Emission Factors for Gas-Powered Vehicles Traveling Through Road Tunnels in São Paulo, Brazil

Transcription

1 Environ. Sci. Technol. 2006, 40, Downloded by UNIV DE SAO PAULO USP on October 29, Emission Fctors for Gs-Powered Vehicles Trveling Through Rod Tunnels in São Pulo, Brzil LEILA D. MARTINS,*, MARIA F. ANDRADE, EDMILSON D. FREITAS, ANGEÄ LICA PRETTO, LUCIANA V. GATTI, EÄ DLER L. ALBUQUERQUE, EDSON TOMAZ, MARIA L. GUARDANI, MARIA H. R. B. MARTINS, AND OLIMPIO M. A. JUNIOR Deprtment of Atmospheric Sciences, Institute of Astronomy, Geophysics nd Atmospheric Sciences, University of São Pulo, São Pulo, , Brzil, Atmospheric Chemistry Lbortory, Institute for Energy nd Nucler Reserch, Deprtment of Chemicl Process Development, School of Chemicl Engineering, Stte University of Cmpins, nd São Pulo Stte Environmentl Regultion Agency The objective of this study ws to improve the vehiculr emissions inventory for the light- nd hevy-duty fleet in the metropolitn re of São Pulo (MASP), Brzil. To tht end, we mesured vehicle emissions in rod tunnels locted in the MASP. On Mrch 22-26, 2004 nd My 04-07, 2004, respectively, CO, CO 2,NO x,so 2, nd voltile orgnic compounds (VOCs) emissions were mesured in two tunnels: the Jnio Qudros, which crries light-duty vehicles; nd the Mri Mluf, which crries light-duty vehicles nd hevy-duty diesel trucks. Pollutnt concentrtions were mesured inside the tunnels, nd bckground pollutnt concentrtions were mesured outside of the tunnels. The men CO nd NO x emission fctors (in g km -1 ) were, respectively, 14.6 ( 2.3 nd 1.6 ( 0.3 for light-duty vehicles, compred with 20.6 ( 4.7 nd 22.3 ( 9.8 for hevy-duty vehicles. The totl VOCs emission fctor for the Mri Mluf tunnel ws 1.4 ( 1.3gkm -1. The min VOCs clsses identified were romtic, lkne, nd ldehyde compounds. For the hevy-duty fleet, NO x emission fctors were pproximtely 14 times higher thn those found for the lightduty fleet. This ws ttributed to the high levels of NO x emissions from diesel vehicles. Introduction It is well-known tht vehiculr emissions mke significnt contribution to ir contmintion in urbn res. In the metropolitn re of São Pulo (MASP), megcity locted in southern Brzil, vehiculr emissions re responsible for pproximtely 98, 97, nd 96%, respectively, of ll emissions of crbon monoxide (CO), hydrocrbons (HCs) nd nitrogen oxides (NO x) (1). There re three different methods of estimting such emissions: dynmometer tests of individul * Corresponding uthor phone: ; fx: ; e-mil: leildro@model.ig.usp.br. University of São Pulo. Institute for Energy nd Nucler Reserch. Stte University of Cmpins. São Pulo Stte Environmentl Regultion Agency. vehicles; mesurements tken in tunnels; nd on-bord monitoring (2). Emission fctors obtined from in-tunnel mesurements represent the bulk of emissions from lrge number of vehicles of vrious types used in urbn res. In ddition, these emission fctors cn be useful s reference points in discussions regrding emissions inventories nd vehiculr emission control strtegies. Estimting emission fctors from in-tunnel mesurements hs been proposed by vrious uthors (2-7). In-tunnel mesurements llow the ctul emissions for vriety of vehicles to be obtined under certin driving conditions (hot strt) nd provide informtion tht complements dynmometer test results. The MASP hs vehicle fleet of pproximtely 7.8 million vehicles, of which 84.9% re light-duty vehicles, 5.8% re hevy-duty vehicles, nd 9.3% re motorcycles. This fleet is unique in tht most of the vehicles re fueled by ethnol or by gsoline-ethnol mixture: 14.5% of the fleet burns hydrted ethnol (95% v/v); nd 69.5% burns mixture contining 75-78% gsoline (by volume) nd 22-25% ethnol ( blend referred to s gsohol). Therefore, pproximtely 34% of the fuel burned by the fleet is ethnol (1). Recently, flex-fuel vehicles (cpble of burning either gsohol or ethnol) nd converted vehicles (which burn compressed nturl gs) were introduced into this vehiculr fleet. At the time of this study, the concentrtion of sulfur in diesel fuel sold in lrge metropolitn res of Brzil, such s the MASP (1100 ppm), ws lower thn tht in diesel fuel sold in other res of the country (3500 ppm). In summry, the MASP presents n unconventionl mixture of vehicle types nd fuels, nd the profile of its vehicle fleet hs chnged substntilly in the lst severl yers. Due to their toxicity, vehicle-emitted voltile orgnic compounds (VOCs), especilly those such s benzene, directly or indirectly influence humn helth, s do ldehydes, romtics, nd most lkenes, which re precursors of secondry pollutnts such s ozone nd secondry orgnic erosol. This study is the first ttempt to estimte the VOCs emission fctors from vehicles in rod tunnels in Brzil. The significnce of this study is lso relted to the previously mentioned chrcteristics of the MASP vehicle fleet. The most widely consumed fuel is blend of gsoline (22-25% ethnol) nd there re more thn 1 million vehicles running on pure ethnol. The number of diesel-powered vehicles is lso significnt, minly due to high the levels of NO x nd prticulte mtter they emit. This study reports the results of field mesurements of ir pollutnts relesed in rod tunnels. The min gols of this study were s follows: to identify the chrcteristics of gseous pollutnts in rod tunnel; to mesure in-tunnel emission fctors for the MASP fleet in terms of gseous NO x, CO, nd VOCs; nd to chrcterize the VOCs emitted by the MASP fleet. Experimentl Section Smpling Loction. Field mesurements were crried out in two rod tunnels within the MASP. From Mrch 22 to 26, 2004, mesurements were tken in the Jnio Qudros (JQ) tunnel, nd from My 04 to 07, 2004, mesurements were tken in the Mri Mluf (MM) tunnel. The JQ tunnel is locted in the southwest portion of São Pulo. It is twolne tunnel tht is 1900 m in length, nd the trffic in both lnes flows in the sme direction. The in-tunnel emissions re minly from gsohol- nd ethnol-powered vehicles. The speed limit inside the tunnel is 70 km h -1, nd the ventiltion system opertes continuously. In ddition, there is slight ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 21, /es052441u CCC: $ Americn Chemicl Society Published on Web 10/04/2006

2 TABLE 1. Species Mesured in the Jnio Qudros nd Mri Mluf Tunnels, Together with Dt Regrding Methods nd Smpling Sites species mesured method nd nlyzer mesurement site NO x chemiluminescences in-tunnel nd fresh ir thermo electron (42B), 0.5 ppbv detection limit CO nondispersive infrred in-tunnel nd fresh ir techniquesthermo electron (48B), 0.1 ppm detection limit SO 2 pulse fluorescences in-tunnel thermo electron (43B), 1 ppbv detection limit CO 2 combined infrred gs in-tunnel nd fresh ir nlysissli-cor 6262 system HCs GC/FID nd GC/MS in-tunnel nd fresh ir nlysissstinless steel cnisters HCs GC/FID detections in-tunnel nd fresh ir tubes crbonyls HPLC nlysiss DNPH-crtridge in-tunnel nd fresh ir No mesurements of SO 2 in bckground mbient ir were tken in either tunnel. downhill grde. The MM tunnel is locted in the southest portion of the city. It crries light-duty nd hevy-duty vehicles burning gsohol, ethnol, or diesel. It is 1020 m in length, hs four 5.15-m wide lnes (two trveling in ech direction), nd the terrin is level. A concrete wll pproximtely 16 m in width seprtes the west-bound trffic from the est-bound trffic. At the tunnel midpoint, there is n open corridor connecting the two sides, which is where the mesurements were tken. Ventiltion fns on the roof of the tunnel operte continuously to provide fresh ir. For both tunnels, in-tunnel pollutnt concentrtions were mesured t the midpoint, nd the bckground ir ws mesured outside the tunnels. The sites outside the tunnels were locted fr from the tunnels in order to void their influence. In ddition, in the region surrounding ech tunnel, there re no other potentil sources of pollutnt emissions. Smpling nd Anlysis. On workdys, ir smples for nlysis of CO, NO x, nd SO 2 were collected continuously by the São Pulo Stte Compnhi de Tecnologi de Snemento Ambientl (CETESB, Environmentl Regultion Agency) [The officil nnul report of ir qulity in the MASP provided by CETESB is vilble t nd ir smples for nlysis of CO 2 were collected from 8:00 to 18:00. In-tunnel mesurement nd smpling methods re summrized in Tble 1. In ddition, VOCs were smpled nd nlyzed using three different methods. In the first method, HCs were smpled by drwing ir through stinless steel, vcuum cnister t 50 ml min -1. In the second method, the HCs were smpled in tube contining Tenx TA dsorbent (Supelco, Bellefonte, PA), t 35 ml min -1, using n utomted sequentil tube smpler (STS-25; Perkin-Elmer, Norwlk, CT). For mesurements of crbonyl compounds (third method), the ir ws drwn into crtridge contining 2,4-dinitrophenylhydrzine (Sep-Pk DNPH-Silic; Wters, Milford, MA) t 2 L min -1. All smplings were conducted over 2-h periods between the hours of 8:00 nd 18:00. Mesurements tken outside of the tunnels were lso conducted within the sme time frmes. Ozone scrubber filters (Wters) were used to void rtifct formtion on mbient smples. The cnisters were nlyzed in the Lbortory of Atmospheric Chemistry of the Instituto de Pesquiss Energétics e Nucleres (IPEN, Institute for Energy nd Nucler Reserch). Gs chromtogrphy-mss spectrometry (CG-MS, Vrin 3800, Sturn 2000) ws used to identify HCs, nd simultneously in the sme chromtogrphy flme ioniztion detection ws used to quntify HCs s described in detil by Pretto (8). The HCs were seprted on DB-1 column (J & W Scientific, Folsom, CA) with length of 60 m, n inside dimeter (i.d.) of 0.32 mm nd film thickness of 1 µm. After seprtion, the smples were divided between the two detectors. The quntifiction of HCs ws performed using four certified gs mixtures (Scott Specilty Gses, Plumstedville, PA). The nlysis conditions were s follows: 200 C injector; Helium, s the crrier gs, t constnt flow rte of 2.2 ml min -1 ; oven initil temperture of -50 C, incresed by 6 C min -1 until reching 200 C, then llowed to return to -50 C; 1.3-kV detector tension. Possible rtifcts due to interctions of HCs with components of mbient ir, such s ozone, wter vpor, or free rdicls re sources of errors in determintion of HCs concentrtions (9). The uncertinty of nlysis is dependent upon compound. For this nlysis, detection limits rnged from 0.03 to 0.07 ppbv for the GC- MS used, nd the reltive stndrd devition (RSD) for 40 VOCs were between 2.4 nd 8.1% (8). The tubes were nlyzed t the Stte University of Cmpins (São Pulo, Brzil) using therml desorber (ATD- 400; Perkin-Elmer) coupled to the GC-FID system (Auto- System XL; Perkin-Elmer). The HCs were seprted on CIOLA-1 column (100% dimethylpolysiloxne; 60-m length; 0.25-mm i.d.; 0.2-µm film thickness). The nlysis conditions used were s follows: crrier gs: helium 5.0; desorption temperture: 300 C; desorption flow: 60 ml min -1 ; split: 25 ml min -1 ; temperture progrm: 35 C for 10 min, Ct5 C min -1, 100 C for 2 min; detector temperture: 250 C. In ddition, experimentl error rnged from 4 to 20% for compounds nd intervls of concentrtions obtined by the method in ref 10. The nlysis of crbonyls ws performed t the University of São Pulo Institute of Chemistry using high-performnce liquid chromtogrphy (HPLC) system (SCL-10A; Shimdzu, Tokyo, Jpn). The nlyticl pprtus includes Rheodyne injection vlve with 20-µL smple loop, two pumps (LC- 10AD; Shimdzu) nd n UV/visible detector (SPD-10AV UVvis; Shimdzu) t 365 nm. Hydrzones were seprted using Shimdzu Shim-pck CLC-ODS column (15 cm 6mm 5 µm) connected to Shim-pck G-ODS precolumn operted t mbient temperture (25 C). Concentrtions of crbonyls in ir smples were clculted using the clibrtion dt of the externl crbonyl DNPH stndrds (11-13). The RSD ws estimted t 6.3% for formldehyde nd 4.4% for cetldehyde, with detection limits of ppbv nd ppbv, respectively. For the other crbonyls nlyzed, the degree of uncertinty of nlysis vried on the sme order, with detection limits rnging from 1 pptv to 0.4 ppbv. Trffic Volume. Trffic volume through the tunnel ws counted mnully t 15-minute intervls throughout the smpling dy (8:00-18:00), using supplementl video system. The dt represents every other intervl. Vehicles using the JQ tunnel were clssified s motorcycles, light pssenger vehicles, or light-duty trucks/vns, wheres those using the MM tunnel were clssified s motorcycles, light pssenger vehicles, or hevy-duty vehicles. Clssifictions were mde in function of fuel burned nd vehicle clss. Results nd Discussion Diurnl Profiles. During typicl working dy, pproximtely vehicles pss through the JQ tunnel. Most re light pssenger vehicles, pproximtely 95% of which burn gsohol or pure ethnol. Including ll four lnes, pproximtely light nd hevy vehicles pss through the MM tunnel on dily bsis. Dt describing the trffic profile for both tunnels is presented in Tble 2. Gsohol is the principl fuel burned by vehicles using either tunnel. VOL. 40, NO. 21, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

3 TABLE 2. Averge Composition of the Vehicle Fleets Using the Jnio Qudros or Mri Mluf Tunnels in 2004 dte vehicle type fuel vehicle count trffic volume (veh. h -1 ) % of the fleet Mrch 23 Mrch 24 Mrch 25 light pssenger vehicle gsohol/ethnol ( motorcycle gsohol ( light-duty truck or vn diesel ( light pssenger vehicle gsohol/ethnol ( motorcycle gsohol ( light-duty truck or vn diesel ( light pssenger vehicle gsohol/ethnol ( motorcycle gsohol ( light-duty truck or vn diesel ( Mrch 26 b motorcycle gsohol ( light pssenger vehicle gsohol ( light-duty truck or vn diesel ( My 5 light pssenger vehicle gsohol/ethnol ( motorcycle gsohol ( hevy-duty vehicle diesel ( My 6 light pssenger vehicle gsohol/ethnol ( motorcycle gsohol ( hevy-duty vehicle diesel ( Txis were counted s light pssenger vehicles, lthough most burn nturl gs. Txis were estimted to ccount 0.7% of the fleet (14). b Only mesured from 8:00 until 14:00. FIGURE 1. Men workdy in-tunnel diurnl profiles of CO, NO x, nd SO 2 in the Jnio Qudros tunnel (JQT) nd Mri Mluf tunnel (MMT). On verge, 13% of the vehicles using the MM tunnel were diesel-powered. In Figure 1, the CO, NO x, nd SO 2 concentrtions observed in both tunnels re presented. Totl NO x concentrtions were converted to µg m -3 using moleculr weight of 46 g mol -1 nd re reported s NO 2 mss equivlents even though most of the NO x ws emitted in the form of NO. The mrked difference between the two tunnels in terms of the concentrtions of NO x nd SO 2 indictes the significnt emissions of such gses by the hevy-duty fleet. Pollutnt concentrtions were t their lowest during the night (t pproximtely 3:00), nd peked during the morning (t pproximtely 9:00). For the JQ tunnel, the men in-tunnel concentrtions were higher thn those obtined outside the tunnel: 5.4 ( 2.2 times higher for CO; 5.4 ( 2.6 times higher for NO x; nd 3.5 ( 1.4 times higher for the sum of ll VOCs nlyzed. For the MM tunnel, in-tunnel concentrtions were 7.3 ( 4.6 times higher thn concentrtions outside the tunnel for CO, 4.4 ( 3.9 times higher for ll VOCs nlyzed, nd 25.9 ( 21.1 times higher for NO x. The lrge stndrd devitions in the MM tunnel concentrtions re relted to differences in flow intensity nd trffic driving ptterns during the study period. On My 6, hevy trffic congestion ws observed in the MM tunnel. Tht ws minly due to the poor wether conditions in the MASP on tht dy, s well s to dy-to-dy vritions in trffic volumes. In ddition, the differences between the two tunnels in terms of the rtio of in-tunnel concentrtions to outdoor (bckground) concentrtions were much more pronounced for NO x thn for other gses, finding tht is primrily ttributed to the contribution mde by the hevyduty diesel-powered fleet. As cn be seen in Figure 2, the CO profile in the JQ tunnel (which crries lmost exclusively light pssenger vehicles) is relted to the light trffic flow, with the exception of the morning rush-hour pek. This morning pek is due to frequent trffic congestion t this time. In both tunnels, the trffic flow velocity ws extremely low during the first hours of the morning, resulting in fewer numbers of vehicles thn those counted in mid-fternoon, when the velocity ws higher. Middy concentrtions of NO x were higher thn those of CO during the sme period. This is relted to the fct tht NO x emission is dependent on speed, but less thn is tht of CO, s hs been stted by Keen et l. (15) ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 21, 2006

4 FIGURE 2. Workdy diurnl profile of the correltion between CO nd the number of vehicles in the Jnio Qudros tunnel. Emission Fctors for Regulted Gseous Pollutnts. Emission fctors for light-duty vehicles trveling through the JQ tunnel were computed directly from the light-duty fleet mesurements of CO, CO 2, nd NO x concentrtions using the following eq (6, 7, 16): [P] E p ) ( [CO 2 ] + [CO]) w c, (1) where E p is the emission fctor of the pollutnt (kg of pollutnt emitted per kg of fuel burned), [P] is the bckgroundsubtrcted in-tunnel pollutnt concentrtion (µgm -3 ), [CO 2] nd [CO] re the bckground-subtrcted concentrtions of CO 2 nd CO, respectively, given in µg of crbon m -3 (i.e., when converting concentrtions of CO 2 nd CO from mol frctions to mss units, moleculr weight of 12 g mol -1, rther thn 44 g mol -1 nd 28 g mol -1 for CO 2 nd CO, respectively, ws used), nd w c is the crbon weight frction of the fuel (0.87 for diesel nd 0.85 for gsohol). Orgnic compounds were ignored in the eq 1 denomintor becuse their contribution to totl crbon concentrtions in the tunnels is negligible in comprison to those mde by CO 2 nd CO. Emissions from hevy-duty diesel-powered vehicles trveling through the MM tunnel could not be computed directly since the trffic in tht tunnel consisted of both light-duty vehicles nd hevy-duty vehicles (diesel-powered busses nd trucks). Therefore, it ws necessry to pportion pollutnt emissions in the MM tunnel into the two vehicle clsses in function of fuel burned: light-duty (burning gsohol or pure ethnol) nd hevy-duty (burning diesel). Previous rod tunnel studies hve shown tht hevy-duty diesel-powered trucks nd light-duty gsoline-powered vehicles emit comprble mounts of CO per unit of distnce trveled (6, 7). The CO 2 emissions in the MM tunnel were pportioned using trffic counts nd fuel economies (of light-duty gsolinepowered vehicles nd hevy-duty diesel-powered trucks) using the following eqution: [CO 2 ] D [CO 2 ] ) f D U D F D w D (f D U D F D w D ) + ((1 - f D ) U G F G w G ), (2) where [CO 2] D is the component of [CO 2] ttributble to emissions from hevy-duty diesel-powered vehicles, f D is the proportion of trffic identified s hevy-duty diesel-powered trucks, U is the fuel consumption rte, F is fuel density (740 gl -1 for gsohol nd 840 g L -1 for diesel), nd w is the crbon weight frction in fuel. The subscripts D nd G denote diesel nd gsohol, respectively. The contributions mde by other pollutnts in the MM tunnel were expressed using the eqution: [P] D ) [P] - [CO] (1 - f D )( [P] 2 [CO] 2), (3) where [P] D is the component of [P] ttributble to hevyduty vehicle emissions, nd [CO] (1 - f D) is the frction of [CO] ttributed to light-duty vehicle emissions. For the JQ tunnel, the light-duty vehicle pollutnt emission rtio ( [P] 2/ [CO] 2) ws mesured. These fuel-bsed emission fctors, expressed s mss emitted per unit volume of fuel burned, re representtive of typicl sitution with different types of vehicles nd under different vehicle operting conditions (speed, mss, ccelertion, engine size, etc.). Emission fctors cn be expressed per kilometers driven. Averge fuel consumption ws ssumed to be 12 L per 100 km for the light-duty fleet nd 47 L per 100 km for the hevyduty fleet (7). There re uncertinties ssocited with the fuel economy vlues used in pportioning the MM tunnel emissions (see eq 3) s well s with the fuel densities nd crbon frctions used to clculte the emission fctors (6, 7). The SO 2 emission fctors were not clculted due to problems tht occurred in the SO 2 mesurements tken outside the tunnels. The men emission fctors obtined for light-duty vehicles were 14.6 ( 2.3gkm -1 for CO nd 1.6 ( 0.3gkm -1 for NO x. For hevy-duty vehicles, the men emission fctors were 20.6 ( 4.7gkm -1 nd 22.3 ( 9.8gkm -1 for CO nd NO x, respectively. The lrge stndrd devitions for the hevyduty vehicle emission fctors cn be ttributed to differences in the trffic ptterns on My 5 nd My 6 nd, s previously mentioned, to the dy-to-dy vritions in trffic volume. In the nlysis of the JQ tunnel dt, the highest CO emission fctors were found to be correlted with trffic congestion nd lower trffic flow rtes. However, the highest NO x emission fctors were observed between 11:00 nd 13: 00, which ws not found to be true for CO, indicting tht NO x emission is less dependent on velocity thn is tht of CO. The NO x emission fctors for the hevy-duty fleet were pproximtely 14 times higher thn those clculted for the light-duty fleet. Kristensson et l. found tht the contribution of the hevy-duty fleet to NO x levels is on the order of 20-50% (2). The men rtio of NO x concentrtions to CO concentrtions ws 0.12 ( 0.03 in the JQ tunnel nd 0.26 ( 0.10 in the MM tunnel, indicting tht hevy-duty diesel-powered vehicles emitted pproximtely 54% of the totl NO x. In ddition, such vehicles were responsible for n estimted 25% of ll SO 2 emissions recorded. Tble 3 shows the comprisons between emission fctors obtined in the present study nd those reported by the uthors of other tunnel studies. The CO nd NO x emission fctors for the JQ nd MM tunnels were much higher thn those found in studies performed in other tunnels. These discrepncies cn be ttributed to the emission controls put on vehicles. The current emission limits for new crs in Brzil, s estblished by the ntionl Progrm de Controle d Poluição do Ar por Veículos Automotores (PROCONVE, Progrm for the Control of Air Pollution Emission by Motor Vehicles), re 2.0 g km -1 nd 0.25 g km -1 for CO nd NO x,respectively. For diesel-powered vehicles, the limits re bsed on dynmometer mesurements nd re given in grms emitted per kilowtt-hour of engine power: 2.1 g kwh -1 nd 5.0 g kwh -1 for CO nd NO x, respectively (1). In ddition, the motorcycle fleet in the MASP hs incresed considerbly over the pst few yers, nd motorcycle emission controls (which hve only been in force since 2003) pply exclusively to new sles. In Brzil, there is s yet no progrm of emission inspection for current vehicles. The ge of the fleet ssocited with the lck of vehicle mintennce nd common ltertion VOL. 40, NO. 21, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

5 TABLE 3. Emission Fctors (in g km -1 ) in Comprison with Other Tunnel Studies tunnel vehicle type CO NO x Wshburn tunnel LD (Houston, TX, USA) b Söderleds tunnel LD 1.07 ( 0.03 (Stockholm, Sweden) c HD 8.0 ( 0.8 ll vehicles 5.27 ( ( 0.03 Tuscror tunnel LD 3.04 ( ( 0.16 (Tuscror, PA, U.S,) d HD 3.75 ( ( 0.53 Fort McHenry tunnel LD 3.95 ( ( 0.06 (Bltimore, MD, USA) c HD 6.11 ( ( 0.28 Jnio Qudros nd LD 14.6 ( ( 0.3 Mri Mluf tunnels (current study) HD 20.6 ( ( LD ) light-duty fleet; HD ) hevy-duty fleet. b Ref 7. c Ref 2. d Ref of the ctlysts my lso be the cuse of lrge differences on CO emission fctors when compred with tht obtined in other studies. The dissimilrities between our results nd those of other uthors might lso be relted to differences in the proportion of hevy-duty to light-duty vehicles nd in driving ptterns, s well s in the type nd qulity of fuel burned. In comprison with other tunnels studied, the CO emission fctors obtined for the JQ nd MM tunnels re resonbly similr (with some discrepncies) to those reported for the Wshburn nd Fort McHenry tunnels, lthough the NO x-relted differences re greter (7, 17). Previous studies of emission fctors in Brzil were bsed on dynmometer tests. Since the implementtion of new technology bsed on PROCONVE directives, there hs been significnt decrese in emission fctors for gsohol-powered vehicles. In ddition, the mndtory ddition of ethnol to gsoline in Brzil brought reduction (of pproximtely 50%) in CO emissions when it ws implemented (18). Emission Fctors for Orgnic Compounds. The emission fctors of individul VOCs were not clculted for light-duty or hevy-duty vehicles due to the difficulties involved in the specition nd in the differentition between those emitted by gsohol-powered vehicles nd those emitted by dieselpowered vehicles. These difficulties rose due to the smll size of the VOC frction in reltion to the totl pollutnts observed in the MM tunnel s well s to our limited dt. However, bulk estimte of the emission fctors of the VOCs nlyzed in the MM tunnel ws performed, tking into considertion ll vehicle nd engine types. The men VOC emission fctors for ll vehicles using the MM tunnel re presented in Tble 4. The mens were clculted for smples collected during the morning period (8:00-12:00) nd during the fternoon period (12:00-18:00) on My 5 nd My 6. In ddition, the men emission frction TABLE 4. Averge Emission Fctors, by Dy nd by Period, for the Voltile Orgnic Compounds Anlyzed in the Mri Mluf Tunnel pollutnt My 5 (mg kg -1 ) 8:00-12:00 My 5 (mg kg -1 ) 12:00-18:00 My 6 (mg kg -1 ) 8:00-12:00 My 6 (mg kg -1 ) 12:00-18:00 Period men for the entire fleet (mg km -1 ) toluene ( butene ( 82.2 n-pentne ( 86.1 cyclohexne ( benzene ( 72.0 n-butne ( 70.3 m+p-xylene ( 62.8 n-hexne ( ,2,4-trimethylbenzene ( 49.6 formldehyde ( 35.1 cetldehyde ( 29.1 o-xylene ( 39.8 n-heptne ( ethyl-4-methylbenzene ( 23.8 ethylbenzene ( 33.0 n-octne ( 28.1 methyl pentne ( 27.7 ldehydes >C ( 21.2 n-nonne ( 22.5 isobutne ( ,3,5-trimethylbenzene ( pentene ( methylhexne ( ethyl-3-methylbenzene ( 11.5 cumene ( ethyl-2-methylbenzene ( 16.0 decne ( 15.1 n-propylbenzene ( 6.0 methylcyclopentne ( 4.6 n-undecne ( 10.4 cetone ( 4.4 methylcyclohexne ( methylethylbenzene ( 4.5 2,3-dimethylpentne ( 7.4 isoprene ( butnone ( hexene ( 7.3 n-dodecne ( 5.6 styrene ( 6.0 2,2-dimethylbutne ( 4.6 2,4-dimethylpentne ( 3.1 2,3- dimethylhexne ( 2.7 selected VOCs ( Results from second method ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 21, 2006

6 FIGURE 3. Men contributions of the 12 most bundnt VOCs emitted in the Mri Mluf tunnel in (by period) of ech VOC nlyzed is presented in reltion to the men CO emission fctor (for the light-duty fleet). The vlues re presented in mss units of VOCs emitted per mss of fuel burned (mg kg -1 ) nd in mss units of VOCs emitted per kilometer driven (mg km -1 ), tking into considertion the proportions of light-duty vehicles nd hevy-duty vehicles trveling through the MM tunnel. The significnt differences between the men VOC vlues recorded on the morning of My 5, nd those recorded on the morning of My 6 re ttributble considerble vrition in trffic ptterns, which were lso responsible for the differences observed in the men vlues for regulted gseous compounds recorded on those 2 dys. My 5 ws considered dy with good trffic conditions, wheres record congestion occurred in the city nd in the MM tunnel on My 6. Emission fctors, which re strongly dependent on driving conditions such s ccelertion nd speed, were influenced by the wide vritions in trffic congestion over the course of given dy, s well s by dy-to-dy vritions in trffic congestion. The men emission fctor for the sum of ll VOCs nlyzed in the MM tunnel ws pproximtely 1.4 ( 1.3gkm -1. Figure 3 shows the individul contributions of the twelve VOCs tht, mong the VOCs nlyzed, were found to be the most bundnt in the MM tunnel. Those twelve compounds ccounted for 65.3% of the VOCs emissions recorded in the MM tunnel. Such VOC emission chrcteristics cn reflect the specific composition of fuel (minly gsohol, pure ethnol, nd diesel) burned in the tunnel smpled. However, significnt frctions of light HCs (C 2-C 3), olefins nd ethnol were not nlyzed. Cólon et l. (19) mesured the concentrtions of ethnol in the city of São Pulo nd found high concentrtions of ethnol nd methnol. Among the VOCs nlyzed in the MM tunnel, lknes, romtic compounds, lkenes nd ldehyde compounds were identified s the min clsses emitted (without considering the number of species nlyzed in ech clss). Xylene compounds, s well s toluene nd benzene, mde significnt contributions, findings tht re in greement with those of Hsieh et l. (3) The ldehyde compounds: cetldehyde nd formldehyde, which re importnt species for ozone formtion, ccounted for 3.4 nd 3.6%, respectively, of ll the VOCs emitted (by the fleet s whole) in the MM tunnel. There is evidence tht ldehyde concentrtions in Brzil hve decresed s reported by Grosjen et l. (for instnce, in reltion to 1990s for cetldehyde nd formldehyde) (20). This decrese hs occurred in prllel with technologicl dvnces in utomobile mnufcturing nd chnges in the proportion of crs in the MASP fleet tht burn pure ethnol. Tble 5 presents comprison with other tunnel experiments in terms of the percentges of orgnic compounds most emitted in reltion to the CO emission fctor. With the exception of the Söderleds tunnel in Sweden, the percentges recorded in the MM tunnel were typiclly higher thn those recorded in other tunnels locted in the United Sttes, Tiwn, nd Europe. The resons for those differences re probbly similr to those previously cited for the differences in CO nd NO x vlues. Comprisons with studies conducted in other prts of the world re problemtic due to the gret number of fctors tht cn influence the results (trffic conditions, driving ptterns, vehicle mintennce, etc.) Differences relted to the type of fuel used mke the tsk of evluting nd compring dt even more complex. Anlysis performed of liquid fuel sold by the leding fuel distributors in the MASP during 1999 nd reported by Andrde et l. showed tht gsohol contins 24.9% ethnol, 22.4% olefins, 12.2% romtics, nd 0.37% benzene (23). Annul Vehiculr Emissions. The clcultion of totl nnul vehiculr emissions, sptil distribution, nd temporl vritions in concentrtions plys very importnt role in ir qulity models. In Figure 4 is presented n estimte of the 2004 vehiculr emissions in the MASP, bsed on emission fctors obtined from tunnel mesurements. To estimte the nnul vehiculr emissions, vrious fctors were tken TABLE 5. Percentges of VOCs Emitted in Reltion to CO Emission Fctor: Comprisons Among Tunnel Studies pollutnt Tuscror tunnel (U.S.) b LD nd HD percentge of orgnic pollutnt emitted in reltion to CO (%) So1derleds tunnel (Sweden) c ll vehicles Gubrist tunnel (Switzerlnd) d LD Tipei tunnel (Tiwn) e ll vehicles JQ nd MM tunnels (Brzil) f ll vehicles CO 100 nd toluene 0.47 nd butene n-pentne cyclohexne 0.54 benzene 0.30 nd n-butne m+p-xylene 0.35 nd n-hexne ,2,4-trimethylbenzene formldehyde 0.13 nd cetldehyde o-xylene 0.13 nd JQ ) Jnio Qudros; MM ) Mri Mluf; LD ) light-duty fleet; HD ) hevy-duty fleet. b Refs 2 nd 17 (dt from September of 1992). c Ref 2 (dt from the winter of 1998/1999). d Refs 21 nd 22 (dt from 2002). e Ref 5. f Current study (CO nd NO x emission fctors re for ll vehicles using the MM tunnel). VOL. 40, NO. 21, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

7 FIGURE 4. Percentge contributions of CO, NO x, nd HCs mde by the light-duty nd hevy-duty fleets in 2004: Comprison between nnul vehiculr emissions dt provided by CETESB nd dt obtined in the present study. into considertion. First, we ssumed tht the emission fctors re representtive of verge emissions from the MASP fleet. Second, the dily trffic flow distribution for 2004 ws clculted on the bsis of 252 working dys t 100% of mximum flow, 50 Sturdys t 80% of mximum flow nd 63 Sundys nd holidys t 50% of mximum flow. Finlly, we took s premise, the estimte tht light-duty vehicles trvel n verge of 36.7 km dy -1 nd tht hevy-duty vehicles trvel n verge of 167 km dy -1. These vlues were dpted from technicl report published by Murgel et l. (24), which ws sponsored by the Instituto de Pesquiss Tecnológics (IPT, Institute for Technologicl Reserch) (25). The uthors derived their dt from the percentges of ech vehicle type, s reported in December of 2004 by the Deprtmento Estdul de Trânsito/Processmento de Ddos do Estdo de São Pulo (DETRAN/PRODESP, Stte Deprtment of Motor Vehicle Trnsit/Stte of São Pulo Dt Processing Center) (14). Figure 4 shows the comprison between the dt collected in the present study nd those vilble from CETESB for 2004, which is the officil inventory for the MASP. The CETESB inventory ws obtined using different methodology, bsed on the number of vehicles, consumption of fuel per kilometer, nd the verge number of kilometers per dy tht the vehicles trvel, s well s on interntionl emission fctors dpted for the PROCONVE. The results were converted from tons of pollutnts per dy into kilotons of pollutnts per yer, nd nnul emissions were clculted for the MASP. The vlues obtined in the present study re generlly higher thn those presented by CETESB in terms of the emission fctors, except for HC emission fctors (due to the frction mesured). For CO, the nnul emissions for 2004 were 4.0% higher in this study thn in the CETESB dt. For HCs (here the term HCs indictes ll orgnic compounds nlyzed), not including evportive emissions, the vlue recorded in the present study ws 25% lower thn the corresponding CETESB finding, nd nnul vehicle emissions for NO x were found to be 96% higher thn those reported by CETESB. The differences between our dt nd the CETESB dt re minly ttributble to the lterntive pproch used in the clcultion of the emission fctors. In ddition, the difference for HCs is ssocited with the specific frction mesured, s well s with the fct tht species such s ethne, ethene, propne, propene, nd ethnol not were mesured. In ddition, the emission fctors estimted from tunnel mesurements most likely do not ccount for cold-strt emissions. The comprison between the light-duty nd hevy-duty fleets in terms of their 2004 emissions of CO, NO x, nd HCs, s well s the comprison between our dt nd the CETESB dt, cn lso be seen in Figure 4. Although different methodologies were used, the estimted contributions mde by ll compounds were reltively similr in both dt sets, lthough the differences mong emission fctors for motorcycles, txis, nd crs were not ccounted for. In ddition, the emission fctors used to estimte of the contribution of HCs were the sme for the light-duty fleet s for the hevyduty fleet. The results indicte tht the diesel-powered fleet, lbeit representing only 5.8% of the totl fleet, mkes significnt contribution to NO x emissions in the MASP. Therefore, it is recommended tht efforts be mde to reduce emissions from diesel-powered vehicles. Acknowledgments This study received finncil support in the form of grnt from the Fundção de Ampro à Pesquis do Estdo de São Pulo (FAPESP, Foundtion for the Support of Reserch in the Stte of São Pulo; grnt no. 02/ to L.D.M.). We thnk Helber Freits for ssisting us with the CO 2 mesurements, s well s Rosn Astolfo nd ll of the other prticipnts for their experimentl work. We re lso grteful to Lilin R. F. de Crvlho nd Pérol de Cstro Vsconcellos for providing the ldehydes nlysis lbortory, s well s to the Compnhi de Engenhri Tráfego (CET, Trnsit Engineering Agency) nd the Consldel Compny for providing the necessry infrstructure for the tunnel mesurements. Finlly, we thnk Lynn J. Scrnton for reviewing this pper. Literture Cited (1) CETESB. Reltório Anul de Qulidde do Ar no Estdo de São Pulo 2004 [in portuguese]; CETESB-Compnhi de Tecnologi de Snemento Ambientl: São Pulo, Brzil, (2) Kristensson, A.; Johnsson, C.; Westerholm, R.; Swietlicki, E.; Gidhgen, L.; Wideqvist, U.; Vesely, V. Rel-world trffic emission fctors of gses nd prticles mesured in rod tunnel in Stockholm, Sweden. Atmos. Environ. 2004, 38, (3) Hsieh, C.; Chng, K.; Ko, Y. Estimting the ozone formtion potentil of voltile romtic compounds in vehicle tunnels. Chemosphere 1999, 39, (4) Hsu, Y.; Tsi, J.; Chen, H.; Lin, W. Tunnel study of on-rod vehicle emissions nd the photochemicl potentil in Tiwn. Chemosphere 2001, 42, (5) Hw, M.; Hsieh, C.; Wu, T.; Chng, L. W. Rel-world vehicle emissions nd VOCs profile in the Tipei tunnel locted t Tiwn Tipei re. Atmos. Environ. 2002, 36, ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 21, 2006

8 (6) Kirchstetter, T. W.; Hrley, R. A.; Kreisberg, N. M.; Stolzenburg, M. R.; Hering, S. V. On-rod mesurement of fine prticle nd nitrogen oxide emissions from light- nd hevy-duty motor vehicles. Atmos. Environ. 1999, 33, (7) McGughey, G. R.; Desi, N. R.; Allen, D. T.; Seil, R. L.; Lonnemn, W. A.; Frser, M. P.; Hrley, R. A.; Pollck, A. K.; Ivy, J. M.; Price, J. H. Anlysis of motor vehicle emissions in s Houston tunnel during the Texs Air Qulity Study Atmos. Environ. 2004, 38, (8) Pretto, A. O. Estudo do Comportmento dos gses-trço O 3, NO x, CO, SO 2 e de COVs n tmosfer d cidde de São Pulo [in portuguese]. Ph.D. Thesis of Institute of Nucler Energy nd ReserchsIPEN, 2005, São Pulo, Brzil, bibliotec/informtion.htm. (9) Plss-Dülmer, C.; Michl, K.; Ruf, R.; Berresheim, H. C 2-C 8 Hydrocrbon mesurement nd qulity control procedures t the Globl Atmosphere Wtch Observtory Hohenpeissenberg. J. Chromtogr. A 2002, 953, (10) Albuquerque, E.L. de; Andriett, L. C. G.; Souz, K.R.P. de; Tomz, E.; Concentrtion of Selected VOC in Metropolitn Region of São Pulo. In Air Pollution XII; Brebbi, C. A., Ed.; WIT Press: UK, 2004; ISBN: (11) Montero, L.; Vsconcellos, P. C.; Souz, S. R.; Pires, M. A. F.; Snchez-Ccoyllo, O. R.; Andrde, M. F.; Crvlho, L. R. F. Mesurements of tmospheric crboxylic cids nd crbonyl compounds in São Pulo city, Brsil. Environ. Sci. Technol. 2001, 35, (12) Souz, S. R.; Vsconcellos, P. C.; Mntovni, W.; Crvlho, L. R. F. Emissão por folhs de Ficus benjmin L. (Morcee) de compostos orgânicos voláteis oxigendos. Rev. Brs. Bot. 2002, 25(4), (13) Pisni, S. O.; Bruns, R. E.; Vsconcellos, P. C.; Crvlho, L. R. F. Artifct evidence in crbonyl compound smpling using the enclosure technique with cuvette system. J. Environ. Monitor. 2003, 5, (14) DETRAN/PRODESP (Depto. de Análises) Arquivo: Frot Circulnte-2004, São Pulo, [ ]. (15) Ken, A. J.; Hrley, R. A.; Kendll, G. R. Effects of vehicle speed nd engine lod on motor vehicle emissions. Environ. Sci. Technol. 2003, 37, (16) Mrr, L. C.; Kirchstetter, T. W.; Hrley, R. A.; Miguel, A. H.; Hering, S. V.; Hmmond, S. K. Chrcteriztion of polycyclic romtic hydrocrbons in motor vehicle fuels nd exhust emissions. Environ. Sci. Technol. 1999, 33, (17) Pierson, W. R.; Gertler, A. W.; Robinson, N. F.; Sgebiel, J. C.; Zielinsk, B.; Bishop, G. A.; Stedmn, D. H.; Zweidinger, R. B.; Ry, W. D. Rel-world utomotive emissionsssummry of studies in the fort McHenry nd Tuscror Mountin tunnels. Atmos. Environ. 1996, 30, (18) Moreir, J. R.; Goldemberg, J. The lcohol progrm. Energy Policy 1999, 27, (19) Colón, M.; Pleil, J. D.; Hrtlge, T. A.; Gurdni, M. L.; Mrtins, M. H. Survey of voltile orgnic compounds ssocited with utomotive emissions in the urbn irshed of São Pulo, Brzil. Atmos. Environ. 2001, 35, (20) Grosjen, D.; Grosjen, E.; Moreir, L. F. R. Specited Ambient Crbonyls in Rio de Jneiro, Brzil. Environ. Sci. Technol. 2002, 36, (21) Stemmler, K.; Bugmnn, S.; Buchmnn, B.; Reimnn, S.; Stehelin, J. Lrge decrese of VOC emissions of Switzerlnd s cr fleet during the pst decde: results from highwy tunnel study. Atmos. Environ. 2005, 39, (22) Colberg, C. A.; Ton, B.; Ctone, G.; Sngiorgio, C.; Sthel, W. A.; Sturm, P.; Stehelin, J. Sttisticl nlysis of the vehicle pollutnt emissions derived from severl Europen rod tunnel studies. Atmos. Environ. 2005, 39, (23) Andrde, M. F.; Ynoue, R. Y.; Hrley, R.; Miguel, A. H. Air qulity model simulting photochemicl formtion of pollutnts: the São Pulo Metropolitn Are, Brzil. Int. J. Environ. Pollut. 2004, 22, (24) Murgel, E. M.; Szwrc, A.; Sntos. M.D.S.R.; Brnco, G.M.; Crvlho, H. Inventário de emissão veiculr metodologi de Cálculo. Congresso Brsileiro de Engenhri Snitári e Ambientl. São Pulo, Brzil [in portuguese]. Rev. Eng. Snitári 1987, (25) IPT, Avlição Técnico-Econômic de Modelos de Implntção d Inspeção Técnic de Veículos. [in portuguese]; Technicl report no. 45, 745; IPT: São Pulo, Brzil, Received for review December 5, Revised mnuscript received April 4, Accepted August 22, ES052441U VOL. 40, NO. 21, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY