lable at ScienceDirect

Environmental Pollution 158 (2010) 1451–1460

Contents lists avai

Environmental Pollution

journal homepage: www.elsevier .com/locate/envpol

Effects of low concentration biodiesel blend application on modern passengercars. Part 1: Feedstock impact on regulated pollutants, fuel consumption andparticle emissions

Georgios Fontaras a, Marina Kousoulidou a, Georgios Karavalakis b, Theodoros Tzamkiozis a,Panayotis Pistikopoulos a, Leonidas Ntziachristos a, Evangelos Bakeas c, Stamoulis Stournas b,Zissis Samaras a,*

a Laboratory of Applied Thermodynamics, Aristotle University Thessaloniki, P.O. Box 458, GR 54124 Thessaloniki, Greeceb Laboratory of Fuels Technology and Lubricants, National Technical University of Athens, 9 Iroon Polytechniou str., Zografou Campus, 15780 Athens, Greecec Laboratory of Analytical Chemistry, Chemistry Department, National and Kapodistrian University of Athens, Greece

Fleet-wide biodiesel application on passenger cars may affect pollutan

t emissions and the impact can vary depending on biodiesel feedstock.

a r t i c l e i n f o

Article history:Received 22 July 2009Received in revised form9 December 2009Accepted 16 December 2009

Keywords:Exhaust emissionsBiodieselRegulated pollutantsParticle emissions

* Corresponding author. Tel.: þ30 2310996014; faxE-mail address: zisis@auth.gr (Z. Samaras).

0269-7491/$ – see front matter � 2009 Elsevier Ltd.doi:10.1016/j.envpol.2009.12.033

a b s t r a c t

Five biodiesels from different feedstocks (rapeseed, soy, sunflower, palm, and used fried oils) blendedwith diesel at 10% vol. ratio (B10), were tested on a Euro 3 common-rail passenger car. Limited effects(�2% to þ4%) were observed on CO2 emissions. CO and HC emissions increased between 10% and 25% onaverage, except at high speed – high power where emissions were too low to draw conclusions. NOxemissions increased by up to 20% for two out of the five blends, decreased by up to 15% for two otherblends, and remained unchanged for one blend. Particulate matter (PM) was reduced for all blends by upto 25% and the reductions were positively correlated with the extent of biodiesel saturation. PMreductions are associated with consistent reductions in non-volatile particle number. A variablebehaviour in particle number is observed when volatile particles are also accounted.

� 2009 Elsevier Ltd. All rights reserved.

1. Introduction

Biofuel use in transportation is promoted in the European Unionin an effort to tackle climate change, diversify energy sources, andsecure energy supply. Directive 2003/30/EC stipulated that 5.75% oftotal petrol and diesel consumption in transport should be replacedby biofuels by 2010. In 2009 the new Directive 2009/28/EC intro-duced new targets stating that each member state shall ensure thatthe share of energy from renewable sources in all forms of transportin 2020 is at least 10% of the final consumption of energy intransport in that Member State. Since the use of other renewableenergy sources in the transport sector is rather difficult to achieve,it is expected that the goal set by the new regulation will be mainlymet through biofuels application. For various reasons, biodiesel hasbeen the biofuel of choice in Europe, with its share reaching 75% oftotal biofuel consumption in transport in 2007 (Eurobserver, 2008).

Biodiesel differs from petroleum derived diesel (petro-diesel) asthe former consists mainly of mono-alkyl esters (fatty-acid

: þ30 2310996019.

All rights reserved.

methylesters – FAME), which contain two oxygen atoms permolecule. As a result, biodiesel contains 10–12% wt. oxygen, whichleads to proportionally lower energy density (Graboski andMcCormick, 1998). In addition, due to the ester content, biodieseldiffers from petro-diesel in that it has a zero or ultralow naturalsulphur content, it contains no aromatic or polyaromatic hydro-carbons, it has a higher cetane value, a lower heating value, betterlubricity, higher viscosity, and a higher flash point.

Previous evidence has shown (Graboski and McCormick, 1998;Lapuerta et al., 2008a; Kousoulidou et al., 2009) that petro-dieselblending with biodiesel rather decreases all regulated pollutantsbut NOx. The increase in NOx is not yet clear, but several mecha-nisms have been proposed, including the different fuel chemistryand fuel spray characteristics (Marshall et al., 1995; EPA, 2002; Tat,2003; Knothe and Steidley, 2005; Sze, 2007). Such NOx increasesmay particularly aggravate air quality conditions in European cities.

In addition, more information is required with respect to bio-diesel effects on the concentration and characteristics of exhaustparticulate matter (PM). PM is the pollutant responsible for mostpremature deaths in Europe and significant concentrationexceedances still occur in several urban environments (WHO,2005). In particular, ultrafine particles (<0.1 mm diameter), such as

G. Fontaras et al. / Environmental Pollution 158 (2010) 1451–14601452

the ones produced by mobile combustion, are deposited with muchhigher efficiency in the pulmonary alveoli than fine particles(Kawanaka et al., 2004) because of their small size. Biodiesel mayalter particle characteristics, as the different fuel chemistry, thehigher viscosity, and the different surface tension may change thein-cylinder fuel spray characteristics. This has been shown to leadto higher particle number emissions in some studies, although theextent of this phenomenon is not yet clear (Mathis et al., 2005b;Lapuerta et al., 2008b).

The previous evidence shows that biodiesel effects on engineemissions may be quite significant. However, most studies havebeen conducted on heavy-duty and, generally, older technologyengines. Only few studies are available on modern diesel enginesand passenger cars, employing common-rail engine systems andaftertreatment technologies. Most of these studies show that bio-diesel application on modern engines and particularly on vehiclesequipped with some form of exhaust aftertreatment system canlead to results not consistent with what is generally reported(Durbin et al., 2007; Martini et al., 2007; Mazzoleni et al., 2007;Ropkins et al., 2007; Fontaras et al., 2009; Karavalakis et al., 2009;Lujan et al., 2009). Therefore, a recent technology diesel passengercar has been employed in the current study to investigate theeffects of five different biodiesel blends on exhaust emissions. Theblends have been produced by mixing 10% vol. biodiesel (B10) ofdifferent feedstock in petro-diesel. The study aims at providinga representative picture of biodiesel effects on gaseous andparticulate emissions of passenger cars, shedding some light on theinfluence of biodiesel feedstock and properties on emissions.

2. Methods

2.1. Fuels and test vehicle

Five biodiesel blends were prepared by mixing neat biodiesel, originating fromfive different feedstocks, with a base petro-diesel at a 10% vol. ratio (B10). The mostcommonly used feedstocks in Europe were selected, i.e. rapeseed oil, sunflower oil,soybean oil, palm oil and used frying oils, and their methyl ester blends are referredto as RME, SUME, SME, PME, and UFOME, respectively. Each biodiesel was froma single batch and prior to the measurements all necessary handling and storageactions were taken in order to avoid degradation. The base fuel was a typicalautomotive, low sulphur diesel (<50 ppm S) complying with EN590:2004 (Directive2003/17/EC). This base fuel may already contain a minimal biodiesel content (<2%),blended at the refinery. Hence, the total biodiesel content in the test fuels is esti-mated in the range between 11% and 12%.

The main properties of the test (blended) fuels are shown in Table 1. All fuelscomply with EN590:2004. In addition, all neat biodiesels were tested according tothe EN14214 standard, and their properties are presented in Table 2. Iodine numberand CFPP, were slightly above the EN14214 limits. These deviations are ratherexpected as the EN14214 standard was originally introduced for RME and thestandard’s limits were adjusted to the properties of that particular fuel.

The blends were tested on a Renault Laguna 1.9 dCi passenger car, complyingwith Euro 3 emission standard. The vehicle was equipped with a high-pressure

Table 1Properties of the base diesel fuel and the biodiesel blends.

Properties Diesel B10 PME B10 RME B10

Viscosity (40 �C, mm2 s�1) 3.61 3.81 4.0 3.89Density, (15 �C, g cm�3) 0.834 0.838 0.841 0.83Flash point, (�C) 71 67 65 67Sulphur content, (mg g�1) 24 22 19 20Water content, (mg g�1) 52 65 78 62CFPP, (�C) �9 �7 �8 �9Cetane number 55.5 55.5 � 0.5 (within the accuracy limiDistillationIBP 179 182 185 18310 220 225 228 23150 281 290 298 29790 344 346 349 348FBP 367 355 360 359Gross heating value, (kJ kg�1) 45 605 44 923 44 635 44 7

common-rail diesel engine, exhaust gas recirculation (EGR) for NOx control,a closed-coupled pre-catalyst and an underfloor main oxidation catalyst also usedfor PM control. This combustion and emission control technology is the mostwidespread one in diesel cars today. The main engine and vehicle characteristics areprovided in Table 3.

2.2. Testing facility – particle sampling

Emission measurements were conducted following the European regulations(Directive 70/220/EEC and amendments). The exhaust gas was primarily diluted andconditioned by means of Constant Volume Sampling (CVS). A 6 m long corrugatedstainless steel tube transferred the exhaust from the tailpipe to the CVS tunnel inlet.The tube was insulated to minimize heat losses and particle thermophoresis andwas clamped onto the vehicle exhaust pipe with a metal-to-metal connection toavoid exposing the hot exhaust gas to any synthetic material connectors. A flowrateof 500 Nm3/h was maintained in the CVS tunnel by a positive displacement pump.The dilution air was filtered through a HEPA class H13/EN1822 filter at the inlet ofthe dilution tunnel. Proportional diluted exhaust samples were collected in bags forgaseous pollutants measurements. Gaseous pollutants were measured with labo-ratory analyzers (chemiluminescence for NOx, flame ionization detector for HC andnon dispersive infrared for CO and CO2).

Fuel consumption was derived by means of the exhaust-to-fuel carbon balance,taking into account the oxygen content of fuels. Although regulations foreseecorrection values for other fuels, in addition to diesel and gasoline, there is nomethodology proposed for biodiesel hence the following formula was used:

FC ¼�

ECO2

44:011þ ECO

28:011

�ð12:011þ 1:008rH:C þ 16:000rO:CÞðg=kmÞ

Where: E: stands for the emissions of CO2 and CO as calculated from the bag samplesin (g/km); rH:C is the hydrogen to carbon ratio in the fuel molecule. This was 1.85 forthe base fuel and 1.865 for the biodiesel blends; rO:C is the oxygen to carbon ratiowhich was zero for the base fuel and 0.011 for the biodiesel blend.

PM samples were collected on 47 mm PTFE-coated fibre filters (PallflexTX40HI20-WW) which were conditioned for 24 h in a constant temperature (22 �C)and humidity (40%) room before and after the particulate collection and prior toweighing.

Fig. 1 presents the setup used for airborne particle sampling. Aerosol sampleswere taken from the CVS through an FPS-4000 dilutor (Dekati Ltd, Finland), oper-ating at a nominal dilution ratio of 12:1. This was followed by two calibrated ejector-type diluters (Giechaskiel et al., 2004) used in series, in order to reduce the particlenumber concentration within the measuring range of the instruments. A Conden-sation Particle Counter (CPC, Model 3010, TSI Inc, Shoreview, MN) was employedduring the transient driving cycles to monitor the total particle number concen-tration. The CPC was replaced by a scanning mobility particle sizer (SMPS, Model3936L, TSI Inc, Shoreview, MN) during the steady-state tests (50–90–120 km/h), tomonitor the mobility size distribution. It is obvious that once the mobility sizedistributions are obtained, the total particle number can also be calculated. Ona different sampling branch, an Electrical Low Pressure Impactor (ELPI) (Dekati Ltd,2001) provided the aerodynamic size distribution in real time. The ELPI was con-nected downstream of a Dekati thermodenuder, operating at 250 �C, which removedvolatile and semi-volatile exhaust components to sample non-volatile particles only.The ELPI operated with oil-soaked sintered plates and a filter stage that extended thelower cutpoint to w7 nm (Marjamaki et al., 2002).

In the following sections, particle number emissions measured by the CPC, theELPI and the SMPS are expressed per unit of distance driven (km�1). The conversionfrom cm�3 to km�1 is done by multiplying the total number of particles measuredwith the diluted exhaust volume through the CVS, and dividing with the corre-sponding distance driven by the car during the measurement. The particle

SUME B10 UFOME B10 SME EN 590 limits Test methods

4.21 3.97 2.00–4.50 EN ISO 31049 0.839 0.838 0.820–0.845 EN ISO 3675

64 66 55 min EN 2271918 21 50 max EN ISO 2084697 118 200 max EN ISO 12937�6 �8 þ5 max EN 116

ts of the measurement) 51 min EN ISO 5165EN ISO 3405

178 180219 226276 289342 342358 361

48 44 814 44 777 IP 12

Table 2Fuel properties of the methylesters used in the study and limit values according to EN 14124.

Properties PME RME SUME UFOME SME EN14214 limits Test methods

Viscosity (40 �C, mm2 s�1) 3.7 4.5 4.4 4.9 4.4 3.5–5.0 EN ISO 3104Density, (15 �C, g cm�3) 0.876 0.884 0.886 0.879 0.882 0.860–0.900 EN ISO 12185Flash point, (�C) 125 182 174 167 189 120 min EN ISO 3679Sulphur content, (mg g�1) 2.6 4 4 6 3 10.0 max EN ISO 20846CFPP, (�C) 0 �3 �7 0 �5 �5max EN 116Cetane number 56.7 54.6 52 59.6 52 51 min EN ISO 5165Iodine number 42 119 130 92 124 120 max EN 14111Acid value (mg KOH g�1) 0.22 0.14 0.48 0.45 0.49 0.50 max EN 14104Monoglyceride content (% m/m) 0.60 0.33 0.43 0.43 0.51 0.80 max EN 14105Diglyceride content (% m/m) 0.23 0.09 0.12 0.14 0.07 0.20 max EN 14105Triglyceride content (% m/m) 0.16 0.08 0.09 0.24 0.08 0.20 max EN 14105Free glycerol (% m/m) 0.02 0.003 0.01 0.03 0.002 0.02 max EN 14106Total glycerol (% m/m) 0.21 0.11 0.23 0.18 0.15 0.25 max EN 14105Ester content (% m/m) 97.1 96.5 99.6 96.5 96.5 96.5 min EN 14103Linolenic acid methyl ester (% m/m) 0.12 3.3 0.9 0.7 5.7 12 max EN 14103Gross heating value (kJ kg�1) 40 166 37 928 38 827 39 329 39 036 – IP 12

G. Fontaras et al. / Environmental Pollution 158 (2010) 1451–1460 1453

concentration was corrected for the total dilution ratio downstream of the CVS,depending on the instrument considered (ELPI 12:1, CPC and SMPS 1200:1). In thiscalculation, the ELPI concentration has been estimated with the assumption of unitparticle density, which is a common assumption in relevant studies to report ELPIresults (e.g. Shi et al., 1999; Maricq et al., 2000; Mathis et al., 2005a). Thisassumption may lead to an overestimation of the absolute level of particle emissions(Ahlvik et al., 1998), which means that interpretation of the ELPI results should bemade on a relative scale (i.e. comparison of measurements conducted by the ELPI)rather than on an absolute scale (i.e. comparison with other instruments).

2.3. Test cycles and protocol

Both the type-approval (the New European Driving Cycle – NEDC) and real-world driving cycles were involved for the complete assessment of the biodieselimpact on emissions. The real-world driving cycles tested were developed in theframework of the ARTEMIS project (Andre et al., 2006) and they are consideredrepresentative of city (URBAN), rural (ROAD), and highway (MOTORWAY) drivingconditions in Europe. The speed vs. time profiles of the type-approval and the real-world cycles are presented in Fig. 2.

The daily measurement protocol started with the NEDC, which is a cold-startdriving cycle. This comprises two parts; an urban part (UDC) where the engine startsfrom room temperature and an extra-urban part (EUDC), which aims at testing thecar at higher than urban speeds. In our protocol, the NEDC was followed by a hot-start UDC to compare with the cold-start UDC. The hot-start UDC was then followedby the three Artemis cycles, and the measurement day was concluded with testsover steady-speeds (50, 90, and 120 km/h). Each steady-speed test consisted of10 min of conditioning and 15 min of testing. No regulated pollutants weremeasured during the steady-speed modes. This protocol was repeated twice per fuelwhile two sets with the base fuel (petro-diesel) were conducted at the beginningand end of the campaign.

2.4. Statistical analysis

The results of the measurements were analysed using one-way analysis ofvariance (ANOVA), in an effort to reveal if the test fuels led to statistically significantdifferences on the emission levels. This method examines the null hypothesis thatthe mean values of the test fuels emissions are the same as those of the baseline fuel,against the alternate hypothesis of inequality. Because the number of test repetitions(observations) for each fuel-cycle combination was small (2 for each blend and 4 for

Table 3Test vehicle characteristics.

Vehicle model Renault Laguna

Emission standard compliance 98/69/EC Step I (Euro III)Model year 2001–2002Engine model 1.9 dCi (F9Q718)Capacity [cm3] 1870Cylinders/Valves per cylinder 4/2Max. power [kW/rpm] 79/4000Max torque [Nm/rpm] 250/1750Fuel system Common rail direct injection (1350 bar)Vehicle weight [kg] 1315Exhaust aftertreatment Pre-catalyst and an underfloor

main oxidation catalyst

the base fuel over 5 driving cycles), there was a high risk of type-b error, i.e.accepting the null hypothesis although this may be false. In order to increase thenumber of observations, each individual result over a particular cycle was normal-ized against the average value of the reference fuel results over the same cycle. Thisprocedure resulted in actually comparing the effect of biodiesel application ratherthan the actual emissions allowing for a global assessment of biodiesel impact in theexpense of more detailed per cycle analysis. In this approach, tests over differentdriving cycles are pooled in the same sample, increasing the number of observationsto ten per fuel (8 in the case of HC and CO as the cold-start cycle was not consideredfor these pollutants).

3. Results and discussion

3.1. Absolute emission levels with base diesel and biodiesel blends

Mean emission levels per driving cycle with the base diesel andthe B10 blends are shown in Fig. 3(a–f). Limited differences of CO2

and fuel consumption were observed over all driving cycles (Fig. 3aand b). Mean tailpipe CO2 emissions were 153 g/km with the basediesel, compared to 151 g/km for B10 over NEDC. Overall, thedifferences ranged from �0.6% (NEDC) to þ0.3% (Artemis), whichare practically within the measurement uncertainty. The derivedfuel consumption more or less followed the same pattern with CO2.However, due to the oxygen content and the lower energy content,about 1% higher fuel consumption on a volumetric basis is calcu-lated for the B10 fuels.

HC, and in particular CO, are emitted at significant quantitiesonly over the cold-start UDC and NEDC (Fig. 3c and d). Their levelsare much lower over the hot-start cycles. Use of the biodiesel leadsto an increase of both CO and HC, on the order of 25% for cold-startcycles. Marginal increases, up to 10% are observed for HC over hotcycles.

Biodiesel effects on NOx emissions (Fig. 3e) are more interesting,as diesel vehicles are prime contributors to urban NOx concentra-tions. Base and B10 NOx emissions were 0.46 g/km and 0.44 g/kmover the NEDC, respectively. Highest NOx emissions were observedover the Artemis Urban cycle, at w0.97 g/km for both the basediesel and the B10 fuels. Over the Artemis Road and ArtemisMotorway cycles, base diesel NOx emissions were 0.48 and 0.61 g/km, compared to 0.50 and 0.63 g/km on average respectively for theB10 fuels. The B10 effect appears therefore quite variable, witha slightly positive effect over the certification test and a rathernegative effect over the higher-speed Artemis driving cycles.Overall the trend recorded was towards slightly higher NOx whenoperating with B10, observation that agrees with most studies inthe field (EPA, 2002; Agarwal, 2007; Kousoulidou et al., 2008).

On the other hand, PM (Fig. 3f) consistently decreased whenusing biodiesel blends. B10 fuels led to average reductions of the

Fig. 1. Particle sampling experimental setup.

G. Fontaras et al. / Environmental Pollution 158 (2010) 1451–14601454

order of 22% (0.039–0.032 g/km) over the NEDC. A maximumreduction of 34% was observed over Artemis Urban, dropping to16% over Artemis Road, while a zero effect was seen over ArtemisMotorway. The decrease in PM mass was also associated withchanges in the particle number concentration (Fig. 4). Non-volatileand total particle number emissions were on the order of 1014 for allcycles and fuels tested, and they seemed to decrease for the B10fuels. In principle, the particle number followed similar trends toPM mass over the various driving cycles. Over the Artemis

Fig. 2. Speed profile vs. time of driving c

Motorway in particular, the B10 led to the same total numberconcentration with the base diesel, followed by a large measure-ment variability, as indicated by the wide error bars.

3.2. Effect of feedstock on emissions

Biodiesel fuels are mainly comprised of six different FAMEs atdifferent proportions. These are derivatives of palmitic acid, stearicacid, oleic acid, linoleic acid, and linolenic acid, with palmitoleic

ycles employed in the test protocol.

Fig. 3. Average emissions with diesel fuel and the B10 blends expressed in g/km. Error bars correspond to the standard deviation.

G. Fontaras et al. / Environmental Pollution 158 (2010) 1451–1460 1455

acid alkyl ester present in small amounts. The fatty acid profilevaries with feedstock. Generally, tropical oils, such as palm, mainlyconsist of saturated fatty acids, rapeseed oils are high in mono-unsaturated fatty acids, while soybean and sunflower oils mainlyconsist of poly-unsaturated fatty acids. Due to the variable chemicalcharacter, it is necessary to study how the different componentsinfluence the biodiesel properties and, in turn, how these affectemissions.

In order to establish statistically significant conclusions for theanalysis, the significance levels (p-values) calculated by the ANOVA

method are summarised in Table 4. Values lower than 0.05 indicatethat there is sufficient evidence to reject the null hypothesis ata 95% confidence level, meaning that the B10 values are statisticallydifferent to the base fuel. In addition, Table 5 presents the meandifferences between B10 and the base diesel in cases of p < 0.05.

The results of the statistical analysis show that practically allfuels led to differences in the emission levels for at least some of thepollutants. All differences are better seen in Fig. 5, distinguished perdriving cycle. Results are presented normalized over the base fuelwhich therefore always receives the value 1. The error bars stand

Fig. 4. (a) Non-volatile and (b) total (volatile, semi-volatile, non-volatile) particlenumber emissions. Error bars correspond to the standard deviation. Non-volatilenumber appears higher than total one, due to the assumption of unit particle density inthe ELPI calculation.

Table 5Relevant emission change with use of B10 blends over the base fuel, in case ofstatistically significant differences.

Blend CO2 CO HC NOx PM PN (solid) PN (total)

PME – – þ20% þ8% �19% �17% �24%RME – – – – �15% �21% –SUME þ4% – þ28% þ11% �17% �16% �29%UFOME �2% – – �6% �24% – �20%SME – – þ22% �7% �13% – –

G. Fontaras et al. / Environmental Pollution 158 (2010) 1451–14601456

for the maximum and minimum values observed in each case. Thefollowing sections discuss the trends established by the statisticalanalysis.

3.2.1. Regulated pollutants and CO2

The analysis showed limited correlation between biodieselfeedstock and emissions in the case of CO2 emissions. Only UFOMEand SUME led to statistically significant effects, by�2% and þ4% onaverage, respectively. These blends showed a consistent behaviourbetween driving cycles, with UFOME being consistently lower andthe SUME consistently higher than the base fuel. It is difficult to

Table 4Significance levels (p-values) of the ANOVA test. Values below 0.05 designatea statistically significant difference with the base fuel at a 95% confidence level.

Blend CO2 CO HC NOx PM PN (solid) PN (total)

PME 0.11 0.78 0.01 0.03 0.00 0.00 0.00RME 0.21 0.07 0.14 0.37 0.03 0.00 0.20SUME 0.00 0.77 0.01 0.00 0.00 0.00 0.00UFOME 0.01 0.29 0.74 0.02 0.00 0.47 0.04SME 0.51 0.08 0.02 0.01 0.01 0.31 0.38

draw solid conclusions that would explain this behaviour based onthe properties of each blend. Moreover, results with the otherbiodiesel blends lean towards a rather neutral effect on CO2. Furtherinvestigation on the factors associated with biodiesel combustionand efficiency will prove valuable to understand biodiesel behav-iour in this context.

Use of PME, SUME and SME seems to negatively affect HCemissions, causing an increase in the range of 20–28%, for differentdriving cycles. The reasons for HC increase may be twofold. First,poor volatilization and combustibility of specific fuel componentsmay be responsible for higher in-cylinder HC production. Thepresence of methyl ester components in diesel fuel may alter thevolatility and distillation characteristics of the final blend. It wasfound that the distillation curve of the blends changed with theaddition of biodiesel. A comparison showed that PME, RME, SMEand SUME blends had increased boiling points compared to dieselfuel, with the exception of the final boiling point which decreasedin all cases. On the other hand UFOME presented the same orslightly lower boiling points as diesel fuel (Table 1). Contrary to theother blends which led to increases in HC emissions, the use ofUFOME over most cycles led to reduced emissions raising the issueof the impact of the boiling point in HC emissions. Second, thecatalyst selectivity in specific species may also have an effect on theincreased HC tailpipe concentrations. It should be commented thatHC increase does not agree with what is generally known aboutbiodiesel. HC emissions reduction may be a fact when it comes toengine exhaust but is not necessarily the case when it comes tovehicle tailpipe emissions where the effect of aftertreatmentsystems takes also place. Studies performed on diesel vehiclesequipped with oxidation catalysts show that biodiesel may havea deteriorating effect on HC, particularly over cold-start conditions(Martini et al., 2007; Mazzoleni et al., 2007; Fontaras et al., 2009). Inany case, HC emission levels of this car were in general very low andthe limited increases do not raise particular concerns regarding airquality. This is also the case with CO: the absolute emission levelwas too low to identify statistically significant effects of the bio-diesel blends. It should be noted that in the case of an older vehiclethat lacks exhaust aftertreatment system or has a non-functioningone, biodiesel application may reduce CO and HC emissions(reduced engine out emissions). Even so emission levels willremain much higher than those foreseen by recent Europeanemission standards so the application of aftertreatment systems isabsolutely necessary.

It is more important from a diesel emissions perspective tostudy the impact of B10 fuels on the NOx and PM emissions. Allblends but RME were found to result in NOx emission changes.SUME and PME led to increases of 11% and 8% on average, whereasSME and UFOME led to a decrease of w7%. These decreases in NOxemissions are not unexpected as other researchers also mentionthat NOx emissions may decrease with biodiesel over certaindriving conditions (Wang et al., 2000; Martini et al., 2007; Lapuertaet al., 2008a; Fontaras et al., 2009). Similar trends regarding NOxwere observed over most driving cycles.

The impact of biodiesel on NOx may be potentially explained byseveral mechanisms. According to Fernando et al. (2006) feedstocks

Fig. 5. Emissions of each fuel tested reduced to the average baseline emissions (Diesel ¼ 1). Error bars correspond to the maximum and minimum values.

G. Fontaras et al. / Environmental Pollution 158 (2010) 1451–1460 1457

containing unsaturated fatty acid chains produce significantlyhigher NOx emissions than more saturated components. SUMEconsists of highly unsaturated components, as indicated by the highiodine number (Table 2) of this fuel. Therefore, the increase in NOxmay result from the participation of the double bonds in thecombustion chemistry. On the other hand, PME also led to NOxincreases despite its mostly saturated character (lowest iodinenumber in Table 2). A different mechanism of NOx increase may be

associated with glycerol presence, as concentrations of glycerolhave been associated with changes in the ignition delay andsubsequent NOx increases (Hamasaki et al., 2004). A glycerol-basedmechanism is plausible for both SUME and PME fuels, as they areassociated with the highest total glycerol (free and bonded glyc-erols) content of all blends tested. At this point someone couldargue that the increased cetane number (e.g. palm-oil biodiesel)could also be affecting NOx as it may result in delayed combustion

G. Fontaras et al. / Environmental Pollution 158 (2010) 1451–14601458

and increased peak cylinder temperature. However as presented inTable 1 all blends tested had almost the same cetane number. Inaddition tests performed on a similar test cell engine (Kousoulidouet al., 2009) indicated limited differentiations in ignition delay.Therefore the evolution of NOx is unlikely to be related with fuelcetane number.

The effect of B10 blends on PM was more consistent than NOx.All blends led to a decrease in PM on the order of 13–24%. UFOMEand PME led to the highest reductions, while SME and RME to thelowest ones. It appears that the more saturated the blend, thelowest the PM emissions, an observation also reported by Schon-born et al. (2009).

3.2.2. Airborne particle characteristicsThe biodiesel effect on PM may be better explained by sepa-

rately looking on the non-volatile and total (non-volatile, volatile,and semi-volatile) particle emission rates. All biodiesel blends,except SME, affected either the non-volatile or the total particlenumber concentration. PME, RME, and SUME led to reductions inthe non-volatile particle number concentration, which wereconsistent throughout most driving cycles. In general, thesereductions are consistent with the corresponding reduction in PMmass for all cycles except the Artemis Motorway, which means thatthe fuel may change the combustion chemistry towards reducingthe soot (non-volatile) particle emissions. This may be an effect ofthe oxygen inhibition to soot formation (Jung et al., 2006; Di et al.,2009). The reduction in non-volatile particle number and PM maybe attributed to increased combustion temperatures due to thepresence of oxygen in the ester molecules, differentiations in air-fuel ratio, and possibly by changes occurring in the combustionchemistry. Similar observations regarding particle mass andnumber are reported in other studies (Martini et al., 2007; Lapuertaet al., 2008a).

No consistent effect of biodiesel on total particle number wasidentified for any of the fuels over all cycles. For low power cycles(NEDC, Artemis Urban and Road) the total number concentration ofall fuels rather follows the non-volatile particle number concen-trations. The highest reductions originate from SUME, as in the caseof non-volatile particles. On the other hand, the reductions ach-ieved with RME are marginal, although RME reduced the non-volatile particle number by w20%. This indicates that a significantnumber of volatile or semi-volatile particles is produced with RMEeven under these low power cycles. Both increases and decreases oftotal particle number have been observed in other studies. Lin et al.(2008) tested palm-biodiesel in a heavy-duty engine and founda slight increase in particles with diameter <310 nm. Chung et al.

4.0E+13

8.0E+13

1.2E+14

1.6E+14

2.0E+14

10

Mobility

To

ta

l N

[#

/km

/dlo

gD

p]

120 kph

0.0E+00

Fig. 6. Mobility size distribution of total (volatile, semi

(2008) also found increases in small particles emitted duringoperation of a small non-road engine fuelled with biodiesel.However, Bagley et al. (1998) and Lapuerta et al. (2008b) observedeither an increase or a decrease in nanoparticles, as a function ofthe operation point.

The variable effect of biodiesel on total particle number can beexplained by better studying the differences between non-volatileand total particle number. Our study shows that most blendsdecrease the non-volatile particle number at low-load conditionsand this is rather followed by the total particle number. However,this observation changes when looking at the high-power ArtemisMotorway cycle. Two of the five blends led to a much higher totalparticle number (by up to 3 times) compared to the base diesel.Such high particle number concentration can be only explained bythe formation of nucleation mode particles downstream of thecombustion chamber. In order to obtain a better understanding ofthe characteristics of these new particles, Fig. 6 shows the mobilitysize distribution with the vehicle running at 120 km/h. This oper-ation condition is a high-speed one and stands for the steady-speedequivalence to the Artemis Motorway cycle. This figure shows thatthe RME and the PME blends result in the formation of a secondnucleation mode at <50 nm size range, which is responsible for thehigh total number concentration. This may actually form asa consequence of the reduction in non-volatile particle numberfrom these fuels (Fig. 5e).

Non-volatile particles act as condensation sites which reducethe super-saturation of volatile components. In the absence of suchsites, volatile and semi-volatile species tend to condense to formnew particles (nucleation). RME in particular contains unsaturatedcomponents, and specifically methyl oleate (C18:1). Unsaturatedcomponents have been shown to increase nucleation mode parti-cles, possibly because of the emission of organic species with highboiling points which tend to be more condensable than fossil dieselcomponents (Schonborn et al., 2009). Such nucleation may befavoured at high speed to the higher emission rate of suchcomponents and the high ion production of the catalyst (ion-induced nucleation). In particular, engine load increases at highspeed and this raises the temperature of the exhaust gas and, byturn, the temperature of the aftertreatment devices. This increasesthe catalyst conversion rate of sulphur to sulphates, which may actas active nanosites for volatile and semi-volatile material tocondense and develop a distinct nucleation mode (Vouitsis et al.,2005). Increase in the number of particle number with the use ofRME blends compared to fossil diesel, has been also seen in most ofthe relevant studies. A good summary of these is provided by(Lapuerta, et al., 2008a) and has been attributed to the emission

100 1000

Diameter [nm]

Baseline

SME

UFOMEPME

SUME

RME

-volatile, and non-volatile) particles, at 120 km/h.

Table 6Ratio of average emissions over UDC cold and UDC hot (cold/hot).

Fuel CO HC PM

Base diesel 52 3.6 1.2Average B10 28 4.6 1.2

G. Fontaras et al. / Environmental Pollution 158 (2010) 1451–1460 1459

and condensation of semi-volatile components. Finally, it should bestated that the variability in biodiesel effects as a function of theengine operation point is potentially the reason for the inconclusiveresults with regard to biodiesel impacts on total particle numberfrom the various literature studies. In this direction, samplingconditions also have a role to play, as they may affect the relevantstrength of nucleation mode vs. accumulation mode particleconcentration. Therefore, the particle concentration in the nucle-ation mode might be different for atmospheric dilution or withdifferent setting of the sampling conditions (dilution ratio,temperature, residence time) in the laboratory, than what isreported in this study.

3.3. Biodiesel and cold-start

Emission levels of modern diesel cars before the engine warm-up and the catalyst light-off may be substantially higher thanduring normal driving (Mathis et al., 2005a). It is therefore inter-esting to explore the effect of biodiesel under such transientconditions. Fig. 7 presents the NEDC emission ratios of B10 over thebase diesel for CO and HC. All blends led to rather proportionalemission increases for both pollutants over this cold-start drivingcycle. Such an increase may originate either from an increase inengine out emissions or from a reduction in the aftertreatmentefficiency when using biodiesel.

In order to further explore the effect of cold-start, Table 6 showsthe UDC cold over UDC hot emission ratios both in the case of basediesel and the B10 fuel. This comparison shows that the propor-tional increase of emissions during cold-start are less for CO andmore for HC when using biodiesel, compared to base diesel. Theopposite trends in CO and HC ratios between B10 and base dieselindicate a different operation of the oxidation catalyst in the twocases. It shows that CO is preferentially oxidized in the case ofdiesel, and the catalyst efficiency increases for HC in the case ofbiodiesel. One can assume several potential mechanisms for thistrade-off, including catalyst selectivity to biodiesel HC, differentextent of HC adsorption in the two cases, or oxidation competitionbetween the two pollutants in the two cases. However, theimportant message from Table 6 is that oxidation aftertreatmentmay perform differently than designed when using biodieselblends. Our results indicate that this is more relevant for HC and COthan PM, which is reduced by w20% by the fully warmed-upcatalyst in both cases.

SME-10

UFOME-10

PME-10

SUME-10

RME-10

1

1.1

1.2

1.3

1.4

1.5

1.6

1.7

1.8

1 1.2 1.4 1.6 1.8

B10 average CO / Baseline average CO

B10

ave

rage

HC

/ B

asel

ine

aver

age

HC

Fig. 7. Comparison of B10 effect on CO and HC emissions over NEDC.

4. Conclusions

Five different biodiesel blends were tested in this study in orderto examine their impact on the emissions and consumption ofa common-rail passenger car. Limited effects were observed on CO2

emissions, with only two (PME, SUME) of the five blends providingstatistically significant differentiations. The results confirm that nodifferences should be expected in direct CO2 emissions, on a fuelenergy content basis (over the same driving conditions), as can alsobe predicted from combustion stoichiometry.

The effect of biodiesel on HC and CO was more prominent overthe cold-start driving cycles, where absolute emission levels arehigh. Over such cycles, B10 fuels resulted in w25% higher HC andCO emissions than base diesel. The cold-start vs. hot driving cyclesemission ratios for CO and HC are different for the base diesel andthe biodiesel blends. This is an indication of variance in theoxidation aftertreatment performance when using biodiesel.

The effect of different biodiesel blends on NOx emissions wasvariable, ranging from �7% to þ11% on average, depending on theorigin of the feedstock. Total glycerol content in the fuel seemed topositively correlate with NOx emissions, an observation whichpotentially calls for better control of this property by the fuelstandards.

The biodiesel blends had a consistent positive effect in reducingthe mass of PM, from 13% to 25%. This was accompanied by a similarreduction in non-volatile particle number, except over the ArtemisMotorway cycle. Having in mind the forthcoming European legis-lation which poses a limit in total non-volatile particle number,diesel blending with biodiesel may have a positive effect inachieving the regulated limits. The lower PM mass emission ratemay also have a positive effect in cars equipped with particle filters,as this may extend the filter loading mileage and may reduce theregeneration frequency increasing the filter’s mileage expectancy.

The effect of biodiesel on total particle number, includingvolatile and semi-volatile particles, was more variable. Althoughreductions were observed over the low power cycles, the PME andRME blends were associated with up to 3 times higher particlenumber emissions that the base diesel over the Artemis Motorwaytest. The fact that this occurs with the most widespread biodiesel inEurope today (RME) raises some concerns and its extent needs to bethoroughly studied.

Acknowledgements

The authors would like to thank Vert Oil SA for their support andMr. Christos Samaras for his help in the data analysis. This paper isdedicated to the memory of Prof. Stamoulis Stournas.

References

Andre, M., Joumard, R., Vidon, R., Tassel, P., Perret, P., 2006. Real-world Europeandriving cycles, for measuring pollutant emissions from high- and low-poweredcars. Atmospheric Environment 40, 5944–5953.

Agarwal, A.K., 2007. Biofuels (alcohols and biodiesel) applications as fuels forinternal combustion engines. Progress in Energy and Combustion Science 33,233–271.

Ahlvik, P., Ntziachristos, L., Keskinen, J., Virtanen, A., 1998. Real Time Measurementsof Diesel Particle Size Distribution with an Electrical Low Pressure Impactor,SAE Technology Paper 980410.

G. Fontaras et al. / Environmental Pollution 158 (2010) 1451–14601460

Bagley, S.T., Gratz, L.D., Johnson, J.H., McDonald, J.F., 1998. Effects of an oxidationcatalytic converter and a biodiesel fuel on the chemical, mutagenic, and particlesize characteristics of emissions from a diesel engine. Environmental Science &Technology 32, 1183–1191.

Chung, A., Lall, A.A., Paulson, S.E., 2008. Particulate emissions by a small non-roaddiesel engine: biodiesel and diesel characterization and mass measurementsusing the extended idealized aggregates theory. Atmospheric Environment 42,2129–2140.

Dekati Ltd, 2001. Sampling Automotive Exhaust with a Thermodenuder. DekatiTechnical Note.

Di, Y., Cheung, C.S., Huang, Z., 2009. Experimental investigation on regulated andunregulated emissions of a diesel engine fueled with ultra-low sulfur diesel fuelblended with biodiesel from waste cooking oil. Science of the Total Environ-ment 407, 835–846.

Durbin, T.D., Cocker Iii, D.R., Sawant, A.A., Johnson, K., Miller, J.W., Holden, B.B.,Helgeson, N.L., Jack, J.A., 2007. Regulated emissions from biodiesel fuels fromon/off-road applications. Atmospheric Environment 41, 5647–5658.

EPA, 2002. A comprehensive analysis of biodiesel impacts on exhaust emissions,Draft Technical Report. In: Agency, U.S.E.P. (Ed.).

Eurobserver, 2008. Biofuels barometer, p. 18.Fernando, S., Hall, C., Jha, S., 2006. NOx reduction from biodiesel fuels. Energy and

Fuels 20, 376–382.Fontaras, G., Karavalakis, G., Kousoulidou, M., Tzamkiozis, T., Ntziachristos, L.,

Bakeas, E., Stournas, S., Samaras, Z., 2009. Effects of biodiesel on passenger carfuel consumption, regulated and non-regulated pollutant emissions over leg-islated and real-world driving cycles. Fuel 88, 1608–1617.

Giechaskiel, B., Ntziachristos, L., Samaras, Z., 2004. Calibration and modelling ofejector dilutors for automotive exhaust sampling. Measurement Science andTechnology 15, 1–8.

Graboski, M., McCormick, R., 1998. Combustion of fat and vegetable oil derived fuelsin diesel engines. Progress in Energy and Combustion Science 24, 125–164.

Hamasaki, K., Kinoshita, E., Tajima, H., 2004. Effects of water droplet size and crudeglycerin content as an emulsifier on diesel combustion with emulsified biodiesel.Transactions of the Japan Society of Mechanical Engineers 70, 2907–2913.

Jung, H., Kittelson, D.B., Zachariah, R., 2006. Characteristics of SME biodiesel-fueleddiesel particle emissions and the kinetics of oxidation. Environmental Science &Technology 40, 4949–4955.

Karavalakis, G., Alvanou, F., Stournas, S., Bakeas, E., 2009. Regulated and unregulatedemissions of a light duty vehicle operated on diesel/palm-based methyl esterblends over NEDC and a non-legislated driving cycle. Fuel 88, 1078–1085.

Kawanaka, Y., Matsumoto, E., Sakamoto, K., Wang, N., Yun, S.J., 2004. Size distri-butions of mutagenic compounds and mutagenicity in atmospheric particulatematter collected with a low-pressure cascade impactor. Atmospheric Environ-ment 38, 2125–2132.

Knothe, G., Steidley, K., 2005. Kinematic viscosity of biodiesel fuel components andrelated compounds, influence of compound structure and comparison to pet-rodiesel fuel components. Fuel Processing Technology 84, 1059–1065.

Kousoulidou, M., Fontaras, G., Ntziachristos, L., Mellios, G., 2008. Effect of Biodieseland Bioethanol on Exhaust Emissions. ETC/ACC Technical Paper 2008/5. Euro-pean Topic Center on Air and Climate Change.

Kousoulidou, M., Fontaras, G., Ntziachristos, L., Samaras, Z., 2009. Evaluation ofBiodiesel Blends on the Performance and Emissions of a Common-rail Light-duty Engine and Vehicle. SAE Technical Paper No.2009-01-0692.

Lapuerta, M., Armas, O., Rodrnguez-Fernandez, J., 2008a. Effect of biodiesel fuels ondiesel engine emissions. Progress in Energy and Combustion Science 34, 198–223.

Lapuerta, M., Rodriguez-Fernandez, J., Agudelo, J.R., 2008b. Diesel particulateemissions from used cooking oil biodiesel. Bioresource Technology 99, 731–740.

Lin, Y.C., Lee, C.F., Fang, T., 2008. Characterization of particle size distribution fromdiesel engines fueled with palm-biodiesel blends and paraffinic fuel blends.Atmospheric Environment 42.

Lujan, J.M., Bermıdez, V., Tormos, B., Pla, B., 2009. Comparative analysis of a DIdiesel engine fuelled with biodiesel blends during the European MVEG-A cycle:performance and emissions (II). Biomass and Bioenergy 33, 948–956.

Maricq, M.M., Podsiadlik, D.H., Chase, R.E., 2000. Size Distributions of Motor VehicleExhaust PM: a Comparison Between ELPI and SMPS Measurements. Taylor &Francis, pp. 239–260.

Marjamaki, M., Ntziachristos, L., Virtanen, A., Ristimaki, J., Keskinen, J., Moiso, M.,Palonen, M., Lappi, M., 2002. Electrical Filter Stage for the ELPI. SAE Paper 2002-01-0055.

Marshall, W., Schumacher, L., Howell, S., 1995. Engine Exhaust Emissions Evaluationof a Cummins L10E When Fueled with a Biodiesel Blend, Society of AutomotiveEngineers. SAE Technical Paper 952363.

Martini, G., Astorga, C., Farfaletti, A., 2007. Effect of Biodiesel on Pollutant Emissionsfrom Euro 3 LD Diesel Vehicles. Institute for Environment and Sustainability –Joint Research Center.

Mathis, U., Mohr, M., Forss, A.-M., 2005a. Comprehensive particle characterizationof modern gasoline and diesel passenger cars at low ambient temperatures.Atmospheric Environment 39, 107–117.

Mathis, U., Mohr, M., Kaegi, R., Bertola, A., Boulouchos, K., 2005b. Influence of dieselengine combustion parameters on primary soot particle diameter. Environ-mental Science & Technology 39, 1887–1892.

Mazzoleni, C., Kuhns, H.D., Moosmoller, H., Witt, J., Nussbaum, N.J., Oliver Chang, M.C.,Parthasarathy, G., Nathagoundenpalayam, S.K.K., Nikolich, G., Watson, J.G., 2007. Acase study of real-world tailpipe emissions for school buses using a 20% biodieselblend. Science of the Total Environment 385, 146–159.

Ropkins, K., Quinn, R., Beebe, J., Li, H., Daham, B., Tate, J., Bell, M., Andrews, G., 2007.Real-world comparison of probe vehicle emissions and fuel consumption usingdiesel and 5% biodiesel (B5) blend. Science of the Total Environment 376, 267–284.

Schonborn, A., Ladommatos, N., Williams, J., Allan, R., Rogerson, J., 2009. Theinfluence of molecular structure of fatty acid monoalkyl esters on dieselcombustion. Combustion and Flame 156, 1396–1412.

Shi, J.P., Khan, A.A., Harrison, R.M., 1999. Measurements of ultrafine particleconcentration and size distribution in the urban atmosphere. Science of theTotal Environment 235, 51–64.

Sze, C., 2007. Impact of Test Cycle and Biodiesel Concentration on Emissions. SAE2007-01-4040.

Tat, M., 2003. Investigation of Oxides of Nitrogen Emissions from Biodiesel-fueledEngines. Mechanical Engineering Department, Iowa State University, Iowa.

Vouitsis, E., Ntziachristos, L., Samaras, Z., 2005. Modelling of diesel exhaust aerosolduring laboratory sampling. Atmospheric Environment 39, 1335–1345.

Wang, W.G., Lyons, D.W., Clark, N.N., Gautam, M., 2000. Emissions from nine heavytrucks fueled by diesel and biodiesel blend without engine modification.Science of the Total Environment 34, 933–939.

WHO, 2005. In: organization, W.H. (Ed.), Particulate matter air pollution: how itharms health, p. 4. Berlin, Copenhagen, Rome.