Technical Report EPA-AA-SDSB-85-5q Draft Recommended Test Procedure for the Measurement of Refueling Emissions By Lisa D. Snapp July 1985 NOTICE Technical Reports do not necessarily represent final EPA decisions or positions. They are intended to present technical analysis of issues using data which are currently available. The purpose in the release of such reports is to facilitate the exchange of technical information and to inform the public of technical developments which may form the basis for a final EPA decision, position or regulatory action. Standards Development and Support Branch Emission Control Technology Division Office of Mobile Sources Office of Air and Radiation U. S. Environmental Protection Agency ------- Table of Contents I. Introduction 1 II. History of the Procedure . 1 III. Recommended Procedure 3 A. Equipment 3 B. Procedure 4 C. Parameters 7 1. Test Fuel Volatility 12 2. Dispensed Fuel Temperature 16 3. Temperature Differential 20 4. Fuel Level Prior to Refueling 24 5. Fuel Dispensing Rate 25 D. Vehicle Preconditioning 28 E. Canister Purging and Vehicle Drivedown 30 F. Testing Requirements for In-Use Vehicles .... 33 1. Preconditioning Problems 33 2. Testing Problems 36 ------- I. INTRODUCTION The emissions which occur as a result of vapors displaced during the refueling of motor vehicles are currently of public and Agency concern. These vapors have possible adverse health effects, as does benzene, a substance in gasoline vapors linked to cancerous and mutagenic effects. They also contribute to the formation of photochemical smog, which is of great concern in areas which exceed the National Ambient Air Quality Standard (NAAQS) for ozone. Because of these concerns, EPA has undertaken a detailed evaluation of various regulatory strategies which could be used to reduce emissions due to refueling.[1] One aspect of this evaluation involves the measurement of refueling emissions from both current vehicles and vehicles equipped with prototype refueling vapor control systems. EPA has been involved in the development of a test procedure designed to measure these emissions for quite some time. EPA began working with test procedure issues during the summer of 1984. A workshop was held on October 17, 1984 at which EPA shared its ideas on the test procedure and on critical test parameters with workshop participants. Those present at the workshop were urged to submit comments on the draft recommended practice and some of them did so. A follow-up meeting was held on February 15, 1985 at which industry representatives further detailed their thoughts about the draft test procedure as proposed by EPA. EPA has considered the comments submitted in response to the initially proposed test procedure in development of this current version of the recommended practice. This report summarizes the development of the recommended test procedure including discussions of equipment needed and values for critical test procedure parameters. II. HISTORY OF THE PROCEDURE In 1973, the Society of Automotive Engineers (SAE) approved Recommended Practice J1045, "Instrumentation and Techniques for Vehicle Refueling Emissions Measurement," which was designed to measure refueling emissions from uncontrolled vehicles.[2] The procedure was designed mostly for experimental purposes, in that it was to be used to determine the effects of various parameters, such as temperatures, fuel volatility, and so forth, on refueling emissions. With these parameters held constant, it could also be used to measure the control efficiency of refueling emission control devices. ------- -2- The SAE procedure incorporated four key parameters for use when measuring refueling emissions. These were the fuel volatility, fuel tank temperature, dispensed fuel temperature, and fuel level prior to refill. Also specified was the testing equipment to be used, most notably the sealed housing for evaporative determination (SHED), and procedures for conducting the tests. In the SAE recommended practice, the test fuel specified had a volatility of 9.0+0.5 psi as measured by Reid Vapor Pressure (RVP), which is comparable to normally specified values in Federal test procedures and SAE recommended practices for evaporative and tailpipe emissions testing. Fuel tank temperature was recommended to be 95+2°F, which SAE found to be a reasonable temperature to expect in summer driving conditions; no supporting data was cited. Dispensed temperature was 82+1°F, an average summer dispensed temperature reported by Scott Research Laboratories.[3] Fuel level prior to refill was suggested at one-fourth tank capacity; again, no data was cited in support of this value. The equipment to be used, particularly the SHED, was also prescribed by the SAE procedure. It was specified to be the same as the SHED generally used for evaporative emissions testing, with minor alterations such as an added fuel dispensing hose and nozzle. Also specified in some detail were the equipment components necessary for heating the fuel tank, analyzing the HC emissions, mixing and purging the air, and recording temperatures. Procedures for using the equipment were given, along with tolerances of applicable parameters. The basic test procedure involved draining the fuel tank to nominally empty, filling it to one-fourth of tank capacity, and pushing the vehicle into the SHED. Fuel tank temperature was then to be raised to the specified test temperature by means of an electric heating pad applied beneath the tank, generally referred to as single blanket heating. When the desired temperature was reached, the SHED was to be closed and, upon stabilization of background hydrocarbon concentration, the operator was to enter the SHED and refuel the vehicle, allowing the nozzle to automatically shut off when the tank was full. Readings of hydrocarbon concentration, elapsed time, ambient temperature, and barometric pressure were to be taken, and the mass of emissions due to the refueling could then be calculated. The SAE procedure described above and the Federal test procedure for evaporative emissions were used to develop the ------- -3- EPA recommended test procedure for refueling emissions outlined below. This procedure uses the basic SHED test approach for measuring HC refueling emissions, and integrates it into the existing federal test procedure (FTP) for exhaust and evaporative emissions. The remaining sections of this report discuss this procedure and are organized as follows. First, the equipment to be used to perform the testing will be described in Section III.A. This will be followed in Section III.B. by a review of the test procedure sequence itself. Sections III.C.1-5 will deal with the derivation of specific test conditions, including test fuel volatility, dispensed fuel temperature, temperature differential, fuel level prior to refueling, and fuel dispensing rate. Vehicle preconditioning is discussed in Section III.D, followed by a more detailed discussion of the canister purging and vehicle drivedown portion of the test sequence in Section III.E. Finally, the application of this test procedure to in-use vehicles is described in Section III.F. III. RECOMMENDED PROCEDURE A. Equipment The equipment to be used in the EPA refueling test is similar to that recommended by SAE and that specified in the evaporative emissions portion of the Federal test procedure. This equipment consists of a SHED, fuel dispenser, meters, and recording devices. The SHED itself is essentially the same, except that the ^uel line opening in the SHED wall must be lower than the fuel tank opening on the vehicle, to eliminate the potential for siphonage and spillage onto the floor. Other pieces of equipment - the flame ionization detector (FID) hydrocarbon analyzer, purge and mixing blowers, thermocouples and recording devices - are also essentially the same in the EPA test procedure as in the SAE procedure. The fuel dispensing system varies in that the EPA procedure requires a self-supporting nozzle and a fuel flow shut-off switch on the fuel cart. This eliminates the need for a nozzle operator inside the SHED when vapors are accumulating. Another difference is in the fuel tank heating system used during preconditioning. The SAE procedure prescribes a maximum liquid-vapor temperature differential of 6F°, which limits the speed of liquid heating, while the EPA procedure allows heating as quickly as possible, with the temperatures then held in equilibrium for ten minutes after completion of heating. This equipment is used to conduct the refueling test, which is integrated into the FTP as described in the following section. ------- -4— B. Procedure The proposed test sequence for exhaust, evaporative, and refueling emissions uses the current sequence for exhaust and evaporative emissions and adds refueling tests at the beginning and end. These two new tests are designed to check the capacity and purge capability of the refueling control system separately. Their integration into the FTP is shown in Figure 1 and described below. The test begins with a refueling test performed to ensure that the overall vapor control capacity of the canister is sufficient for a complete fill-up. Certification test vehicles are expected to arrive at the test site purged to a level commensurate with a nearly empty fuel level. Prior to testing, the fuel tank will be drained and filled to 10 percent of capacity with standard test fuel. The refueling canister will be disconnected during this operation. In-use vehicles may arrive in any condition, and will be preconditioned by 50 miles of driving on standard test fuel using the durability driving schedule contained in Appendix IV of Part 86 of the Code of Federal Regulations (CFR), or equivalent urban driving. Following the 50 miles of driving, the fuel tank will again be drained and fueled to 10 percent of capacity. As with certification vehicles, the refueling canister will be disconnected for this step. To prevent any canister loading while any vehicle awaits testing at the facility, the gas cap will be loosened or removed when the vehicle is not in operation. Immediately prior to testing, the vehicle will be pushed into the SHED and the doors and luggage compartments will be opened. The fuel in the vehicle fuel tank will then undergo a heat build to the prescribed test temperature, followed by a ten minute equilibrium period. The fuel dispensing nozzle will be placed in the vehicle fill neck, and the SHED closed and sealed. A switch outside of the SHED will then be used to start the flow of fuel. Gasoline shall be dispensed into the fuel tank until the automatic shutoff feature on the fuel nozzle is tripped. At the time that fuel flow is terminated, the fuel tank must be filled to at least 95 percent of nominal tank cagaqlty. The total vehicle refueling emissions will be defined asi the hydrocarbon level in the SHED (as measured by FID) at the time of termination of the fueling event. For baseline testing of vehicles equipped with current technology, some difficulties with spillage at nozzle shutoff may occur. In order to avoid excessive emissions and high ------- Figure 1 § $*.130-71 THta 40—Protection •# lnvironnt«nt START Refueling Capacity Test - 85% fill test - canister saturatio: Fua drain t rm I HOUR MAX. DYNO PMCONDITIONtNO S MM. MAX. C010 SOAK PARKING Disconnect Canister OIISH ONLY Reconnect ~ '7 J4 HOURS (no ntoa. diotolf) . (or 0IURNA1 HIAT IUIt.0 • HIAT full-1 HOUR • to-uv IVAf. TIST NOT RtQ COID START IXHAUST TIST HC RUNNING IOSSCS-AS *10 10 MIN HOT START IXHAUST TIST HOT SOAK INCIOSUM TEST Refueling Purge Test - LA-4 drivedown - 30% fill test CNO Plfl« VI 10 TtlT SIOUINCI ------- -6- test-to-test variability caused by spillage, manual nozzle shutoff at the 95 percent full level is sometimes necessary in this case. For new vehicles equipped with onboard controls, however^j^fueling systems should be designed such that no spillagaf^iil occur when a properly functioning nozzle is used to fuel the- vehicle. Any emissions caused by spillage would be included as part of the vehicle's total refueling emissions. Next, the vehicle fuel tank will be drained and refilled repeatedly until the canister is saturated and hydrocarbon breakthrough occurs. "Breakthrough" will be defined as that point in time that the concentration indicated by the FID hydrocarbon trace increases by a factor of at least 3 in a one minute period. At this point, the vehicle will begin the existing test sequence for exhaust and evaporative emissions. Currently, the test procedure specifies a fuel drain and fill at this point. In the refueling test procedure the vehicle manufacturer will have the option to adjust the fuel level in the tank to the specified level of 40 percent following canister saturation, or perform the existing drain and 40 percent fill operation. In either case the refueling canister will be disconnected. The next steps in the procedure follow the existing FTP. First, a single dynamometer preconditioning cycle is done, followed by a cold soak for between 10 and 35 hours. Additional preconditioning, allowed under the current FTP, will no longer be permitted. After the soak, the FTP calls for the fuel tank to be drained and filled to 40 percent of tank capacity. For the refueling procedure, the canister will be disconnected before this fuel drain and fill in order to avoid additional canister loading.* The canister will then be reconnected and the FTP will continue as usual, with the diurnal breathing loss SHED test, cold and hot start dynamometer exhaust tests, and hot soak evaporative SHED test. This aspect of the procedure is still subject to change. Leaving the canister connected would require substantially increased purge capacity during the dynamometer preconditioning cycle and might be unnecessarily severe. On the other hand, it may happen that a purge strategy capable of adequately purging all the collected refueling vapors during normal driving would have a purge capability at saturated canister conditions which approximates this higher rate. This subject is discussed further below in connection with the drivedown portion of the test. ------- -7- The& final... stage in the refueling test procedure is a purge test sequence involving dynamometer operation of the vehicle to driven down* approximately 30 percent of the fuel in the tank, followedjr^bXi a partial refueling operation to replace the consumed^ fuel. The purpose of this stage is to demonstrate that the refueling control system has adequate purge capacity to purge accumulated refueling vapors. For reasons to be discussed further below, the drivedown sequence in this draft recommended practice is initially being applied only to vehicles employing refueling emissions control systems which are completely separate from any evaporative emissions controls. Vehicles having refueling emissions control integrated with either the diurnal fuel tank emissions control or all evaporative emissions controls are not being required to undergo the drivedown portion of the test at this time. However, it is not yet clear to EPA that the refueling procedure lacking the drivedown is, in fact, completely adequate for testing integrated refueling and evaporative control systems. In fact, there are indications that a 30 percent drivedown operation may itself be inadequate to verify proper canister purge for a full refueling. These uncertainties are discussed further below in Section E, Canister Purging and Vehicle Drivedown. The purge test will begin with a drivedown sequence on the dynamometer, consisting of sequential Urban Dynamometer Driving Schedules, contained in Appendix I of Part 86 of the CFR and normally referred to as the LA-4, alternating with one hour hot soaks. This sequence is 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. The LA-4/hot soak sequence will be repeated until approximately 30 percent of the tank fuel capacity has been used. The vehicle will then be put back into the SHED, and a refueling test conducted as at the beginning of the proposed procedure, except that the refueling amount shall approximately correspond to the amount of drivedown. C. Parameters In the draft procedure, five key test parameters affecting refueling emissions were identified: test fuel volatility, dispensed fuel temperature, differential between dispensed fuel temperature and vehicle liquid fuel tank temperature, initial and final fuel tank volumes, and fuel dispensing rate. The values for these test parameters have been chosen so that roughly 90 percent of the refueling events under ozone-prone ------- -8- summer conditions would be covered.* Test parameter values are shown in Table 1. In order to choose the appropriate values, it was necessary to know typical levels or have frequency distributions of the levels of the key parameters during the months and in the regions of interest. Information used to select the appropriate values for the key parameters was taken from several sources. These sources are briefly described below and summarized in Table 2. One important study was conducted by the American Petroleum Institute (API) Product Temperature Task Force in 1975. In this survey, gasoline temperature data was collected at 56 service stations throughout the United States.[4,5] The data analysis report by Radian Corporation under contract to API, and the final API report generated from this data, provided information which EPA used to establish frequency distributions of dispensed fuel temperature and temperature differential for the country, divided into six regions as shown in Figure 2. Distributions of the fuel level in a vehicle before it is taken in for a fill-up were derived from six studies done as part of California's service station Stage II vapor control program.[6,7,8,9,10,11] These studies recorded amount of gasoline dispensed per refueling at a variety of service stations. Typical flow rates used in dispensing such fuel were provided to EPA by an informal API survey of member companies.[12] Maximum fuel volatility prescribed in the summer months of the appropriate regions was taken from the American Society for Testing and Materials (ASTM) standards.[13] The effects of weathering on volatility were studied in the EPA baseline refueling testing program and in an API study conducted by Stanford Research Institute to develop a test procedure for service station vapor control systems.[14] The values which these sources provided for the key test parameters are discussed in the following sections. Since refueling emissions control is also important for reasons other than ozone reductions, the test parameters must also ensure adequate control requirements for non-summer conditions. Based on the refueling emission factors developed in EPA's baseline study[16], the choice of summertime condition will ensure year-round control. ------- Designation Volatility, RVP Table 1 Critical Test Parameters Meaning dispensed temperature, Td temperature differential, AT fuel level dispensing rate test fuel volatility, expressed in Reid Vapor Pressure temperature of dispensed fuel tank temperature minus dispensed temperature level of fuel in vehicle prior to refueling, percent of capacity to nearest 0.1 U.S. gal flow rate of fuel as it is dispensed Value 11.5 + 0.5 psi 88 + 2°F +2 to +5 F° 10% 8-10 gal/min ------- Oonch 1975 1984 1974 1984 1976 1976 1979 1982 1976 1981 1984 Table 2 Sources for Parameter Values Source Parameter Ifedian/API[4,5] ASTML13] Stanford[14] SwRI[17] Union 76[6] Union 76[7] Union 76[8] Healy£9] ExxonC10] Emco VJheaton[ll] API Letter[12] To, AT RVP RVP EVP fuel level fuel level fuel level fuel level fuel level fuel level dispensing rate Comments Measured underground, dispensed, and tank fuel and airtoient temperatures Prescribed fuel volatility classes by state and month Studied effect of variables, including tank fuel RVP (weathering) cn refueling emissions Studied amount of fuel weathering experienced by in-use vehicles Gave fill fraction for 105 refuelings, fill-upe only Gave gallons of fuel dispensed for 117 refuelings, fill-ups only Gave gallons of fuel dispensed for 103 refuelings, with 89 fill-ups Gave gallons of fuel dispensed for 99 refuelings, with 18 fill-ups Gave gallons of fuel dispensed for 573 refuel ings, with 327 fill-ups Gave gallons of fuel dispensed for 99 refuelings, with 40 fill-ups Surveyed API member companies informally for typical fuel flow rafao ------- V- J 9 i ------- -12- 1. Test Fuel Volatility Fuel volatility has a significant impact on the level of refueling emissions, and as such is one of the critical parameters to be specified in this test procedure. Commercial fuels with volatilities that are higher than that of EPA certification fuels are currently sold in areas and months that are ozone-prone. Ozone-prone states currently include any state with a Standard Metropolitan Statistical Area (SMSA) which had a violation of the NAAQS for ozone in 1982 or 1983.[15] The months of violation are mainly May through September, and the states are predominately in the Midwest, Northeast, Southeast and Southwest areas of the United States, as can be seen in Figure 3. The 30 ozone-prone states fall predominantly within four of the six regions which Radian had designated in its study of gasoline temperature[4], also shown in Figure 3. These regional designations are: 1 (Northeast), 2 (Southeast), 3 (Southwest), and 4 (West), and since they correspond so closely to the ozone-prone regions, they will be used as the relevant regions in subsequent choices of parameter values. ASTM specifies, by state and month, maximum volatility levels which can be used by gasoline distributors. These levels are sufficiently high to ensure ease of engine start-up, but are not so high as to cause vapor lock. They are identified as classes A through E, with A being the lowest maximum volatility at 9.0 psi, and E being the highest at 15.0 psi. Generally, classes A, B and C are for use in the summer months, while D and E are for winter use; in each season, the lower volatility classes are specified for the hotter regions. Class C volatility, with a maximum allowable Reid Vapor Pressure (RVP) of 11.5 psi, is the highest level prescribed for the summer. It is widely prescribed in those months in states which EPA has identified as ozone-prone, as can be seen in Figure 4. It is clear, at least from the ASTM specification, that summertime fuels can have significantly higher volatility than the 9 psi RVP of EPA's current certification fuel. EPA is now in the process of examining fuel volatility to more accurately establish the characteristics of summertime fuels for the , foreseeable future, and is therefore not- in a position to definitively establish its test fuel RVP for this procedure. In the interim, EPA plans to require the use of 11.5 psi for test fuel volatility. This volatility represents the RVP of summer commercial fuel as used in current EPA emission factors ------- ** o, O •» > •• Correlation of Ozone-prone States (•) and ------- Number of Class C (11.5 RVP) Months in Mav thru September ------- -15- test programs, as well as the ASTM class C volatility limit which shows a widespread correlation with the ozone prone areas and months. Weathering of the . fuel in the vehicle tank is also a factor in refueling emissions. Agitation, thermal cycling, temperature extremes, and time aging can all lower the RVP of the fuel in the vehicle. The effects of this weathered fuel are only beginning to be studied; however, it seems that two counteracting effects occur. The first is that the weathered fuel in the tank has a lower RVP, and hence a lower concentration of vapors. This would result in lower refueling emissions compared to non-weathered fuel, if refueled with similar RVP fuel. However, the fuel being added is non-weathered, and there is thus a balancing of the pressures of the new, non-weathered fuel being added and the older, weathered fuel in the tank. The higher RVP fuel vaporizes as it enters the tank in order to equalize the vapor pressures. Excess vapors are then vented from the tank, causing higher emissions. Of the two effects, the second is apparently dominant under most conditions, according to limited data. These data come from weathering effects tests conducted as part of EPA's baseline testing for refueling emissions[16], and from a report by Stanford Research Institute concerning refueling.[14] The EPA tests involved dispensing commercial fuel (RVP of approximately 12 psi) into a tank filled to 10 percent of capacity with an Indolene-commercial blend (RVP of approximately 10 psi). The emissions were about 2 g/gal higher than when dispensing blended into blended fuel. This was apparently due to vaporization of the commercial fuel, with higher RVP, as it was dispensed into the tank. In fact, the grams per gallon emissions when adding commercial to blend fuel (12 psi to 10 psi) were approximately 1 g/gal higher than those generated when adding commercial to commercial fuel (12 psi to 12 psi). In the Stanford study, increases in emissions for weathered fuel were also seen. However, these increases were significant only with large positive AT, created by increasing tank temperature while holding dispensed temperature constant. The difference in emissions between weathered and non-weathered fuel, increasing with increasing ATs, reached a high of 1.7 g/gal at the highest tested AT, 35F°. At AT between 0 and 10F°, there was no noticeable difference in emissions with weathered or non-weathered fuel initially in the tank. It could be that the lower volatilities used in the ------- -16- Stanford program (6.9 and 8.6 psi) account for the smaller difference seen at small ATs. For the Stanford study, tank fuel RVP was about 20 percent lower than dispensed RVP, at 6.9 and 8.6 psi, respectively. For the EPA study, the difference was about 17 percent. In a project by Southwest Research Institute, the amount of weathering a tank of fuel actually experiences was studied. [17] Two vehicles were each driven 50 miles a day and parked overnight until a full tank was used. Three fuels were tested, with initial RVPs of 9.0, 10.5, and 12.0. For each fuel, the RVP dropped approximately 9 percent as the tank of fuel was consumed. Very limited data indicates this value may increase if the fuel sits in the tank for extended periods, especially when approaching empty. Using the Southwest data as an indicator of typical fuel weathering to be expected in-use, it appears that fuel weathering could increase refueling emissions by up to about 0.5 g/gal. However, because of the uncertainty of this amount, coupled with the difficulty of requiring the handling of two test fuels and the inherently conservative nature of the rest of the test parameters, no specific provisions to account for fuel weathering are included in the recommended practice. 2. Dispensed Fuel Temperature SAE recommended a dispensed fuel temperature of 82+1°F, which was the average dispensed fuel temperature observed for summer months, as reported by Scott Research Laboratories.[33 However, EPA desires to test at conditions which will assure control for the majority of cases, rather than at average conditions. Data described below shows a 90th percentile point of 88°F for daily average dispensed temperatures in May through September for ozone-prone areas. Therefore, the recommended practice uses 88+2°F for the dispensed fuel temperature. This 90th percentile point was determined by EPA using data taken from the Radian report analyzing the gasoline temperature data gathered in 1975 by the API Task Force. [4]* For the months of interest (May-September), 1975 temperatures were only approximately 0.2°F higher than the recorded historical average (over about 40 years) for the cities in which data was gathered. Therefore, the ambient temperatures for the Radian study are considered adequately representative of normal conditions. ------- -17- In this study, temperatures recorded were of fuel in: 1) the bulk delivery tank, 2) the underground storage tank, 3) the dispensing nozzle, and 4) the vehicle fuel tank. From this data, a mathematical model of dispensed temperature was formulated, in which the dispensed fuel temperature exponentially approaches either the underground tank temperature (during fill cycles), or ambient temperature (between fill cycles), as a function of cycle time. This model has been subsequently validated by API, and may be useful in retail dispensed fuel temperature adjustment in the future. Dispensed temperatures have been shown to be predicted to within +0.5 percent. The dispensed fuel temperatures actually measured in this study were used to determine the appropriate value for the test procedure. Of special significance are temperatures in ozone problem areas during months in which ozone effects are significant. These are the four regions which correspond to the ozone-prone states, shown previously in Figure 3, with the critical months being May through September. Distributions of dispensed fuel temperatures in each of these four regions show that 5-day average dispensed fuel temperatures range from 59 to 93°F during the months of interest.- Higher temperatures predominate in Region 2 (SE), as can be see in Figure 5. This region also has a majority of states with four or five of the five ozone-prone months classified as volatility class C, as indicated previously in Figure 4. It might be argued that inclusion of Region 2 would unduly bias the data toward higher dispensed temperatures. However, there are two other factors which must also be considered. First, the figure shows 5-day averages, and so does not show the peaks from days which exceed the average. Second, the maximum dispensed temperature of the day, calculated from the model described above and using typical ambient temperature differentials[4,18], would generally be 4 to 7°F above the average, again indicating a need to include higher temperatures (such as those found in Region 2) in the test procedure. The 90th percentile point covering all four regions is 88°F for the five ozone-prone months of May-September, as shown in Figure 6. More stringent conditions could be found by looking at just the two major ozone months, July and August. For such conditions, the 90th percentile for all four regions combined is 9l°F. However, the five month dispensed ------- Figure 5 Regional Dispensed Fuel Temperature Five-Day Average, Sunnier 1975 95 . 90 . 85 80_ Dispensed ramperature 7C- (°F) 70- 65. 60- 55- 2 _ / 1* Region 1 (NE) Region 2 (SE) —• Region 3 (SW) —— Region 4 ( W) ' 5 10 15 20 25 | 5 » IS 20 24 | 5 K> IS 2D 25 1 S 10 IS 20 23 | S U 15 20 25 MAY JUNE JUIiY AUGUST SEPTEMBER Ozone-Prone Months ------- Figure 6 Distribution of Dispensed Fuel Temperatures May thru Sept 1975. lily Averages, Regions 1,2,3 and 4 100 -t~L2 80 60. 40- 20- - Cumulative Percent 5 Day Average Percent f t 50th percentile - 80°F 90th percentile a 88 °F Dispensed Temperature (F°) ------- -20- temperature value of 88°F was deemed more representative for the ozone-prone months overall. 3. Temperature Differential The difference between fuel tank temperature and dispensed temperature, AT, is a critical parameter. It directly affects the magnitude of vapor growth or shrinkage in the fuel tank, and thus the amount of vapor displaced from the tank. The SAE recommended practice does not indicate any particular AT, although it does state a desired fuel tank temperature of 95+2°F. This value, in conjunction with the recommended dispensed temperature, gives a AT of +10 to +16 F°. Either parameter could be specified in the procedure, but AT is the parameter which EPA feels should be directly constrained, since it is so critical to the level of refueling emissions. It is also the parameter that was recorded for individual refueling events in the Radian/API study, used above to determine the dispensed temperature value.[4,5] It is clearly desirable to use the same study for such closely related parameters, as the data were gathered at the same stations and times. In the API report, AT values were tabulated separately from dispensed temperature values. In general, direct correlation of fuel tank to dispensed temperature cannot be made, because dispensed temperature is strongly a function of the slowly fluctuating underground temperature, while fuel tank temperature varies with ambient temperature, trip length, number of recent trips, and so on. Thus, there are many fuel tank temperatures which could occur for each dispensed temperature. It is impossible to tell, from the available independent distributions, exactly how the AT distribution would be affected by choosing a given dispensed temperature. However, the AT distributions by region (Figure 7), derived from the API report, give the indication that this problem may not be significant. Of particular interest is the distribution for Region 2, which was identified earlier as having the highest dispensed fuel temperatures. In terms of AT, Region 2 falls in the middle ground of the regional distributions, indicating that it is reasonable to use the average AT distribution with the dispensed fuel temperature already developed. That is, the figure shows that the high average dispensed temperatures found in Region 2 do not unduly influence the corresponding AT distribution. If anything, the inclusion of Region 1 data (representing the lowest recorded dispensed temperatures) will tend to shift the AT distribution toward more positive (less ------- Figure 7 emulative Distribution of AT = T^-f^ Cumulative Percent 100 90 80 70 60 50 ¦ 40 30 20 10 20 25 30 35 A Region 1, Northeast 878 Vehicles Q Region 2, Southeast 296 Vehicles q Region 3, Southwest 324 Vehicles + Region 4, West 391 Vehicles -20 -15 -10 -5 50 55 60 Tank Tamp. - Dispensed Tenp. (°F) ------- -22- stringent) AT values than would otherwise be found using only high dispensed temperature data.* Therefore, the average AT distribution for the four regions combined was used to determine the 90th percentile value for the recommended practice. The basic distributions were given in the API report in tabular form by increments of 5F° for each region by month, for those cases when the ambient temperature was between 65 and 85°F**. From these, a distribution was developed for the ozone-prone months (May through September) for each ozone-prone region. The range of AT for these distributions is from -20 to +55 F°, with negative values giving vapor growth as the warmer dispensed fuel enters the cooler tank. Since vapor growth is the more stringent condition in that it creates the largest amount of emissions, the 90th percentile point was found by accumulating from the top down, that is, from positive AT. An examination of the tabulated data shows that the 90th percentile point falls in the 0 to +5 F° range for all four regions together, as can be seen in Figure 8. This conclusion has been confirmed, by subsequent analysis performed by EPA on the raw data used for the API report. These data indicate that, for high dispensed fuel temperatures between (85 and 95°F), the 90th percentile AT would be slightly more negative than that derived below, with a value near -3°F. However, EPA has decided not to adjust the test conditions in order to counter any tendency for the overall test parameter set to become unreasonably stringent. This might occur through the use of a 90 percent limit of one parameter which is itself derived from a 90 percent limit of another parameter, thus defining a condition which might only occur one percent of the time. While it would be preferable to have the AT distribution based upon all ambient temperature conditions, it was not possible to derive this information from the tabulated data. Subsequent analysis by EPA of the raw data files has verified that this limitation did not substantially affect the AT distributions. ------- r xyui.ti o Distribution of at = Tt - Tq May thru Sept 1975, regions 1,2,3 and 4 25, 100 .Cumulative Percent 80 Percent per 5° Increment 60 i i i 40 I i 20 20 Tank Temp. - Dispensed Temp. (°F) ------- -24- It was also determined that the AT distribution for all six regions - the 48 contiguous states - indicates a mild vapor growth condition, with a 90th percentile value in the -5 to 0 F° range. The desire for a worst case condition might therefore indicate the choice of the nationwide AT value rather than that found for the ozone-prone regions only. However, there is an important reason why this approach was not adopted. In 1975, when data for the Radian/API study was gathered, very few fuel injected (FI) vehicles were in the fleet. Such vehicles, which are becoming increasingly popular and will dominate the fleet in future years, presumably have higher fuel tank temperatures due to fuel return lines, which recirculate heated fuel to the fuel tank. Therefore, data without FI vehicles gives a somewhat smaller AT than might be expected for FI vehicles. To provide some compensation for this situation, EPA has decided not to use the 48-state data, and to use the upper end of the 0 to +5 F° range from the ozone-prone regions. Providing a range for testing variability, the recommended practice employs a AT value of +2 F° to +5 F°. 4. Fuel Level Prior to Refueling Fuel level is important as a vapor control canister design parameter; the canister must be sized to contain all of the vapors displaced during the refueling test in order to avoid breakthrough. Vapor quantities involved are 7 to 8 g of HC per gallon of dispensed fuel, at the temperature and RVP conditions specified in the test procedure. As an example of sensitivity, this translates into approximately an 11 to 13 gram difference in emissions per 10 percent increment in fuel level, for a medium size 16 gallon tank. Therefore, accurately specifying a fuel level prior to refill is important to assure in-use control. Data on fuel level prior to refill comes from studies conducted as part of California's Stage II vapor control program.[6,7,8,9,10,11] These studies observed refueling events at various service stations in California to determine control efficiency of Stage II systems, and, as part of the program, reported amount of gasoline dispensed for each refueling event that took place. One study, by Union 76, provided fill fractions in the report, while five other studies, by Union 76 (2 studies), Exxon, Healy, and Emco Wheaton did not. However, these studies did provide information on vehicle make, model, model year and fill amount. EPA estimated fill fraction for these last five reports by assuming 13, 18 and 23 gallon tanks for compact, mid and large size cars, respectively. These estimates were made by averaging the tank sizes, by vehicle size category, of 44 vehicles using the service stations in the studies. ------- -25- The vehicle sample for these six reports includes 696 vehicles and light-duty trucks receiving complete fill-ups. The distribution of fuel level prior to refill for these vehicles is shown in Figure 9. It can be seen that, in order to cover 90 percent of fill-up events, tanks should be filled from about 10 percent of tank capacity before refueling. The above value is taken from data for fill-ups only; partial fills are likely to have sufficient control since purge rates are higher when the canister is fully loaded, and hence have been deleted from this data base. The question arises, however, of how much total fuel is actually dispensed in fill-ups, and how many events are fill-ups, in order to ensure that basing control on fill-up conditions is not unnecessarily stringent. The reports mentioned above show that a substantial amount of the fuel dispensed is part of a fill-up event, such that full control by the canister is necessary: in the four reports which tabulated both full and partial fills, 54 percent of the refueling events and 62 percent of the gasoline dispensed was used in fill-ups. Also noteworthy is the fact that most of the events — 86 percent — at an entirely full-service station used in one Union 76 report are fill-ups. Therefore, it is important that control be adequate for a full refueling. The SAE recommended practice indicates that the fuel tank should be filled to one-fourth capacity before the refueling. It is EPA's desire to give a more precise specification, as well as one which duplicates 90 percent of actual refueling events. On the basis of the above data, this value has been chosen to be 10 percent of tank capacity, to the nearest one-tenth of a U.S. gallon. The refueling capacity test as defined in this test procedure will be terminated when automatic nozzle shutoff occurs. In order to ensure adequate canister capacity, this automatic shutoff must occur when the tank is at a level between 95 and 100 percent of its nominal capacity. Thus the total amount of fill for a valid test will be between 85 and 90 percent of the nominal capacity of the tank. 5. Fuel Dispensing Rate No value for a fuel dispensing rate is specified in the SAE recommended practice. However, this is an important parameter because it directly influences the rate of vapor displacement and also the occurrence of nuisance shut-offs, and ------- Figure 9 Distribution of Fuel Level Prior to Refill 696 Vehicles $ , 20 15 - % of Events 10 _ 13.5% 5 _ 18.7% 19.4% 18.2% 0.1 0.2 0.3 15.7% 10.1% 3.2% JLiU. 0.4 0.5 0.6 0.7 0.8 Fuel Level Prior to Refill (fraction of full tank) ------- -27- perhaps to some degree the amount of fuel spillage. Excessive back pressure and premature nozzle shut-off occurs in a system which is too small to handle the volume of fuel being dispensed. Thus, the dispensing rate affects the design of the refueling control system, and it is necessary that this parameter have a realistic value. A value of 8 to 10 gallons per minute (gpm) has been chosen to cover the majority of fill events. A 1984 informal survey by API of member companies and nozzle manufacturers indicates flow rate extremes of 6.5 to 12 gpm. [12] Lower rates of 6.5 to 8 gpm are found at full-service pumps, where the attendant may purposely choose the low notch setting on the fuel nozzle to allow enough time to service the vehicle. The low rate also reduces the chance of nuisance shut-off, which may cause the nozzle to fall off the vehicle and cause large fuel spills. Higher flow rates of 9 to 11 gpm are typically found at self-serve pumps. Facility type also affects flow rates. Newer, larger facilities generally have submersible pumps which serve all of the dispensers at each underground tank, and their flow rates tend to be high, with as much as 10. to 12 gpm capability. Older, smaller stations usually have individual suction pumps on each dispenser, with normal rates of 8 to 9 gpm. All flow rates are affected by such factors as the number of vehicles being refueled, distance from the pump, age of the pump, and vehicle design constraints which cause nuisance shut-off. Such shut-offs begin to occur generally at flow rates above 8 to 9 gpm. The nozzle does not appear to be a significant factor in altering flow rate, although in conjunction with flow rate it can affect fuel spitting. From the available surveys, it was not possible to quantify a distribution of fuel dispensing rates. However, the information generally available indicates that most refuelings take place at 10 gpm or less. Higher rates are possible, especially at newer, self-serve facilities, but often result in nuisance shut-off, which should be minimized below 10 gpm for a well designed onboard system. Therefore, the specified value of 8 to 10 gpm should cover the majority of the refueling events while minimizing nuisance shut-off. It is worth noting that California specifies a 10 gpm maximum flow rate with its Stage II vapor control systems, making the two sets of specifications compatible, which is important for fuel system design purposes. ------- -28- D. Vehicle Preconditioning A variable with potential for affecting the outcome of the test is vehicle preconditioning. For this test sequence, there are several aspects of preconditioning, including initial canister purging, canister saturation, and fuel tank heating. Initial preconditioning has been described earlier, and involves ensuring a well but not abnormally purged canister and a nearly empty fuel tank, so that all vehicles will have both canister and fuel tank capacity for the 85% refueling test. Certification vehicles are expected to arrive at the test facility with canisters purged to a level commensurate with a nearly empty fuel tank. Preparation for testing will involve draining the fuel tank and fueling to one-tenth of fuel tank capacity (with the. refueling canister disconnected). In-use vehicles, on the other hand, will arrive at the facility in various conditions; therefore, such vehicles will be driven for 50 miles on the EPA durability driving schedule or equivalent so that the canister can be adequately purged. Prior to this mileage accumulation, the fuel tank will be drained and fueled with test fuel. Following mileage accumulation, the fuel tank will again be drained and then fueled to one-tenth of tank capacity. The refueling canister will be disconnected for this fueling just as for certification vehicles. Both certification and in-use vehicles will also have their gas caps loosened while awaiting testing in order to avoid any canister loading during that time. The specification of a 50 mile driving schedule for in-use vehicles is still tentative at this time. EPA believes that this amount of continuous driving should be sufficient to purge even a fully loaded canister for a refueling control system capable of passing the rest of the test procedure. However, data is still being gathered to verify this conclusion and it remains subject to change. The Agency desires to allow sufficient preconditioning to adequately purge the system, but also desires to minimize the overall testing time and expense. Following the initial capacity test, it is important that vehicles be standardized in a condition which will permit the subsequent testing to verify proper system purging. For this purpose, the recommended practice requires that the canister be fully loaded following the capacity test. This is accomplished by putting the vehicle in the SHED and repeatedly filling and draining the tank; in most cases, one or two fills should be enough. Saturation is detected by monitoring the SHED ------- -29- hydrocarbon concentration with a FID; a sudden rise indicates break-through of vapors from the canister. Based upon limited EPA testing to date, the recommended practice defines this condition^ as producing a change in the FID reading by a factor of at least 3 in a one minute period. A third aspect of preconditioning for refueling testing is the heating of the fuel in the fuel tank; this brings the liquid and vapor of the fuel into equilibrium as well as simulates the warmer fuel of a vehicle which has been driven. Four tank heating methods were tested by EPA as part of the baseline refueling emissions study. The four procedures tested included three types of heating with an electric blanket, and driving over an actual road route. "Single blanket" heating is the simplest of the preconditioning methods. It applies an electric heating pad to the bottom of the fuel tank, which imparts heat directly to the liquid fuel. This procedure was initially used with the constraint from the original SAE recommended practice that the temperature difference between vapor and liquid in the tank must remain within 6F° in order to maintain equilibrium throughout the heat build. It was, therefore, a relatively long procedure, taking one and a half to three hours to reach the desired temperature, because the fuel heating must be gradual in order stay within this constraint. "Dual blanket" heating was therefore evaluated, using a second heat blanket applied to the top of the tank. This method heats the vapor as well as the liquid and allows more rapid heating within the 6F° temperature difference constraint, typically requiring under one hour for heat build. It has the drawback of requiring the removal of the fuel tank to gain access to its top surface. A final type of blanket preconditioning was tested by EPA in an attempt to eliminate the problems encountered with the first two methods: excessively slow heat builds for the single blanket, and the necessity of dropping the tank for the dual blanket. This preconditioning uses a single blanket but removes the constraint of no more than a 6F° temperature difference between the vapor and liquid during the heat build. Instead, the fuel is heated to the desired temperature as quickly as possible, which allows the vapor to be heated much more rapidly than with the original single blanket approach. Once the vapor temperature comes within 3F° of the liquid temperature and the temperature conditions prescribed in the test procedure have been reached, a ten minute waiting period is begun so that the vapor and liquid can attain equilibrium. ------- -30- This fast heat build single blanket method of preconditioning gives emission results similar to the slow heat build and dual blanket methods as shown in Figure 10, and has therefore been adopted for the recommended practice. The fourth method of preconditioning allows heating of the fuel by having the vehicle driven over an actual road route. The route involves three hours of rural and urban driving in the area near Ann Arbor, Michigan, and is run in order to approximate a real life situation. While not practical in itself due to its length and potential for variability, it confirms the appropriateness of blanket preconditioning by giving similar emissions results, as described in the baseline report.[16] E. Canister Purging and Vehicle Drivedown As mentioned previously, there are two key aspects of the refueling vapor control system which must both be tested in order to assure that emissions are being properly controlled. These are the capacity of the canister which captures the vapors and the capability of the system to purge these vapors during vehicle operation. These aspects must be tested without affecting the results of other portions of the test sequence. It is also desirable to keep costs low and testing time as short as possible, without placing an excessive burden on manufacturers. All this has to be accomplished while maintaining a representative and standardized test. To ensure adequate purge design, EPA has developed a vehicle drivedown sequence consisting of consecutive LA-4's and one hour hot soaks which are repeated until approximately 30 percent of the fuel tank capacity has been consumed. This drivedown is then followed by a 30 percent refueling test to verify that the system has purged sufficient vapors to handle a proportionate refueling. Subsequent analysis of this drivedown requirement indicates that it may not be needed in all cases. Depending on the design of the refueling control system, the test sequence from initial canister capacity testing and subsequent saturation through the remainder of the existing FTP may be sufficient to ensure adequate refueling system performance. For those refueling control systems which are fully or partially integrated with the vehicle's evaporative emission controls, a substantial amount of system purge is required to pass the diurnal evaporative test following the initial ------- Figure 10 < a c o 8.0 COMPARISON OF REFUELING LOSSES ESCORT - DISP.TEMR.= 80 RVP=11.9 1 2.0 1 0.0 1 4.0 -2.0 0.0 2.0 ~ STANDARD HEAT BUILD 4.0 6.0 DELTA T (DEGREES F) + FAST HEAT BUILD ------- -32- canister saturation step, and it may be that the stringency of this requirement is sufficient to ensure that the vehicle will also purge collected refueling vapors over the course of the drivedown sequence. As an initial test of this hypothesis, EPA evaluated the amount of vapors required to be purged for compliance with the diurnal evaporative emissions test in comparison with the amount of purge required to satisfy the drivedown requirement. This was initially done on the assumption that the refueling canister had a linear purge response as a function of canister loading. With this assumption the calculated purge rate for passing the diurnal test was approximately ten times the purge rate needed to pass the drivedown, and the drivedown appeared to be unnecessary. EPA then repeated the analysis, incorporating an estimate of the actual non-linear purge characteristics expected from the refueling canister. On this basis, the diurnal test requirement was only marginally adequate for ensuring sufficient refueling purge. The analysis also indicated that, since the canister purge rate continues to decline as the canister is purged, a 30 percent drivedown may not be adequate to account for system performance as an empty tank is approached. At this time it is unclear whether the drivedown is needed for integrated systems, or if it is, whether a 30 percent drivedown amount is sufficient. EPA is currently in the process of gathering more data on the purge characteristics of refueling and evaporative canisters and on ways to avoid the need for an extended vehicle drivedown. One option being considered is to replace the drivedown with the requirement that the refueling canister remain connected during the 40 percent fill preceding the diurnal evaporative emissions test. The added purge required by this approach may turn out to correspond well with the performance characteristics of a refueling system capable of fully purging collected refueling vapors, and thus eliminate the need to further demonstrate purge capability via a drivedown. For refueling control systems which employ separate canister systems for refueling emissions and evaporative emissions, the drivedown sequence appears to be unavoidable. The completely separate refueling system incurs no loading from evaporative emissions, so that the adequacy of the purge cannot be tested during the preconditioning cycle and evaporative ------- -33- loading sequence in the FTP. In these instances, then, the purge test will have to be conducted, involving the 30% drivedown/refueling sequence appended to the usual sequence. At this time, the draft recommended practice includes a drivedown requirement for only those vehicles employing a non-integrated refueling control system. It remains possible, however, that the drivedown requirement, or some other modification to the test sequence, may be needed for integrated systems. F. Testing Requirements for In-Use Vehicles It is EPA's intent that motor vehicles meet emissions standards throughout their useful life, in order to best protect the public and environment from the harmful effects of mobile source pollutants. Certification of new motor vehicles is, of course, one major aspect of assuring that only vehicles likely to have low emission levels in-use are produced. In order to ensure that production vehicles continue to meet the standards throughout their useful lives, EPA routinely performs tests on in-use vehicles, following the certification testing procedure as closely as possible. However, some adjustments to the procedure are needed to accommodate in-use situations which are distinct from certification testing. EPA's proposals for handling the unique aspects of in-use emissions testing are discussed in this section. Included are the procedures to standardize the vehicles via preconditioning, as well as potential testing problems not encountered with specially-equipped certification vehicles. These issues have largely been identified by manufacturers, and EPA's approaches to resolving them are outlined here. The Agency remains open to further comments on preconditioning and testing of refueling emissions from in-use vehicles. 1. Preconditioning Problems Residence Time Before Testing It is important, for fully or partially integrated systems, that vehicles not stand with a closed system for extended periods of time prior to the refueling test, as this may cause loading of the canister which is unrepresentative of a vehicle which is driven regularly. However, while vehicles are tested as soon as possible, it is not possible for EPA to assure that every vehicle brought to the test site will be tested soon after arrival, due to fluctuations in vehicle ------- -34- procurement and test site/technician availability. Therefore, in order to keep the level of canister loading down, the refueling system will be opened by loosening or removing the gas cap to allow the escape of vapors. It should be noted that certification vehicles can also have their gas caps loosened upon arrival at the test site. Residual Fuel The variety of the RVP of fuels in the tanks of in-use vehicles could cause excessive variation in test results if such fuels were allowed to remain in the tank. Therefore, vehicles will have their tanks drained of residual fuel and will be preconditioned and tested only with the specified test fuel. For present testing, such fuel will have an RVP of 11.5+0.5 psi; the fuel to be used in actual certification and in-use enforcement testing will be specified after further investigation, and will be similar to average in-use fuels. Manufacturers have also expressed concern that consumer use of high RVP fuels will cause long-term effects on the refueling and/or evaporative system which will significantly affect test results. These concerns are, in the Agency's opinion, unwarranted. First of all, as already noted, EPA is changing its test fuel to more closely match in-use fuels. Therefore, normally available fuels should not differ widely from certification fuels. Secondly, vehicles should be designed so that systems are not permanently disabled by the use of any normally available fuel, and in fact should remain largely operational. Any loss in function should be restored by adequately preconditioning with the test fuel. This is not to say that concern about such things as alcohol additives, which may damage emission controls if used improperly, is unwarranted. Such an issue is simply beyond the scope of this proposal, and must be dealt with separately. Finally, manufacturers have expressed a desire that the test fuel used for residual tank fuel be weathered, i.e., have a lower RVP than the dispensed fuel. EPA has considered the use of weathered fuel, as described earlier, and decided against it. Weathered tank fuel increases emissions slightly, but not enough to warrant the added complexity of handling two test fuels. Such a decision should not be objectionable to manufacturers since it reduces the stringency of the test somewhat. ------- -35- In-Use Driving Patterns The driving patterns encountered in-use can affect refueling emissions control by altering the purge capability of the refueling system. Permanent damage to the system caused by a low trip frequency and corresponding low level of purge is a manufacturer concern, but/ in Agency opinion, should not occur. Like the effects of various fuels, a range of normal driving patterns should not cause any irreversible disablement of systems; that is, vehicles should be designed to withstand the vast majority of in-use conditions. A proper preconditioning cycle should be able to standardize vehicles which will have arrived at the test site under a variety of driving conditions. The procedure currently to be used involves driving the vehicle for 50 miles on the durability driving schedule or equivalent, on a tank of standard test fuel. This is intended to allow the system to purge the canister, thereby restoring the system, and should adequately standardize the purge level of in-use vehicles for subsequent testing. As noted in earlier discussion of vehicle preconditioning, EPA is open to comment should manufacturers have information showing that this preconditioning cycle will not meet their needs, or have suggestions for alternative preconditioning for canister purge. Fuel Tank Heating An additional aspect of preconditioning involves fuel tank heating, which further standardizes the vehicles by attaining a uniform tank temperature and liquid/vapor equilibrium. Tank heating is often done using the dual blanket method discussed in Section III.D., which was originally presented by EPA as the chosen method for refueling testing. However, this method presents practical problems for in-use vehicles, since it involves dropping the fuel tank from its normal position. On the other hand, the traditional single blanket method, with a constraint of a maximum 6F° difference between liquid and vapor fuel temperatures, requires several hours to complete, and so is also undesirable. In response to these problems, a new method was devised which is much quicker than the original single blanket method and requires no major alterations of the. vehicle. This method allows rapid heating of the liquid with the use of a single blanket applied beneath the tank. Upon arrival of the liquid fuel at the specified temperature, the vapor and liquid are ------- -36- allowed to come into equilibrium during a 10 minute waiting period, once they are within a 3F° difference. Rapid single blanket heating gives results comparable to other methods, and is the method that will be used on in-use vehicles to save time and keep consumer vehicle fuel systems intact. 2. Testing Problems Canister Disconnect The recommended practice contains several steps which require the refueling canister to be disconnected in order to avoid excessive loading with vapors. Manufacturers have expressed the concern that the canister may be inaccessible for disconnect, and that the integrity of the vapor line seal upon reconnect may be suspect, causing leaks or spillage. It is EPA's expectation that the easiest way to allow for the refueling canister to be disconnected for testing will be through the insertion of a valving mechanism in the vapor line. This would be done during initial vehicle preparation, allowing adequate opportunity for the integrity of the vapor line seals to be verified. Thereafter, the required canister disconnects can be produced through simple operation of the valve. As for accessibility, EPA does not believe that the requirement for vapor line accessibility poses any serious design constraint to manufacturers. Test Voids Tests which do not meet specifications will need to be aborted and rerun, and it is apparent that these vehicles will need to be restandardized and reenter the test sequence at an appropriate point. For the refueling test there are several convenient places where partial testing can be restarted to avoid having to rerun the entire sequence in the case of a test void. For those vehicles which have not completed the initial refueling capacity test, the prime concern will be to return the canister to a purged condition similar to that with which it entered the test. This can be accomplished by recycling the vehicle through the preconditioning sequence for in-use vehicles (50 mile drive sequence) and beginning the testing anew. ------- -37- For vehicles which have completed the refueling capacity- test and subsequent canister saturation, but which have not yet completed the diurnal test, it will be sufficient, in the event of a test void, to return to the canister saturation step and proceed from there. Subsequent to the diurnal test and prior to beginning the drivedown test, the current FTP remains essentially unchanged and test voids may be handled according to current practice. The last portion of the refueling test is the drivedown and partial refueling test. A test void during the refueling operation here can be accommodated by completing the prescribed amount of fill, repeating the drivedown sequence (appropriately calculating the required amount of driving and refueling) and repeating the refueling test. As the above discussions indicate, EPA believes that the recommended practice as currently designed adequately deals with issues of concern relative to testing in-use vehicles. Of course, EPA is always open to further comments concerning the appropriateness of these solutions, and to alternate solutions should these be deemed unsatisfactory. ------- References 1. "Evaluation of Air Pollution Regulatory Strategies for Gasoline Marketing Industry," (EPA-450/3-84-012a), U.S. EPA, OAR, OAQPS and OMS, July 1984. 2. "Instrumentation and Techniques for Vehicle Refueling Emissions Measurement," SAE Recommended Practice J1045, August 1973. 3. "Investigation of Passenger Car Refueling Losses," (SRL 2874-12-0972), Scott Research Laboratories. 4. "Summary and Analysis of Data from Gasoline Temperature Survey Conducted by American Petroleum Institute," Radian Corporation, May 1976. 5. "Analysis of Temperature Effects on Gasoline Marketing Operations," API Publication No. 1625, American Petroleum Institute, January 1979. 6. "Testing of Union Oil's Vapor Balance Service Station Vapor Control System by the California ARB Test Procedure," Memorandum from Donald C. Gearhart, Assistant Counsel, Union Oil Company of California, to Members of API Subcommittee on Environmental Lav, September 10, 1976. 7. "Testing of a Vapor Balance Service Station Vapor Control System by the California ARB Test Procedure - Brea, June 1976," Technical Memorandum from M.J. Dougherty, Refining and Products Research, Union Oil Company of California, to Mr. Cloyd P. Reeg, August 19, 1976. 8. "Efficiency Evaluation of Union 76 Balance-Type Stage II Vapor Control System with OPW 7VC Dispensing Nozzles - Attendant Serve Operating Mode," Scott Environmental Technology, August 30, 1979. 9. "Healy Phase II Vapor Recovery System Certification Report," Scott Environmental Technology, June 1982. 10. "A Service Station Test of a Vapor Balance System for the Control of Vehicle Refueling Emissions," A.M. Hochhauser and L.S. Bernstein, Exxon Research and Engineering Company, July 1, 1976. 11. "Efficiency Evaluation of Emco Wheaton A3006 Dispensing Nozzle," Scott Environmental Technology, July 8, 1981. ------- References Cont'd. 12. Letter from Edward P. Crockett, American Petroleum Institute, to Charles L. Gray, U.S. EPA, regarding fuel flow rates, August 8, 1984. 13. Annual Book of ASTM Standards, Part 23, Standard D439-83, American Society for Testing and Materials, 1983. 14. "A Study of Variables that Affect the Amount of Vapor Emitted During the Refueling of Automobiles," Edward M. Liston, Standford Research Institute, May 16, 1975. 15. "1981-83 Standard Metropolitan Statistical Area (SMSA) Air Quality Data Base for Use in Regulatory Analyses," Technical Memorandum from Richard 6. Rhoads, U.S. EPA, MD-14, to Charles L. Gray, U.S. EPA, OMS, February 25, 1985. 16. "Refueling Emissions from Uncontrolled Vehicles," Dale Rothman and Robert Johnson, EPA-AA-SDSB-85- , July 1985. 17. "Fuel Volatility Trends," Final Report, Letter from Bruce B. Bykowski, Southwest Research Institute, to Craig Harvey and Amy Brochu, U.S. EPA, September 28, 1984. 18. The Weather Almanac, Third Edition, James A. Ruffner and Frank E. Bair, Ed., Gale Research Company: Detroit, 1981. ------- |