fill-ups
of empty tanks.  Given that some portion of refueling events are likely not fill-ups of empty
tanks, this assumption will tend to underestimate spillage emissions. If we knew, on average, the
percentage of the tank volume that was actually being filled, we could develop  a better estimate
of spillage emissions.  However, we are not aware of any data on which to base an assumption.
We welcome comments or data submissions on this issue.

Tank Volumes

       Previous versions of the draft NONROAD Model used the method contained in the
NEVES report [1]. For more information, please see the previous version of this technical report
in the appendix.

       For draft NONROAD2002, EPA expanded the regression approach used by NEVES for
larger equipment to include all applications. The regression equation calculates proper tank
volume for each horsepower bin using the appropriate ratio of tank size to horsepower (in gallons
per horsepower) for each application. The resulting tank sizes have been included in the input
data of NONROAD.  Since actual tank size values are used in draft NONROAD2002 instead of
the calculated values that could have resulted in unrealistically large fuel tank sizes in previous
versions of the model, the 50 gallon cap on fuel tank size has been removed.

       To revise recreational marine fuel tank sizes, EPA used data from an October 1999
database of specifications for new pleasure boats over 25 feet long from Ovation Digital
Productions.1 These data were then analyzed by looking only at the boats with  gasoline engines
    CD-ROM's containing these data are available on the web at www.boatshow.com. In obtaining these data from Ovation,
    the US EPA has agreed that the data contained in the database are provided under license by Ovation digital Productions for
    internal use by the US Environmental Protection Agency, Office of Transportation and Air Quality, and that this EPA office
    will not resell or redistribute this data, including to other offices of the US government, without the consent of Ovation.

-------
and separating outboards from sterndrive/inboards.  Linear regression of these data yielded the
following equations.

       Sterndri ve/Inb oard
             0-300hp     y = 0.3335x          r-squared = 0.1996 (forced through 0,0)
             over 300 hp   y = 1.5871x - 354.1   r-squared = 0.4206

       Outboard
             0 - 100 hp     y = 0.4244x          r-squared = 0.2473 (forced through 0,0)
             over 100 hp   y = 1.2218x - 74.45   r-squared = 0.4246

       where:       y = fuel tank size (gallons)
                    x = engine horsepower

Vapor Displacement

       For the draft NONROAD2002 model, we revised the methodology using the following
formula from the Onboard Refueling Vapor Recovery Rule to calculate vapor displacement
emissions:
             Displacement (g/gal) =
                    EXP(-1.2798-0.0049 x (Td - Ta) + 0.0203 x Td + 0.1315

             where Td = dispensed fuel temperature (degrees F)
                    Ta = ambient temperature (degrees F)
                    RVP = Reid Vapor Pressure (psi)
       This formula relies on user-supplied input for temperature and RVP.  The temperature of
the equipment tank is assumed to be equal to the ambient temperature supplied by the user.  The
temperature of the dispensed fuel depends in part on the refueling mode. For equipment refueled
by portable container, we assume that the temperature of the dispensed fuel equals the ambient
temperature. For equipment refueled from a gasoline pump, NONROAD uses the following
equation (derived from the relationship between equipment tank temperature and dispensed fuel
temperature in the NEVES report) to calculate the temperature of dispensed fuel based on the
ambient temperature:

       Dispensed Fuel Temperature (F) = 62 + 0.6 x (ambient temperature - 62)

Effect of Stage II Vapor Recovery Systems

       Many ozone nonattainment areas are subject to Clean Air Act requirements for Stage n
vapor recovery systems on gasoline pumps.  These systems are designed to capture gasoline
vapors displaced during refueling from a gasoline pump, preventing their release into the
atmosphere.  In general, the overall effectiveness of Stage II systems at controlling refueling

-------
emissions depends on a number of factors including the baseline efficiency of the system used,
the amount of refueling done at stations exempt from Stage n requirements, and the frequency
and stringency of enforcement programs.  For nonroad equipment, the effectiveness of Stage n
systems will also depend on the refueling mode (refueling from a portable container would not be
affected by Stage n controls), the frequency at which nonroad equipment is refueled at exempt
stations (some categories of nonroad equipment may be more likely to be refueled at private
refueling depots exempt from Stage n requirements), and the efficacy of Stage n systems when
refueling nonroad equipment (fuel tank filler openings on nonroad equipment may not be
compatible with Stage II nozzles designed for refueling highway vehicles).

       Given these uncertainties, we propose the following approach to account for Stage II
controls in draft NONROAD2002. When the user specifies that Stage II controls are in place,
they would also specify the effectiveness of the controls as a percent reduction in refueling
emissions.  That percent reduction would only be applied to vapor displacement emissions for
equipment refueled from a gasoline pump.

Diesel Refueling Emissions

       The draft 2002 NONROAD model assumes zero refueling emissions for diesel
equipment. Because diesel fuel has  a higher boiling point than gasoline, refueling emissions
from diesel equipment tend to be much less significant than from gasoline equipment. As a
result, very little refueling emissions data exist for diesel equipment. The NEVES report used a
single emission factor of 0.041 g/gal for vapor displacement from diesel equipment under all
conditions. However, this rate was based on  a study conducted at fuel tank temperatures of
approximately 80 F.  The actual rate at other temperatures was not identified. In addition, EPA
has received no comments or information that supports the use of the NEVES value or suggests
any alternatives.

References

1.  "Nonroad Engine and Vehicle Emission Study", Appendix I, U.S. EPA Office of Air and
Radiation, November,  1991.

2.  "Offroad Equipment Refueling Emissions", Presentation by  California Air Resources staff,
Emissions Inventory Workshop, December 16, 1997.

-------
                                    Appendix

               Refueling Emissions for Nonroad Engine Modeling

                                Report No. NR-013

                                    August 20, 1998

                                     Gary J. Dolce
                       Nonroad Engine Emissions Modeling Team
                           Assessment and Modeling Division
                           US EPA, Office of Mobile Sources
  This technical report describes the methods and assumptions used in NONROAD to estimate
  refueling emissions from nonroad equipment. The discussion primarily focuses on refueling
 emissions for gasoline fueled equipment. Refueling emissions for diesel fueled equipment are
                         discussed briefly at the end of this report.

                                     Background

  Refueling emissions can be divided into two components: spillage and vapor displacement.
  Spillage emissions result when fuel is spilled during the refueling process.  Some or all of the
  spilled fuel will subsequently vaporize, adding hydrocarbon compounds to the atmosphere.
 Vapor displacement emissions result when new liquid fuel being added to a fuel tank displaces
 fuel vapors already present in the tank. For example, if one gallon of gasoline is added to a fuel
  tank already containing some gasoline, one gallon of gasoline vapor will be displaced to the
                         atmosphere by one gallon of liquid fuel.

 Although spillage and vapor displacement both occur during a single refueling event, the final
    version of NONROAD will show them as separate output. This is to provide additional
 flexibility in using NONROAD output for atmospheric modeling. Spillage emissions and vapor
displacement emissions will typically have different chemical characteristics.  Spillage emissions
will be composed of all of the compounds found in gasoline, while vapor displacement emissions
  will be primarily composed of the lighter compounds that have vaporized in the gas tank (the
 heavier compounds will remain as liquids in the tank).  Separating refueling emissions into the
  two components in the output allows atmospheric modelers to account for the effects of these
                          differences on atmospheric chemistry.

  For both spillage and vapor displacement, the model initially calculates an emission factor in
  terms of grams of emissions per gallon of fuel consumed. Fuel consumption is then used to
               calculate total emissions based on the g/gal emissions factors.

-------
               Refueling modes - Gasoline Pump vs. Portable Container

Many types of nonroad equipment are commonly refueled from a portable container rather than
   from a gasoline pump.  Refueling nonroad equipment from a portable container results in
   different emissions for both spillage and vapor displacement compared to refueling from a
gasoline pump.  These differences are described in detail below.  In addition, the use of portable
containers also results in extra refueling events.  Both spillage and displacement emissions will
also occur when the container is filled from a gasoline pump. However, due to lack of data, we
     have not attempted to quantify this extra set of refueling emissions in this version of
 NONROAD. We welcome comments and data submissions that would help us quantify these
                  refueling emissions in future versions of NONROAD.

Because the different refueling modes result in different emissions, we must make assumptions
 in NONROAD about which equipment will be refueled predominantly using a gasoline pump
 and which will be refueled predominantly from a portable container. Table 1 shows refueling
mode assumptions that have been used in the past.  Please note that the simplified method used
   in the draft version of NONROAD was in part necessitated by the fact that draft version of
    NONROAD was not designed to easily handle differences in refueling mode based on
 horsepower or tank volume cutoffs.  The final version of NONROAD will allow the refueling
 mode to be based on more complicated criteria such as horsepower or tank volume. We could
 use the Nonroad  Engine and Vehicle Emission Study (NEVES)1 approach, the California Air
 Resources Board (ARE) small engine2 approach, or some other approach. We invite comment
   on what to assume in the final version of NONROAD. We are also interested in any data
 submissions, such as survey results, that would be useful in separating equipment by refueling
                        mode in future versions of NONROAD.
                   Table 1. Alternative refueling mode assumptions.
 Source
Portable container
Gasoline pump
NEVES
Lawn and garden (except
chippers/stump grinders),
recreational, light
commercial, all other
equipment with tank volumes
less than 6 gallons.
Chippers/stump grinders, all
other equipment with tank
volumes greater than 6
gallons except lawn and
garden, recreational, and light
commercial equipment.
ARE Small Engine Model
All gasoline 2-stroke engines
and all equipment less than
15 hp
All gasoline 4-stroke engines
greater than 15 hp
Draft NONROAD
All lawn and garden,
recreational equipment,
outboards and personal
watercraft.
All others.

-------
                                  Spillage emissions

   We have received no significant new information on spillage emissions since the Nonroad
 Engine and Vehicle Emission Study (NEVES) was published in 1991. NEVES described two
 refueling factors; a value of 17 grams of fuel spilled per refueling event for equipment refueled
    from a portable container, and a value of 3.6 grams of fuel spilled per refueling event for
 equipment refueled from a gas pump. The first value was derived from an OPEI study and the
   second value was derived from MOBILE4 estimates for refueling of on-highway vehicles.
NEVES gives the following reasons for the difference in these two values: (1) fuel containers are
   more difficult to use than gas pumps, and (2) fuel containers do not have automatic shutoff
capability.  Given the lack of new information, we have incorporated the NEVES values into the
  draft version of NONROAD using the following equations (all gasoline spilled is assumed to
                             evaporate into the atmosphere):

                         For refueling from portable containers:
                          Spillage (g/gal.) = 17.0 + tank volume

                           For refueling from gasoline pumps:
                           Spillage (g/gal.) = 3.6 ^ tank volume

  By using tank volumes in these equations, we assume that all refueling events are fill-ups of
 empty tanks. Given that some portion of refueling events are likely not fill-ups of empty tanks,
   this assumption will tend to underestimate spillage emissions.  If we knew, on average, the
 percentage of the tank volume that was actually being filled, we could develop  a better estimate
 of spillage  emissions.  However, we are not aware of any data on which to base an assumption.
                 We welcome comments or data submissions on this issue.

                                    Tank Volumes

   The NEVES report includes a lengthy discussion of gasoline fuel tank volumes. Fuel tank
   volumes in NEVES were primarily derived from manufacturer supplied information with
 adjustments made to deal  with mismatches between the way manufacturers and EPA describe
 equipment  categories.  For a few equipment categories where manufacturer information was not
    available, tank volumes in NEVES were based on EPA judgement. For larger farm and
 construction engines, NEVES estimated tank volumes using a regression equation that related
     engine horsepower to tank volume.  Based on this equation, NEVES used  the average
             horsepower of each application to estimate an average tank volume.

While developing the draft version of NONROAD, we considered modifying and expanding the
regression approach used by NEVES for larger equipment to include all applications and to have
 the model calculate the proper tank volume for each horsepower bin using the appropriate ratio
     of tank size to horsepower (in gallons per horsepower) for each application. We did a

-------
 preliminary analysis of a small amount of available data and incorporated the resulting ratios in
                            the draft version of NONROAD.

    After further review of the data and methodology used in the draft NONROAD, we have
concluded that this new approach needs a considerable amount of additional work before we can
   be certain that it improves on the NEVES methodology.  Therefore, for the final version of
  NONROAD, we propose to go back to using the original NEVES methodology and data. We
     invite comment or data submissions that would help us to develop a better tank volume
                                    methodology.

                                 Vapor Displacement

   For the final version of NONROAD, we propose to adopt the methodology described in the
NEVES report with some minor modifications. NEVES used the following formula to calculate
                             vapor displacement emissions:

          Displacement (g/ gal) = -5.909 - 0.0949 x dt + 0.0884 x T + 0.485 x RVP

       where:dt=  Temperature of Equipment Tank - Temperature of Dispensed Fuel (F)
                        T  = Temperature of Dispensed Fuel (F)
                              RVP= Reid Vapor Pressure

 NEVES used this formula to create a simple lookup table of average summer and winter season
emissions based on average temperatures and RVP values. The final version of NONROAD will
     incorporate this formula and rely on user supplied input for temperature and RVP. The
temperature of the equipment tank is assumed to be equal to the ambient temperature supplied by
   the user. The temperature of the dispensed fuel depends in part on the refueling mode.  For
 equipment refueled by portable container, we assume that the temperature of the dispensed fuel
     equals the ambient temperature.  For equipment refueled from a gasoline pump, we will
   incorporate the following equation (derived from the relationship between equipment tank
temperature and dispensed fuel temperature in the NEVES report) to calculate the temperature of
                     dispensed fuel based on the ambient temperature:

          Dispensed Fuel Temperature (F) = 62 + 0.6 x (ambient temperature - 62)

                       Effect of Stage II Vapor Recovery Systems

  Many ozone nonattainment areas are subject to Clean Air Act requirements for Stage II vapor
  recovery systems on gasoline pumps. These systems are designed to capture gasoline vapors
displaced during refueling from a gasoline pump, preventing their release into the atmosphere.  In
 general, the overall effectiveness of Stage n systems at controlling refueling emissions depends
   on a number of factors including the baseline efficiency of the system used, the amount of
refueling done at stations exempt from Stage n requirements, and the frequency and stringency of
  enforcement programs. For nonroad equipment, the effectiveness of Stage n systems will also

-------
   depend on the refueling mode (refueling from a portable container would not be affected by
   Stage II controls), the frequency at which nonroad equipment is refueled at exempt stations
  (some categories of nonroad equipment may be more likely to be refueled at private refueling
 depots exempt from State n requirements), and the efficacy of Stage n systems when refueling
 nonroad equipment (fuel tank filler openings on nonroad equipment may not be compatible with
                  Stage II nozzles designed for refueling highway vehicles).

 Given these uncertainties, we propose the following approach to account for Stage n controls in
 the final version of NONROAD.  When the user specifies that Stage n controls are in place, they
would also specify the effectiveness of the controls as a percent reduction in refueling emissions.
  That percent reduction would only be applied to vapor displacement emissions for equipment
 refueled from a gasoline pump.  EPA will develop guidance as to the appropriate effectiveness
                     levels for Stage II controls for nonroad equipment.

                               Diesel Refueling Emissions

   Because diesel fuel  has a higher boiling point than gasoline, refueling emissions from diesel
 equipment tend to be much less significant than from gasoline equipment. As a result, very little
 effort has been spent  studying diesel refueling emissions.  NEVES did not include any spillage
 emissions for diesel equipment.  NEVES used a single emission factor of 0.041 g/gal for vapor
  displacement from diesel equipment under all conditions. However, this rate was based on a
    study conducted at fuel tank temperatures of approximately 80 F. The actual rate at other
 temperatures was not  identified.  We invite comment on whether to use the single emission rate
  used in NEVES for all diesel refueling emissions due to vapor displacement, or to assume no
               diesel refueling emissions until further information is available.
                                      References

1.  "Nonroad Engine and Vehicle Emission Study", Appendix I, U.S. EPA Office of Air and
Radiation, November, 1991.

2.  "Offroad Equipment Refueling Emissions", Presentation by California Air Resources staff,
Emissions Inventory Workshop, December 16, 1997.
                                           10

-------