Light-Duty Alternative Fuel Vehicles: Federal Test Procedure Emissions Results September 1999 • NREL/TP-540-25818 K. Kelly, L. Eudy, and T. Coburn National Renewable Energy Laboratory 1617 Cole Boulevard Golden, Colorado 80401-3393 NREL is a U.S. Department of Energy Laboratory Operated by Midwest Research Institute • Battelle • Bechtel Contract No. DE-AC36-98-GO10337
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Light-Duty Alternative FuelVehicles: Federal TestProcedure Emissions Results
September 1999 • NREL/TP-540-25818
K. Kelly, L. Eudy, and T. Coburn
National Renewable Energy Laboratory1617 Cole BoulevardGolden, Colorado 80401-3393NREL is a U.S. Department of Energy LaboratoryOperated by Midwest Research Institute •••• Battelle •••• Bechtel
Contract No. DE-AC36-98-GO10337
National Renewable Energy Laboratory1617 Cole BoulevardGolden, Colorado 80401-3393NREL is a U.S. Department of Energy LaboratoryOperated by Midwest Research Institute •••• Battelle •••• Bechtel
Contract No. DE-AC36-98-GO10337
September 1999 • NREL/TP-540-25818
Light-Duty Alternative FuelVehicles: Federal TestProcedure Emissions Results
K. Kelly, L. Eudy, and T. CoburnPrepared under Task No. FU905010
ACKNOWLEDGMENTS
This work was sponsored by the Office of Technology Utilization, which is part of the U.S. Department of Energy’s(DOE) Office of Transportation Technologies in Washington, D.C. Mr. Dana O’Hara is DOE’s program manager for thelight-duty vehicle evaluation projects at the National Renewable Energy Laboratory. Appreciation is expressed to thethree emissions laboratories that performed the testing: Environmental Research and Development, in Gaithersburg,Maryland; Automotive Testing Laboratory, in East Liberty, Ohio; and ManTech Environmental, in Denver, Colorado.We also thank Phillips Chemical Company and Compressed Gas Technologies for supplying the test fuels for this project.
TP-25818
NOTICE
This report was prepared as an account of work sponsored by an agency of the United States government. Neither theUnited States government nor any agency thereof, nor any of their employees, makes any warranty, express or implied,or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, appara-tus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference hereinto any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does notnecessarily constitute or imply its endorsement, recommendation, or favoring by the United States government or anyagency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of theUnited States government or any agency thereof.
Available to DOE and DOE contractors from:Office of Scientific and Technical Information (OSTI)P.O. Box 62Oak Ridge, TN 37831
Prices available by calling (423) 576-8401
Available to the public from:National Technical Information Service (NTIS)U.S. Department of Commerce5285 Port Royal RoadSpringfield, VA 22161(703) 605-6000 or (800)-553-6847orDOE Information Bridgehttp://www.doe.gov/bridge/home.html
Printed on paper containing at least 50% wastepaper, including 20% postconsumer waste
In 1988, the federal governmentenacted the Alternative Motor FuelsAct (AMFA) to encourage the devel-opment and use of methanol, ethanol,and natural gas as transportation fuelsfor consumers. This was followed by the Clean Air Act Amendments(CAAA) in 1990 and the EnergyPolicy Act (EPAct) in 1992. As partof AMFA and EPAct, the Departmentof Energy (DOE) is required to pro-mote the use of alternative fuels toaddress environmental concerns andenergy security. As a result of thesefederal actions and the lack of con-clusive information on in-use emis-sions from alternative fuel vehicles(AFVs), DOE, through the NationalRenewable Energy Laboratory(NREL), conducted an extensiveseries of emissions tests on AFVsbeing used in the federal governmentfleet.
The goal of the NREL emissions test-ing program was to provide a highquality, objective assessment of thein-use emissions from commerciallyavailable AFVs. This report summa-rizes the results from 1,280 emissionstests performed on 413 vehiclesbetween 1994 and 1997, includingtests on methanol and ethanol flexible-fuel vehicles (FFVs), dedicated com-pressed natural gas (CNG) vehicles,and matching standard gasoline vehi-cles. Many vehicles were tested sev-eral times at approximately 1-yearintervals. The data sets for each yearare referred to as test "rounds." Alltests followed the U.S. EnvironmentalProtection Agency’s (EPA) existingFederal Test Procedures (FTP-75) foremissions certification. Measurement
of regulated emissions included non-methane hydrocarbons (NMHC),carbon monoxide (CO), oxides ofnitrogen (NOx), and evaporativehydrocarbons. Measurements of non-regulated emissions includedformaldehyde (HCHO), acetaldehyde(CH3CHO), carbon dioxide (CO2),and methane (CH4). The vehiclestested were original equipment manufacturer (OEM) models takenfrom the pool of vehicles used in theGeneral Service Administration’s(GSA) federal fleet. The testing wasperformed at private emissions labo-ratories in Ohio, Colorado, andMaryland. Each laboratory used theEPA’s FTP-75 for exhaust emissionsand evaporative emissions with testfuels that were blended specificallyfor this program. The gasoline fuelthat was used for comparison wasCalifornia Phase II reformulatedgasoline (RFG). This fuel was chosenin order to make a comparisonbetween alternative fuel emissionsand a "best case" scenario for gaso-line. One might expect that the com-parison of emissions betweenalternative fuels and an industry aver-age gasoline would be slightly morefavorable for alternative fuels than thecomparison in this report becauseRFG is a cleaner burning fuel thanthe industry average gasoline. Severalvehicles were randomly selected formore extensive tests that includeddetailed analysis of the hydrocarbonemissions. The test results were usedto assess differences in the composi-tion of hydrocarbon emissions interms of their relative toxicity andreactivity or propensity to form ozone in the atmosphere.
In general, this study found that fuelis an important factor in vehicle emis-sions. However, the study also showsthat vehicle-to-vehicle variability is significant, and that engine andemissions controls system design andcalibration are also critical factors. In other words, the fuel is important,but individual vehicle differences(resulting from, for example, manu-facturing tolerances, vehicle servicehistory, or duty cycle) and vehiclemodel design differences also play amajor role in the measured emissionsreductions.
A comparison of the regulated emis-sions from the FFVs tested on alcoholfuels and RFG tended to fall into oneof two categories:
(1) Compared to RFG, the alcoholemissions showed a decrease forone or two of the regulated emis-sions constituents coupled with anincrease in the other constituents,or
(2) There was no significant differ-ence in the emissions from thetwo test fuels.
In both cases, the average resultstended to be well within the applica-ble emissions standards. The lack of a clear benefit in regulated emissionsfor the alcohol tests may be a result of FFV design. FFVs are designed to meet customer performance andemissions certification requirementson any blend of alcohol and gasolinefrom 85% alcohol with 15% gasolineup to 100% gasoline. This designstrategy allowed FFVs to be placed in the market with only a limited
vi
TP-25818
EXECUTIVE SUMMARY
alcohol refueling infrastructure, but it required compromises to be madein engine design and calibration. Forexample, FFVs cannot take advan-tage of the higher octane rating ofalcohol fuel because they must bedesigned to accommodate the lowestoctane rating of all possible fuelblends (i.e., 100% gasoline). Otherstudies have shown that more sub-stantial emissions benefits can beachieved from a vehicle that has beenoptimized to run on a single blend ofalcohol fuel1.
Two areas where the alcohol fuelemissions did show clear advantagesover RFG were in reducing the toxic-ity and the ozone-forming potential(OFP) of the hydrocarbon emissions.It could be expected that these bene-fits would be even more pronouncedif a comparison were made to indus-try average gasoline, because RFGhas been shown to reduce emissionsof toxic constituents and be less reac-tive in forming ozone2. Tests on thealcohol fuels also showed a small butconsistent reduction in CO2 emis-sions compared to RFG tests.
Comparison of the average resultsfrom the CNG vehicles tended to bemore straightforward. The dedicatedCNG vehicles tested in this programexhibited significantly lower regulat-ed emissions compared to similargasoline vehicles tested on RFG. Thetoxicity and reactivity of the hydro-carbon emissions from CNG vehicleswere also significantly lower.
The rapid development of emissionscontrol technology continues, pushedby tougher regulations designed tohelp meet the National Ambient AirQuality Standards of the CAAA. Theresults presented here are representa-tive of the alternative fuel technolo-gies that were available during thestudy (1992 to 1995). More recentdevelopments include both alternativefuel and gasoline vehicle designs that
have been shown to meet more stringent emissions standards such as the state of California’s ultra low-emission vehicle (ULEV) require-ments. Dedicated CNG vehicles haverecently been produced that advertisesuper-ULEV (or 1/10 below ULEV)capabilities. At the same time, auto manufacturers are producing bi-fuelCNG/gasoline vehicles that may runinto similar design constraints as theFFVs (i.e., compromises are requiredto allow an engine to run on differentfuels). Emissions certification testshave also evolved to address issuessuch as cold temperature emissions,emissions resulting from real-worldor more aggressive driving behaviors,extended and running loss evapora-tive emissions, and emissions duringoperation of the vehicle’s air condi-tioner. These changes may affect thecomparison of emissions from alter-native fuel to gasoline vehicles. Theability for AFVs to maintain emis-sions benefits at high mileage is alsoa question. Most of the AFVs in thefederal fleet do not accumulate highmileage levels. Some of these issuesare being addressed in other parts ofthis DOE/NREL program, and willbe covered separately.
SUMMARY OF RESULTS
Methanol
One-hundred and one M85 FFVs,including 1995 Dodge Intrepids and1993 Dodge Spirits, were testedalong with similar numbers of stan-dard gasoline control vehicles. Mostof the results from these vehicleswere very consistent across vehiclemodels, test laboratories and testrounds. Non-methane hydrocarbonequivalent (NMHCE), CH4, and CO2were significantly lower for the M85tests than for the tests on RFG.Results for NOx, CO, and evaporativeemissions were not as consistent.Although CO emissions were slightly
higher for one vehicle model andwere lower for the other model tested,these results tended to be not statisti-cally significant. NOx results tendedto be higher for the FFVs tested onM85 than when those same vehicleswere tested on RFG. The evaporativeemissions results for one vehiclemodel were consistently higher forthe M85 tests; results for the othermodel were varied. Fuel economy forthe M85 tests was significantly lowerthan the gasoline tests because of thelower energy content of the fuel, butwas slightly higher when comparedon an energy equivalent basis. Resultsfor the more detailed tests show thatboth vehicle models tested on M85emit significantly less potency-weighted toxics (PWT), and the OFPand specific reactivity is lower.
There are several possible reasons forfinding mixed results and fuel effectsthat are not statistically significant for FFVs. One is that these vehiclesare not optimized for either alcoholfuel or gasoline, but are designed toperform acceptably on a wide rangeof fuel blends. Another reason forvarying results is calibration andhardware differences between vehiclemodels.
Ethanol
Forty-nine E85 FFVs, including the1995 Ford Taurus and the 1993Chevrolet Lumina, were tested alongwith similar numbers of standardgasoline control vehicles. The regu-lated emissions results for the twoethanol FFV models were not as con-sistent as the methanol results. Ingeneral, the regulated emissions fromthe FFV Taurus tested on E85 werenot significantly different from emis-sions from the same vehicles testedon RFG. For the FFV Lumina, theNOx emissions were significantlylower on E85, the CO emissions were significantly higher, and thehydrocarbon emissions were mixed
vii
TP-25818
from round to round (total hydrocar-bon and NMHCE). Non-regulatedemissions for both vehicle modelstested tended to be consistent, and thedifferences tended to be statisticallysignificant. Average CO2 was consis-tently lower when tested on E85 com-pared to RFG. Average aldehydeswere consistently higher from theE85 test compared to the RFG tests.
When comparing the FFVs tested onE85 to the same vehicles tested onRFG, results of the detailed hydro-carbon analysis showed that averagealdehyde emissions and OFP tendedto be higher, while average 1,3-buta-diene, benzene, total PWT, and specific reactivity tended to be signif-icantly lower.
As with the methanol vehicles, theethanol vehicles are flexible-fueldesigns that are not optimized foreither gasoline or ethanol. The differ-ences in results between vehicle models and the lack of clear differ-ences in regulated emission resultsmay be due, in part, to engine hard-ware choices and calibrations thatmust be flexible to accommodate awide range of fuel blends.
Compressed Natural Gas
In all, 67 dedicated CNG vehicles(1992/94 Dodge B250 vans and 1994Dodge Caravans) were tested alongwith 69 similar gasoline control vehi-cles. Results for the CNG vehiclesshow that there tend to be statisticallysignificant differences between theaverage emissions from the CNG andRFG tests, and that these results tendto be fairly consistent for both vehiclemodels from lab to lab and fromround to round. The average NMHC,CO, CO2, and acetaldehyde resultswere significantly lower from theCNG tests than from the RFG tests.Average CH4 emissions were consis-tently higher from CNG than from
RFG. NOx and "evaporative" hydro-carbons tended to be lower from theCNG tests, but in some cases the dif-ferences were not significant. A mod-ified "evaporative" emissions test wasperformed to measure the hydrocar-bons emanating from the vehiclesduring two 1-hour soaks in a sealedenclosure with the engine off.Dedicated gaseous fuel vehicles typi-cally do not have evaporative controlsystems because the fuel system issaid to be "sealed" under pressure.Nevertheless, hydrocarbons (mostlymethane) were found emanating fromgaseous fuel vehicles. In all cases, theaverage total hydrocarbons measuredduring the CNG evaporative testswere lower than those from the RFGtests, but in a few cases the differencewas not statistically significant. Thefuel economy results for the CNGvehicles were lower than those of thegasoline vehicles. This was consistentfor both models.
Results from the detailed analysis of hydrocarbon emissions were veryconsistent for the two labs where thisanalysis was performed. At both labs, the CNG emissions had loweraverage values of the four toxic emissions that were quantified, hadlower PWT, lower average OFP andlower average specific reactivity.These differences were all deemedstatistically significant at the 95%confidence level.
viii
TP-25818
For the past few years, the NationalRenewable Energy Laboratory(NREL) has managed a series oflight-duty vehicle chassis dynamo-meter emissions tests on alternativefuel vehicles (AFVs) for the U.S.Department of Energy (DOE). Thesetests are part of a larger program todemonstrate the use of AFVs thatwas mandated by the AlternativeMotor Fuels Act of 1988 (AMFA)and the Energy Policy Act of 1992(EPAct). One of the major objectivesof these legislative actions is to pro-mote the use of alternative trans-portation fuels in order to addressenergy security and environmentalissues. As part of the AMFA pro-gram, vehicle performance, opera-tional costs, maintenance, and fueleconomy data are also being collect-ed by NREL’s Alternative FuelsUtilization Program and disseminat-ed through the Alternative Fuels Data Center (AFDC). This report isdesigned to present a detailed evalua-tion of the emissions test results collected in this program.
The principal phase of the AMFA testprogram was initiated in 1994. Itspurpose was to determine relativeemissions from AFVs compared tootherwise identical gasoline vehiclestaken from actual service. Approxi-mately 25 each of several AFV mod-els from several locations (includinghigh altitude) around the countrywere randomly selected for participa-tion in this program. All vehicleswere selected from those available inthe U.S. federal fleet. Test vehicleswere scheduled for emissions testingonce per year. The test matrix ofvehicles, locations, and mileage
levels was statistically designed tooptimize reliability of the data and tocontrol variability in the emissionsresults.
In addition to testing all vehicles forregulated exhaust and evaporativeemissions, we conducted a detailedspeciation of the hydrocarbon (HC)emissions on a subset of the test vehicles. Speciation of the HC emissions allows for an evaluation of the relative level of air toxic emis-sions and the reactivity or ozoneforming potential (OFP) of the HC.Additionally, we also tested a smallnumber of vehicles using new or pro-posed chassis dynamometer drivingcycles. These "off-cycle" emissionstests are still in progress and theresults will be discussed in a laterreport.
A BACKGROUND ON VEHICLEEMISSIONS AND FUELECONOMY
As a result of fuel combustion, auto-mobiles emit various compounds intothe atmosphere in the form ofexhaust. The U.S. EnvironmentalProtection Agency (EPA) regulatessome of these compounds; theamounts of the compounds that areemitted by vehicles cannot exceedcertain levels. Other compounds,although not officially regulated, areimportant contributors to adverseatmospheric conditions such asambient ozone and global climatechange.
The emissions compounds regulatedby the EPA include carbon monoxide(CO), oxides of nitrogen (NOx), HC,
and non-methane hydrocarbons(NMHC). Methane (CH4) is not cur-rently regulated because it is consid-ered to be relatively non-reactive informing ozone in the atmosphere.Exhaust from alcohol fuel vehiclesalso includes unburned alcohol andaldehydes, which are partial combus-tion products. For alcohol fuels, suchas the ones investigated in this study,these compounds are regulated alongwith non-methane hydrocarbons asnon-methane hydrocarbon equivalent(NMHCE). NMHCE is calculated by modifying the measured NMHCfraction to account for the alcoholand aldehyde emissions that areprevalent in emissions from alcoholfuels. More recent standards use non-methane organic gases (NMOG) asthe regulated compound. NMOG isthe sum of non-oxygenated and oxy-genated HC in a gas sample. Thisincludes all oxygenated organicgases with 5 or less carbon atoms(such as aldehydes, ketones, andalcohols) and all known alkanes,alkynes, alkenes, and aromatics with12 or less carbon atoms.3 The EPA’semissions standards applicable to thelight-duty vehicles tested in this pro-gram are given in Table 1. Table 2shows the EPA standards applicableto the heavy light-duty vehicles thatwere tested. EPA defines heavy light-duty vehicles as those with grossvehicular weight ratings between6,000 and 8,500 lb.
Hydrocarbons can also escape from a vehicle through evaporation of theliquid fuel. Such evaporation occursin several ways. Diurnal evaporativelosses are emissions that occur dur-ing the day as the temperature rises.
1
TP-25818
INTRODUCTION
As the fuel tank temperature increas-es, fuel evaporation increases andvapors are vented. Hot soak lossesoccur after the vehicle is turned off—the engine and fuel tank remains hotfor a period of time, allowing furtherfuel evaporation. While the vehicle isrunning, the hot engine and exhaustsystem cause additional fuel to bevaporized. These emissions are calledrunning loss emissions. Finally, dur-ing refueling, fuel vapors present inthe tank are forced out as the tank isfilled, resulting in refueling losses.5
Since this test program began, theEPA has expanded its Federal TestProcedures for evaporative emissionsto include procedures for each of theevaporative sources listed above.However, all the evaporative emis-sions results discussed in this reportare from the previous EPA test proce-dures that were limited to two (onediurnal and one hot soak) 1-hourevaporative emissions tests.
Modern light-duty vehicles includeevaporative control systems that con-tain and redirect much of the vapor-ized fuel back into the engine. Onenotable exception is compressed nat-ural gas (CNG) vehicles. For vehiclesdesigned to operate exclusively onCNG, the fuel remains in a gaseousstate, and the entire fuel system is
sealed under pressure. Therefore, aseparate evaporative control system isnot necessary for these vehicle types.
The non-regulated emissions evaluat-ed in this study include carbon diox-ide (CO2), CH4, and air toxics. CO2and CH4 are greenhouse gases thattrap the earth’s heat and may con-tribute to global warming. Air toxicsare pollutants that EPA classifies asknown or probable human carcino-gens—in other words, componentsconsidered to have adverse affects onhuman health. The air toxics evaluat-ed in this study include benzene(C6H6), formaldehyde (HCHO),acetaldehyde (CH3CHO), and 1,3-butadiene (C4H6). Benzene is aknown carcinogen, and the latterthree compounds are probable car-cinogens.
Hydrocarbon emissions from vehi-cles may be made up of hundreds ofindividual hydrocarbon compoundsor species. A gas chromatograph canbe used to quantify the amounts ofthe individual HC species in aprocess known as detailed HC speciation. In this report, the specia-tion of hydrocarbon emissions is used to gain additional insight intoHC emissions. Air toxics emissionsare reported directly and as potency-weighted toxics (PWT). Potency
weighting gives an indication of therelative level of risk for each of thetoxic compounds emitted. The EPAhas calculated an inhalation unit riskfactor for each of the hazardous com-pounds. The weighting factor foreach compound is determined bydividing its individual unit risk factorby the unit risk factor that is the high-est of the four (in this case, 1,3-buta-diene). The resulting number ismultiplied by the mass emissions forthe respective compound to calculatethe PWT value. For example,acetaldehyde has a risk factor that is127 times lower than 1,3-butadiene.The total PWT is the sum of the individual potency weighted values.These EPA risk factors are listed inTable 3.6
Results from the HC speciation arealso used to evaluate the tendency forHC emissions to react in the atmos-phere and form ozone. These resultsare reported here as OFP and specificreactivity (SR). Regulations inCalifornia assign a maximum incre-mental reactivity (MIR) value to individual compounds emitted inautomobile exhaust. The MIR valueis the predicted contribution of thecompound to ozone formation in cer-tain urban atmospheres, and isexpressed in units of milligrams of
2
TP-25818
Table 1. Intermediate Useful Life (5 years, 50,000 miles) Standards for Light-Duty Vehicles (g/mi)4
Fuel Standard THC NMHC NMOG HCE NMHCE CO NOx
Gasoline Tier 0 0.41 3.4 1.0
Gasoline Tier 1 0.41 0.25 3.4 0.4
Alcohol Tier 0 0.41 3.4 1.0
Alcohol Tier 1 0.41 0.25 3.4 0.4
TLEV 0.125 3.4 0.4
Table 2. Intermediate Useful Life Standards for Heavy Light-Duty Vehicles (g/mi)4
Standard THC NMHC CO NOx
Tier 0 (120,000 mi full useful life) 0.80 0.67 10 1.7
Tier 1 (5-yr or 50,000 mi intermediate useful life) 0.32 4.4 0.7
Tier 1 (100,000 mi intermediate useful life) 0.4 5.5 0.97
ozone formed per milligram of thecompound emitted. The MIR value isdetermined in a laboratory experi-ment in which a small increment ofthe compound is added to a simulatedurban background mixture and thenet increase in ozone is measured.Taking into account the MIR valuesfor all measured exhaust compounds,an OFP for the fuel in question maybe calculated. Specific reactivity for agiven fuel may also be calculated bycombining the respective mass ofcompound emissions per mile withthe OFP, which results in units of mil-ligrams of ozone per milligram oftotal organic emissions. In California,SR is based on NMOG emissions.Specific reactivity is usually constantfor a given fuel and engine technolo-gy. To clarify the difference betweenthem, OFP gives an estimate of theamount of ozone formed per miletraveled; SR gives an estimate of theamount of ozone formed per gram ofNMOG emitted. OFP and SR are rel-ative numbers associated with partic-ular atmospheric conditions.
Fuel economy is also calculated fromthe results of the emissions testingprocedures. For vehicles tested ongasoline, fuel economy is reported inmiles per gallon (mpg). For vehiclestested on alcohol fuels, fuel economyis expressed both as miles per gallonand miles per equivalent gallon(mpeg). The mpeg measurementgives an estimate of how far the vehi-cle can travel on an amount of fuelthat has the same energy as a gallonof gasoline. Both are reported foralcohol tests because alcohol fuelshave a lower volumetric energy content than gasoline. The energycontent of the methanol test fuel(M85) is approximately 58% of gaso-line; the energy content of the ethanoltest fuel (E85) is approximately 73%of gasoline (M85 and E85 are furtherdescribed below). For vehicles testedon CNG, fuel economy is reportedonly in miles per equivalent gallons.
This is used for CNG tests becauseCNG is stored in a compressedgaseous state, which is not typicallymeasured in gallons. For transporta-tion applications, CNG is often dis-pensed and priced per gasoline gallonequivalent.
TEST VEHICLES FOR THESTUDY
This report presents emissions testresults on a number of different vehi-cle models. Table 4 lists these vehiclemodels, along with the numbers ofvehicles of each model that were test-ed, and the total numbers of tests thatwere performed on all vehicles ofeach model. For every AFV modeltested, an equivalent number of vehi-cles of the corresponding standardgasoline model (controls) were alsotested. Because many vehicles weretested more than once over the courseof the program (at increased mileagelevels) more tests than vehicles arereported in Table 4. Replicate testswere also conducted on some vehi-cles. All the vehicles discussed hereare original equipment manufacturer(OEM) vehicles. The test vehiclesinclude four passenger car models,one full-size passenger van, and oneminivan.
In order to provide information onemissions deterioration over time, thevehicles were scheduled for testingapproximately once per year. Thefirst set of tests on a particular vehiclemodel was designated as "Round 1,"the second set as "Round 2," and soforth.
Both alcohol-fueled and CNG-fueledAFVs were included in the testingprogram. The principal alcohol fuelsof interest were M85 (a blend of 85%methanol and 15% gasoline) and E85(a blend of 85% ethanol and 15%gasoline). The alcohol-fueled vehi-cles are flexible-fuel vehicles (FFVs),which means that they are capable ofoperating on unleaded gasoline, orany blend of the alcohol and gasolineup to 85% alcohol and 15% gasoline.All the CNG models included in thisreport are dedicated CNG vehicles,which means they are designed tooperate on CNG only.
As noted above, all test vehiclesincluded in this program were part of the federal vehicle pool leased tovarious government fleets by theGeneral Services Administration(GSA). A relatively large number of vehicles were selected for testingto account for the high variabilityobserved in emissions from vehiclespulled directly from fleet service.These differences may be caused byphysical differences inherent in anymanufacturing process, or becausevehicle usage and care vary fromdriver to driver and fleet manager tofleet manager. For instance, vehicleservice applications may vary fromshort delivery routes to highway driving, and the degree to which thepreventive maintenance schedule isfollowed depends, to a certain extent,on the diligence of the fleet manager.For these and other reasons, vehicle-to-vehicle variability in emissionslevels was expected to be fairly high,even at the outset of the testing program.
3
TP-25818
Table 3. EPA Unit Risk Factors for Emissions Air Toxics
CompoundEPA Risk EPA Factor(µg/m3)-1 (Normalized)
1,3-butadiene 2.8 x 10-4 1.000
Benzene 8.3 x 10-6 0.030
Formaldehyde 1.3 x 10-5 0.046
Acetaldehyde 2.2 x 10-6 0.008
TEST FACILITIES
All testing was performed at privatecommercial laboratories with chassisdynamometer exhaust and evapora-tive emission test equipment that iscapable of performing EPA emissionscertification test procedures. Adetailed description of the type of test procedures and equipment used canbe found on the AFDC Web site(http://www.afdc. doe.gov). The labo-ratories were selected on the basis ofa federal government competitivebidding process in which experiencewith performing the Federal TestProcedures (FTP)—in particular, FTPtesting of alcohol and natural gasvehicles—was stressed. Three organi-zations were awarded emissions test-ing subcontracts: Automotive TestingLaboratories (ATL) in East Liberty,Ohio, which tested vehicles fromOhio, Michigan, and Illinois;Environmental Research andDevelopment (ERD), which testedvehicles in the Washington D.C. and
New York City regions; and ManTechEnvironmental Technology, Inc.(ManTech), which tested vehiclesfrom Colorado (at a high altitude ofapproximately 5,300 feet). For theremainder of the report, these labs arereferred to as Lab 1, Lab 2, andLab 3, respectively. Before any test-ing began, a coordination meetingwas held between all the participatinglaboratories and NREL to ensure con-sistency in the test procedures. NRELand EPA employees subsequentlyconducted laboratory site visits.
TEST FUELS
Table 5 summarizes the physicalproperties of the liquid test fuels usedin this study. The baseline gasolineused was California Phase 2 reformu-lated gasoline, or RFG. This fuel waschosen because it represents a "bestcase" scenario for gasoline emissions.If alternative fuels are to compete,they must be compared to the bestgasoline available. RFG has a lower
sulfur, olefin, and aromatic contentthan standard unleaded gasoline. TheAuto/Oil Air Quality ImprovementResearch Program (AQIRP) con-ducted extensive testing that com-pared emissions from vehicles testedon various fuel blends, including certification test fuel, industry-aver-age gasoline, and RFG2. In general,the AQIRP study found that vehiclestested on RFG tended to showreduced regulated emissions. There-fore, one might expect that the com-parison between alternative fuels andan industry-average gasoline wouldbe slightly more favorable for alterna-tive fuels than the results discussedhere. The alcohol blends were pre-pared using 85% alcohol (methanolor ethanol) and 15% RFG. PhillipsPetroleum Company blended andsupplied the alcohol and gasolinefuels. Compressed Gas Technologies,Inc., supplied the CNG fuel that wasdesigned to represent a nationalindustry-average fuel composition.
Table 4. Emissions Tests Completed
4
TP-25818
Vehicle Model Vehicle Type Number of NumberModel Year Vehicles Tested of Tests
Methanol
Dodge 1995
M85 FFV 24 89
Intrepid Standard 25 47
Dodge1993
M85 FFV 77 373
Spirit Standard 72 145
Ethanol
Ford 1994/95 E85 FV 24 88
Taurus 1995 Standard 24 45
Chevrolet 1992/93 E85 FFV 25 144
Lumina 1993 Standard 16 45
Compressed Natural Gas
Dodge1992/94
Dedicated CNG 54 144
B-250 Standard 53 138
Dodge1994
Dedicated CNG 13 16
Caravan Standard 6 6
Total 413 1,280
Methanol
Ethanol
Compressed Natural Gas
Table 6 lists the specifications and asample analysis of the CNG fuel usedthroughout the study.
TEST PROCEDURES
This program used the EPA’s emis-sions certification test procedure,known as the FTP-75. The FTP-75includes measurement of exhaustemissions on a chassis dynamometerand two 1-hour evaporative emissionstests. Details of the test proceduresare described in the Code of FederalRegulations4. Once a vehicle wasidentified for testing, the laboratorynotified the fleet representative andscheduled a convenient test date. Thelab also verified that the vehicle hadreceived all scheduled maintenanceand was operating properly. Onarrival at the test laboratory, the vehi-cle was inspected for any problems.Once the vehicle was approved fortesting, it was subjected to an exten-sive procedure designed to minimizeresidual effects from resident fuels.Figure 1 outlines the complete proce-dure for testing a vehicle, includingthe fuel changeover procedure. Thefuel changeover procedure was per-formed before every test, includingthe first test in the sequence. Thisprocess follows the AQIRP’s vehicletesting procedures.7 The main elements of the fuel changeover pro-cedure are a 60-minute purge of thevehicle’s evaporative canister, severalfuel tank drain and fill sequences, a
chassis dynamometer driving cycleusing the test fuel, and several enginestart-up and idle sequences. Anotherpart of the vehicle preconditioningprocedure is the Urban DynamometerDriving Schedule (UDDS), alsocalled the LA4. The UDDS wasderived from an actual driving routethrough LA that was selected to rep-resent a typical city driving pattern.
Once the fuel changeover procedurewas complete, the vehicle was testedfollowing the FTP-75 for light-dutyvehicle chassis dynamometer testing(including evaporative testing).Figure 2 shows the FTP-75 drivingcycle. Alcohol fuel vehicles weretested on both alcohol fuel (M85 or
E85) and RFG. The correspondingcontrol vehicles were tested on RFG.All CNG vehicles were tested only onCNG fuel, and their correspondinggasoline controls were tested onRFG.
The emissions samples collected dur-ing the FTP were analyzed for HC,CH4, NOx, CO, and CO2. Alcohols(ethanol and methanol) in the emis-sions were collected using primaryand secondary impingers. Gas chromatography was used to analyzethe alcohols. Aldehydes were collect-ed on dinitrophenylhydrazine(DNPH) coated silica cartridges orimpingers filled with an acetoni-trile/DNPH solution, and analyzed
Net Heat of Combustion (Btu/gal) 64,600 81,825 111,960
Reid Vapor Pressure 7.5 6.15 6.9
Table 6. Composition of CNG
% Volume
Component Specification Analysis
Methane 93.05 93.15
Ethane 3.47 3.52
Nitrogen 1.67 1.47
Carbon Dioxide 0.81 0.82
Propane 0.66 0.68
N-Butane 0.12 0.13
I-Butane 0.08 0.07
N-Hexane 0.06 0.06
I-Pentane 0.04 0.06
N-Pentane 0.03 0.04
Oxygen 0.00 0.00
using high-performance liquid chro-matography. Appendix A contains theentire FTP data set.
The emissions from a subset of testvehicles were subjected to full hydro-carbon speciation. Speciation is thequantification of individual HC com-ponents using gas chromatography.Table 7 lists the numbers and types ofvehicles for which hydrocarbon emis-sions were speciated. Up to 288 HCconstituents in the emissions sampleswere identified; a complete list isgiven in Appendix B. Appendix Ccontains the speciated HC data set.
DATA ANALYSIS APPROACH
Raw data files of the emissions tests from each laboratory were electronically submitted and loaded
into the AFDC at NREL. Before con-ducting any analyses of the data, anumber of checks and edits wereundertaken to ensure data quality.The data sets were sorted by vehiclemodel, test fuel, and test round.Repeat tests were reviewed for prob-lems or outliers. In most cases, theseduplicate tests were averaged andreturned to the data set. Each data setwas then analyzed for outliers, whichwere removed. Outliers were definedas any value that was +/- 3 standarddeviations from the mean. An excep-tion was made with the evaporativeemissions results. Because of thehigh variability of evaporative data,no outliers were removed from thedata sets.
After all checks and edits wereapplied, the data were imported into
the JMP® software, which is a com-prehensive PC-based statistical dataanalysis package developed by SASInstitute. Using this software, a multi-variable analysis of variance(ANOVA) was performed to deter-mine the statistical significance ofvarious factors on emissions. The pri-mary effects of interest include fuel,vehicle, and test round. Secondaryeffects include the fuel by vehicle,fuel by test round, and vehicle by testround interactions. All data were ana-lyzed at the 95% confidence level.Appendix D gives a detailed explana-tion of the data compilation and theANOVA statistical approach.
PRESENTATION OF ANALYSISRESULTS
The following sections contain dis-cussions of the results from each ofthe individual vehicle models tested.Sections on each alternative fuelbegin with an overview comparingthe fuel with RFG, followed bydetails on each model. The discus-sions on each vehicle model are sub-divided into sections on regulatedemissions, evaporative emissions,greenhouse gases, and aldehydes.Separate tables and graphs cover theair toxics, OFP, and SR. Each of thesesections concentrates on the compari-son between the emissions and theEPA standard, fuel differences, andround-to-round differences.
The results are presented in tablesthat include regulated and non-regu-lated emissions constituents for eachvehicle model. These tables containdescriptive statistics for emissionsresults obtained for each fuel onwhich the vehicle model was tested.Average emissions are reported asgrams per mile. Of particular interestis the percent difference between theemissions from the alternative fueland the RFG tests (e.g., M85 versusRFG).
6
TP-25818
Yes
Fuel Changeover Procedure
60-minute canister purge (40 ft3/H)
Drain fuel
3-gallon fill – new fuel(room temp.)Idle 1 minute
Drain fuel
40% fill – new fuel(cold temp.)Diurnal heat build
Engine off 5 minutes
Idle 1 minute
Engine off 1 minute
Idle 1 minute
Engine off 1 minute
Certification Proceedure
Same testfuel?
Drain fuel
40% fill (room temp.)
Drain fuel
40% fill (cold temp.)
Matrix complete? Finished
YesNo
No
LA-4 precondition
Test sequence start
Diurnal 60˚–84˚F (16˚–19˚C)
FTP mass emissionstailpipe & converter
LA-4 precondition
Cold soak 12–36 h
Hot soak
0258
1801
m
Figure 1. Vehicle testing procedure
For each vehicle model tested, a summary table of results shows theaverage results, percent differencesbetween the averages, and an indication of which differences inaverage values are statistically signif-icant at the 95% confidence level.Percent difference was calculatedusing the following formula:
where U is the average of emissionstest results obtained on the fuel inquestion. Statistical significance wasdetermined through ANOVA proce-dures, applying the appropriate datamodel for each particular case. Anexample ANOVA table is shown inAppendix D.
In addition to the tables, each sectioncontains a series of graphs depictingthe average emissions results (byfuel, lab, and/or round) for the
different fuels tested. Bar charts orline graphs are used to illustrate thedifferences between fuels. The textaccompanying the tables and graphsdescribes the various trends depictedin them, and discusses the statisticalsignificance (if any) of those trends.
For the alcohol-fuel vehicle models,the comparisons discussed concen-trate on the difference between thealcohol and the gasoline tests on theFFV. This eliminates any discrepan-cies in the results that could resultfrom large differences in odometer readings for the FFV and gasolinecontrol vehicles. The results for thegasoline control model are shown in the graphs for reference. Becausethe CNG vehicles are dedicated vehi-cles, the comparison must be madebetween the AFV and the gasolinecontrol. Odometer range differencesbetween these vehicles could play apart in the test results.
7
TP-25818
Table 7. Number and Type of Vehicles with HC Speciation
Model Fuel Type Number of Number ofVehicles Tests
Dodge M85 FFV 6 16
Intrepid RFG Standard 4 7
Dodge M85 FFV 10 28
Spirit RFG Standard 9 14
Ford E85 FFV 6 16
Taurus RFG Standard 5 8
Dodge CNG Dedicated CNG 8 17
B250 RFG Standard 8 16
Total 56 122
70
60
50
40
30
20
10
0
Veh
icle
Sp
eed
, MP
H
Bag 1 Bag 2 Bag 3
0258
1802
m
ColdStart Hot
Start
10 minuteSoak
Test Time (s)0 240 480 720 960 1200 1440 1680 1920 2160 2400
Figure 2. EPA’s FTP-75 driving cycle
AlternateFuel Gasoline x 100,U – U
GasolineU
—
8
TP-25818
For this study, three different M85FFV models were tested: the DodgeSpirit, the Dodge Intrepid, and theFord Econoline van. Because theresults for the Ford van were reportedin a previous publication,8 they arenot included in this report.
Table 8 provides a summary compari-son of the emissions from the FFVstested on M85 to the same vehiclestested on RFG. In the table, the high-lighted blocks indicate that there wasa 95% statistically significant differ-ence (based on the ANOVA) in emis-sions from the two fuels tested. Aplus sign in the block means that theemissions from the M85 test werehigher than those from the RFG test,and a minus sign means that the M85emissions were lower. These resultsare shown for all of the measuredemissions from the Dodge Spirit andthe Intrepid at the respective test lab-oratories. For instance, during thefirst round (Round 1) of testing, theCO emissions from the DodgeIntrepid were higher for M85 thanRFG (plus sign), but the differencewas not statistically significant at the95% confidence level (not highlight-ed). A more detailed and quantitativediscussion of the specific results foreach vehicle is presented in the fol-lowing sections, but it may also beuseful to consider a more qualitativeview of the general trends for themethanol tests.
Some of the results (such as HC,greenhouse gases, aldehydes, and thefuel economy calculation) were veryconsistent across vehicle models, testlaboratories and test rounds, others
(CO, NOx, and evaporative HC) weremore mixed. Although both vehiclemodels are FFVs produced byDodge, the two models may employdifferent engine calibrations in orderto meet differing performance andemissions expectations.
In general, both vehicles tended tohave significantly (evaluated at 95%)lower NMHCE, total hydrocarbon(THC), CO2, CH4, and CH3CHO
emissions, as well as lower fuel econ-omy, when tested on M85. On theother hand, both vehicles tended tohave significantly higher HCHOemissions and energy equivalent fueleconomy (mpeg) when tested onM85. There appeared to be very littledifference (not statistically significantat 95%) in CO and evaporative HCemissions between the two fuels. TheNOx emissions tended to be higher
METHANOL VEHICLES
Table 8. Summary Comparison of Average Emission Resultsfrom M85 versus RFG
Dodge Intrepid Dodge SpiritLab 1 Lab 1 Lab 3
Round 1 Round 2 Round 1 Round 2 Round 1 Round 2
Regulated Emissions
NMHCE - - - - - -
THC - - - - - -
CO + + - - - -
NOx + + + - + +
Evaporative Emissions
THC + + - - - +
Greenhouse Gases
CO2 - - - - - -
CH4 - - - - - -
Aldehydes
HCHO + + + + + +
CH3CHO - - - - - -
Fuel Economy
mpg - - - - - -
mpeg + + + + - +
Regulated Emissions
Evaporative Emissions
Greenhouse Gases
Aldehydes
Fuel Economy
“+” Indicates results from M85 tests were higher than RFG tests“-” Indicates results from M85 tests were lower than RFG testsHighlighted blocks indicate a significant statistical difference.
from M85, but this result was notconsistent across all test categories.
One possible reason for findingmixed results and fuel effects that arenot statistically significant is that aFFV is not optimized for either fuel,but is instead designed to performacceptably on a wide range of fuelblends. An inherent benefit of theflexible fuel design is the capabilityfor convenient fueling on gasoline ormethanol where it is available. Aninherent drawback to this design isthat the vehicle cannot be optimizedto take advantage of some of the ben-eficial properties of methanol. Oneobvious example of this is that thesevehicles are designed with a com-pression ratio that is suitable forgasoline. A vehicle optimized formethanol could be designed with an increased compression ratio thatwould take advantage of methanol’shigher octane rating and provideincreased power and efficiency.
A similar evaluation of the generaltrends from the more limited set ofHC speciation tests (shown in Table9) is very consistent across vehiclesand labs. These results give an indica-tion of how the chemical compositionof the hydrocarbon emissions differbetween the two fuels. With regard tothe four air toxic HC covered here,
the vehicles tested on M85 tended toemit much higher levels of HCHO,and significantly lower levels ofCH3CHO, 1,3-butadiene, and ben-zene compared to the same vehiclestested on RFG. When the potencyweighting factors are applied to theseemissions levels and totaled as thetotal PWT emissions, the M85 resultswere significantly lower than theRFG results.
The detailed speciation of the HCwas also used to compare the tenden-cy for HC emissions to react and
form ozone. The OFP and the SR ofthe HC emissions from the M85 testswere significantly lower than thosefrom the same vehicles tested onRFG. The detailed evaluation ofhydrocarbon emissions from M85and RFG was consistent for both thetoxic emissions and the parametersrelated to ozone formation for bothvehicle models at the two laboratoriesthat performed hydrocarbon speciation.
DODGE INTREPID
The 1995 Dodge Intrepid (shown inFigure 3) is a passenger car equippedwith a 3.3 L V6 engine. This vehiclemodel employs electronically con-trolled multi-point fuel injection andis equipped with a three-way catalyst for exhaust emissions con-trol. The flexible-fuel version wascertified to the EPA federal Tier 0emissions standard and the standardgasoline version was certified to fed-eral Tier 1 levels (refer to Table 1,page 2). We performed two rounds oftests on the Dodge Intrepids at Lab 1.There were 17 standard gasolineIntrepids and 16 FFVs tested in bothrounds. Mileage ranges and average
9
TP-25818
Table 9. Summary Comparison of Average SpeciatedHydrocarbon Results from M85 versus RFG
Intrepid Spirit
Air Toxics Lab 1 Lab 1 Lab 3
HCHO + + +
CH3CHO - - -
1,3-butadiene - - -
Benzene - - -
Total PWT - - -
Ozone Reactivity
OFP - - -
SR - - -
Figure 3. 1995 Dodge Intrepid
Arg
onne
Nat
iona
l Lab
orat
ory/
PIX
0
Ozone Reactivity
“+” Indicates results from M85 tests were higher than RFG tests“-” Indicates results from M85 tests were lower than RFG testsHighlighted blocks indicate a significant statistical difference.
odometer readings for the Intrepidsare shown in Table 10. The completelisting of the vehicles tested and thedetailed emissions test results areincluded in Appendix A.
Regulated Emissions
Table 11 shows the average emissionsresults for the Dodge Intrepid. Thevalues shown include the averages forthe FFV model tested on M85 and
RFG and the percent differencebetween the averages. An indicationis also given on whether the differencebetween the average results is statisti-cally significant as determined by theANOVA. All average regulated emis-sions shown here were well below theTier 1 emissions standards. Figure 4shows the regulated and CO2 emis-sions for the Intrepid along with theTier 1 50,000-mile certification
standard. In general, when comparingthe M85 and RFG regulated emis-sions for the FFV Dodge Intrepid,NMHCE emissions from the M85tests were lower, there was very littledifference in CO emissions, and theNOx emissions from the M85 testswere substantially higher.
More specifically, the FFV Intrepidshowed a statistically significantdecrease in HC emissions when test-ed on M85. In Round 1, the averageemissions from the M85 tests were16% lower; in Round 2, they were19.6% lower than those from thesame vehicles tested on RFG. Whencomparing the FFV tested on RFG tothe standard Intrepid, the FFV hadhigher NMHCE emissions in bothtest rounds. For the FFVs, there tended to be a small, but statisticallysignificant increase in NMHCE emissions from Round 1 to Round 2.
10
TP-25818
Table 10. Odometer Readings for the Dodge Intrepid
FFV Gasoline
Round 1 2 1 2
No. vehicles tested 16 16 17 17
Odometer (miles)
Average 5,128 14,332 5,661 17,231
Maximum 9,558 26,084 18,783 42,738
Minimum 3,047 9,653 3,336 5,929
Odometer (miles)
Table 11. Average Emissions Results from the Dodge Intrepid
Round 1 Round 2
FFV- FFV- Percent Sig. Fuel FFV- FFV- Percent Sig. FuelM85 RFG Difference Effect? M85 RFG Difference Effect?
Regulated Emissions (g/mi)
NMHCE 0.107 0.127 -15.7% y 0.127 0.158 -19.62% y
THC 0.112 0.149 -24.7% y 0.132 0.182 -27.6% y
CO 1.01 0.99 2.0% n 1.16 1.12 3.9% n
NOx 0.328 0.245 33.9% y 0.283 0.239 18.2% y
Evaporative Emissions (g)
Total Evaporative 0.876 0.669 30.9% y 0.816 0.712 14.6% n
Greenhouse Gases (g/mi)
CO2 413.9 452.3 -8.5% y 395.0 431.2 -8.4% y
CH4 0.016 0.028 -42.7% y 0.017 0.031 -43.6% y
Aldehydes (mg/mi)
HCHO 16.0 1.9 742.1% y 17.62 2.52 604.8% y
CH3CHO 0.17 0.45 -62.0% y 0.23 0.59 -60.9% y
Fuel Economy
mpg 11.66 19.19 -39.2% y 12.16 20.13 -39.6% y
mpeg 20.21 19.19 5.3% y 21.07 20.13 4.7% y
Regulated Emissions (g/mi)
Fuel Economy
Aldehydes (mg/mi)
Greenhouse Gases (g/mi)
Evaporative Emissions (g/Test)
11
TP-25818
The CO and NOx emissions do not show the same trend asNMHCE. The CO emissions from the FFVs tested on M85were not statistically different from the results of the FFVstested on RFG and there was little difference between rounds.There was a statistically significant increase in NOx emis-sions for the FFV tested on M85. In Round 1, the NOx emis-sions from the M85 tests were 33.9% higher; in Round 2, theywere 18.2% higher than those from the RFG tests on the same vehicles. The NOx emissions for the FFV Intrepid show adecrease in the second round that was significant for M85, butwas not statistically significant for RFG. NOx emissions fromthe standard gasoline vehicles tested on RFG were substan-tially lower than those from the FFVs tested on the same fuel.
Evaporative Emissions
The average evaporative emissions for the FFV Intrepid arelisted in Table 11 and shown graphically in Figure 5. Theaverage evaporative HC were well below the 2-g standard forthe FFVs and the gasoline vehicles. When comparing evapo-rative emissions results for the FFV Intrepid tested on M85 to the same vehicles tested on RFG, the M85 evaporativeemissions were 30% higher in Round 1, and 14.6% higher inRound 2. The higher evaporative emissions for the FFV testedon M85 is expected, because the Reid vapor pressure (RVP)of the methanol fuel is higher than that of RFG (see Table 5).The difference in evaporative emissions was statistically sig-nificant in Round 1, but was not in Round 2. The averageevaporative emissions for the conventional Intrepids werelower than the averages for the FFV on both fuels. There wasno significant difference between Round 1 and 2 for the FFVon either fuel.
Greenhouse Gases
The average CO2 emissions for the Intrepids are listed inTable 11 and shown in Figure 4d. Results from Rounds 1 and 2
0
100
200
300
400
500
STD-RFGFFV-RFGFFV-M85
4a: Non-Methane Hydrocarbon Equivalent
4b: Carbon Monoxide
4d: Carbon Dioxide
4c: Oxides of Nitrogen
NM
HC
E E
mis
sion
s (g
/mi)
CO
2 E
mis
sion
s (g
/mi)
NO
x E
mis
sion
s (g
/mi)
CO
Em
issi
ons
(g/m
i)
Round 1 Round 2
Round 1 Round 2
Round 1 Round 2
Round 1 Round 2
0.30
0.25
0.20
0.15
0.10
0.05
0.00
4.0
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0
0.45
0.40
0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
EPA Tier 1
EPA Tier 1
EPA Tier 1
0258
1804
m
Figure 4. Emissions results from theDodge Intrepid
STD-RFGFFV-RFGFFV-M85
Round 1 Round 2
2.5
2.0
1.5
1.0
0.5
0.0
EPA Tier 1 & 0
0258
1805
m
Total Evaporative Hydrocarbons
Eva
pora
tive
Em
issi
ons
(g/T
est)
Figure 5. Evaporative emissions results from theDodge Intrepid
12
TP-25818
followed the same trend between thefuels and vehicle types, with very little difference between the rounds.The CO2 emissions from the FFVtested on M85 were approximately8.5% lower than those from the samevehicles tested on RFG. The resultsfor the standard model were similarto the FFV on RFG. Average CH4emissions were very low (less than0.05 g/mi). For the FFV tested onM85, the CH4 emissions wereapproximately 43% lower than thosefrom the FFV tested on RFG in bothrounds.
Aldehydes
Figure 6 shows the comparison ofaldehyde emissions for the DodgeIntrepid. This graph shows that theformaldehyde emissions were muchhigher from the FFV when tested onM85. Formaldehyde is a primarydecomposition product frommethanol combustion; therefore,the higher numbers are expected. For Round 1, average formaldehydeemissions were 742% higher in theM85 tests, and for Round 2, the M85results were 605% higher than theRFG results. Acetaldehyde emissionlevels for the FFV tested on M85were approximately 61% lower thanthe results for the same vehicles tested on RFG, but the levels ofacetaldehyde emissions were verylow (less than 0.6 mg/mi).
Over the two rounds of emissionstests performed, full HC speciationwas performed on a total of six FFVIntrepids and four standard gasolinevehicles. Table 12 lists the averagemeasured toxic emissions and thePWT values and percent differencefor the four air toxic compounds. Thepotency weighting is discussed onpage 2 and the factors are shown inTable 3. The aldehyde values listedare the averages for the speciatedvehicles only. Figure 7 shows thecomparison of these compounds andthe total PWT for the DodgeIntrepids. When comparing PWT forthe FFV Dodge Intrepids tested onM85 compared to the same vehiclestested on RFG, the HCHO emissionswere significantly higher, butCH3CHO, 1,3-butadiene, and ben-
zene were significantly lower whentested on M85. Total PWT emissionsfor the FFVs tested on M85 were16.2% lower than those from thesame vehicles tested on RFG.
Table 13 lists the average OFP andSR for the FFV Intrepid. Figure 8illustrates an important considerationwhen comparing HC emissions forthe two test fuels. Both OFP and SRwere significantly lower for the FFVwhen tested on M85. Although theaverage NMOG emissions from theM85 tests were 85% higher than theRFG tests, the OFP was 33.7% lowerand the SR was 65.2% lower for theM85 tests. In other words, althoughthe NMOG emissions from this sub-set of vehicles were higher, the poten-tial to form ozone based on theexhaust composition is significantlylower. The exhaust from M85 is lessreactive in forming ozone in the
STD-RFGFFV-RFGFFV-M85
Round 1
0258
1806
m
Aldehyde Emissions
Ald
ehyd
e E
mis
sion
s (m
g/m
i)
0
5
10
15
20
FormaldehydeAcetaldehyde Round 1
Round 2Round 2
Figure 6. Aldehyde emissions from the Dodge Intrepid
Table 12. Toxic Emissions from the Dodge Intrepid
FFV-M85 FFV-RFG
MeasuredPWT
MeasuredPWT
Percent Sig. Fuel
Value (mg/mi) Value (mg/mi)Difference Effect?
HCHO 15.65 0.72 2.00 0.092 682.5% y
CH3CHO 0.20 0.0016 0.488 0.0039 -59.0% y
1,3-butadiene 0.113 0.113 0.813 0.813 -86.2% y
Benzene 0.919 0.028 3.956 0.119 -76.8% y
Total 16.882 0.861 7.257 1.027 -16.2% y
13
TP-25818
atmosphere. The OFP and the SR for the gasoline model tested onRFG were similar to those of the FFVtested on RFG.
Fuel Economy
The fuel economy for the FFVIntrepid was approximately 12 mpgwhen operating on M85 and 20 mpgon gasoline. This is a decrease ofapproximately 39% for the FFV test-ed on M85 for both rounds. This is
expected because methanol has alower volumetric energy content than gasoline. The energy content of theM85 (64,600 Btu/gal) is 58% of theRFG (111,960 Btu/gal). In otherwords, it takes approximately 1.7 gal-lons of M85 to travel the same dis-tance as 1 gallon of gasoline. Whenthe values are adjusted to account forthis difference, the average fuel econ-omy for the FFV Intrepid on M85 is20 mpeg in Round 1 and 21 mpeg in
Round 2. In other words, the M85tests showed a 5% improvement inenergy equivalent fuel economy overRFG for Round 1 and a 4.7%improvement for Round 2. The FFVon gasoline had similar fuel economynumbers to the conventional model.An important consideration for mostdrivers is the range of the vehicle.Because of the difference in energycontent of the fuels, the FFV operat-ing on M85 will not travel as far aswhen using gasoline. For this reason,many manufacturers increase the size of the tank to help offset this dif-ference. The FFV Intrepid and thegasoline control Intrepid tested here,however, both had 18-gallon fueltanks. Based on the fuel economy for the FTP-75, the gasoline controlvehicle has an approximate range of356 miles; the FFV has a range of214 miles on M85 and 354 miles ongasoline.
DODGE SPIRIT
The 1993 Dodge Spirit (shown inFigure 9) is a passenger car equippedwith a 2.5 L, I6 engine with multi-point fuel injection. Although boththe FFV and gasoline Spirits werecertified to federal Tier 0 emissionsstandards, the majority of the emis-sions results are below the more strin-gent Tier 1 levels. This report coversthe two rounds of testing performedon the Dodge Spirits at Labs 1 and 3.Lab 2 tested the Dodge Spirit in only1 round and the results can be foundin a previous publication.8 At Lab 1,21 FFV Spirits and 24 gasoline con-trols were tested in both rounds. AtLab 3, the FFV Spirits totaled 22 andthe gasoline controls 20 in bothrounds. Mileage ranges and averageodometer readings for each vehicletype and round are listed in Tables 14and 15. The complete data set can befound in Appendix A.
0258
1807
m
Toxic EmissionsP
WT
(m
g/m
i)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
STD-RFGFFV-RFGFFV-M85
TotalBenzene1,3 butadieneCH3CHOHCHO
Figure 7. PWT emissions from the Dodge Intrepid
Table 13. OFP for the Dodge Intrepid
FFV- FFV- Percent Sig. FuelM85 RFG Difference Effect?
NMOG (mg/mi) 257.94 139.76 84.6 y
OFP (mg O3/mi) 319.5 481.69 -33.7% y
SR (mg O3/mg NMOG) 1.248 3.587 -65.2% y
0258
1808
m
Ozone Formation and Reactivity
OF
P (
mg
O3/
mi)
SROFP
STD-RFGFFV-RFGFFV-M85
600
500
400
300
200
100
0
4.0
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0
SR
(m
g O
3/m
g N
MO
G)
Figure 8. OFP and SR for the Dodge Intrepid
Regulated Emissions
Tables 16 and 17 list the averageemissions results for the FFV DodgeSpirits tested at Lab 1 and Lab 3.Included in the tables are the averagesfor the FFV tested on M85 and RFG,along with the percent differencebetween the averages. The statisticalsignificance of the fuel effect was
determined using the ANOVA analy-sis. All average regulated emissionsfor the Spirits tested at both labs werewell below the Tier 0 emission stan-dard and in most cases, also belowthe more stringent Tier 1 levels. (TheEPA emissions certification standardsare shown in Table 1 on page 2.)Figures 10 and 11 show the regulated
and CO2 emissions for the Spiritstested at Labs 1 and 3. In general,when comparing the regulated emis-sions for M85 and RFG tests for theDodge Spirit, NMHCE emissionsfrom the M85 tests were lower, COemissions from the M85 tests wereslightly lower, and NOx emissions forthe M85 tests tended to be higher.
Average HC emissions showed simi-lar patterns on the vehicles tested atboth labs. The NMHCE emissions forthe FFV operating on M85 were sig-nificantly less than those from thesame vehicles tested on gasoline(Figures 10a and 11a). For Lab 1, thedifference was 17% during Round 1and 27% in Round 2. For Lab 3, thedifference between the fuels was evenlarger, approximately 30.5% in bothrounds. NMHCE emissions for theconventional Spirits tested at bothlabs were lower than the levels of theFFV operating on either fuel. The dif-ference in NMHCE emissions fromRound 1 to Round 2 tended to be notsignificant at the 95% confidencelevel.
The CO emissions from both labs areshown in Figures 10b and 11b. Theaverage values at Lab 3 were higherthan the averages at Lab 1, but theyfollow the same pattern. At both labsthe standard gasoline model hadlower CO emissions than the FFV oneither fuel. The FFV had lower COemissions when tested on M85, butthe difference between the two fuelswas only significant for Round 2 atLab 1. At Lab 1, the FFV on M85 was1% lower in Round 1 and approxi-mately 11% lower in Round 2. TheFFVs tested at Lab 3 showed a differ-ence of approximately 10% lower onM85 for both rounds. Average COemissions showed increases fromRound 1 to Round 2 that were statisti-cally significant for both fuels at bothlabs. All CO emissions averages werewell below the Tier 0 and Tier 1 stan-dard of 3.4 g/mi.
14
TP-25818
Figure 9. The 1993 M85 Dodge Spirit
War
ren
Gre
tz, N
RE
L/P
IX02
481
Table 14. Odometer Readings for the Dodge Spirit Tested at Lab 1
FFV Gasoline
Round 1 2 1 2
No. vehicles tested 21 21 24 24
Odometer (miles)
Average 8,803 17,073 12,208 27,834
Maximum 18,203 29,679 35,757 61,638
Minimum 3,704 7,683 4,339 10,036
Odometer (miles)
Table 15. Odometer Readings for the Dodge Spirit Tested at Lab 3
FFV Gasoline
Round 1 2 1 2
No. vehicles tested 22 22 20 20
Odometer (miles)
Average 14,030 24,240 16,063 28,035
Maximum 26,058 38,506 28,005 47,989
Minimum 4,080 8,746 5,743 9,467
Odometer (miles)
15
TP-25818
Table 16. Average Emissions Results from the Dodge Spirit Tested at Lab 1
Round 1 Round 2
FFV- FFV- Percent Sig. Fuel FFV- FFV- Percent Sig. FuelM85 RFG Difference Effect? M85 RFG Difference Effect?
Regulated Emissions (g/mi)
NMHCE 0.108 0.130 -16.9% y 0.104 0.142 -26.9% y
THC 0.112 0.151 -25.8% y 0.111 0.168 -33.8% y
CO 1.43 1.45 -1.2% n 1.61 1.81 -10.9% y
NOx 0.212 0.151 40.4% y 0.182 0.219 -16.9% y
Evaporative Emissions (g)
Total Evaporative 0.708 0.724 -2.21% n 0.78 0.887 -12.1% n
Greenhouse Gases (g/mi)
CO2 350.3 379.5 -7.7% y 348.6 376.8 -7.5% y
CH4 0.015 0.026 -43.1% y 0.016 0.031 -49.8% y
Aldehydes (mg/mi)
HCHO 12.7 1.47 763.9% y 12.4 1.42 771.8% y
CH3CHO 0.31 0.50 -37.8% y 0.19 0.39 -50.9 y
Fuel Economy
mpg 13.56 22.82 -40.6% y 13.8 23.02 -40.1% y
mpeg 23.5 22.82 3.0% y 23.92 23.02 3.9% y
Table 17. Average Emissions Results from the Dodge Spirit Tested at Lab 3
Round 1 Round 2
FFV- FFV- Percent Sig. Fuel FFV- FFV- Percent Sig. FuelM85 RFG Difference Effect? M85 RFG Difference Effect?
Regulated Emissions (g/mi)
NMHCE 0.113 0.162 -30.6% y 0.128 0.184 -30.4% y
THC 0.061 0.188 -67.5% y 0.061 0.220 -72.5% y
CO 1.63 1.80 -9.6% n 1.98 2.11 -10.5% n
NOx 0.207 0.166 24.7% y 0.251 0.236 6.4% n
Evaporative Emissions (g)
Total Evaporative 0.371 0.48 -22.7% n 1.207 1.067 13.1% n
Greenhouse Gases (g/mi)
CO2 331.3 357.2 -7.3% y 331.5 357.9 -7.4% y
CH4 0.014 0.028 -48.5% y 0.015 0.031 -52.0% y
Aldehydes (mg/mi)
HCHO 9.15 1.16 688.8% y 10.4 1.63 538.0% y
CH3CHO 0.19 0.35 -45.7% y 0.29 0.47 -38.3% y
Fuel Economy
mpg 12.78 24.07 -46.9% y 14.46 24.0 -39.8% y
mpeg 22.15 24.07 -8.0% y 25.06 24.0 4.4% y
Regulated Emissions (g/mi)
Evaporative Emissions (g/Test)
Greenhouse Gases (g/mi)
Aldehydes (mg/mi)
Fuel Economy
Evaporative Emissions (g/Test)
Greenhouse Gases (g/mi)
Aldehydes (mg/mi)
Fuel Economy
16
TP-25818
STD-RFGFFV-RFGFFV-M85
10a: Non-Methane Hydrocarbon Equivalent
10b: Carbon Monoxide
10d: Carbon Dioxide
10c: Oxides of Nitrogen
NM
HC
E E
mis
sion
s (g
/mi)
CO
2 E
mis
sion
s (g
/mi)
NO
x E
mis
sion
s (g
/mi)
CO
Em
issi
ons
(g/m
i)
Round 1 Round 2
Round 1 Round 2
Round 1 Round 2
Round 1 Round 2
0.45
0.40
0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
4.0
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0
1.2
1.0
0.8
0.6
0.4
0.2
0.0
EPA Tier 0
EPA Tier 1 & 0
EPA Tier 1
0258
1810
m
400
350
300
250
200
150
100
50
0
EPA Tier 0 (THC)
EPA Tier 1
Figure 10. Emissions results from theDodge Spirit tested at Lab 1
STD-RFGFFV-RFGFFV-M85
11a: Non-Methane Hydrocarbon Equivalent
11b: Carbon Monoxide
11d: Carbon Dioxide
11c: Oxides of Nitrogen
NM
HC
E E
mis
sion
s (g
/mi)
CO
2 E
mis
sion
s (g
/mi)
NO
x E
mis
sion
s (g
/mi)
CO
Em
issi
ons
(g/m
i)
Round 1 Round 2
Round 1 Round 2
Round 1 Round 2
Round 1 Round 2
0.45
0.40
0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
4.0
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0
1.2
1.0
0.8
0.6
0.4
0.2
0.0
EPA Tier 0
EPA Tier 1 & 0
EPA Tier 1
0258
1811
m
400
350
300
250
200
150
100
50
0
EPA Tier 0 (THC)
EPA Tier 1
Figure 11. Emissions results from theDodge Spirit tested at Lab 3
The NOx emissions for the Spiritstested at Lab 1 showed different pat-terns in the two rounds. DuringRound 1, the NOx emissions from theFFV operating on M85 were 40%higher than those from the same vehicles tested on RFG. The standardmodel tested on RFG had an evenhigher NOx average. In Round 2, theaverage NOx emissions for the FFVtested on M85 were 17% lower thanthe average when tested on RFG. Thestandard model again tested higherthan the FFV on both fuels. TheSpirits tested at Lab 3 showed similartrends. In Round 1, the FFV Spiritstested on M85 had 25% higher NOxemissions than when they were testedon RFG. In Round 2, the average forM85 was only 6% higher than theaverage for RFG. The values for thestandard model Spirits were muchhigher than the FFV Spirits in bothrounds. All NOx values were wellbelow the Tier 0 levels.
Evaporative Emissions
Average evaporative emissions forthe Dodge Spirits are listed in Tables16 and 17. Figures 12 and 13 graphi-cally illustrate these values. The aver-age evaporative HC for the FFV andstandard gasoline Spirits were wellbelow the standard of 2 g per test.When comparing the FFV Spiritstested on M85 to the same vehiclestested on RFG, both labs showed nosignificant difference between thetwo fuels. The conventional Spiritstested lower than the FFV Spirits oneither fuel with one exception. At Lab3 during Round 1, the conventionalSpirits had higher evaporative emis-sions than the FFV. There was anincrease in evaporative emissionsbetween Rounds 1 and 2 for the FFVtested on both fuels at Lab 1, but thedifference was not significant at the95% confidence level. At Lab 3, theFFV on both fuels showed statistical-ly significant increases in Round 2.
Because of the high variability ofevaporative results, outliers were not deleted from the data sets. Round 2evaporative results for the FFVSpirits tested at Lab 3 increased significantly over Round 1 for bothfuels. This was not consistent withthe results from Lab 1, and warranteda closer look. The evaporative resultsfor the FFV Spirits tested at Lab 1showed only 2 outliers, which had little effect on the final averages. Theevaporative results from the Spiritstested at Lab 3, however, revealedseveral apparent outliers. Most ofthose data points were well above theEPA limit of 2 g per test; the highest
was 6.9 g total. When those outlierswere removed from the data set, theresults were more consistent from labto lab and round to round.
Greenhouse Gases
The average CO2 emissions areshown in Figures 10d and 11d. Bothlabs showed the same patterns, withthe FFV on M85 having the lowestCO2 emissions and the FFV on RFGthe highest. The percent differencebetween the FFV on M85 and onRFG was approximately 7% for bothlabs during both rounds. These werestatistically significant differences atthe 95% confidence level. Average
17
TP-25818
STD-RFGFFV-RFGFFV-M85
Round 1 Round 2
2.5
2.0
1.5
1.0
0.5
0.0
EPA Tier 1 & 0
0258
1812
m
Total Evaporative Hydrocarbon
Eva
pora
tive
Em
issi
ons
(g/T
est)
Figure 12. Evaporative emissions results from theDodge Spirit tested at Lab 1
STD-RFGFFV-RFGFFV-M85
Round 1 Round 2
2.5
2.0
1.5
1.0
0.5
0.0
EPA Tier 1 & 0
0258
1813
m
Total Evaporative Hydrocarbon
Eva
pora
tive
Em
issi
ons
(g/T
est)
Figure 13. Evaporative emissions results From theDodge Spirit tested at Lab 3
CO2 emissions at Lab 1 showed adecrease between Round 1 andRound 2 that was not significant forM85, but was significant for RFG.Average CO2 emissions at Lab 3showed an increase from Round 1 toRound 2 for both fuels that was notstatistically significant at the 95%confidence level.
Although the differences in CH4emissions between fuels were statisti-cally significant for both rounds atboth labs, the measured amountswere all below 0.04 g/mi. The average CH4 values for the FFV tested on M85 were 43% to 52% less than those from the same vehi-cles tested on RFG. Both labs show
increases in CH4 during Round 2 forM85 and RFG. These differencesbetween rounds were not significantfor M85, but they were significant forRFG at Lab 1.
Aldehydes
The average aldehyde emissions forthe Dodge Spirits are shown inFigures 14 and 15. For both labs, theformaldehyde emissions were six toeight times higher in the FFVs testedon M85. As with the Intrepid, this isexpected, because formaldehyde is aprimary decomposition product frommethanol combustion. At Lab 1, thepercent increase for the M85 testswas 764% and 772% for Rounds 1
and 2, respectively. At Lab 3, theincrease was 689% and 538% for thetwo rounds. The average formalde-hyde emissions for the FFV and thestandard model (both tested on RFG)were similar.
Acetaldehyde emissions from theM85 and RFG tests were quite low(all below 0.005 g/mi). The acetalde-hyde emissions were lowest on theFFV tested on M85 for both labs. At Lab 1, the FFV tested on M85 in Round 1 showed a decrease inacetaldehyde emissions of 38% andin Round 2, the decrease was 51%when compared to the FFV tested onRFG. Lab 3 showed similar decreasesfor M85 compared to RFG of 46%and 38% in Rounds 1 and 2, respec-tively. The average acetaldehydeemissions for the standard modelswas higher than the FFV tested onM85, but lower than those from theFFV tested on RFG for both labs.
During this study, full speciation wasperformed on 10 FFV Spirits and 9standard gasoline Spirits. Tables 18and 19 list the average measuredtoxic emissions and the average PWTfor the FFV Dodge Spirits tested atLabs 1 and 3. Aldehyde values are theaverage of the speciated vehiclesonly. Figures 16 and 17 illustrate thedifferences graphically. When com-paring the FFV tested on M85 to thesame vehicles tested on RFG, therewas a significant increase in formal-dehyde emissions, and significantdecreases in acetaldehyde, 1,3-buta-diene, and benzene. Total PWT forLab 1 FFV Spirits tested on M85 was23% lower than the total PWT for theRFG tests. At Lab 3, the differencewas 46% lower for the M85 tests. Allof these differences between fuelswere statistically significant at the95% confidence level. The total PWTfor the gasoline control Spirits wassubstantially lower than the PWT for
18
TP-25818
STD-RFGFFV-RFGFFV-M85
Round 1
0258
1815
m
Aldehyde Emissions
Ald
ehyd
e E
mis
sion
s (m
g/m
i)
FormaldehydeAcetaldehyde Round 1
Round 2Round 2
12
10
8
6
4
2
0
Figure 15. Aldehyde emissions from the Dodge Spirit tested at Lab 3
STD-RFGFFV-RFGFFV-M85
Round 1
0258
1814
m
Aldehyde Emissions
Ald
ehyd
e E
mis
sion
s (m
g/m
i)
FormaldehydeAcetaldehyde Round 1
Round 2Round 2
14
12
10
8
6
4
2
0
Figure 14. Aldehyde emissions from the Dodge Spirit tested at Lab 1
the FFV Spirit tested on either fuel.This trend was consistent among labs.The decrease in PWT appears to be a direct result of the decrease inNMHCE for the gasoline Spiritscompared to the FFV Spirit tested onRFG. The decrease in HC may resultfrom the differences in calibration ofthe vehicle models.
Tables 20 and 21 list the NMOG,OFP, and SR for the Spirits at bothlabs. Figures 18 and 19 graphicallyillustrate these averages. The NMOGemissions from the M85 tests werehigher than those from the RFG testson this vehicle subset, but the OFPand SR were lower. As with theIntrepids, although the NMOG emis-sions were higher, the ozone formedfrom these emissions would tend tobe less than that formed from theRFG emissions. The OFP and SRwere significantly lower for the FFVwhen tested on M85. The FFV Spirits
19
TP-25818
Table 18. Toxic Emissions from the Dodge Spirit Tested at Lab 1
FFV-M85 FFV-RFG
MeasuredPWT
MeasuredPWT
Percent Sig. Fuel
Value (mg/mi) Value (mg/mi)Difference Effect?
HCHO 14.035 0.646 1.687 0.078 731.0% y
CH3CHO 0.252 0.002 0.488 0.004 -50.0% y
1,3-butadiene 0.10 0.10 0.80 0.80 -87.5% y
Benzene 1.042 0.031 4.40 0.132 -90.2% y
Total 15.429 0.779 7.375 1.013 -23.1% y
Table 19. Toxic Emissions from the Dodge Spirit Tested at Lab 3
FFV-M85 FFV-RFG
MeasuredPWT
MeasuredPWT
Percent Sig. Fuel
Value (mg/mi) Value (mg/mi)Difference Effect?
HCHO 9.725 0.447 1.538 0.071 532.3% y
CH3CHO 0.275 0.0022 0.475 0.0038 -42.1% y
1,3-butadiene 0.174 0.174 0.997 0.997 -82.6% y
Benzene 1.695 0.051 6.023 0.181 -71.9% y
Total 11.869 0.674 9.033 1.252 -46.2% y
Table 20. OFP for the Dodge Spirit Tested at Lab 1
FFV- FFV- Percent Sig. FuelM85 RFG Difference Effect?
NMOG (mg/mi) 191.70 151.80 26.3% y
OFP (mg O3/mi) 263.74 380.63 -30.7% y
SR (mg O3/mg NMOG) 1.385 2.908 -52.4% y
Table 21. OFP for the Dodge Spirit Tested at Lab 3
FFV- FFV- Percent Sig. FuelM85 RFG Difference Effect?
NMOG (mg/mi) 242.56 219.18 10.7% n
OFP (mg O3/mi) 332.66 749.19 -55.6% y
SR (mg O3/mg NMOG) 1.387 3.581 -61.9% y
20
TP-25818
tested at Lab 1 on M85 showed a31% reduction in OFP and a 52.4%reduction in SR. Lab 3 values showeda similar finding; OFP was 55.6%lower in the M85 tests and SR was61.8% lower.
Fuel Economy
When tested on M85, the fuel econo-my on the Dodge Spirits was signifi-cantly less than when the samevehicles were tested on gasoline. ForLab 1, there was a decrease of about40% for both rounds. The DodgeSpirits tested at Lab 3 averaged 47%lower in Round 1 and 40% lower inRound 2 when tested on M85. Aswith the Intrepids, the energy equiva-lent fuel economy for the Spirits onM85 was much higher. On an energyequivalent basis, the FFV tested onM85 was 3% to 4% more energy effi-cient than when it was tested on RFGat Lab 1. The Spirits tested at Lab 3during Round 1 were approximately8% less energy efficient in Round 1,but were 4.4% more energy efficientin Round 2. Unlike the Intrepid,Dodge increased the tank size of theFFV Spirit to help offset the differ-ence in energy content of the fuels.The tank on the gasoline controlholds 16 gallons for a range ofapproximately 390 miles. The FFVtank holds 18 gallons for a range ofapproximately 245 miles on M85 and 420 miles on gasoline.
0258
1816
m
Toxic EmissionsP
WT
(m
g/m
i)
0.0
0.2
0.4
0.6
0.8
1.0
1.4
1.2
TotalBenzene1,3 butadieneCH3CHOHCHO
STD-RFGFFV-RFGFFV-M85
Figure 16. PWT emissions from the Dodge Spirit tested at Lab 1
0258
1817
m
Toxic Emissions
PW
T (
mg/
mi)
0.0
0.2
0.4
0.6
0.8
1.0
1.4
1.2
TotalBenzene1,3 butadieneCH3CHOHCHO
STD-RFGFFV-RFGFFV-M85
Figure 17. PWT emissions from the Dodge Spirit tested at Lab 3
0258
1818
m
Ozone Formation and Reactivity
OF
P (
mg
O3/
mi)
SROFP
STD-RFGFFV-RFGFFV-M85
400
350
300
250
200
150
100
50
0
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0
SR
(m
g O
3/m
g N
MO
G)
Figure 18. OFP and SP for the Dodge Spirittested at Lab 1
0258
1819
m
Ozone Formation and Reactivity
OF
P (
mg
O3/
mi)
SROFP
STD-RFGFFV-RFGFFV-M85
800
700
600
500
400
300
200
100
0
4.0
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0
SR
(m
g O
3/m
g N
MO
G)
Figure 19. OFP and SR for the Dodge Spirittested at Lab 3
21
TP-25818
Two vehicle models were tested onethanol during this study: the FFVFord Taurus and the FFV ChevroletLumina. The Taurus was tested at Lab1 over two rounds and the Luminawas tested at Lab 2 over three rounds.Full hydrocarbon speciation was notperformed on the Lumina emissions.The following sections of this reportprovide a detailed discussion of theresults for both vehicles. A briefoverview with a more qualitative
discussion of the results is presentedin this section.
Table 22 and Table 23 provide a summary comparison of the averagemass emissions and the hydrocarbonspeciation, respectively, from E85compared to RFG tests. As in the pre-vious section on methanol, the shad-ed blocks represent a statisticallysignificant difference (at the 95%confidence level) between average
results from the two fuels. A plus signindicates that the average E85 resultswere higher, and a minus sign indi-cates that the average E85 resultswere lower than the RFG results.
The most obvious trend displayed inTable 22 is that the comparison ofnon-regulated emissions (greenhousegases, aldehydes, and fuel economy)tended to be consistent across testrounds and vehicle types, and the differences tended to be statisticallysignificant. Average CO2 and mpgwere consistently lower when testedon E85 compared to RFG. Averagealdehydes (HCHO and CH3CHO)and gasoline equivalent fuel economy(mpeg) were consistently higher fromthe E85 tests compared to the RFGtests. On the other hand, the compari-son of average regulated emissionsresults tended to be less consistent.
Table 23. Summary Comparisonof Average Speciated
Hydrocarbon Results forE85 versus RFG
Table 22. Summary Comparison of Average Emission Resultsfrom E85 versus RFG
ETHANOL VEHICLES
Ford Taurus Chevrolet LuminaLab 1 Lab 2
Round 1 Round 2 Round 1 Round 2 Round 3
Regulated Emissions
NMHCE - + - - +
THC + + - + -
CO + + + + +
NOx - + - - -
Evaporative Emissions
THC - + - - -
Greenhouse Gases
CO2 - - - - -
CH4 + + + + +
Aldehydes
HCHO + + + + +
CH3CHO + + + + +
Fuel Economy
mpg - - - - -
mpeg + + + + +
Regulated Emissions
Evaporative Emissions
Greenhouse Gases
Aldehydes
Fuel Economy
FordTaurus
Air Toxics Lab 1
HCHO +
CH3CHO +
1,3-butadiene -
Benzene -
Total PWT -
Ozone Reactivity
OFP +
SR -
Ozone Reactivity
“+” Indicates results from E85 tests were higher than RFG tests“-” Indicates results from E85 tests were lower than RFG testsHighlighted blocks indicate a significant statistical difference.
22
TP-25818
Results from the FFV Taurus tendedto show higher regulated emissionsfrom E85, but the differences werenot statistically significant. For theLumina, some of the regulated emis-sions were significantly lower on E85 (NOx), some tended to be signifi-cantly higher (CO), and others weremixed from round to round (THC andNMHCE).
Similar to the methanol vehicles, theethanol vehicles are flexible-fueldesigns that are not fully optimizedfor either gasoline or ethanol. Thedifferences in results between vehiclemodels and the lack of clear regulatedemissions differences may result, inpart, from engine hardware choicesand calibrations that must be flexibleto accommodate a wide range of fuelblends.
The results from the detailed specia-tion of hydrocarbon emissions on theTaurus are summarized in Table 23.This table combines the results fromthe two rounds because the differencebetween the two rounds was not sig-nificant. The general trends that areevident in Table 23 include:
• Average aldehyde emissions(HCHO and CH3CHO) tended to be higher from the E85 testscompared to the RFG tests
• Average 1,3-butadiene, benzene,and total PWT emissions tendedto be significantly lower from theE85 tests compared to the RFGtests
• Average OFP tended to be higher,but SR tended to be significantlylower from the E85 tests com-pared to the RFG tests.
This last point was a bit surprisingand deserves additional explanation.Although the OFP (expressed in milligrams of ozone per mile) washigher for the ethanol tests, the SR(expressed in terms of milligrams of
ozone per milligram of non-methaneorganic gases) was lower. This wasthe case because, although the hydro-carbon emissions from the E85 testswere significantly less reactive, thetotal hydrocarbons from this subset of test vehicles were significantlyhigher when tested on E85 comparedto the same vehicles tested on RFG.However, this was not the case for the larger sample of test vehicles. As was mentioned earlier, for all theFord Taurus test vehicles, there wasnot a statistically significant differ-ence between the average NMHCEemissions from E85 compared to thesame vehicles tested on RFG.
FORD TAURUS
The 1995 FFV Ford Taurus (Figure20) tested in this project was actuallydesigned to run on methanol, butGSA obtained approval to operate thevehicles on ethanol. The Taurus is apassenger car equipped with a 3.0 LV6 engine. The FFV Taurus was cer-tified to transitional low emissionvehicle (TLEV) standards and thegasoline model was certified to EPATier 1 levels (Table 1). Two rounds of testing were completed on the FFV Ford Taurus at Lab 1. Therewere 14 FFV Tauruses and 16 gaso-line controls tested in both rounds.Mileage ranges and average odometer
Figure 20. The 1995 E85 Ford Taurus
Arg
onne
Nat
iona
l Lab
orat
ory/
PIX
0Table 24. Odometer Readings for the Ford Taurus
FFV Gasoline
Round 1 2 1 2
No. vehicles tested 14 14 16 16
Odometer (miles)
Average 5,069 16,095 4,859 14,201
Maximum 10,253 29,184 12,822 31,503
Minimum 3,067 8,158 3,027 8,055
Odometer (miles)
readings for the Taurus are listed inTable 24. The complete data set forthe Taurus is found in Appendix A.
Regulated Emissions
Table 25 shows the average emissionsresults for the FFV Ford Taurus.Figure 21 illustrates the average regu-lated emissions and CO2 values. In general, when comparing the regulat-ed emissions from the FFV Taurustested on E85 to the same vehiclestested on RFG, there was no signifi-cant difference between fuels. InRound 1, the emissions levels fromthe FFV on E85 and RFG were simi-lar to the conventional Taurus testedon RFG. In Round 2, the FFV on E85had slightly higher values for all threeregulated compounds.
When comparing the NMHCE emis-sions for the Taurus (Figure 21a),
there was not a significant differencebetween the FFV on either fuel andthe conventional model for Round 1.In Round 2, the NMHCE emissionsfor the FFV on E85 were 12.5% high-er than on RFG, but this differencewas not significant at the 95% confi-dence level. The average for the stan-dard model in Round 2 was lowerthan the FFV on both fuels. All thesevalues were below the Tier 1 limit of0.25 g/mi. The FFV Taurus is certi-fied to the TLEV emissions standard,which is written in terms of NMOG(see explanation on page 1). AlthoughNMOG was not evaluated for theentire set of vehicles, it appears thatthe FFV in Round 2 exceeded theTLEV standard.
When comparing the average COemissions for the Taurus (Figure21b), the FFV on E85 had slightlyhigher values than the same vehicles
tested on RFG in both rounds, but thedifference was not statistically signif-icant at the 95% confidence level. In Round 1, the increase for the FFVtested on E85 was 8% higher and inRound 2 the average was approxi-mately 2% higher. Once again, allaverages were well below the Tier 1and TLEV limit of 3.4 g/mi.
NOx emissions for the Taurus areshown in Figure 21c. When compar-ing the FFV on E85 to the same vehi-cles on RFG, there was a decrease in average NOx in Round 1, but anincrease in Round 2. Neither of thesedifferences was statistically signifi-cant, and all values remained wellbelow the Tier 1 and TLEV limit of0.4 g/mi. The averages for all threeregulated compounds showed signifi-cant increases from Round 1 toRound 2, but all were below the Tier1 certification limit.
23
TP-25818
Table 25. Average Emissions Results from the Ford Taurus
Round 1 Round 2
FFV- FFV- Percent Sig. Fuel FFV- FFV- Percent Sig. FuelE85 RFG Difference Effect? E85 RFG Difference Effect?
Regulated Emissions (g/mi)
NMHCE 0.089 0.091 -2.2% n 0.163 0.144 12.5% n
THC 0.103 0.101 2.4% n 0.184 0.156 17.9% y
CO 1.162 1.075 8.1% n 1.522 1.486 2.4% n
NOx 0.104 0.125 -16.8% n 0.183 0.178 2.8% n
Evaporative Emissions (g)
Total Evaporative 0.328 0.332 -1.2% n 0.362 0.319 13.5% n
Greenhouse Gases (g/mi)
CO2 405.5 426.5 -4.9% y 398.5 422.9 -5.8% y
CH4 0.025 0.012 107.4% y 0.035 0.016 122.9% y
Aldehydes (mg/mi)
HCHO 2.03 1.29 57.4% y 2.96 1.54 92.2% y
CH3CHO 9.0 0.37 2332.4% y 13.6 0.37 3575.7% y
Fuel Economy
mpg 15.22 20.4 -25.4% y 15.46 20.49 -24.6% y
mpeg 20.82 20.4 2.1% y 21.15 20.49 3.2% y
Regulated Emissions (g/mi)
Fuel Economy
Aldehydes (mg/mi)
Greenhouse Gases (g/mi)
Evaporative Emissions (g/test)
24
TP-25818
Evaporative Emissions
Figure 22 shows the comparison of the average evaporativeemissions for the Taurus. Evaporative emissions for the FFVon both fuels and the conventional Tauruses were well belowthe EPA limit of 2 g of hydrocarbon per test. When comparing the evaporative emissions for the FFV Taurus,there was not a significant difference in the FFV tested onE85 and the same vehicles tested on RFG. The conventionalTaurus had lower average evaporative emissions than the FFVon either fuel. The round-to-round comparison for the FFVshowed a small increase for the E85 tests, and a smalldecrease for the RFG tests. Neither of these differences wasstatistically significant at the 95% confidence level.
Greenhouse Gases
Carbon dioxide emissions for the Taurus are shown inFigure 21d. When comparing the FFV on E85 to the samevehicles tested on RFG, the E85 CO2 emissions were approximately 5% lower in both rounds. This difference was statistically significant at the 95% confidence level. The conventional Taurus tested on RFG showed very similar values to the FFV tested on RFG. There was a small decrease in CO2 emissions for Round 2 that was statistically significant for both fuels.
Methane emissions for the FFV tested on E85 were signifi-cantly higher than when the same vehicles were tested onRFG. The average CH4 emissions were 107% higher inRound 1 and 123% higher in Round 2. It is important to note,however, that the values for both fuels are very small (0.012to 0.035 g/mi). There was a small increase in CH4 emissionsfrom Round 1 to Round 2 that was significant for both fuels.
0
100
200
300
400
500
STD-RFGFFV-RFGFFV-E85
21a: Non-Methane Hydrocarbon Equivalent
21b: Carbon Monoxide
21d: Carbon Dioxide
21c: Oxides of Nitrogen
NM
HC
E E
mis
sion
s (g
/mi)
CO
2 E
mis
sion
s (g
/mi)
NO
x E
mis
sion
s (g
/mi)
CO
Em
issi
ons
(g/m
i)
Round 1 Round 2
Round 1 Round 2
Round 1 Round 2
Round 1 Round 2
0.30
0.25
0.20
0.15
0.10
0.05
0.00
4.0
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0
0.45
0.40
0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
EPA Tier 1 & TLEV
EPA Tier 1 & TLEV
EPA Tier 1
0258
1821
m
TLEV (NMOG)
Figure 21. Emissions results from theFord Taurus
STD-RFGFFV-RFGFFV-E85
Round 1 Round 2
2.5
2.0
1.5
1.0
0.5
0
0258
1822
mTotal Evaporative Hydrocarbon
Eva
pora
tive
Em
issi
ons
(g/T
est)
EPA Tier 1 & 0
Figure 22. Evaporative emissions results from theFord Taurus
Aldehydes
Aldehyde emissions for the FordTaurus are shown in Figure 23.Formaldehyde emissions were higherfor the FFV tested on E85 in bothrounds. The percent differencebetween the FFV tested on E85 andthe same vehicles tested on RFG was 57% for Round 1 and 92.2% forRound 2. Acetaldehyde is a primarydecomposition product from ethanolcombustion; therefore, the muchhigher values were expected when the vehicle was operating on E85. Thepercent increase in the FFV acetalde-hyde emissions when tested on E85was 2,332% for Round 1 and 3,575%for Round 2. The acetaldehyde levelsfor RFG were very low—less than0.5 mg/mi. Although both fuels showincreases in aldehyde emissions from Round 1 to Round 2, only theincreases for the E85 tests were sta-tistically significant.
During this project, full hydrocarbonspeciation was performed on six ofthe FFVs and five of the standardTauruses. Table 26 summarizes theaverage measured toxic emissionsand the PWT results for the Taurus.When comparing the FFV tested onE85 to the same vehicles tested onRFG, there were significant increasesin aldehyde emissions and significantdecreases in 1,3-butadiene and
benzene. Figure 24 shows this differ-ence graphically. Although the totalmeasured toxics were higher, thepotency weighted values were signifi-cantly lower for the E85 tests. TotalPWT for the FFV tested on E85 were44% lower than the same vehiclestested on RFG. Although acetalde-hyde is the highest measured valuefor E85 tests, it is the least toxic ofthe four. The conventional model test-ed on RFG showed results similar tothe FFV tested on RFG.
Table 27 and Figure 25 show theNMOG, OFP, and SR results for theTaurus. The OFP for the FFV testedon E85 was significantly higher(19%) than the same vehicles testedon RFG, but the SR was significantlylower (approximately 38%) for theE85 tests. The OFP for the FFV
tested on E85 was higher than thesame vehicles tested on RFG becausethe total HC from this subset of vehi-cles were substantially higher. Thelower SR indicates that the FFV test-ed on E85 was less reactive per unitmass.
Fuel Economy
Table 25 gives the actual and equiva-lent fuel economy for the FFV FordTaurus. Average fuel economy for the FFV Taurus on E85 was approxi-mately 15 mpg. The average whentested on RFG was approximately25% higher, at 20 mpg. As withmethanol, E85 has a lower volumetricenergy content than RFG. The volu-metric energy content for E85(81,825 Btu/gal) is approximately73% of RFG (111,960 Btu/gal). This
25
TP-25818
STD-RFGFFV-RFGFFV-E85
Round 1
0258
1823
m
Aldehyde Emissions
Ald
ehyd
e E
mis
sion
s (m
g/m
i)
0
5
10
15
20
FormaldehydeAcetaldehyde Round 1
Round 2Round 2
Figure 23. Aldehyde emissions from the Ford Taurus
Table 26. Toxic Emissions from the Ford Taurus
FFV-E85 FFV-RFG
MeasuredPWT
MeasuredPWT
Percent Sig. Fuel
Value (mg/mi) Value (mg/mi)Difference Effect?
HCHO 2.223 0.102 1.30 0.06 70.9% y
CH3CHO 9.854 0.079 0.275 0.002 3,490.9% y
1,3-butadiene 0.175 0.175 0.544 0.544 -67.8% y
Benzene 1.013 0.03 2.863 0.086 -65.1% y
Total 13.265 0.386 4.982 0.692 -44.2% y
26
TP-25818
means that it takes 1.3 gallons of E85to travel the same distance as 1 gallonof gasoline. On an energy equivalentbasis, the FFV Taurus was 2% to 3%more energy efficient when tested onE85. Like the Spirit, the fuel tank ofthe FFV Taurus was increased toaccount for the differing energy con-tent of the fuel. The gasoline Taurushas a 16-gallon tank for a range ofapproximately 326 miles. The FFVhas a tank that holds 20.4 gallons for
a range of 313 miles on E85 and 417miles on gasoline.
CHEVROLET LUMINA
The 1993 FFV Chevrolet Lumina(shown in Figure 26) is a passengercar equipped with a 3.1 L V6 enginewith multi-point fuel injection. TheLumina was certified to EPA federalTier 0 emissions levels. This reportcovers the three rounds of testingcompleted on the Chevrolet Lumina
at Lab 2. Ten FFV Luminas and 11gasoline controls were tested in all 3rounds. Mileage ranges and averageodometer readings for the Luminasare listed in Table 28. Lab 1 tested alimited number of FFV Luminas dur-ing Round 1 only. The results forthose tests were reported in anotherpublication and are not included inthis paper.9 Hydrocarbon speciationwas not performed on the vehiclesincluded in this analysis. The entiredata set is located in Appendix A.
Regulated Emissions
The average emissions results for theLumina are listed in Table 29. Theregulated and CO2 emissions for theFFV Lumina are shown in Figure 27.In general, when comparing the FFVtested on E85 to the same vehiclestested on RFG, there tended to be aslight decrease in NMHCE, a largerdecrease in NOx, and an increase inCO emissions. The average regulatedemissions for the FFV Lumina wereall well below the Tier 0 standard, aswell as the more stringent Tier 1 stan-dard, shown here for reference. Theregulated emissions for the gasolinemodel did not follow the same trend.NMHCE and NOx emissions for thegasoline Lumina were below theTier 0 levels, but CO emissions wereover the limit for all 3 rounds.
Although NMHCE values for theFFV tested on E85 were lower thanthe RFG tests in Rounds 1 and 2 (seeFigure 27a), the difference was notsignificant in Round 2. There was nosignificant difference in NMHCEemissions between the two fuels forRound 3. All the values for the FFVLumina were below the EPA Tier 1limit of 0.25 g/mi. Round-to-roundcomparison for the E85 tests showedan increase in NMHCE over time thatwas statistically significant. Thesmaller increase in NMHCE for theRFG tests on the FFV was not statis-tically significant at the 95% confi-dence level. The standard gasoline
0258
1824
m
Toxic EmissionsP
WT
(m
g/m
i)
STD-RFGFFV-RFGFFV-E85
TotalBenzene1,3 butadieneCH3CHOHCHO
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
Figure 24. PWT emissions from the Ford Taurus
Table 27. OFP for the Ford Taurus
FFV- FFV- Percent Sig. FuelE85 RFG Difference Effect?
NMOG (mg/mi) 171.97 92.30 86.3% y
OFP (mg O3/mi) 377.58 318.06 18.7% y
SR (mg O3/mg NMOG) 2.215 3.57 -38.0% y
0258
1825
m
Ozone Formation and Reactivity
OF
P (
mg
O3/
mi)
SROFP
STD-RFGFFV-RFGFFV-E85
400
350
300
250
200
150
100
50
0
4.0
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0
SR
(m
g O
3/m
g N
MO
G)
Figure 25. OFP and SR for the Ford Taurus
27
TP-25818
model showed a small but significantincrease in each round.
CO emissions follow a different trendthan NMHCE (Figure 27b). In allthree rounds, the FFV tested on E85showed higher CO emissions thanwhen the same vehicles were testedon RFG. The percent increases were7.5% for Round 1, 33% for Round 2,and 22% for Round 3. This increasewas statistically significant forRounds 2 and 3, but not for Round 1.The standard gasoline model testedsignificantly higher than the FFV oneither fuel. The average CO for theFFV tested on E85 and RFG werebelow the Tier 0 emissions standard,but the gasoline Lumina exceeded the limit for all three rounds. TheRound 3 average for the gasolineLuminas was approximately 50%higher than the 3.4 g/mi standard.
NOx emissions for the FFV tested onE85 were significantly lower thanthose from the same vehicles testedon RFG for all 3 rounds. There was a decrease of 40%, 37%, and 34% forRounds 1, 2, and 3, respectively. Aswith the CO emissions, NOx averagesfor the standard model were muchhigher than the averages for the FFV.Both the FFV tested on E85 and RFGand the standard model tested onRFG had NOx levels below the Tier 0standard of 1 g/mi. The FFV on eachfuel was also below the more strin-gent Tier 1 level.
Evaporative Emissions
Evaporative emissions for the FFVLumina are listed in Table 29 andgraphically illustrated in Figure 28.When comparing the average evapo-rative emissions for the FFV tested onE85 to the averages for the samevehicles tested on RFG, there was asmall reduction in evaporative emissions for all three rounds. How-ever, only the reduction for Round 2was statistically significant. The con-ventional Lumina tested higher than
the FFV on both fuels. All averages were well below the 2 g per test stan-dard. Round-to-round differencesshow small increases over time forthe FFV on both fuels. These differ-ences tended not to be significant atthe 95% confidence level.
Greenhouse Gases
Figure 27d shows the average CO2emissions levels for the Lumina. TheCO2 average for the FFV tested onE85 was approximately 6% lowerthan when tested on RFG in all threerounds. These differences were allstatistically significant at the 95%confidence level. CO2 emissions forthe standard Lumina tested on RFG
were lower than the FFV on RFG.Round-to-round comparisons for theFFV tested on E85 and RFG showedsignificant decreases in CO2 duringRound 2 and significant increases inRound 3. This held true for both theE85 and RFG tests on the FFVLumina.
Although emissions of CH4 for theFFV are small (less than 0.08 g/mi),the results for the tests on E85 aresignificantly higher than those fromthe RFG tests. Round-to-round com-parisons of CH4 emissions for theE85 tests show a small but significantincrease in Round 2 and a small butsignificant decrease in Round 3. The FFV tests with RFG show no
Figure 26. The 1993 E85 Chevrolet Lumina
War
ren
Gre
tz, N
RE
L/P
IX02
479
Table 28. Odometer Readings for the Chevrolet Lumina
FFV Gasoline
Round 1 2 3 1 2 3
No. vehicles tested 10 10 10 11 11 11
Odometer (miles)
Average 10,111 22,568 30,883 6,344 12,434 19,403
Maximum 12,409 35,842 42,538 10,713 18,970 37,902
Minimum 8,218 12,991 19,700 2,903 6,826 11,365
Odometer (miles)
28
TP-25818
significant difference betweenrounds. Average CH4 values for thegasoline Lumina also show no signif-icant difference between rounds.
Aldehydes
Aldehyde emissions for the Luminaare shown in Figure 29. Formalde-hyde emissions from the FFV testedon E85 were significantly higher than those from the same vehiclestested on RFG. In Round 1, formalde-hyde emissions from the FFV on E85were 50% higher than those fromRFG, Round 2 results were 42%higher, and Round 3 results were60% higher. Formaldehyde emissionsfor the standard Lumina were higherthan those from the FFV on RFG,but lower than those from the FFV on E85. The average acetaldehyde (a primary decomposition product ofethanol combustion) emissions for
the FFV tested on E85 were 2,483%,2,030%, and 2,469% higher thanthose from the same vehicles testedon RFG, respectively. The differencesbetween rounds were not statisticallysignificant.
Because full hydrocarbon speciationwas not performed on the Luminasduring this project, PWT and OFPwere not evaluated.
Fuel Economy
Table 29 gives actual and equivalentfuel economy for the FFV Lumina.Actual fuel economy for the Luminatested on E85 over the 3 roundsranged from 13.5 to 14 mpg. Thiswas 25% to 26% lower than the samevehicles when tested on RFG. Thestandard models tested slightly higher
than the FFV on RFG. Because of the difference in energy content betweenE85 and RFG, gasoline energy equiv-alent fuel economy was calculated forthe E85 tests. The energy equivalentfuel economy for the E85 tests rangedfrom 18.6 mpeg to 19.3 mpeg. Takingthis into account, the fuel economyfor the FFV tested on E85 was 1.3%to 2.6% higher than when tested onRFG. The fuel tanks for the gasolineand FFV Lumina are similar in size.The gasoline Lumina has a tank thatholds 17.1 gallons for a range ofapproximately 330 miles. The FFVLumina has a 16.5-gallon fuel tankfor a range of 228 miles on E85 and306 miles on gasoline.
Table 29. Average Emissions Results from the Chevrolet Lumina
Round 1 Round 2 Round 3
FFV FFV Percent Sig. Fuel FFV FFV Percent Sig. Fuel FFV FFV Percent Sig. Fuel
NMHCE 0.087 0.102 -14.7% y 0.105 0.109 -3.7% n 0.118 0.117 0.8% n
THC 0.106 0.125 -14.5% y 0.140 0.134 4.5% n 0.141 0.1414 -0.3% n
CO 2.22 2.07 7.5% n 3.08 2.32 32.9% y 2.84 2.33 21.3% y
NOx 0.156 0.261 -40.4% y 0.206 0.329 -37.4% y 0.233 0.352 -34.1% y
Evaporative Emissions (g)
Total Evaporative 0.153 0.162 -5.6% n 0.159 0.242 -34.3% y 0.163 0.207 -21.3% n
Greenhouse Gases (g/mi)
CO2 454.2 485.9 -6.5% y 435.9 462.5 -5.7% y 443.9 468.9 -5.3% y
CH4 0.056 0.028 100% y 0.074 0.031 141.6% y 0.066 0.031 110.6% y
Aldehyde Emissions (mg/mi)
HCHO 6.98 4.66 49.8% y 5.56 3.92 41.8% y 5.38 3.36 60.1% y
CH3CHO 18.08 0.73 2482.9% y 17.04 0.78 2030% y 17.98 0.70 2468.6% y
Fuel Economy
mpg 13.57 18.09 -25.0% y 14.1 18.99 -25.8% y 13.86 18.72 -26% y
mpeg 18.57 18.09 2.6% y 19.29 18.99 1.6% y 18.96 18.72 1.3% y
Regulated Emissions (g/mi)
Evaporative Emissions (g/test)
Greenhouse Gases (g/mi)
Aldehydes (mg/mi)
Fuel Economy
29
STD-RFGFFV-RFGFFV-E85
Round 1 Round 2 Round 3
2.5
2.0
1.5
1.0
0.5
0.0
EPA Tier 1 & 0
0258
1828
m
Total Evaporative Hydrocarbon
Eva
pora
tive
Em
issi
ons
(g/T
est)
Figure 28. Evaporative emissions results from theChevrolet Lumina
STD-RFGFFV-RFGFFV-E85
27a: Non-Methane Hydrocarbon Equivalent
27b: Carbon Monoxide
27d: Carbon Dioxide
27c: Oxides of Nitrogen
NM
HC
E E
mis
sion
s (g
/mi)
CO
2 E
mis
sion
s (g
/mi)
NO
x E
mis
sion
s (g
/mi)
CO
Em
issi
ons
(g/m
i)
0.45
0.40
0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
6
5
4
3
2
1
0
1.2
1.0
0.8
0.6
0.4
0.2
0.0
EPA Tier 0
EPA Tier 0
0258
1827
m
600
500
400
300
200
100
0Round 2 Round 3Round 1
EPA Tier 1
EPA Tier 1
Round 2 Round 3Round 1
Round 2 Round 3Round 1
Round 2 Round 3Round 1
EPATier 1 & 0
Figure 27. Emissions results from theChevrolet Lumina
STD-RFGFFV-RFGFFV-E85
Round 1
0258
1829
m
Aldehyde Emissions
Ald
ehyd
e E
mis
sion
s (m
g/m
i)
0
5
10
15
20
FormaldehydeAcetaldehyde Round 1
Round 2Round 2
Round 3Round 3
Figure 29. Aldehyde emissions from theChevrolet Lumina
30
TP-25818
Table 30. Summary Comparison of Average Emissions Results from CNG versus RFG
Two different CNG vehicle modelswere tested during this study. Thesemodels include the Dodge B250 vanand the Dodge Caravan minivan.Both vans are dedicated natural gasvehicles, which means they aredesigned to operate on CNG only. To make fuel-to-fuel emissions com-parisons, it was necessary to testclosely matched gasoline vehicles.The AFV and the gasoline models are
both classified by the EPA as "heavylight-duty vehicles." See Table 2 onpage 2 for the EPA intermediate use-ful life standards for the vans.
As with the other fuels, an overviewof the general trends is presented first and then the detailed results foreach of the test vehicles are presentedin subsequent sections. Table 30 and Table 31 show summary
comparisons of the average CNGemissions compared to the averageRFG emissions. As in the sections onmethanol and ethanol, the shadedblocks indicate differences betweenthe averages that were statisticallysignificant (at the 95% confidencelevel). Plus signs indicate that theaverage CNG emissions were higherthan the average RFG emissions,and the minus signs indicate that the
“+” Indicates results from CNG tests were higher than RFG tests“-” Indicates results from CNG tests were lower than RFG testsHighlighted blocks indicate a significant statistical difference.
31
TP-25818
average CNG emissions were lowerthan the average RFG emissions.Table 30 includes mass emissionsresults from the B250s that were test-ed over multiple rounds at all 3 labs,and more limited results from theCaravans that were only tested duringa single round at Lab 1. Table 31includes results from detailed hydro-carbon speciations of emissions fromthe B250 tests performed at Labs 1and 3.
Table 30 shows that there tend to bestatistically significant differencesbetween the average emissions fromthe CNG and RFG B250 vans, andthat these results tend to be fairlyconsistent from lab to lab and fromround to round. The average NMHC,CO, CO2, CH3CHO, and fuel econo-my results were significantly lowerfrom the CNG tests than the RFGtests for all three labs and in all threetest rounds. Average CH4 emissionswere consistently higher from CNGthan from RFG. NOx and "evapora-tive" hydrocarbons tended to be lowerfrom the CNG tests, but in somecases the differences were not significant, and in one case (Lab 2,Round 3) the average NOx emissionswere higher from CNG. The evapora-
tive emissions test is a measure of thehydrocarbons emanating from two,1-hour soaks in a sealed room withthe engine off. Dedicated gaseousfuel vehicles typically do not haveevaporative control systems becausethe fuel system is said to be sealedunder pressure. Nevertheless, hydro-carbons (mostly methane) may stillbe found emanating from gaseousfuel vehicles. In all cases, the averageTHC measured during the evapora-tive tests were lower than from theRFG tests, but in a few cases the difference was not statistically significant.
Results from a subset of the vehicles(on which detailed speciation of thehydrocarbon emissions was per-formed) are summarized in Table 31.The general trend of these results wasvery consistent for the 2 labs wherethis analysis was performed. At bothlabs, the CNG emissions had loweraverage values of the four toxic emis-sions that were quantified, had lowerPWT, lower average OFP, and lower
average SR. These differences wereall deemed statistically significant atthe 95% confidence level.
DODGE B250 VAN
The CNG and the gasoline DodgeB250 vans are full-size passengervans equipped with a 5.2 L V8engine. Both models have multi-pointfuel injection and 4 speed automatictransmissions. The gasoline modelwas certified to EPA Tier 0 standards.The CNG model had received a waiv-er on emissions certification. Thevehicles tested in this project were amixture of 1992 and 1994 model yearvans. Figure 30 shows the 1992model year CNG Dodge van.
The gasoline model has a 35-gallon fuel tank, and the CNG model wasequipped with 3 or 4 fuel cylindersmounted under the vehicle. The 3-cylinder configuration gives a capaci-ty of 11.1 equivalent gallons and the4-cylinder configuration gives acapacity of 15.7 equivalent gallons.
Table 31. SummaryComparison of AverageSpeciated Hydrocarbon
Results from CNG versus RFG
Dodge B250 Van
Air Toxics Lab 1 Lab 3
HCHO - -
CH3CHO - -
1,3-butadiene - -
Benzene - -
Total PWT - -
Ozone Reactivity
OFP - -
SR - -
Ozone Reactivity
Figure 30. The 1992 CNG Dodge B250 van
War
ren
Gre
tz, N
RE
L/P
IX02
485
Two rounds of testing were complet-ed on the Dodge B250 vans at Lab 1,and three rounds were completed atLabs 2 and 3. At Lab 1, 10 CNG vansand 8 gasoline controls were tested inboth rounds. The vans tested at Lab 2in all 3 rounds totaled 12 CNG vehi-cles and 13 gasoline vehicles. AtLab 3, 15 CNG vehicles and 14 gaso-line models were tested in all 3rounds. Mileage ranges and averageodometer readings for the B250 vanstested at the three labs are listed in
Tables 32, 33, and 34. All data for the Dodge B250 vans can be found inAppendix A.
Regulated Emissions
Table 35 lists the average emissionsvalues for the B250 vans tested atLab 1 along with the percent differ-ence and an indication of whether the differences are statistically signif-icant at the 95% confidence level.Table 36 lists the values for the vans
tested at Lab 2 and Table 37 for thosetested at Lab 3. Figures 31–33 showthe graphical representation of theaverage regulated and CO2 exhaustemissions for the Dodge B250 vanstested at Labs 1, 2, and 3, respective-ly. Regulated emissions results forboth the CNG and gasoline vans werewell below the Tier 0 standard. TheCNG vans, although not certified,tended to be below the more stringentTier 1 standard.
Average NMHC emissions are shownin Figures 31a, 32a, and 33a for Labs1, 2, and 3, respectively. All NMHCvalues were not only below the Tier 0full useful life standard of 0.67 g/mi,but were also below the more strin-gent Tier 1 full useful life standard of0.4 g/mi. NMHC emissions for theB250 vans were significantly lower inthe CNG model for all 3 labs. Lab 1showed the largest percent differenceat approximately 94% lower for theCNG model during both rounds.Lab 2 showed a 76% to 85% decreasein NMHC for the CNG model. Lab 3showed a decrease in NMHC of 81%in Round 1, 41% in Round 2, and45% in Round 3. The higher percent-age for Lab 1 could be due partiallyto the discrepancy in odometer read-ing between the CNG and gasolinemodels. The average odometer for theCNG vans was 5,412 miles inRound 1 and 12,154 miles in Round2. In contrast, the average odometerfor the gasoline model was 39,749miles and 45,755 miles for Rounds 1and 2, respectively. All of the vanstested at Lab 1 were from the 1994model year. Round-to-round compar-isons at Lab 1 showed a significantincrease in NMHC for the RFG testsin Round 2, but no significant differ-ence between rounds for the CNGtests at Lab 1. At Lab 2, the CNGtests showed a significant increasefrom Round 1 to Round 2, and theRFG tests increased significantlyfrom Round 2 to Round 3. Lab 3CNG tests increased significantly
32
TP-25818
Table 33. Odometer Readings for the Dodge B250 Van Tested at Lab 2
CNG Gasoline
Round 1 2 3 1 2 3
No. vehicles tested 12 12 12 13 13 13
Odometer (miles)
Average 7,246 11,778 15,633 11,429 18,327 27,037
Maximum 15,026 24,824 30,050 22,195 32,165 57,099
Minimum 3,951 5,377 6,243 3,527 3834 9,363
Odometer (miles)
Table 34. Odometer Readings for the Dodge B250 Van Tested at Lab 3
CNG Gasoline
Round 1 2 3 1 2 3
No. vehicles tested 15 15 15 14 14 14
Odometer (miles)
Average 6,978 12,051 18,515 13,321 17,338 19,670
Maximum 22,245 29,585 45,147 30,493 36,629 38,485
Minimum 2,121 3,455 6,782 3,875 5,210 6,720
Odometer (miles)
Table 32. Odometer Readings for the Dodge B250 Van Tested at Lab 1
CNG Gasoline
Round 1 2 1 2
No. vehicles tested 10 10 10 8
Odometer (miles)
Average 5,412 12,154 39,749 45,755
Maximum 6,611 15,527 107,350 60,261
Minimum 3,455 8,047 23,991 33,050
Odometer (miles)
33
TP-25818
Table 35. Average Emissions Results from the Dodge B250 Van Tested at Lab 1
Round 1 Round 2
CNGSTD- Percent Sig. Fuel
CNGSTD- Percent Sig. Fuel
RFG Difference Effect? RFG Difference Effect?
Regulated Emissions (g/mi)
NMHC 0.018 0.323 -94.3% y 0.022 0.362 -93.8% y
THC 0.288 0.387 -25.7% y 0.383 0.431 -11.1% y
CO 0.651 5.615 -88.4% y 0.734 6.846 -89.3% y
NOx 0.287 0.858 -66.6% y 0.521 0.888 -41.3% y
Evaporative Emissions (g)
Total Evaporative 0.0684 0.6999 -90.2% y 0.4501 0.8749 -48.5% y
Greenhouse Gases (g/mi)
CO2 539.16 637.87 -15.5% y 526.54 617.84 -14.8% y
CH4 0.27 0.078 244.8% y 0.362 0.085 325.2% y
Aldehydes (mg/mi)
HCHO 2.08 6.45 -67.7% y 2.31 6.13 -62.3% y
CH3CHO 0.17 1.25 -86.7% y 0.26 1.38 -80.9% y
Fuel Economy 12.97 13.49 -3.9% y 12.5 13.73 -9.0% y
Aldehydes (mg/mi)
Greenhouse Gases (g/mi)
Evaporative Emissions (g/test)
Table 36. Average Emissions Results from the Dodge B250 Van Tested at Lab 2
Round 1 Round 2 Round 3
CNG STD Percent Sig. Fuel CNG STD Percent Sig. Fuel CNG STD Percent Sig. Fuel
NMHC 0.045 0.306 -85.4% y 0.071 0.325 -78.1% y 0.083 0.352 -76.3% y
THC 0.759 0.367 106.6% y 1.017 0.387 163.2% y 1.273 0.416 205.7% y
CO 1.747 5.994 -70.9% y 1.604 5.954 -73.1% y 1.393 7.079 -80.3% y
NOx 0.547 0.762 -28.3% n 0.757 0.810 -6.5% n 1.290 0.853 51.2% y
Evaporative Emissions (g)
Total Evaporative 0.406 0.621 -34.7% y 0.317 0.803 -60.5% y 0.267 1.060 -74.9% y
Greenhouse Gases (g/mi)
CO2 559.5 667.9 -16.2% y 547.2 644.5 -15.1% y 548.1 644.4 -14.9% y
CH4 0.716 0.075 853.7% y 0.94 0.077 1,127.7% y 1.192 0.080 1,386.7% y
Aldehyde Emissions (mg/mi)
HCHO 8.14 7.41 9.9% n 6.09 6.43 -5.4% n 8.79 5.79 51.9% y
CH3CHO 0.37 1.71 -78.3% y 0.37 1.56 -76.3% y 0.50 1.96 -74.6% y
Fuel Economy 11.64 13.08 -11.0% y 11.89 13.45 -11.6% y 11.86 13.51 -12.2% y
Regulated Emissions (g/mi)
Evaporative Emissions (g/test)
Greenhouse Gases (g/mi)
Aldehydes (mg/mi)
34
TP-25818
from Round 1 to 2, but the RFG testsdid not show a significant differencebetween the rounds.
The average CO emissions for theB250 vans tested at the 3 labs areshown in Figures 31b, 32b, and 33b.Average results were below the Tier 0full useful life standard for CO.Although the CNG vans were not cer-tified, the average CO emissions forthese vehicles were below the morestringent Tier 1 levels at all 3 labs.The average CO emissions from theCNG vehicles at Lab 1 were 88% and89% lower than the RFG emissionsfor Rounds 1 and 2, respectively. Lab2 showed a decrease in CO for theCNG vans of 71% in Round 1, 73% in Round 2, and 80% in Round 3. Lab 3 showed a decrease of 35.5% inRound 1, 48% in Round 2, and 53%in Round 3. Round-to-round compar-isons of CO emissions at Lab 1 showa significant increase in Round 2 forthe RFG tests, but no significant dif-
ference between rounds for the CNGtests. The only significant increase inCO emissions at Lab 2 was for theRFG tests from Rounds 2 to 3. TheCO emissions for the CNG vans atLab 2 showed a slight downwardtrend that was not significant at the95% confidence level. This sametrend was seen with the CNG vanstested at Lab 3. The RFG vans atLab 3 showed a significant COincrease in Round 2 and a significantdecrease in Round 3.
Average NOx emissions for theB250s tested at the 3 labs are shownin Figures 31c, 32c, and 33c. Theaverage NOx emissions for the B250vans were below the federal Tier 0standard of 1.7 g/mi. The averageNOx emissions for the CNG vanswere lower than that of the gasolinemodels except for the third round atLab 2. At Lab 1, the CNG emissionswere 66.5% lower in Round 1 and41% lower in Round 2. Lab 3 also
followed this trend; Round 1 CNGemissions were 45.5% lower,Round 2 were 31% lower, andRound 3 were 10.7% lower. The aver-age NOx emissions for both van mod-els were below the Tier 0 as well asthe more stringent Tier 1 limits. Theexception to this trend was seen atLab 2. Rounds 1 and 2 showed adecrease in NOx emissions for theCNG model, but this difference wasnot significant. In Round 3, the CNGaverage for NOx was 51% higher thanthe average for the gasoline model.This was mainly caused by one high-emitting van, which was not taggedas an outlier. During Bag 3 of theFTP on this van, the check enginelight came on, indicating a possibleproblem. If this value is removed, theCNG average is lowered to 0.997g/mi, but this is still higher than thegasoline average by 16.9%. Round-to-round comparisons of NOx emis-sions at all 3 labs showed anincreasing trend for the CNG vans
Table 37. Average Emissions Results from the Dodge B250 Van Tested at Lab 3
Round 1 Round 2 Round 3
CNG STD Percent Sig. Fuel CNG STD Percent Sig. Fuel CNG STD Percent Sig. Fuel
NMHC 0.049 0.257 -80.9% y 0.179 0.304 -41.1% y 0.170 0.310 -45.2% y
THC 0.710 0.311 128.1% y 0.741 0.353 109.9% y 0.797 0.365 118.4% y
CO 2.563 3.974 -35.5% y 2.458 4.713 -47.9% y 1.828 3.877 -52.9% y
NOx 0.379 0.695 -45.5% y 0.506 0.738 -31.4% y 0.709 0.794 -10.7% n
Evaporative Emissions (g)
Total Evaporative 0.571 1.041 -45.2% n 0.524 1.39 -62.3% y 0.764 1.35 -43.4% y
Greenhouse Gases (g/mi)
CO2 502.3 616.0 -18.5% y 494.1 604.8 -18.3% y 488.1 606.25 -19.5% y
CH4 0.66 0.054 1,134.5% y 0.557 0.049 1,030.6% y 0.617 0.055 1,026.3% y
Aldehyde Emissions (mg/mi)
HCHO 1.68 3.62 -53.6% y 1.82 3.85 -52.7% y 1.86 3.87 -51.9% y
CH3CHO 0.089 1.03 -91.3% y 0.196 1.06 -81.4% y 0.2 1.08 -81.4% y
Fuel Economy 13.32 13.93 -4.4% y 13.66 14.16 -3.5% y 13.86 14.15 -2.0% y
Regulated Emissions (g/mi)
Evaporative Emissions (g/test)
Greenhouse Gases (g/mi)
Aldehydes (mg/mi)
35
TP-25818
700
600
500
400
300
200
100
0
STD-RFGCNG
31a: Non-Methane Hydrocarbon
31b: Carbon Monoxide
31d: Carbon Dioxide
31c: Oxides of Nitrogen
NM
HC
Em
issi
ons
(g/m
i)C
O2
Em
issi
ons
(g/m
i)N
Ox
Em
issi
ons
(g/m
i)C
O E
mis
sion
s (g
/mi)
Round 1 Round 2
Round 1 Round 2
Round 1 Round 2
Round 1 Round 2
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
12
10
8
6
4
2
0
EPA Tier 0
EPA Tier 0
EPA Tier 0
0258
1831
m
1.8
1.6
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.0
EPA Tier 1
EPA Tier 1
EPA Tier 1
Figure 31. Emissions results from theDodge B250 van tested at Lab 1
STD-RFGCNG
32a: Non-Methane Hydrocarbon
32b: Carbon Monoxide
32d: Carbon Dioxide
32c: Oxides of Nitrogen
NM
HC
Em
issi
ons
(g/m
i)C
O2
Em
issi
ons
(g/m
i)N
Ox
Em
issi
ons
(g/m
i)C
O E
mis
sion
s (g
/mi)
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
12
10
8
6
4
2
0
1.8
1.6
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.0
EPA Tier 0
EPA Tier 0
0258
1832
m
700
600
500
400
300
200
100
0Round 2 Round 3Round 1
EPA Tier 1
EPA Tier 1
Round 2 Round 3Round 1
Round 2 Round 3Round 1
Round 2 Round 3Round 1
EPA Tier 0
EPA Tier 1
Figure 32. Emissions results from theDodge B250 van tested at Lab 2
36
TP-25818
that tended to be significant. The differences between roundsfor the RFG vans showed no significant difference at any ofthe labs.
Evaporative Emissions
CNG vehicles were designed with sealed fuel systems. Todetermine if the test vans were experiencing any leaks or"weepage" at any point in the natural gas fuel system, a mod-ified evaporative test was performed. The gasoline vansreceived the standard evaporative test, which includes a heatbuild on the fuel tanks. The CNG vans were placed in theSHED for the two prescribed 1-hour tests, but without heat-ing the tanks.
Average evaporative emissions for Labs 1, 2, and 3 are listedin Tables 35–37 and shown in Figures 34–36. The averageevaporative emissions for the B250 van were well below theTier 1 and Tier 0 limit of 2 g per test for all rounds at eachlab. "Evaporative" HC emissions from the modified evapora-tive tests on the CNG vans were significantly lower than theevaporative emissions for the standard models for all labsduring all test rounds. Evaporative emissions for the CNGvans tested at Lab 1 were 90% lower than those from thegasoline vans in Round 1 and 48.5% lower in Round 2. TheCNG vans tested at Lab 2 showed larger differences of 35%,61%, and 75% lower than the gasoline controls for Rounds 1,2, and 3, respectively. Lab 3 also showed decreases for theCNG vans, from 43% to 62%. These differences tended to bestatistically significant at the 95% confidence level.
Round-to-round comparisons showed significant increasesfor both fuels at Lab 1. The CNG vans at Lab 2 showed nosignificant difference between rounds and the control vansshowed a steady increase in evaporative emissions that wasonly significant between Round 2 and Round 3. The CNGvans tested at Lab 3 also showed no significant differencebetween rounds. The evaporative emissions for the controlvans at Lab 3 showed an increase in Round 2 and a decreasein Round 3. Neither of these differences, however, was statis-tically significant.
Greenhouse Gases
The average CO2 emissions for the CNG vans were consis-tently lower than the average for the gasoline controls. Labs 1and 2 showed a decrease of around 15% for all rounds. Lab 3had a slightly higher percent decrease at approximately 19%for the 3 rounds. The differences in CO2 emissions betweenCNG and RFG were statistically significant. The differencesbetween rounds for both van types at all 3 labs tended not tobe significant at the 95% confidence level.
Because CNG is 95% CH4, emissions of this greenhouse gasare expected to be significantly higher for the CNG vans.
STD-RFGCNG
33a: Non-Methane Hydrocarbon
33b: Carbon Monoxide
33d: Carbon Dioxide
33c: Oxides of Nitrogen
NM
HC
Em
issi
ons
(g/m
i)C
O2
Em
issi
ons
(g/m
i)N
Ox
Em
issi
ons
(g/m
i)C
O E
mis
sion
s (g
/mi)
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
12
10
8
6
4
2
0
1.8
1.6
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.0
EPA Tier 0
EPA Tier 0
0258
1833
m
700
600
500
400
300
200
100
0Round 2 Round 3Round 1
EPA Tier 1
Round 2 Round 3Round 1
Round 2 Round 3Round 1
Round 2 Round 3Round 1
EPA Tier 1
EPA Tier 1
EPA Tier 0
Figure 33. Emissions results from theDodge B250 van tested at Lab 3
37
TP-25818
Differences in CH4 emissions between the CNG testsand the RFG tests range from 245% higher to 1,387%higher. Round-to-round comparisons of CH4 emis-sions at Labs 1 and 2 showed significant increases forthe CNG tests over time. Lab 3 showed a significantincrease in CH4 for the CNG tests in Round 2, but nosignificant difference in Round 3. The RFG testsshowed no significant difference in CH4 emissionsbetween rounds at any of the labs.
Aldehydes
Figures 37–39 present the average aldehyde emis-sions for the Dodge B250 vans at each lab. In general,aldehyde emissions from the CNG vans were muchlower than those from the gasoline vans. The excep-tion to this was the formaldehyde emissions at Lab 2.Labs 1 and 3 showed similar values between fuels forboth formaldehyde and acetaldehyde with the CNGvans testing significantly lower than the gasoline con-trol vans. Reductions in formaldehyde at Lab 1 wereapproximately 68% in Round 1 and 62% in Round 2.Lab 3 showed reductions in formaldehyde of approxi-mately 54%, 53%, and 52% in Rounds 1, 2, and 3respectively. Acetaldehyde emissions for the CNGvans at Lab 1 were 87% lower than those from theconventional vans in Round 1 and 81% lower inRound 2. Lab 3 showed similar reductions inacetaldehyde of 91% in Round 1 and 81% inRounds 2 and 3.
Average formaldehyde emissions for the B250 vanstested at Lab 2 were not significantly differentbetween fuels for the first 2 rounds. The CNG vanstested 9.9% higher than the gasoline controls inRound 1 and 5.4% lower in Round 2. Round 3, how-ever, showed a significant increase in formaldehydeemissions for the CNG vans (51.8%). This could bedue in part to the van mentioned earlier (on which thecheck engine light came on during the last phase ofthe FTP). The formaldehyde value for this van wasconsiderably higher than that of the other vans tested.Removal of this value, which was not identified as anoutlier, would reduce the percent difference to 26%,but the CNG average is still greater than that of theconventional model. Acetaldehyde emissions at Lab 2agree with the other 2 labs, with the CNG vans testingsignificantly lower than the gasoline vans. The aver-age acetaldehyde emissions for the CNG vans testedat Lab 2 were 78% lower than those from the conven-tional model van in Round 1, 76% lower in Round 2,and 75% lower in Round 3. The differences betweenrounds for aldehydes at all 3 labs tended to be not significant at the 95% confidence level.
STD-RFGCNG
Round 1 Round 2
2.5
2.0
1.5
1.0
0.5
0.0
EPA Tier 1 & 0
0258
1834
m
Total Evaporative HydrocarbonE
vapo
rativ
e E
mis
sion
s (g
/Tes
t)
Figure 34. Evaporative emissions results from theDodge B250 van tested at Lab 1
STD-RFGCNG
Round 1 Round 2 Round 3
2.5
2.0
1.5
1.0
0.5
0.0
EPA Tier 1 & 0
0258
1835
mTotal Evaporative Hydrocarbon
Eva
pora
tive
Em
issi
ons
(g/T
est)
Figure 35. Evaporative emissions results from theDodge B250 van tested at Lab 2
STD-RFGCNG
Round 1 Round 2 Round 3
2.5
2.0
1.5
1.0
0.5
0.0
EPA Tier 1 & 0
0258
1836
m
Total Evaporative Hydrocarbon
Eva
pora
tive
Em
issi
ons
(g/T
est)
Figure 36. Evaporative emissions results from theDodge B250 van tested at Lab 3
38
TP-25818
Round 1
0258
1837
m
Aldehyde Emissions
Ald
ehyd
e E
mis
sion
s (m
g/m
i)
FormaldehydeAcetaldehyde Round 1
Round 2Round 2
9
8
7
6
5
4
3
2
1
0CNG STD-RFG
Figure 37. Aldehyde emissions from the Dodge B250 van tested at Lab 1
Round 1
0258
1838
m
Aldehyde Emissions
Ald
ehyd
e E
mis
sion
s (m
g/m
i)
FormaldehydeAcetaldehyde Round 1
Round 2Round 2
Round 3Round 3
9
8
7
6
5
4
3
2
1
0CNG STD-RFG
Figure 38. Aldehyde emissions from the Dodge B250 van tested at Lab 2
Round 1
0258
1839
m
Aldehyde Emissions
Ald
ehyd
e E
mis
sion
s (m
g/m
i)
FormaldehydeAcetaldehyde Round 1
Round 2Round 2
Round 3Round 3
9
8
7
6
5
4
3
2
1
0CNG STD-RFG
Figure 39. Aldehyde emissions from the Dodge B250 van tested at Lab 3
Hydrocarbon speciation was per-formed on a percentage of the DodgeB250 vans at Labs 1 and 3. FourCNG and three gasoline control vansspeciated at Lab 1. The vans receiv-ing full speciation at Lab 3 totaledfour CNG and five gasoline controlvans.
Tables 38 and 39 present the compar-isons between van models for PWTemissions at Labs 1 and 3, respective-ly. Figures 40 and 41 show the resultsgraphically. The aldehyde averageslisted include the results for onlythose vehicles that were speciated.These results show a significantadvantage in using CNG fuel overgasoline. All the toxics for the CNGvans tested at Lab 1 were significant-ly lower than the averages for theRFG tests. Lab 1 reported no
1,3-butadiene present in the CNGtests, which represented a 100%decrease over the RFG levels. TotalPWT for the CNG vans was 96.8%lower than that of the gasoline controlvans. Lab 3 showed agreement withLab 1. All toxics for the CNG vanswere significantly lower than thegasoline controls. Total PWT for theCNG vans was 95.4% lower than thatof the gasoline controls.
Tables 40 and 41 present the NMOG,OFP, and SR results for the DodgeB250 vans. Average NMOG for theCNG vans was significantly lowerthan the average for the gasolinemodels. The OFP and SR results aregraphically presented in Figures 42and 43. OFP from the CNG vans wassignificantly lower than that from thegasoline vans by 96.5% at Lab 1 and81% at Lab 3. SR also showed signif-icant reductions for the CNG vans,
approximately 46% at Lab 1 and 56%at Lab 3.
Fuel Economy
Because CNG is a gaseous fuel, itmust be converted to gallons of gaso-line equivalent (gge) in order to makea comparison with a liquid fuel. Anequivalent gallon of CNG is the quan-tity of CNG that has the same energycontent as a gallon of gasoline. A gal-lon of RFG has 111,960 Btu.Approximately 121 standard cubicfoot (scf) of test CNG contains thesame Btu as RFG. Therefore, 121 scfequals one gge.
Fuel economy averages for the CNGvan are listed in Tables 35–37 asmiles per equivalent gallon of gaso-line. Average fuel economy for theCNG Dodge B250 vans was onlyslightly less than that of the conven-tional models. All three labs were in
39
TP-25818
Table 39. Toxic Emissions from the Dodge B250 Van Tested at Lab 3
CNG STD-RFG
MeasuredPWT
MeasuredPWT
Percent Sig. Fuel
Value (mg/mi) Value (mg/mi)Difference Effect?
HCHO 2.007 0.092 3.467 0.159 -42.1% y
CH3CHO 0.171 0.0014 0.989 0.0079 -82.3% y
1,3-butadiene 0.014 0.014 1.985 1.985 -99.3% y
Benzene 0.25 0.0075 11.179 0.335 -97.8% y
Total 2.442 0.115 17.62 2.488 -95.4% y
Table 38. Toxic Emissions from the Dodge B250 Van Tested at Lab 1
CNG STD-RFG
MeasuredPWT
MeasuredPWT
Percent Sig. Fuel
Value (mg/mi) Value (mg/mi)Difference Effect?
HCHO 1.878 0. 086 5.741 0.264 -67.4% y
CH3CHO 0.152 0.001 1.167 0.009 -88.9% y
1,3-butadiene 0 0 2.1 2.1 -100.0% y
Benzene 0.060 0.0018 14.15 0.425 -99.6% y
Total 2.09 0.089 23.16 2.798 -96.8% y
40
TP-25818
0258
1841
m
PW
T (
mg/
mi)
TotalBenzene1,3 butadieneCH3CHOHCHO
3.0
2.5
2.0
1.5
1.0
0.5
0.0
Toxic Emissions
STD-RFGCNG
Figure 41. PWT emissions for the Dodge B250 vantested at Lab 3
Table 40. OFP for the Dodge B250 vanTested at Lab 1
CNGSTD- Percent Sig. FuelRFG Difference Effect?
NMOG (mg/mi) 21.95 354.49 -93.8% y
OFP (mg O3/mi) 45.2 1,305.31 -96.5% y
SR (mg O3/mg NMOG) 2.06 3.836 -46.3% y
0258
1840
m
PW
T (
mg/
mi)
TotalBenzene1,3 butadieneCH3CHOHCHO
3.0
2.5
2.0
1.5
1.0
0.5
0.0
Toxic Emissions
STD-RFGCNG
Figure 40. PWT emissions from the Dodge B250 vantested at Lab 1
41
TP-25818
0258
1842
m
Ozone Formation and ReactivityO
FP
(m
g O
3/m
i)
SROFP
STD-RFGCNG
1400
1200
1000
800
600
400
200
0
4.54.03.53.02.52.01.51.00.50.0
SR
(m
g O
3/m
g N
MO
G)
Figure 42. OFP and SR for the Dodge B250 van tested at Lab 1
0258
1843
m
Ozone Formation and Reactivity
OF
P (
mg
O3/
mi)
SROFP
STD-RFGCNG
1400
1200
1000
800
600
400
200
0
4.54.03.53.02.52.01.51.00.50.0
SR
(m
g O
3/m
g N
MO
G)
Figure 43. OFP and SR for the Dodge B250 van tested at Lab 3
Table 41. OFP for the Dodge B250 Van Tested at Lab 3
CNGSTD- Percent Sig. FuelRFG Difference Effect?
NMOG (mg/mi) 76.48 308.72 -75.2% y
OFP (mg O3/mi) 233.27 1208.9 -80.7% y
SR (mg O3/mg NMOG) 1.768 4.031 -56.1% y
agreement, with percent differencesranging from 2% lower in the CNGvans to approximately 12% lower.These differences in fuel economybetween CNG and RFG were signifi-cant for all rounds at all 3 labs.
DODGE CARAVAN MINIVAN
The 1994 Dodge Caravan is a mini-van equipped with a 3.3 L V6 engine(Figure 44). Both models were certi-fied to EPA Tier 1 emissions levels.Because there was a limited numberof vehicles available, these vans wereonly tested in one round. There were13 dedicated CNG vans and 6 stan-dard gasoline vans tested. Mileageranges and average odometer read-ings for the Caravans tested in thisprogram are listed in Table 42.Detailed hydrocarbon speciation wasnot performed on these vehicles.
Regulated Emissions
Table 43 lists the average emissionsfor the CNG and conventional modelCaravans along with the percent dif-ferences and an indication of whetherthe differences are statistically signif-icant at the 95% confidence level.Figure 45 shows the comparison ofaverage regulated emissions and CO2for these vans. All regulated emis-sions results for the Caravans werewell below the EPA Tier 1 standard.When comparing regulated emissionsfor the CNG Caravan to those of thegasoline control vans, there was a sig-
Table 43. Average Emissions Results from theDodge Caravan Minivan
Round 1
CNGSTD- Percent Sig. FuelRFG Difference Effect?
Regulated Emissions (g/mi)
NMHC 0.022 0.147 -84.8% y
THC 0.166 0.169 -2.1% n
CO 0.364 1.552 -76.5% y
NOx 0.187 0.296 -36.9% n
Evaporative Emissions (g)
Total Evaporative 0.311 0.323 -3.7% n
Greenhouse Gases (g/mi)
CO2 389.54 467.22 -16.6% y
CH4 0.142 0.028 415.0% y
Aldehyde Emissions (mg/mi)
HCHO 4.036 3.468 16.4% n
CH3CHO 0.322 0.902 -64.3% y
Fuel Economy 17.45 18.84 -7.3% y
Regulated Emissions (g/mi)
Fuel Economy
Aldehydes (mg/mi)
Greenhouse Gases (g/mi)
Evaporative Emissions (g/test)
43
TP-25818
nificant decrease in NMHC, a significant decrease in CO, anda decrease in NOx that was not significant at the 95% confi-dence level. NMHC was 85% lower for the CNG model. COemissions were 76.5% lower and NOx emissions were 37%lower for the CNG vans.
Evaporative Emissions
The same modified evaporative emissions test described inthe section on the B250 vans was performed on the CNGDodge Caravans. Results for the Dodge Caravans are listed inTable 36 and graphically illustrated in Figure 46. Average"evaporative" emissions for both CNG and gasoline modelswere well below the Tier 0 and Tier 1 limit of 2 g. As with theB250 van, the CNG Caravan emitted measurable HC duringthe test, but they were lower than the average evaporativeemissions from the gasoline control. The reduction was 3.7%,which was not statistically significant at the 95% confidencelevel.
Greenhouse Gases
As with the regulated emissions, average CO2 emissions weresignificantly lower for the CNG Caravans. Values for theCNG vans were approximately 16% lower than those of theirgasoline counterparts. Average CH4 emissions, as expected,were higher for the CNG Caravans. Although the values foreach van type were quite low, the CNG model showed a415% increase in CH4 over the gasoline model.
Aldehydes
Aldehyde emissions levels for the Dodge Caravans are shownin Figure 47. Although the formaldehyde emissions from theCNG minivans were 16% higher than the gasoline model, thisdifference was not statistically significant at the 95% confi-dence level. Acetaldehyde emissions were 64% lower for the
STD-RFGCNG
45a: Non-Methane Hydrocarbon
45b: Carbon Monoxide
45d: Carbon Dioxide
45c: Oxides of Nitrogen
NM
HC
Em
issi
ons
(g/m
i)C
O2
Em
issi
ons
(g/m
i)N
Ox
Em
issi
ons
(g/m
i)C
O E
mis
sion
s (g
/mi)
0.500.450.400.350.300.250.200.150.100.050.00
6
5
4
3
2
1
0
1.2
1.0
0.8
0.6
0.4
0.2
0.0
EPA Tier 1
0258
1845
m
500
400
300
200
100
0Round 1
EPA Tier 1
Round 1
Round 1
Round 1
EPA Tier 1
Figure 45. Emissions results from theDodge Caravan minivan
STD-RFGCNG
Round 1
2.5
2.0
1.5
1.0
0.5
0.0
EPA Tier 1 & 0
0258
1846
mTotal Evaporative Hydrocarbon
Eva
pora
tive
Em
issi
ons
(g/T
est)
Figure 46. Evaporative emissions results from theDodge Caravan minivan
44
TP-25818
CNG model compared to the RFGresults.
Fuel Economy
Fuel economy comparisons for theDodge Caravan showed very littledifference when compared on a gaso-line gallon equivalent between theCNG and standard models. The fueleconomy for the CNG minivans wasapproximately 7% lower than that ofthe standard gasoline model.
0258
1847
m
Aldehyde Emissions
Ald
ehyd
e E
mis
sion
s (m
g/m
i)
FormaldehydeAcetaldehyde
4.5
4.0
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0CNG STD-RFG
Figure 47. Aldehyde emissions from the Dodge Caravan minivan
45
TP-25818
In conclusion, these tests showedthat, overall, there are emissionsadvantages to using alternative fuelover gasoline. The following pointssummarize the comparison betweeneach alternative fuel and gasoline.
METHANOL VEHICLES
• NMHCE was significantly lower inM85 tests for both the Intrepid andSpirit.
• CO emissions were slightly higherfor the M85 tests on the Intrepidand lower for the M85 tests on the Spirit. These differences, how-ever, tended not to be significantat the 95% confidence level.
• NOx emissions tended to be signif-icantly higher in the M85 tests forboth FFV models.
• Greenhouse gases (CO2 and CH4)were significantly lower in theM85 tests for both FFV models.
• Formaldehyde was significantlyincreased for the M85 tests, butacetaldehyde was significantlydecreased.
• Benzene and 1,3-butadiene levelswere significantly lower for theM85 tests.
• Evaporative results were varied,but tended not to be significant atthe 95% confidence level.
• PWT, OFP, and SR were all signifi-cantly lower in the M85 tests forboth FFV models.
ETHANOL VEHICLES
• Regulated emissions for the Taurusshowed no significant differencebetween fuels.
• Regulated emissions results for theLumina were mixed: NMHCEtended not to be significantbetween fuels, CO emissions werehigher (but not significantly) forthe E85 tests, and NOx emissionswere significantly lower for theE85 tests.
• CO2 emissions were significantlylower for the tests on E85 for bothFFV models.
• CH4 emissions were significantlyhigher for the tests on E85 forboth FFV models.
• Formaldehyde and acetaldehydeemissions were significantly high-er for the tests on E85 for bothFFV models.
• Benzene and 1,3-butadiene levelswere significantly lower for theE85 tests.
• Evaporative emissions for theethanol FFVs tended to show nosignificant difference betweenfuels.
• PWT and SR were significantlylower for the E85 tests on the FFVTaurus.
• OFP was significantly higher forthe E85 tests on the FFV Taurus.
CNG VEHICLES
• NMHC emissions for the CNGmodels were significantly lowerthan those of the gasoline vehicles.
• CO emissions were significantlylower for the CNG vans.
• NOx emissions results were mixed,but tended to be significantlylower for the CNG tests.
• CO2 emissions were significantlylower for the CNG tests.
• CH4 emissions were significantlyhigher for the CNG tests.
• Formaldehyde emissions tended tobe significantly lower for the CNGtests.
• Acetaldehyde emissions from theCNG vehicles were significantlylower than the gasoline tests.
• Evaporative emissions results weresignificantly lower for the CNGtests.
• PWT, OFP, and SR were signifi-cantly lower for the CNG tests.
SUMMARY
46
TP-25818
1. L. Dodge, G. Bourn, T. Callahan,J. Grogan, D. Leone, D. Naegeli,K. Shouse, R. Thring, and K.Whitney. Development of aDedicated Ethanol Ultra-LowEmissions Vehicle (ULEV): FinalReport, NREL/SR-540-24603,Golden, CO: NREL.
2. Auto/Oil Air QualityImprovement Research Program,1997. Program Final Report,Atlanta, GA: CoordinatingResearch Council.
3. Clean Air Act, Part C, Section 241Definitions. EPA Web site: www.epa.gov/airprogm/oar/caa/caa.txt
4. Office of the Federal Register.1993. Code of FederalRegulations, Title 40, Part 86.Washington, DC: U. S.Government Printing Office.
6. Health Effects Notebook forHazardous Air Pollutants, Officeof Air Quality Planning &Standards Unified Air Toxics Web site: www.epa.gov/ttn/uatw/hapindex.html
7. V.R. Burns, J.D. Benson, A.M.Hochhuser, W.J. Koehl, W.M.Kreucher, and R.M. Reuter. 1992."Description of Auto/Oil AirQuality Improvement ResearchProgram," SAE Paper No. 912320in Auto/Oil Air QualityImprovement Research Program.Warrendale, PA: Society ofAutomotive Engineers (SAE).
8. K.J. Kelly, B.K. Bailey,T.C.Coburn, W. Clark, L. Eudy, P.Lissiuk. May 1996. "FTPEmissions Test Results fromFlexible-Fuel Methanol DodgeSpirits and Ford Econoline Vans,"SAE Paper No. 961090,Warrendale, PA: SAE.
9. K.J. Kelly, B.K. Bailey, T.C.Coburn, W. Clark, P. Lissiuk. May 1996. "Federal TestProcedure Emissions Test Resultsfrom Ethanol Variable-FuelVehicle Chevrolet Luminas," SAEPaper No. 961092, Warrendale,PA: SAE.
REFERENCES
A-1
Appendix A:Emissions Data Sets
A-2
Table A-1. 1995 Standard Dodge Intrepid : RFG Tests at Lab 1 Round 1Decal ID Test Date Odometer Fuel MPG CH3CHO CH3OH CH4 CO CO2 HCHO NMHCE NOx THC Evap. THC
Table A-3. 1995 FFV Dodge Intrepid: RFG Tests at Lab 1 Round 1Decal ID Test Date Odometer Fuel MPG CH3CHO CH3OH CH4 CO CO2 HCHO NMHCE NOx THC Evap. THC
Table A-4. 1995 Standard Dodge Intrepid: RFG Tests at Lab 1 Round 2Decal ID Test Date Odometer Fuel MPG CH3CHO CH3OH CH4 CO CO2 HCHO NMHCE NOx THC Evap. THC
Table A-5. 1995 FFV Dodge Intrepid: M85 Tests at Lab 1 Round 2Decal ID Test Date Odometer Fuel MPG CH3CHO CH3OH CH4 CO CO2 HCHO NMHCE NOx THC Evap. THC
Table A-6. 1995 FFV Dodge Intrepid: RFG Tests at Lab 1 Round 2Decal ID Test Date Odometer Fuel MPG CH3CHO CH3OH CH4 CO CO2 HCHO NMHCE NOx THC Evap. THC
Table A-19. 1994-95 Standard Ford Taurus: RFG Tests at Lab 1 Round 1Decal ID Date Odometer Fuel MPG CH3CHO CH4 CO CO2 ETOH HCHO NMHCE NOx THC Evap. THC
Table A-22. 1994-95 Standard Ford Taurus: RFG Tests at Lab 1 Round 2Decal ID Date Odometer Fuel MPG CH3CHO CH4 CO CO2 ETOH HCHO NMHCE NOx THC Evap. THC
Table A-25. 1993 Standard Chevrolet Lumina: RFG Tests at Lab 2 Round 1Decal ID Date Odometer Fuel MPG CH3CHO CH4 CO CO2 ETOH HCHO NMHCE NOx THC Evap. THC
Table A-26. 1992-93 FFV Chevrolet Lumina: E85 Tests at Lab 2 Round 1Decal ID Date Odometer Fuel MPG CH3CHO CH4 CO CO2 ETOH HCHO NMHCE NOx THC Evap. THC
Table A-27. 1992-93 FFV Chevrolet Lumina: RFG Tests at Lab 2 Round 1Decal ID Date Odometer Fuel MPG CH3CHO CH4 CO CO2 ETOH HCHO NMHCE NOx THC Evap. THC
Table A-28. 1993 Standard Chevrolet Lumina: RFG Tests at Lab 2 Round 2Decal ID Date Odometer Fuel MPG CH3CHO CH4 CO CO2 ETOH HCHO NMHCE NOx THC Evap. THC
Table A-29. 1992-93 FFV Chevrolet Lumina: E85 Tests at Lab 2 Round 2Decal ID Date Odometer Fuel MPG CH3CHO CH4 CO CO2 ETOH HCHO NMHCE NOx THC Evap. THC
Table A-30. 1992-93 FFV Chevrolet Lumina: RFG Tests at Lab 2 Round 2Decal ID Date Odometer Fuel MPG CH3CHO CH4 CO CO2 ETOH HCHO NMHCE NOx THC Evap. THC
Table A-31. 1993 Standard Chevrolet Lumina: RFG Tests at Lab 2 Round 3Decal ID Date Odometer Fuel MPG CH3CHO CH4 CO CO2 ETOH HCHO NMHCE NOx THC Evap. THC
Table A-32. 1992-93 FFV Chevrolet Lumina: E85 Tests at Lab 2 Round 3Decal ID Date Odometer Fuel MPG CH3CHO CH4 CO CO2 ETOH HCHO NMHCE NOx THC Evap. THC
shaded areas mark outliers or a data point removed to balance a FFV data sete - missing dataf - instrument problems caused loss of data* - evaporative test not required for re-tests resulting from a regulated emissions component being out of EPA standards> - first tests on CNG vehicles did not include evaporative
Appendix D:Emissions Data Compilation, Editing, and Reduction
andthe Analysis of Variance Approach to Statistical
Treatment of Emissions Data
D-2
Raw data files of the emissions tests from each laboratory were submitted electronically andthen loaded into the Alternative Fuels Data Center at NREL. Before any data analysis wasconducted, checks and edits were undertaken to ensure data quality. In particular, the datawere reviewed for the presence of outliers. To begin this review process, the data sets weresorted by vehicle type, test fuel, and test round. At the first level of data quality checks, thereplicate test results were evaluated. An initial set of replicate tests was conducted on somevehicles to provide information about test repeatability. Additional replicated tests wereperformed on vehicles that exceeded the EPA emissions certification standards. Acomparison of the replicate results helped to identify some individual test results as outliers.These results were then eliminated from further consideration (although, as described below,the established outlier detection procedure involved more than these replicate test results).
The four-stage procedure outlined below was used to identify and eliminate outliers in theexhaust emissions test results, and to compile the final data sets for statistical analysis. Noevaporative emissions results were removed from the data sets because of the high level ofvariability in typical evaporative emissions.
1. Stage One (Replicate Analysis)—For each emissions constituent (e.g., NOx), all pairs ofreplicated test results were first considered. The absolute value of the difference betweeneach pair was computed, and the mean and standard deviation of all such differences werealso computed. Individual differences outside a bound equal to the mean plus threestandard deviations were flagged as excessive. The two test results from each of theflagged pairs were then reviewed, and the one result in each pair furthest from the overallmean was designated as an outlier and eliminated. For all other pairs (those not flaggedas excessive), the two test results were simply averaged to produce a single result. In thismanner, the overall data set was reduced to a single value per vehicle type/fuel/test roundfor each emissions constituent.
2. Stage Two (Among-Vehicle Data Quality Checks)—Having a single set of values for eachvehicle type/fuel/test round, it was then necessary to compare the results for eachcombination of the three (e.g., Dodge Spirit, M85, round 1). Consequently, for everyvehicle type/fuel/test round combination, the mean and the standard deviation of eachemissions constituent were computed. Individual vehicle values outside a bound of themean plus or minus three standard deviations were designated as outliers and removedfrom further consideration.
3. Stage Three (Checks Among Emissions Constituents, or Total VehicleViability)—Depending on the emissions constituent in question, the application of theedits performed in Stage Two left a number of “holes” in the data. In some cases, theprocess resulted in multiple holes (more than one emissions constituent missing) for agiven test. Because each hole is the result of an emissions test value being designated asan outlier, tests (for a given fuel/test round combination) having two or more holes onmajor emissions constituents (HC, NOx, and CO) were deemed to be “not viable” andwere completely eliminated from further consideration.
4. Stage Four (Data Reduction for Multiple Rounds)—Finally, for purposes of this particularreport, only the results on vehicles tested in all rounds (for a particular model/fuelcombination) were retained for data analysis purposes (Note: some vehicles were nottested in all rounds for a number of reasons. For example, some failed the pre-testmaintenance checks and were returned to the agencies, and some were retired fromservice by GSA before all rounds of testing could be completed).
Analysis of variance (ANOVA) was the principal statistical technique used to analyze theemissions data presented in this report. Whereas the t-test—one of the most frequentlyapplied statistical procedures—is used to assess the significance of differences in pairs ofmean values, ANOVA facilitates simultaneous assessment of multiple differences among acollection of two or more means (see, for example, Table D-1).
D-3
Table D-1. Example Table of Mean Values
Round 1 Round 2 Round 3Fuel 1 _11 _12 _13
Fuel 2 _21 _22 _23
Note: See below for explanation of “fuel” and “round.” _ stands for the mean value ofsome emissions constituent of interest (e.g., CO). _11 - _23 is an example of one possibledifference in mean values.
ANOVA is even more useful in that it allows the total variation in a set of data (as measuredby the sum of squared deviations from a mean value) to be subdivided into the portions thatare attributable to various experimental or observational factors. In this manner, thecontributions of various factors to the observed variability in some test result, laboratoryresponse, or property of interest, can be identified and quantified, along with the effects ofsuch factors interacting among themselves.
In the context of the emissions testing program discussed in this report, the experimentalfactors assumed to generate differences in test results are: (1) fuel (alternative fuel versusgasoline); (2) round (a proxy for mileage); (3) laboratory (three different laboratories chosenthrough competitive bidding and employing the same test procedures; one of the three athigh altitude); and (4) vehicle model (Dodge Caravan, Chevy Lumina, etc.). In addition,differences among individual vehicles of the same model contribute to the total variation inemissions test results, with random sampling resulting in such differences. Although otherfactors may affect variability in emissions, these are not explicitly controlled in the testprogram. Contributions to the total variation from these factors cannot be determined.
The arithmetic computations of analysis of variance, which are explained in textbooks onstatistical methods, are usually summarized in a tabular form like the one shown in Table D-2.The first column in the table identifies the experimental factors, or sources of variation, whilethe second lists the corresponding numbers associated with a quantity called the “degrees offreedom.” Typically, the degrees of freedom associated with a particular factor consist of thenumber of “levels” of that factor minus one (or in the case of the category labeled “Total,”the overall number of observations or test results minus one).
D-4
Table D-2. General Form of an ANOVA Table
Degrees of Sums of Mean SignificanceSource Freedom Squares Squares F-Value LevelTotal n-1 *Factor A a - 1 * * * *Factor B b-1 * * * *. . . . . . * * * *Factor Z z-1 * * * *Remainder1 (n-1)-(a-1)-(b-1)-…-(z-1) * **Values to be computed.1In many cases, “Remainder” is denoted as “Error,” which, depending on the context of the analysis, can be either experimental error or sampling error.Note: n is the total number of observations; a is the number of levels of Factor A, b is the number of levels of Factor B, etc.
The third column lists a series of intermediate calculations, referred to as “sums of squares,”which are associated with the respective factors or sources of variation. “Sums of squares” isabbreviated wording for “sum of squared deviations from the mean,” which is the basiccalculation needed for computing a statistical variance. The sums of squares associated withthe different factors in Table D-2 below the “Total” line must, of necessity, add up to thesum of squares shown on the “Total” line (this is the additive property of ANOVA).
The fourth column in Table D-2 lists a series of numbers referred to as the “mean squares.”The mean squares associated with the respective factors or sources of variation are computedby dividing the corresponding sum of squares by the corresponding degrees of freedom. Itis these mean squares that are actual variances.
The fifth column in the table contains a series of numbers under the heading of “F-Value.”These numbers are determined by taking ratios of the mean squares associated with variousfactors. The numbers in this column are referred to as F-values because they adhere to aspecial probability distribution called the F-distribution.
The sixth and final column in the table lists probability values that can be used to assess thesize of the corresponding F-values (or ratios of “mean squares”). These are often referredto as “Significance Levels.”
Typical ANOVA tables based on some of the data presented in this report is shown in TablesD-3 and D-4.
Once the experimental factors, or sources of variation, have been accurately identified, thecalculations necessary to complete an ANOVA table are relatively straightforward. Softwareproducts such as JMP, available from SAS Institute, make it possible to avoid the algebraictedium that would otherwise be required to compute all the numbers. Interpreting the resultsis quite a different matter. To make an appropriate interpretation, we must consider thepopulation of units to which statistical inferences are to be drawn. In addition, we mustdetermine which factors are to be regarded as “fixed” and which are “random.”
Table D-3. ANOVA in CO Measurements Obtained in Emissions Tests on Flexible-FuelDodge Intrepids
Degrees of Sums of Mean SignificanceSource Freedom Squares Squares F-Value Level4
Round x Fuel1 1 0.0023 0.0023 0.2306 0.6385 Vehicles 14 0.5149 0.0368 1.1669 0.3941Vehicle x Round2 14 0.4027 0.0288 2.9077 0.0275Vehicle x Fuel3 14 0.1771 0.0127 1.2783 0.3261Error 14 0.1385 0.00991,2,3Factor interaction terms4Values of .05 or less would ordinarily indicate significant differences. For example, the significance level of 0.0066 associated with the F-value for “Rounds” indicates that the same average value of CO was not obtained in both test rounds.
Table D-4. ANOVA in NOx Measurements Obtained in Emissions Tests on Flexible-FuelDodge Spirits
Degrees of Sums of Mean SignificanceSource Freedom Squares Squares F-Value Level4
Total 83 1.2592Rounds 1 0.0074 0.0074 0.4901 0.4920Fuels 1 0.0031 0.0031 0.2444 0.6264Round x Fuel1 1 0.0500 0.0500 17.5905 0.0004Vehicles 20 0.5851 0.0293 1.1708 0.3381Vehicle x Round2 20 0.3032 0.0152 5.3304 0.0002Vehicle x Fuel3 20 0.2534 0.0127 4.4551 0.0008Error 20 0.0569 0.00281,2,3Factor interaction terms4Values of .05 or less would ordinarily indicate significant differences. For example, the significance level of 0.0004 associated with the F-value for the “Round x Fuel” interaction indicates that the difference in the average values of NOx for the two fuels was not the same from one test round to the next.
Fixed factors are those whose range of values, or levels, are completely encompassed by thespecific population units included in the investigation. In the context of the this emissionstesting study, “fuel” is a fixed experimental factor because there is not interest in, norrationale for, drawing conclusions about fuels other than those being specifically studied. Arandom factor, on the other hand, is one about which conclusions can be extended to a largercollection of units than the ones specifically included in the investigation. In this context,“vehicle” is a random factor because individual vehicles were randomly selected from alarger collection, or population, and projecting the results of the testing program to that largerpopulation is desirable. The determination of fixed and random factors governs the way theF-values are computed (that is, the choice of numerator and denominator in the ratio of meansquares; the denominator always represents an “error” term against which the numerator iscompared) and directly affects interpretation of the results. The bigger the F-value, the morelikely at least one difference among the means being compared is statistically significant.
ANOVA's statistical procedure is constructed on certain mathematical assumptions. The firstassumption—that effects of the various experimental factors are additive—has already beenmentioned (in the sense that the individual sums of squares add up to the total). The secondassumption is that all experimental errors are random, independent, and follow a normal(Gaussian, or bell-shaped) distribution. Violating either of these assumptions will negate theinterpretability of the results.
Statistical software packages such as JMP provide many other capabilities that extend andbuild on the information derived from the basic ANOVA. In particular, it is possible toestimate the actual components of variance attributable to each experimental factor, and toadjust mean values for unequal numbers of observations using a least squares approach. Thedetails of these techniques are beyond the scope of this discussion.
REPORT DOCUMENTATION PAGE Form ApprovedOMB NO. 0704-0188
Public reporting burden for this collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources,gathering and maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of thiscollection of information, including suggestions for reducing this burden, to Washington Headquarters Services, Directorate for Information Operations and Reports, 1215 JeffersonDavis Highway, Suite 1204, Arlington, VA 22202-4302, and to the Office of Management and Budget, Paperwork Reduction Project (0704-0188), Washington, DC 20503.
1. AGENCY USE ONLY (Leave blank) 2. REPORT DATE
September 1999
3. REPORT TYPE AND DATES COVERED
Technical report
4. TITLE AND SUBTITLE
Light-Duty Alternative Fuel Vehicles: Federal Test Procedure Emissions Results
6. AUTHOR(S)
Kenneth Kelly, Leslie Eudy, and Timothy Coburn
5. FUNDING NUMBERS (C) (TA) FU905010
7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES)
National Renewable Energy Laboratory1617 Cole Blvd.Golden, CO 80401
8. PERFORMING ORGANIZATION REPORT NUMBER
NREL/TP-540-25818
9. SPONSORING/MONITORING AGENCY NAME(S) AND ADDRESS(ES)
U.S. Department of Energy1000 Independence Avenue, SWWashington, D.C. 20585
10. SPONSORING/MONITORING AGENCY REPORT NUMBER
11. SUPPLEMENTARY NOTES
12a. DISTRIBUTION/AVAILABILITY STATEMENT
National Technical Information ServiceU.S. Department of Commerce5285 Port Royal RoadSpringfield, VA 22161
12b. DISTRIBUTION CODE
UC-1504
13. ABSTRACT (Maximum 200 words)
In support of the U.S. Department of Energy's development and deployment of alternative fuels for environmental and national security reasons, NREL has managed aseries of light-duty vehicle emissions tests on alternative fuel vehicles (AFVs). The purpose of this report is to give a detailed evaluation of the final emissions test resultson vehicles tested on methanol, ethanol, and compressed natural gas.
15. NUMBER OF PAGES14. SUBJECT TERMS
Alternative fuel vehicles (AFVs), emissions, emissions testing, methanol, ethanol, and compressed natural gas
16. PRICE CODE
17. SECURITY CLASSIFICATION OF REPORTUnclassified
18. SECURITY CLASSIFICATION OF THIS PAGEUnclassified
19. SECURITY CLASSIFICATION OF ABSTRACTUnclassified
20. LIMITATION OF ABSTRACTUL
NSN 7540-01-280-5500 Standard Form 298 (Rev. 2-89)Prescribed by ANSI Std. Z39-18