You are on page 1of 6

Environ. Sci. Technol.

2005, 39, 9595-9600

Vehicle Specific Power Approach to Estimating On-Road NH3 Emissions from Light-Duty Vehicles
T A O H U A I , , T H O M A S D . D U R B I N , * , TED YOUNGLOVE, GEORGE SCORA, MATTHEW BARTH, AND JOSEPH M. NORBECK Department of Chemical and Environmental Engineering, College of Engineering, Center for Environmental Research and Technology (CE-CERT), and Statistical Collaboratory, University of California, Riverside, California 92521

NH3 emissions from motor vehicles have been the subject of a number of recent studies due to their potential impact on ambient particulate matter (PM). Highly timeresolved NH3 emissions can be measured and correlated with specific driving events utilizing a tunable diode laser (TDL). It is possible to incorporate NH3 emissions with this new information into models that can be used to predict emissions inventories from vehicles. The newer generation of modal models are based on modal events, with the data collected at second-by-second time resolution, unlike the bag-based emission inventory models such as EMFAC and MOBILE. The development of an NH3 modal model is described in this paper. This represents one of the first attempts to incorporate vehicle NH3 emissions into a comprehensive emissions model. This model was used in conjunction with on-road driving profiles to estimate the emissions of SULEV, ULEV, and LEV vehicles to be 9.4 ( 4.1, 21.8 ( 5.2, and 34.9 ( 6.0 mg/mi, respectively. We also implement this new NH3 model to predict and evaluate the NH3 emission inventory in the South Coast air basin (SoCAB).

FIGURE 1. TDL vs FTIR NH3 comparison. on modal events is the comprehensive modal emissions model (CMEM) by Barth et al. (3). CMEM uses a physical, power-demand modal modeling approach based on a parametrized analytical representation of emissions production (3). The parameters used in the model vary according to the vehicle type, engine, and emission technology, and include vehicle mass, engine size, aerodynamic drag coefficient, etc. North Carolina State University is conducting an exploratory research to understand the relation between vehicle emissions and traffic control measures (4). The Georgia Institute of Technology also developed a motor vehicle emissions model to account for emissions as a function of vehicle operating profiles (5). Vehicle emission models have primarily focused on regulated emissions. Unregulated emissions, such as ammonia (NH3), have not been included due to limitations in the available data. Recently, experimental work has shown that NH3 emissions from vehicles may be more significant than previously thought (6-8). Higher ambient NH3 emissions could contribute to increased levels of secondary particulate matter (PM), a concern because of adverse health effects related to respiratory and cardiovascular diseases in humans and decreased visibility in both urban and rural areas. For NH3, only in the past few years have techniques been developed that provide sufficient time resolution for modal modeling. Durbin et al. (9) and Huai et al. (10) utilized a sensitive and highly time-resolved tunable diode laser (TDL) system for raw exhaust NH3 measurements from vehicles. The importance of the TDL measurements is shown in Figure 1, which compares the TDL raw emissions with FTIR measurements through the dilution tunnel. It is apparent that FTIR measurements are broader and have an extended tail. This is due to the greater residence time of the FTIR cell compared to that of the TDL and the fact that the FTIR measurements are made through the dilution tunnel where adsorption/desorption effects can be considerable for NH3. This results in the FTIR underestimating the peak NH3 emissions but overestimating emissions in periods where operation is less transient. The TDL is also capable of raw exhaust NH3 measurements, which provides an improvement in sensitivity needed for low-emission vehicles. Many of the previous chassis dynamometer and tunnel studies of NH3 emissions have used filter-based techniques (6, 11, 12) to obtain integrated results and, hence, do not provide data that can be used in modal emission models. Huai et al. (10) showed that a number of factors can contribute to the formation of NH3 in vehicle exhaust by
VOL. 39, NO. 24, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9

Introduction
There is an increasing trend toward the use of models based on modal or instantaneous emissions as models of vehicle emissions continue to evolve. These newer models are based on real-time emissions data that is representative of as many modes of operation as possible, unlike bag-based emission inventory models such as EMFAC and MOBILE. The new U.S. Environment Protection Agency (U.S. EPA) mobile source inventory model, motor vehicle emissions simulator (MOVES), uses vehicle specific power (VSP), a road load-based quantity, as the primary variable in modeling emissions (1, 2). EPA conducted a shoot-out modeling study as part of EPAs MOVES modeling evaluation and found that approaches using binning of data by operational mode as well as VSP were both promising for modeling emissions (1, 2). Currently, some other models are also making efforts to use modal data. One of the newer-generation models based
* Corresponding author phone: (951)781-5794; fax: (951)781-5790; e-mail: durbin@cert.ucr.edu. College of Engineering, Center for Environmental Research and Technology. Statistical Collaboratory. Present address: California Air Resources Board, Research Division, 1001 I Street, P.O. Box 2815, Sacramento, CA 95812.
10.1021/es050120c CCC: $30.25 Published on Web 11/05/2005 2005 American Chemical Society

9595

TABLE 1. Description of Test Vehicles and NH3 Emissions Ratesa


vehicle ID MY OEM model certification displacement (L) mileage (mi) vehicle typeb test wt (lb) FTPc hot running 505d NYCCd US06c MEC01v7d SU1 2000 Honda Accord SULEV 2.3 11958 PC 3250 0.002 0.007 0.019 SU2 2001 Nissan Sentra CA SULEV 1.8 6592 PC 3000 0.003 0.002 0.011 0.030 0.071 U1 2001 Chrysler Sebring ULEV 2.4 19677 PC 3750 U2 2001 Acura CL ULEV 3.2 20523 PC 3750 L1 2000 Jeep Grand Cherokee LEV 4.7 29571 LDT 4500 L2 2001 Ford Taurus LEV 3.0 23553 PC 3625 0.001 0.031 0.078 L3 2001 Chevrolet Cavalier LEV 2.4 22482 PC 3125 0.001 0.002 0.002 0.029 0.081 L4 2001 Chevrolet Silverado LEV 5.3 8380 LDT 4750 0.028 0.029 0.083 0.064 0.065

NH3 Emissions Rates (g/mi) 0.023 0.009 0.029 0.003 0.003 0.021 0.011 0.021 0.028 0.138 0.089 0.231 0.098 0.099 0.190

a Test vehicles were conducted with original equipment (OE) as-received catalyst. The test fuel used for this project was a commercial California Phase 2 gasoline doped to 30 ppmw sulfur, a value close to the average sulfur level of California gasoline. b PC: passenger car; LDT: light-duty truck. c FTP and US06 results are averages for duplicate or triplicate tests. d NYCC, hot running 505, and MEC01v7 results are based on single tests.

analyzing the TDL NH3 results for a set eight modern lightduty vehicles. NH3 emissions vary considerably depending on the vehicle, its associated emission control technology, and driving cycle. Real-time emissions data show that NH3 emissions are primarily generated during acceleration events. More specifically, higher NH3 emissions were found under high-VSP events with rich A/F ratios. These factors indicate that NH3 emissions can be incorporated into models, such as MOVES and CMEM, that can be used to predict emissions inventories from vehicles. The objective of the present work is to incorporate NH3 data into a VSP/modal modeling framework. Two modeling approaches were used in this study. First, second-by-second NH3 emissions data were used to calculate NH3 emission rates using a VSP binning methodology, as proposed for EPAs MOVES model (1). Parameter sets for a piecewise linear regression module were then estimated for each vehicle, as well for a composite vehicle by technology category. Finally, the NH3 emissions module is applied to estimate the current NH3 emission inventory in the South Coast air basin (SoCAB) and demonstrate the trend of NH3 emissions inventories from the future mobile source fleet.

VSP is defined here as the instantaneous power per unit mass of the vehicle. The main advantages of using VSP as an independent variable for studying the hot stabilized emissions of passenger cars and light-duty trucks are three: specific power is directly measurable, it captures most of the dependence of emissions on engine operating parameters, and certification driving cycles are defined as a speed versus time trace and can also be specified in terms of VSP (13). An additional advantage of VSP is it can quantify emissions regardless of the physical parameters of the vehicle, such as vehicle type, weight, engine size, etc. A plot of NH3 emissions against VSP is provided in Figure 2a for one of the LEV vehicles (L2). The results show the bimodal nature of NH3 emissions with respect to VSP that is key to the framework developed here. The equation utilized for VSP is similar to that reported by Jimenez-Palacios (13) with the actual dynamometer road load coefficients. The VSP equation used is as follows:

VSP (kW/metric ton ) m2/s3) ) vmps[a(1 + i) + g (grade) + 9.80665(A + Bvkmph + Cvkmph2)/M]


where vmps ) velocity (m/s); a ) acceleration (m/s2); M ) vehicle test weight (kg); vkmph ) velocity (km/h); A (kg), B (kg/(km/h)), and C (kg/(km/h)2) ) dynamometer road load coefficients; i ) mass factor, which is the equivalent translational mass of the rotating components (wheels, gears, shafts, etc.) of the powertrain. We utilize a factor of 0.1 for i similar to that used by Jimenez-Palacios (13). Also, g ) acceleration of gravity (m/s2); 9.80665 ) 1 kg/N. VSP Modal Binning. A VSP binning methodology, as used to develop EPAs MOVES modeling framework (1), was initially applied to process the raw modal NH3 emissions data. At the second-by-second level of time resolution, there was considerable variability in the emissions. The data were therefore binned by VSP increments of 5 m2/s3 as presented in Figure 2b. The VSP binning methodology demonstrated two different behaviors in NH3 emissions: one flat and one sloped. There is still considerable variation in each bin due to limited data, however, especially in high-VSP zones. The data points in Figure 2b show the average values of NH3 emissions in each VSP bin. Error bars represent one standard deviation of the data in the bin. It is important to note that the breakpoints of all test vehicles are consistently within the range of the 20-25 m2/s3 VSP bin. Since the standard FTP has a maximum VSP of 22 m2/s3, testing over cycles more aggressive than the FTP is needed to develop accurate NH3 estimates for real-world conditions. A summary of NH3 emissions rates for this VSP binning method is presented in Table 2. In general, breakpoints can

NH3 Modal Database Development


The data set used for the modeling approach for this study was based on a subset of eight vehicles. The test matrix was composed of late model vehicles and included two superultra-low-emission vehicles (SULEV), two ultralow-emission vehicles (ULEV), and four low-emission vehicles (LEV), as shown in Table 1. It should be noted that although lightduty trucks and passenger cars in the same emissions category (i.e., LEV) are certified to different standards, they are grouped together in this study due to limited data in each category. All vehicles were tested over a range of cycles including the FTP75, US06, MEC01v7, NYCC, and hot running 505 (10). The data and analysis for this study are restricted to the TDL data. The average NH3 emissions over each cycle are also provided in Table 1. More details on these experimental results can be found elsewhere (10).

Development of the VSP Modeling Framework


Two approaches were used for the development of the NH3 versus VSP framework. The first approach was a VSP binning methodology similar to the approach used by EPA (1). The second approach used a piecewise linear regression with two linear slopes, an intercept, and a break point estimated from the data. Both methods were developed using the eight test vehicles and modal NH3 data over the range of cycles used for each vehicle.
9596
9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 24, 2005

TABLE 2. Summary of NH3 Emissions Rates for VSP Modal Binning Methoda
VSP Bin -20 -15 -10 -5 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80
a

SU1

SU2

U1

U2

L1

L2

L3

L4

0.01 0.01 0.01 0.01 0.01 0.07 0.01 0.01 0.01 0.05 0.09 0.03 0.01 0.07 0.01 0.02 0.02 0.03 0.09 0.03 0.01 0.06 0.01 0.03 0.04 0.06 0.15 0.04 0.01 0.07 0.01 0.02 0.05 0.06 0.18 0.03 0.02 0.07 0.04 0.11 0.21 0.17 0.66 0.11 0.10 0.28 0.14 0.12 0.44 0.83 1.41 0.26 0.09 0.57 0.10 0.33 2.04 0.97 1.78 0.42 0.10 0.86 0.38 1.05 1.34 1.37 3.50 0.72 1.09 0.95 0.64 3.23 2.59 3.54 5.13 1.17 1.68 1.73 1.54 6.82 7.73 8.04 7.50 3.19 5.41 4.03 5.98 10.34 10.57 9.55 16.58 8.15 19.93 7.46 6.89 16.79 13.52 15.09 37.07 10.03 20.67 9.75 6.87 14.67 17.21 25.01 15.84 17.45 5.75 8.54 13.93 21.78 19.17 28.17 5.30 14.19 22.62 21.27 23.95 9.46 15.75 16.66 15.78 11.82 10.77 14.67 23.95 42.37 41.40

Unit: mg/s.

TABLE 3. Composite Vehicle NH3 Regression Parameters


technology group SULEV ULEV LEV intercept 1 0.015 0.101 0.176 slope 1 0.008 0.018 0.031 breakpoint 21.00 22.75 24.48 slope 2 0.450 0.720 0.980

higher VSP range and can provide a better overall representation of the high-VSP region. The linear regression also allows NH3 versus VSP to be defined on a 1 m2/s3 scale as opposed to the 5 m2/s3 scale (see the next section). A piecewise linear regression module with two linear slopes, an intercept, and a break point was used to represent NH3 versus VSP behavior. The intercept is the NH3 emission rate at the point VSP ) 0. The breakpoint represents the point at which the NH3 behavior changes between that found at low VSP, where NH3 is relatively flat as a function of VSP, and that found at higher VSP, where NH3 begins to increase more rapidly with VSP. All regressions were fit using the binaveraged data, otherwise the large number of data points at low power would drive the regression without accounting for the behavior at higher VSP. Averaging NH3 emissions within VSP bins averaged out differences in variability between the two parts of the regression. Composite vehicle regressions were derived by averaging the intercept, slopes, and breakpoints for all vehicles within technology groups (Table 3). Overall, vehicles equipped with more advanced control technologies do show better NH3 emission control behavior, i.e., a smaller slope at high VSP (slope 2). Comparisons Between Actual Test Results and Model Results. The NH3 modeling results were compared with the subset of all LEV, ULEV, and SULEV vehicles measured in our laboratory over the various cycles (9, 10, 15, 16). This includes 12 additional ULEV vehicles and 10 additional LEV vehicles that were measured with the FTIR and not used in the model development. The comparison results are shown in Table 4. Overall, the NH3 emission rates generated by the NH3 modeling module were consistent with test measurements, although the test results show significant vehicle variability, as shown by the error bars. In future studies, more vehicles with advanced control technologies should be tested to improve the statistical validation and refine the NH3 modeling module in these categories.
VOL. 39, NO. 24, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9

FIGURE 2. NH3 emissions vs vehicle specific power (VSP), vehicle L2. be found over nearly all test vehicles in a similar VSP range. The NH3 emissions rates in the low-VSP region are relatively small. Every vehicle shows a nearly linear increasing trend in NH3 when it operates under high-VSP conditions. SU1 shows the smallest increases in NH3 emissions with increasing VSP. This can be attributed to the highly advanced feedback and control technology used on this vehicle to maintain precise air/fuel (A/F) ratio (14). Piecewise Linear Regression. As discussed above, NH3 emissions show two different modes of behavior based on VSP; a nearly flat trend at low VSP and a rapidly increasing trend at high VSP. Since the VSP bins at 5 m2/s3 increments still showed large variations at high VSP, a two part linear regression approach was applied to the VSP data. The linear regression method, as opposed to a straight binning method, ensures increasing NH3 emissions for each region in the

9597

TABLE 4. Validation of NH3 Measurements over 5 Different Test Cycles


modeled (mg/mi) FTP US06 MEC01v7 hot running 505 NYCC 41.8 65.6 86.8 4.5 15.7 27.6 FTP US06 measured (mg/mi)a,b MEC01v7 hot running 505 NYCC

SULEV 4.3 33.6 ULEV 16.3 52.4 LEV 28.7 70.8

5.3 2.5 ( 1.3 (2) 18.3 ( 16.0 (2) 35.6 ( 50.3 (2) 1.2 ( 1.7 (2) 5.3 ( 7.4 (2) 21.4 14.7 ( 19.0 (14) 78.7 ( 49.6 (11) 98.5 ( 0.6 (2) 9.4 ( 15.0 (5) 16.0 ( 6.4 (2) 38.0 13.6 ( 20.5 (14) 66.2 ( 79.7 (12) 103.5 ( 57.8 (4) 12.1 ( 17.4 (8) 28.4 ( 38.7 (4)

a Values are averages (one standard deviation variation. The number of vehicles for each emission factor is provided in parentheses. Results for 30 ppmw sulfur fuel. b Durbin and co-workers (9, 15), Huai and co-workers (10, 16).

FIGURE 3. Real-world driving profile.

FIGURE 4. Estimate process of NH3 modeling emissions rates.

Pilot Implementation of NH3 Emission Model


The NH3 modeling module was combined with activity data from a separate on-road project to provide an estimate of on-road NH3 emissions rates (17). This on-road measurement project included multiple test vehicles and provided the realworld driving behavior from tests run over a road course in Southern California. The road course includes cold start, freeway, arterial, and residential sections. The same road course was driven three times in 1 day: morning, noon, and afternoon. This generates a variety of traffic conditions. A total of 106 on-road driving tests provided valid speed profiles. Both on-board diagnostic (OBD) and global position system (GPS) speed data are available in this project. Actual speed data was determined by comparing the OBD and GPS data.
9598
9

The speed profile for one of the typical real-world tests is presented in Figure 3. This freeway, arterial, and residential combination road course presents real-world driving conditions that are comparable to a combination of the FTP75 and US06. It is also interesting to note that different driving time periods (morning, noon, and afternoon) showed no significant difference in the histogram of VSP. On the basis of the NH3 modeling module, on-road NH3 emissions rates can be estimated/predicated mathematically when the actual speed and emission certification level of the vehicles are obtained. The process for estimating an NH3 on-road modeling emissions rate is demonstrated in Figure 4. Here, the VSP histogram represents the number of seconds spent in each of the 1 m2/s3 bins. NH3 emissions rates were

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 24, 2005

TABLE 5. Modeling Results of NH3 Emissions from Gasoline Vehicle in SoCAB


EPZEV ESULEV EULEV ELEV EEarly 4 mg/mi 9.4 mg/mi 21.8 mg/mi 34.9 mg/mi 97.6 mg/mi EAVE VMT per day NH3 emissions C0 C1 C2 C3 1 - C0 - C1 - C2 - C3 (mg/mi) gasoline vehicles (106 mi) mobile sources (tons/day)
0.0% 3.2% 21.2% 0.0% 4.1% 12.6% 1.5% 5.0% 4.6% 18.3% 32.9% 30.6% 80.2% 54.8% 31.0% 84.99 66.57 43.97 314.6 333.5 355.0 26.7 22.2 15.6

year 2000 2005 2010

generated as the product of the VSP histogram and the piecewise linear regression on a 1 m2/s3 VSP basis. The total NH3 emissions value is the summation of the NH3 emission rate per VSP. It was used to generate the NH3 emission factor by dividing the total distance traveled in each on-road test. The VSP histogram plot shows that the low-VSP zone (011 m2/s3) represents the most real-world driving events. NH3 emissions, however, were primarily generated in either the low- (6-11 m2/s3) or moderate-VSP zone (26-32 m2/s3), as shown by the dark curve in Figure 4. Note that higher VSP events can also provide significant NH3 contribution, such as VSP > 32 m2/s3, even though they are rarely observed. On the basis of the 106 speed profiles obtained from the onroad, and using the NH3 modeling module, the on-road NH3 emission rates for SULEV, ULEV, and LEV vehicles were found to be 9.4 ( 4.1, 21.8 ( 5.2, and 34.9 ( 6.0 mg/mi, respectively. This indicates that on-road NH3 emissions are likely to decline as more advanced vehicle technologies enter the in-use vehicle fleet in larger numbers, consistent with other studies (15).

Modification of Vehicle NH3 Emission Factor for SoCAB NH3 Emission Inventory
The NH3 module developed in this study provides a method for estimating NH3 on-road modeling emission rates for SULEV, ULEV, and LEV vehicles. By adding other vehicle certification level segments into this module, an average NH3 modeling emission rate from mobile sources can be created and used to refine the current NH3 inventory in a specific urban area. In earlier studies, it was found that mobile sources made the third largest contribution of 33.8 tons/day of NH3 emissions in SoCAB, which accounted for about 18% of the total 188 tons/day SoCAB NH3 inventory in the year 2000 (18, 19). This estimation was based on tunnel study (conducted on September 21, 1993) data where an average emission factor of 97.6 mg/mi was found (6). Although modeling data are only available for the most recent vehicles, the modeling estimates can be combined with previous estimates to better understand how the emissions inventory may change going into the future. To calculate an overall NH3 emission rate from mobile sources in SoCAB, the emission rates for each vehicle control technology class were multiplied by the relative vehicle population adjustment factors. The average NH3 emission rate is defined as the following:

Tier 0, Tier 1, TLEV, and NLEV etc. Since no modeling results are available for the vehicles, the previous combined emission rate of 97.6 mg/mi is used here for EEarly. EPZEV is estimated as, the detection limit of NH3 FTIR measurement, 4 mg/mi. Both vehicle miles traveled (VMT) and the relative vehicle population adjustment factors were obtained from California Air Resources Boards (CARB) Emfac2002 V2.2 software program. Table 5 presents the average NH3 emission rates over a 10 year period in SoCAB. The NH3 emission factors from mobile sources, defined as the product of EAVE and VMT, are also listed in Table 5. The results show the SoCAB NH3 emission inventory from mobile sources is estimated to 26.7 tons/day in 2000, which is approximately 15% of the total SoCAB NH3 inventory at year 2000. This value is smaller compared to the NH3 emission factor from Mansell and Koizumi (19), 33.2 tons/day at 2000. Note that the future NH3 emission inventory from mobile sources should continue to decline, from 26.7 tons/day to 15.6 tons/day in SoCAB, due to the introduction of improved vehicle technology into the fleet. One element that is not included in the emission inventory estimates in this work or in the current NH3 estimates for SoCAB (19) is start emissions. Although NH3 is not formed in vehicle exhaust until after catalyst light-off, slightly higher NH3 emissions during the period immediately following catalyst light-off during cold starts have been observed for some vehicles (10). Start emissions were evaluated by Huai et al. for a fleet of vehicles, including many of the same vehicles used in developing this model. FTP bag 1 cold start emissions ranged from 3 to 6 mg/mi for SULEVs, from 9 to 50 mg/mi for ULEVs, and from 1 to 62 mg/mi for LEVs (10). These emissions levels are comparable to those found for the over-the-road testing described in the proceeding section. FTP bag 3 hot start NH3 emissions were found to be no different than emissions during fully warm, stabilized operation. Overall, the effects of start emissions are not expected to have a large impact on the inventory estimates in this study, beyond the other limitations associated with working with small data sets. It is suggested, however, that in the development of more sophisticated, full-scale NH3 emissions inventory models consideration should be given to the contribution of cold start emissions.

Discussion
In this study, both a VSP binning methodology and a piecewise linear regression were applied to develop the first VSP-based modeling framework for NH3. This method generated a simplified NH3 module for each vehicle certification category. On-road NH3 emission rates for SULEV, ULEV, and LEV vehicles were estimated based on actual realworld driving combined with the modeling results. These estimated on-road NH3 emissions were 9.4 ( 4.1, 21.8 ( 5.2, and 34.9 ( 6.0 mg/mi, respectively, for SULEV, ULEV, and LEV vehicles. Finally, in conjunction with earlier work, the on-road NH3 emission rates for the SULEV, ULEV, and LEV vehicles were used to help understand the impact of newer vehicles on the NH3 emissions inventory in the SoCAB area. These results show that as newer vehicles continue to enter
VOL. 39, NO. 24, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9

EAVE ) C0EPZEV + C1ESULEV + C2EULEV + C3ELEV + (1 C0 - C1 - C2 - C3)EEarly


where C0 is the portion of partial zero emission vehicle (PZEV) in the total vehicle fleet; C1 is the portion of SULEV in the total vehicle fleet; C2 is the portion of ULEV in the total vehicle fleet; C3 is the portion of LEV in the total vehicle fleet; EEarly is the early emission certification, such as Tier 0, Tier 1, TLEV, and NLEV. Equipped with more advanced control technologies, PZEVs are recognized to contribute much less NH3 emissions than other certification vehicles. EEarly represents the NH3 emission rate over all older emission certification, such as

9599

the fleet, there should be steady decline in NH3 emission inventories, even when increasing VMT is accounted for. This NH3 emission module was initially developed for the LEV, ULEV, and SULEV vehicle/technology groups. More detailed data for older vehicle/technology groups will be needed to expand the capabilities of the model to other portions of the fleet. The piecewise linear regressions gave promising results on the vehicles tested so far, but additional tests will be necessary. The accuracy of NH3 emissions for specific VSP bins will also improve as more data are gathered, which would also improve the results for the binning methodology. As discussed in a pervious study, rich air/fuel ratios can also contribute higher NH3 emissions (10). Therefore, other factors, such as vehicle maintenance and malfunctioning, should also be considered in future NH3 emission inventories.

Acknowledgments
The authors acknowledge the contribution and support of Dave Martis, Joseph Calhoun, Ross Rettig, Joe Gil, and Joe Valdez of the Vehicle Emissions Research Laboratory at CECERT who performed the emissions testing on the vehicles. We thank the Coordinating Research Council (under Contract E-60) and the U.S. Environmental Protection Agency for their financial and technical support of this project.

Literature Cited
(1) Koupal, J.; Michaels, H.; Cumberworth, M.; Bailey, C.; Brzezinski, D. EPAs Plan for MOVES: A Comprehensive Mobile Source Emissions Model. Proceedings of the 12th CRC On-Road Vehicle Emissions Workshop, San Diego, CA, April 15-17, 2002. (2) Frey, H. C.; Unal, A. Recommended Strategy for On-Board Emission Data Analysis and Collection for the New Generation Model. Proceedings of the 12th CRC On-Road Vehicle Emissions Workshop, San Diego, CA, April 15-17, 2002. (3) Barth, M.; An, F.; Younglove, T.; Scora, G.; Levine, C.; Ross, M.; Wenzel, T. Development of a Comprehensive Modal Emissions Model; Final Report by the University of Californias College of Engineering-Center for Environmental Research and Technology, Riverside, CA for the National Cooperative Highway Research Program under contract NCHRP 25-11, 2000. (4) Rouphail, N. M.; Frey, H. C.; Colyar, J. D.; Unal, A. Vehicle Emissions and Traffic Measures: Exploratory Analysis of Field Observations at Signalized Arterials. Presented at the 80th Annual Meeting of the Transportation Research Board, Washington, DC, Jan, 2001. (5) Bachman, W. H. A GIS-Based Modal Model of Automobile Exhaust Emissions; Final Report by the Civil and Environmental Engineering, Center for Geographic Information Systems, Georgia Institute of Technology, Atlanta, Georgia for the U. S. Environmental Protection Agency, Air Pollution Prevention and Control Division under contract CR823020, 1998. (6) Fraser, M. P.; Cass, G. R. Detection of excess ammonia emissions from in-use vehicles and the implications for fine particle control. Environ. Sci. Technol. 1998, 32, 3535-3539.

(7) Kean, A. J.; Harley, R. A.; Littlejohn, D.; Kendall, G. R. On-road measurement of ammonia and other motor vehicle exhaust emissions. Environ. Sci. Technol. 2000, 32, 3535-3539. (8) Baum, M. M.; Kiyomiya, E. S.; Kumar, S.; Lappas, A. M.; Lord, H. C., III. Multicomponent remote sensing of vehicle exhaust by dispersive absorption spectroscopy. 1. Effect of fuel type and catalyst performance. Environ. Sci. Technol. 2000, 34, 28512858. (9) Durbin, T. D.; Miller, J. W.; Pisano, J. T.; Sauer, C.; Rhee, S. H.; Huai, T.; Younglove, T. The Effect of sulfur and NH3 and other Regulated Emissions from Late Model Vehicles; Final Report for the Coordinating Research Council Project E-60, by the University of California at Riversides Bourns College of Engineerings Center for Environmental Research and Technology, May, 2003. (10) Huai, T.; Durbin, T. D.; Miller, J. W.; Pisano, J. T.; Sauer, C. G.; Rhee, S. H.; Norbeck, J. M. Investigation of NH3 emissions from new technology vehicles as a function of vehicle operating conditions. Environ. Sci. Technol. 2003, 37 (21), 4841-4847. (11) Gertler, A. W.; Sagebiel, J. C.; Cahill, T. A. Measurements of Ammonia Emissions from Vehicles in a Highway Tunnel. Proceedings of the 11th CRC On-road Vehicle Emissions Workshop, San Diego, CA, March 26-28, 2001. (12) Graham, L. Gaseous and Particulate Matter Emissions from InUse Light-duty Gasoline Motor Vehicles; Report no. 99-67; Environment Canada, Environmental Technology Center, Emissions Research and Measurement Division: Canada, 1999. (13) Jimenez-Palacios, J. L. Understanding and Quantifying Motor Vehicle Emissions with Vehicle Specific Power and TILDAS Remote Sensing. Ph. D. Thesis, Massachusetts Institute of Technology, Cambridge, MA, 1999. (14) Kitagawa, H.; Mibe, T.; Okamatsu, K.; Yasui, Y. L4-Engine Development for a Super Ultralow Emissions Vehicle; SAE Technical Paper No. 2000-01-0887; 2000. (15) Durbin, T. D.; Wilson, R. D.; Norbeck, J. M.; Miller, J. W.; Huai, T.; Rhee, S. Estimates of the emission rates of ammonia from light-duty vehicles using standard chassis dynamometer test cycles. Atmos. Environ. 2002, 36, 1475-1482. (16) Huai, T.; Durbin, T. D.; Rhee, S. H.; Miller, J. W.; Norbeck, J. M. The Impact of Gasoline Fuel Sulfur Levels on NH3 and N2O Emissions. Proceedings of the 12th CRC On-Road Vehicle Emissions Workshop, San Diego, CA, April, 2002. (17) Malcolm, C.; Younglove, T.; Barth, M.; Davis, N. Mobile source emissions: Analysis of spatial variability in vehicle activity patterns and vehicle fleet distributions. Transp. Res. Rec. 2003, No. 1842, 91-98. (18) Chitjian, M.; Koizumi, J.; Botsford, C. W.; Mansell, G.; Winegar, E. Final 1997 Gridded Ammonia Emission Inventory Update for the South Coast Air Basin; Final Report to the South Coast Air Quality Management District under contract 99025, August, 2000. (19) Mansell, G. E.; Koizumi, J. Development of an Updated Gridded Ammonia Emission Inventory for the South Coast Air Basin; Prepared for the South Coast Air Quality Management District, Diamond Bar, CA 91765, 2000.

Received for review January 18, 2005. Revised manuscript received September 14, 2005. Accepted October 5, 2005. ES050120C

9600

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 24, 2005

You might also like