Environ. Sci. Technol. 2001, 35, 1755-1764
Exhaust Emissions from Engines of the Detroit Diesel Corporation in Transit Buses: A Decade of Trends JACKY C. PRUCZ,* NIGEL N. CLARK, MRIDUL GAUTAM, AND DONALD W. LYONS Department of Mechanical and Aerospace Engineering, West Virginia University, Morgantown, West Virginia 26506-6106
In the U.S.A., exhaust emissions from city buses fueled by diesel are not characterized well because current emission standards require engine tests rather than tests of whole vehicles. Two transportable chassis dynamometer laboratories developed and operated by West Virginia University (WVU) have been used extensively to gather realistic emission data from heavy-duty vehicles, including buses, tested in simulated driving conditions. A subset of these data has been utilized for a comprehensive introspection into the trends of regulated emissions from transit buses over the last 7 years, which has been prompted by continuously tightening restrictions on one hand, along with remarkable technological progress, on the other hand. Two widely used models of diesel engines manufactured by the Detroit Diesel Corporation (DDC) have been selected as a case-study for such an overview, based on full-scale, on-site testing of actual city buses, driven in accordance with the SAE J1376 standard of a Commercial Business District (CBD) cycle. The results provide solid, quantitative evidence that most regulated emissions from engines produced by DDC have declined over the years, especially with the transition from the 6V-92TA to the Series 50 models. This improvement is remarkable mainly for the emissions of particulate matter (PM), that are lower by over 70%, on average, for the Series 50 engines, though the emissions of nitrogen oxides (NOx) exhibit a reversed trend, showing a degradation of about 6%, on average, with the transition from 6V-92TA to the Series 50 engines. The expected trend of decreasing emission levels with the model year of the engine is clear and consistent for particulate matter (PM), hydrocarbons (HC), carbon monoxide (CO), and nitrogen oxides (NOx), starting with the 1990 models, although it is not conclusive for carbon dioxide (CO2) emissions.
Significance Emission inventories rely on certification data for the prediction of NOx and PM masses released into the atmosphere, but this approach is presently inaccurate for the heavy-duty vehicle sector. Chassis-based emissions data are superior for this purpose, even though the commonly used CBD simulation is not fully representative of actual driving cycles of transit buses. This paper presents a statistical overview of regulated exhaust emissions from transit buses * Corresponding author phone: (304)293-3111, ext.2314; fax: (304)293-6689; e-mail:
[email protected]. 10.1021/es001416f CCC: $20.00 Published on Web 03/31/2001
2001 American Chemical Society
powered by common diesel engines of the Detroit Diesel Corporation. It covers over a decade of engine model years, and it reveals, in particular, that NOx levels released during CBD cycles have not declined in proportion to the certification data. The emissions of particulate matter (PM), hydrocarbons (HC), carbon monoxide (CO), and carbon dioxide (CO2) have been reduced, indeed, with the transition from the 6V-92TA to the Series 50 engine model. Furthermore, with the exception of CO2, they show a clear and consistent trend of decreasing emissions from one engine model year to the next, starting about 1990. The scatter patterns of emission test results display asymmetric distributions in many cases, which limits the accuracy of correlation or trend studies that rely solely on average values. The results discussed here indicate a close relationship between the magnitude levels of test results and their scatter characteristics, where lower average levels of emissions are associated, usually, with smaller variability of data, and more symmetric distributions patterns. The results and conclusions presented in this paper are useful to organizations that may be contemplating replacement of bus fleets.
Introduction Full size transit buses are, usually, 40 ft. in length and operate at about 32 000 lb. vehicle weight. During the past decade, such buses have been powered, predominantly, by diesel engines. However, diesel-electric hybrid buses are now in service, while fleets operating on compressed natural gas have been adopted by many cities, primarily under pressures to satisfy environmental regulations. Detroit Diesel Corporation (DDC) has an established record in manufacturing diesel bus engines. Following extensive use of the 71 series (71 cubic inches per cylinder) DDC engines, the common bus engine in the 1980s was the six cylinder model DDC 6V-92 (92 cubic inches per cylinder), with a displacement of 9.05 L. The 6V-92 was a two-stroke engine, with intake ports through the cylinder walls and exhaust valves in the head. It was both supercharged for scavenging and turbocharged for power density. With tightening emissions standards, electronically managed versions of the 6V-92 gave way to the four stroke, four cylinder, turbocharged DDC Series 50, with a displacement of 8.5 L (130 cubic inches per cylinder), but many 6V-92TA engines are still in use. Exhaust emissions of oxides of nitrogen (NOx), particulate matter (PM), hydrocarbons (HC), and carbon monoxide (CO) are regulated at the present time, but levels of HC and CO from diesel engines are low and receive little attention. Emissions of NOx are produced, primarily, as nitric oxide (NO), although nitrogen dioxide (NO2) levels may exceed 10% of the overall NOx levels at high engine speeds and light traction loads (1, 2). A fierce debate over the health effects of PM has been developing, and attention has focused on the harm caused by the ingestion of ultrafine particulate matter into the lungs. Extensive findings on this issue have been reported in the past by the Health Effects Institute (3, 4), while the state of California has declared diesel PM to be toxic. It is important, therefore, to document actual levels of PM and NOx emissions from transit buses in routine fleet operations and to assess their environmental impact as well as the consequences of tightening emissions standards as newer buses and engine models are commissioned. Tailpipe emissions from city buses are not well characterized, mainly because the emissions standards that are presently promulgated by the U.S. federal government and by the state of California rely on engine certification tests, rather than on measurements performed on actual transit VOL. 35, NO. 9, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1755
buses, in typical conditions of operation. The current regulations specify the emissions levels permitted from such an engine, in units of grams/brake horsepower-hour (g/bhph), as it follows a prescribed sequence of speed and torque variations that constitute a transient test schedule published in the Code of Federal Regulations, title 40. There is no guarantee, though, that, once installed in a bus and operated over an actual bus route, the engine follows a similar history of speed-torque variations. However, for purposes of the U.S. Environmental Protection Agency emissions inventory, the certification levels of emissions are converted to units of g/mile, by using the factors of fuel density, vehicle fuel consumption, and engine efficiency. Although the standards for emissions from transit buses have been tightened in recent years, there is no check that the in-use emissions have been reduced in proportion with the regulated levels. Graboski et al. (5) examined the emissions of 24 heavy duty vehicles as part of the Northern Front Range Air Quality Study and concluded that in-use emissions, in particular emissions of oxides of nitrogen (NOx), have not been reduced in proportion to the tightening engine standards. Yanowitz et al. (6) have surveyed recently a wide range of heavy-duty vehicle emissions data and reached the same conclusion. The central purpose of this paper is to present, in a statistical context, comprehensive results of emissions measured over the past decade by the two Transportable Heavy Duty Emissions Testing Laboratories developed and built at West Virginia University (WVU) (7, 8). The data selected for discussion here have been collected from full size transit buses operated on diesel fuel of type D1 or D2 and powered by the two most common engines manufactured by the Detroit Diesel Corporation (DDC), namely the 6V-92 and the Series 50 models. These data span 7 years of extensive testing, in numerous regions of the country, and provide the most comprehensive basis that is available in the nation to date, for assessing emissions from transit buses.
Test Setup and Procedures for Data Acquisition Emissions data were obtained by testing each vehicle, in its routine operational configuration, on the chassis dynamometer of one of the two Transportable Laboratories mentioned above. Each laboratory is built on a semitrailer that incorporates rollers (to support the drive wheels of the test vehicle), flywheels (to simulate the inertia of the vehicle), and eddy current power absorbers (to mimic the road load). Power is drawn directly from the drive wheels using hub adaptors, while the rollers serve only to equalize speeds on each side of the test vehicle. The total exhaust from each bus is led to a full-scale dilution tunnel, where the dilution air is not preconditioned. From the diluted exhaust, heated lines feed analyzers for NOx, CO, HC, and CO2. The analyzers are logged on a continuous basis, throughout a test run. PM is measured gravimetrically, by using 70 mm filters in a dilute exhaust slipstream. Detailed schematics of construction, along with descriptions of experimental arrangements and procedures for these laboratories, may be found in other references (9, 10). Each bus was tested as received, with the original fuel in its tank, in the prevailing weather conditions, except during heavy rain, when the testing was suspended. The test weights used in the general data analysis approximated the curb weight of the bus, plus its driver and half the passenger load. All the data presented in this paper were obtained as the test vehicle was driven by a trained driver, through the Central Business District (CBD) portion of the transit coach cycle described in the SAE recommended practice J1376. This driving cycle consists of 14 successive acceleration, cruise, and deceleration events that cover a distance of 2 miles in about 10 min. Even though the CBD cycle does not represent actual bus operations, it has been employed widely as a metric 1756
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 9, 2001
for characterization of emissions. The WVU Transportable Laboratories have measured exhaust emissions from transit buses driven also through cycles other than the CBD. Since the measured emission levels are strongly dependent on the driving cycle, whether expressed in terms of distance, fuel, or energy specific units, (5, 7), only results generated through the CBD cycle have been selected for analysis and discussion in this paper. The units employed throughout the paper to quantify levels of tailpipe emissions are g/mile.
Test Samples for Data Analysis During a period of 7 years, since they started to operate in March 1992, and through the summer of 1999, the two transportable laboratories of WVU have performed a total of 732 emissions tests on transit and school buses simulating the CBD driving cycle. Although 18 different types of fuel have been examined in these tests, including diesel, natural gas (in the form of CNG or LNG), and alcohol, only the two types of diesel fuel, namely D1 and D2, have been selected for analysis and discussion in this article. The dominant role of the Detroit Diesel Corporation (DDC) in the market of diesel engines for transit buses is well recognized. Its engines powered about two-thirds (207 out of 327) of the transit buses tested to date by the WVU laboratories on diesel fuels. The sample of transit buses powered by DDC engines is divided between 98 vehicles fueled by D1 (out of a total of 136, or 72%) and 109 vehicles fueled by D2 (out of a total of 191 or 57%). As shown in Table 1, the most common DDC engine of the past decade was the two-stroke, 6V-92TA model, used in 137 of the 207 buses tested on D1 or D2 fuels (about 66%). Its successor, the fourstroke Series 50 engine model, powered 47 of these buses (23%), while all other versions of DDC engines, such as the early 6V-71TA model, or the larger, Series 60 models, were used only in a few of the transit buses tested by WVU so far (the buses powered by Series 60 engines were, in fact, coaches, which are heavier than typical transit buses). The distribution of test samples across the model year of the engine is presented in Table 1 for each of these types of engine, separately for the D1 and the D2 types of diesel fuel. However, only the results from emission tests on the 6V-92TA and the Series 50 engine models are reviewed in this paper, since the small sample sizes of the other DDC models are not significant for a comprehensive analysis of trends. Little emphasis is placed in this paper on potential effects that the specific type of diesel fuel, either D1 or D2, may exert on the measured levels of various emissions. Preliminary investigations have revealed in the past that such effects may depend strongly on the maintenance procedures, quality control, weather, and operation conditions of specific vehicle fleets. Besides a comparison between the 6V-92TA and the Series 50 engine models from the viewpoint of their release of regulated emissions, the attention is focused on potential relationships between emission levels and the model year of the engine. Possible relationships between emission levels from these engines and the transmission configuration of the test vehicle are also considered here, due to the trend of increasing the number of gear ratios from 3-speed to 4-speed, and even to 5-speed, during the 1992-1999 period of testing covered in this paper. The data shown in Table 1 for the 6V-92TA engines indicates clearly the transition from 3-speed to 4-speed transmissions with the 1991-1992 engine model years. A similar trend is displayed in Table 1 for recent models of the Series 50 engines, starting with the 1995 models, where 11 of the 14 buses tested were equipped with 4-speed (rather than 3-speed) transmissions, and continuing with the 1996 engine models, where all of the 11 tested so far were installed in buses equipped with 5-speed transmissions.
TABLE 1. Distributions of 207 CBD Emission Tests on DDC Engines in Transit Buses engine model, type of diesel fuel, transmission configuration 6V-92TA year 1967 1973 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 Totals
6V-71TA D1 or D2
6V-92RH D2 diesel
D2 2-speed
1
1
3-speed D1 D2
4-speed D1 D2
Series 50 alltrans D1 D2
3-speed D1 D2
4-speed D1 D2
D2 5-speed
alltrans D1 D2
Series 60 D1
3 2 1
1 1 1 4
1 3 4
1 1 2 3 3
1
5 8 2 14
9
20 8 11 3
10 13 13
1 2
1 2 3 3
1 4 4 2
5 8 12 27 13
1 2 20 8 12 7 4 2
3 12 1
3 4 2
2 4
1 7
2 12 5
3 5 9 11
19
28
11
9
11
4
38
44
36
15
74
83
13
9
6
8
11
3
TABLE 2. Descriptive Statistics of NOx Emissions (g/mile) from DDC-Powered Buses Using CBD Cycles statistical sample
count
mean
median
range
std dev
coeff variat
skewness
kurtosis
95% conf
combined models 6V-92TA models Series 50 models both models on D1 fuel both models on D2 fuel 6V92TA fueled by D1 6V92TA fueled by D2 Series 50 fueled by D1 Series 50 fueled by D2 6V92TA in 2-speed transm. 6V92TA in 3-speed transm. 6V92TA in 4-speed transm. Series 50 in 3-speed transm. Series 50 in 4-speed transm. Series 50 in 5-speed transm.
860 612 248 438 422 337 275 101 147 9 361 229 123 60 54
33.77 33.20 35.18 32.03 35.58 30.07 37.05 38.57 32.85 46.76 36.55 27.57 36.99 37.62 31.43
30.96 27.98 33.07 27.91 33.51 26.75 38.00 35.78 31.62 42.34 37.47 25.99 33.84 37.82 30.03
41.59 41.59 38.67 38.73 41.59 38.73 41.59 31.48 38.67 14.30 41.59 33.77 31.05 24.74 32.88
9.74 10.23 8.28 9.23 9.94 8.59 10.76 8.27 7.46 6.26 10.50 6.72 7.73 7.50 6.50
0.29 0.31 0.24 0.29 0.28 0.29 0.29 0.21 0.23 0.13 0.29 0.24 0.21 0.20 0.21
0.71 0.74 0.85 1.03 0.43 1.41 0.13 0.73 1.04 0.28 0.15 2.71 1.31 -0.02 3.32
-0.53 -0.73 0.56 -0.02 -0.70 0.83 -1.12 -0.35 1.86 -2.40 -1.10 7.02 0.91 -1.32 11.18
0.65 0.81 1.04 0.87 0.95 0.92 1.28 1.63 1.22 4.82 1.09 0.88 1.38 1.94 1.78
Initial Screening of Test Data Each test of regulated emissions from a certain vehicle consists, usually, of 3 to 4 identical runs, so that the sample size for more than 200 buses powered by DDC engines fueled by D1 or D2 diesel may be approaching 1000 records. Consequently, statistical analysis of emissions data is confined, usually, to smaller subsets of the available database. Representative samples are selected for this purpose so that their sizes are statistically significant to achieve reasonable levels of confidence, while, at the same time, the variability of their data can be assessed, controlled, and analyzed in a credible manner. The importance of such an initial screening of test results for analysis purposes is well illustrated in Tables 2 and 3, along with Figures 1 and 2. They represent statistical summaries of the results available, respectively, for NOx and PM emissions, from all the test runs conducted so far on the population of vehicles defined in Table 1, i.e., transit buses powered by 6V-92TA or Series 50 engines, fueled by D1 or D2 diesel, and driven through CBD simulation cycles. The sample sizes may vary, however, from one type of emissions
to another, because certain results, or vehicle characteristics, are incomplete in some test records. Consequently, each type of emissions data has been analyzed separately, so that results available from incomplete records could be also utilized, rather than ignored all together with such records. That is why the numbers of available test runs, or useful sample sizes, displayed in the “count” column of Table 2 for NOx data measured on various subsets of samples are different than the counts shown in Table 3 for the corresponding levels of PM emissions. The first row in each of the Tables 2 or 3 lists the main statistical parameters describing the corresponding test data for the entire population of vehicles selected for discussion here. A top-down approach is, subsequently, followed systematically to analyze representative subsets of this population. It starts with overall comparisons between the two engine models, the two different types of diesel fuel, and it continues with separate investigations of effects on emissions, that may be associated with either the type of diesel fuel, or with the transmission configuration, for each of the two main engine models under consideration here. VOL. 35, NO. 9, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1757
TABLE 3. Descriptive Statistics of PM Emissions (g/mile) from DDC-Powered Buses using CBD Cycles statistical sample
count
mean
median
range
std dev
coeff variat
skewness
kurtosis
95% conf
combined models 6V-92TA models Series 50 models both models on D1 fuel both models on D2 fuel 6V92TA fueled by D1 6V92TA fueled by D2 Series 50 fueled by D1 Series 50 fueled by D2 6V92TA in 3-speed transm. 6V92TA in 4-speed transm. Series 50 in 3-speed transm. Series 50 in 4-speed transm. Series 50 in 5-speed transm.
817 579 238 423 394 330 249 93 145 337 220 113 61 54
0.91 1.15 0.32 0.81 1.02 0.97 1.40 0.24 0.37 1.19 0.97 0.30 0.36 0.28
0.74 0.95 0.26 0.72 0.75 0.85 1.19 0.20 0.28 0.92 0.93 0.24 0.30 0.27
6.87 6.87 1.95 3.87 6.87 3.84 6.87 0.55 1.83 6.85 3.82 0.83 1.95 0.27
0.89 0.94 0.25 0.67 1.06 0.67 1.16 0.13 0.30 1.02 0.56 0.20 0.36 0.07
0.98 0.81 0.80 0.83 1.04 0.69 0.83 0.56 0.81 0.85 0.57 0.67 1.01 0.25
2.50 2.32 3.44 1.87 2.34 1.84 1.97 0.74 3.12 1.90 1.87 1.41 3.41 0.41
9.05 7.67 15.91 4.65 7.08 4.37 4.91 -0.53 11.46 4.88 8.27 1.39 12.21 -0.49
0.06 0.08 0.03 0.06 0.11 0.07 0.14 0.03 0.05 0.11 0.07 0.04 0.09 0.02
sured emissions. The level of skewness, or asymmetry with respect to the mean, depends on the type of measured exhaust emission as well as on the characteristics of the test sample. The positive values of skewness given in Tables 2 and 3 indicate frequency distributions with asymmetric tails extending toward larger values than the median. Furthermore, the results brought up in Tables 2 and 3 for the values of the “median” parameter are lower, sometimes, by as much as 15-25% than the corresponding values of the “mean” parameter, which points toward the observation that a relatively small number of tests must have generated higher emission readings than the bulk of other similar tests on similar populations of vehicles, engines, and fuel categories.
FIGURE 1. Frequency distributions of NOx measurements from DDC engines fueled by D1 or D2 diesel in transit buses simulating CBD cycles.
The values shown in Tables 2 and 3 for the kurtosis parameter measure the relative “peakedness” or “flatness” of the corresponding plots in Figures 1 and 2, respectively, as compared with the ideal normal distribution (11, 12). When these values are positive the associated distributions are, relatively, peaked, whereas negative values indicate relatively flat distributions. A genuinely random series of experiments, where all deviations from the corresponding means are caused by uncontrollable, uncertain factors, or “white noise”, would yield perfectly symmetric distributions, as described by the ideal normal, or Gauss distribution function (11, 12). If the distribution pattern exhibits a significant level of asymmetry, like in Figures 1 and 2, the analysis ought to consider the shape of the distribution curve, in addition to the conventional measures of average and standard deviation. Various approaches are available for this purpose, such as fitting a Weibull distribution model or dividing the initial population into smaller subsets of samples, each formed of specimens with more common characteristics than the overall set.
FIGURE 2. Frequency distributions of PM emissions from DBC engines fueled by D1 or D2 diesel in transit buses simulating CBD cycles. Further, in-depth analysis of possible relationships between measured levels of NOx, or PM emissions and the model year of the engine, has been concentrated on the 6V-92TA engine models. It considers, separately, the D1 and D2 types of diesel as well as the 3-speed and the 4-speed transmission configurations, which are used frequently in buses powered by such engines.
Statistical Concepts The “tailed” distributions of emission results illustrated by the histograms shown in Figures 1 and 2 depict clearly asymmetries skewed toward unusually large levels of mea1758
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 9, 2001
The specific tests that yielded excessively high values for the various types of emissions can be identified easily with the help of such histograms as those depicted in Figures 1 and 2. Unexpected or unusual test results, whose frequencies of occurrence are low or that deviate significantly from the typical levels of most other data points, ought to be analyzed separately from the rest of that sample. The filtered test results become, thus, suitable for analysis in terms of average values and standard deviations, since their distributions are likely to be symmetric about the corresponding means. No such stepwise filtering of data has been carried out, though, for the results discussed in this paper, where the primary objective is to identify trends and relative effects, rather than to determine certain absolute values with high levels of accuracy. The numerical results analyzed here are, therefore, the complete data sets described by the nonsymmetric frequency distributions of Figures 1 and 2, which are identical with the samples defined and characterized by the statistical parameters presented in Tables 2 and 3.
TABLE 4. Descriptive Statistics of NOx Data (g/mile) from DDC Engines of Various Model Years model 6V-92TA models
Series 50 models
year
count
mean
median
std dev
coeff variat
95% conf
1980 1981 1982 1983 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996
4 4 6 11 5 12 17 83 39 99 75 140 68
35.85 26.56 39.33 38.51 45.28 46.78 48.87 46.10 39.13 30.12 26.91 26.29 26.15
35.49 26.55 39.24 41.45 44.84 47.22 51.19 47.06 43.21 28.11 25.95 25.71 26.38
1.12 0.18 1.17 14.35 0.97 1.55 4.23 7.43 11.25 7.50 3.75 3.35 1.78
0.03 0.01 0.03 0.37 0.02 0.03 0.09 0.16 0.29 0.25 0.14 0.13 0.07
1.78 0.29 1.23 9.64 1.20 0.98 2.17 1.62 3.65 1.50 0.86 0.56 0.43
FIGURE 3. Low bounds and 95% confidence intervals for NOx emissions measured through CBD cycles of 6V-92TA and Series 50 engine models in transit buses.
count
mean
median
std dev
coeff variat
95% conf
21 108 65 54
26.18 35.75 40.24 31.43
21.69 33.37 41.66 30.03
6.70 8.16 6.26 6.50
0.26 0.23 0.16 0.21
3.05 1.55 1.55 1.78
FIGURE 5. Low bounds and 95% confidence intervals for NOx emissions measured through CBD cycles of Series 50 engine models in transit buses with various transmissions.
FIGURE 4. Low bounds and 95% confidence intervals for NOx emissions measured through CBD cycles of 6V-92TA engine models in transit buses with different transmissions.
Emissions of Oxides of Nitrogen (NOx) The statistical results of the 860 test runs that provided relevant NOx data for the 6V-92TA and the Series 50 engine models in buses equipped with either 3-speed, 4-speed, or 5-speed automatic transmissions and fueled by either D1 or D2 diesel are summarized in Tables 2 and 4, along with the Figures 1 and 3-10. In addition to the sample sizes of various subsets of data and their corresponding mean values, these tables and figures display variability characteristics of the test results. The 95% confidence levels are especially mean-
FIGURE 6. Frequency distributions of NOx emissions from 6V-92TA engine models fueled by D1 or D2, in buses with various transmission configurations. ingful, not only due to the fact that they combine the sample size and the data scatter level in one statistical parameter but also that they provide the basis for quality boundaries in the statistical design and process control of emission tests. The value of 0.65 g/mile for the overall 95% level of confidence in the combined NOx data indicates that 95% of any additional results will fall closely to the current average of 33.77 g/mile VOL. 35, NO. 9, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1759
FIGURE 10. Trends of NOx emissions (low bounds and 95% confidence intervals) from 6V-92TA engines in buses with 4-speed transmissions, fueled by either D1 or D2 diesel. FIGURE 7. Frequency distributions of NOx emissions from 6V-92TA engine models fueled by D1 or D2, in buses with various transmission configurations.
FIGURE 8. Variation trends of average NOx emissions over a decade of model years for main DDC engines in transit buses.
FIGURE 9. Trends of NOx emissions (low bounds and 95% confidence intervals) from 6V-92TA engines in buses with 3-speed transmissions, fueled by either D1 or D2 diesel. (first row in Table 2), within an interval ranging from 33.12 g/mile (-1.925%) to 34.42 g/mile (+1.925%). The width of the 95% confidence interval tends to increase, though, for subsamples with high emission levels, such as the group of vehicles equipped with 3-speed transmission configurations ((2.98% about the 36.55 g/mile average for 6V-92TA engines or (3.73% about the 36.99 g/mile average for the Series 50 engines). 1760
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 9, 2001
Contrary to what one may expect, the above results do not indicate any superiority of the modern, Series 50 engine, over its 6V-92TA predecessor, in regard with the emissions of nitrogen oxides. Figures 3-5 complement the numerical values presented in Table 2. They illustrate graphically that the NOx emissions from the Series 50 models are higher by about 6% on average than those of 6V-92TA engines (Figure 3), buses operated on D1 fuel emit, on average, lower NOx levels than those fueled by D2 diesel, but this difference is evident only for the 6V-92TA models (Figure 4) and is even reversed in the case of the Series 50 engines (Figure 5). It is reasonable to assume that testing parameters and procedures as well as the technical conditions of specific fleets are likely to affect measured emission levels at a greater extent than the particular type of diesel fuel that powered the test vehicles. The statistical characteristics given in Table 2 indicate that the distributions of NOx data are more asymmetric for the Series 50 engines (skewness equal to 0.85) than for the 6V92TA model (skewness equal to 0.74), while the 95% confidence intervals are also wider for the Series 50 ((3% about the 35.18 g/mile average) than the 6V-92TA engines ((2.4% about the 33.20 g/mile average). The latter observation is displayed visibly in Figure 3. One example of dividing an asymmetric population of data points into smaller subsets, with a more symmetric distribution, is illustrated in Figures 6 and 7, for the 6V-92TA and the Series 50 models, respectively. Separate studies of data groups divided by the configuration of the vehicle transmission reveal that, for both types of engine, the degree of asymmetry associated with the 3-speed transmissions is much more pronounced than for all the other cases. A drastic reduction, of close to 25%, in the average NOx levels is observed with the transition from 3-speed to 4-speed transmissions in 6V-92TA engines (Figures 4 and 6). However, these two transmission configurations are associated with almost identical NOx emissions in the case of the Series 50 engine model (Figures 5 and 7). It is important to outline, though, that potential effects of transmission characteristics may be coupled with those of the engine model year, since transitions to new models have been accompanied, frequently, by performance enhancements of vehicle transmissions. The trends of changes in NOx emissions over various model years have been investigated from 1980 to 1993 for the 6V-92TA engines and from 1993 to 1996 for the Series 50 engines. The results are shown in Table 4 as well as in Figures 8-10. These tables indicate clearly that the sample sizes for 6V-92TA engines vary dramatically, from only four 1980 and 1981 models, to 140 data points for 1992 models. Most tests on Series 50 engines have involved 1994 models. Unusually large levels of data scatter are noticeable for the 1983 and
1989 model years of 6V-92TA engines, but their causes have not been, yet, identified. One possible cause of this wide scatter may be the coupling that exists between the twostroke operation of 6V-92TA engines and their varying supercharger performance. The average NOx emissions from the 6V-92TA engines have dropped dramatically and consistently from the 1987 to the 1993 model years, as illustrated in Figure 8. Tougher government regulations and restrictions on allowable limits of emissions from diesel engines have, obviously, been a driving factor for technical changes, like injection timing retardation for example, that have lead to such reductions. On the other hand, the new, Series 50 engines appear to produce higher emissions of NOx than the last model years of the 6V-92TA engines, which may be attributed to “off-cycle” timing strategies. It is interesting to note that the model year 1993 appears to be a “reversal” stage for NOx emissions from engines produced by DDC, when the transition from the 6V-92TA to the Series 50 models happens to coincide with the minimum levels of NOx emissions for both of them (see Tables 4 and Figure 8). Stacked bar charts are used in Figures 9 and 10 to display, separately for 3-speed and 4-speed transmissions, respectively, possible effects of the engine model year on NOx emissions from 6V-92TA engines powered by either D1 or D2 diesel fuel. The ranges of 95% confidence levels are stacked in these figures on top of the corresponding “low bound” values of NOx emissions from selected sample groups of test data. While the trend of decreasing emissions for engine models later than 1988 is visible in Figure 9 for 3-speed transmissions, no significant change is noticed in Figure 10 for 4-speed transmissions, in regard with engine models following the 1991 model year. Furthermore, the latter chart shows no noticeable effect of the fuel type, D1 or D2, on the NOx emissions measured for the 1991-1993 model years. The increased levels of scatter measured on buses equipped with 3-speed transmissions are visible in Figure 9, in the form of wider ranges of 95% confidence intervals.
Trends of PM (Particulate Matter) Emissions A visual comparison between the histograms displayed in Figures 1 and 2 reveals that the 6V-92TA models perform better than the Series 50 engines with respect to the emissions of nitrogen oxides, but this comparison is reversed for PM emissions. Unlike the above discussion about NOx emissions, both the levels of magnitude and scatter of PM emissions seem to be significantly lower for the Series 50 than for the 6V-92TA models. Furthermore, the frequency distributions of PM emissions from the Series 50 engines exhibit visibly lower skewness and higher kurtosis (Figure 2) than the corresponding distribution characteristics of NOx data. These observations are supported clearly by a comparison of numerical values between Tables 2 and 3 as well as by comparing the results shown in Figures 3-5 for the NOx emissions, with the equivalent ones displayed, respectively, in Figures 11-13 for the PM emissions from the same data samples. Although the Series 50 engines produce higher levels of NOx emissions than their 6V92-TA predecessors (by 6%, on average), the levels of PM produced by these engines are, on average, less than a third of those released by the 6V92TA models. Although the sample size of PM data is almost 2.5 times larger for 6V-92TA than for Series 50 engines, the range of the corresponding 95% confidence interval is more than 2.5 wider. This indicates a significantly lower level of scatter for the Series 50 models, which could be attributed, most likely, to more accurate process controls, leading to lower variability of air/fuel ratios for the 4-stroke Series 50 engines, in comparison to the early, 2-stroke, 6V-92TA, engines. As in the case of NOx emissions, the overall averages and scatter levels of PM emissions are lower for engines fueled
FIGURE 11. Low bounds and 95% confidence intervals of PM emissions measured through CBD cycles of DDC engines in transit busses.
FIGURE 12. Low bounds and 95% confidence intervals for PM emissions measured through CBD cycles of 6V-92TA engine models in transit buses.
FIGURE 13. Low bounds and 95% confidence intervals for PM emissions measured through CBD cycles of Series 50 engine models in transit buses. by D1, instead of D2 diesel, although by a larger difference of about 20%, as compared to the corresponding 10% difference observed for nitrogen oxides. One may notice, though, that this observation holds for both engine models in regard with the average PM emissions, whereas the NOx emissions from Series 50 models were lower, on average, for D2 than for D1 diesels (see Figure 5). Any comparison between the two main types of diesel fuel, D1 and D2, must consider, as outlined earlier in this paper, many additional factors related both to equipment (engine model year, its maintenance quality) and operation or testing conditions (process VOL. 35, NO. 9, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1761
FIGURE 14. Frequency distributions of PM emissions from 6V-92TA diesel engines in buses with various transmission configurations.
FIGURE 15. Frequency distributions of PM emissions from Series 50 diesel engines in buses with various transmission configurations. control, maintenance procedures, weather, data reduction and analysis, technical skills). It is interesting to note that, unlike Series 50 models, the 6V-92TA engines produce lower emissions in vehicles with 4-speed, rather than 3-speed, transmissions, both in regard with NOx (Figure 4) and PM (Figure 12). The average emissions released by the model Series 50, however, do not appear to depend on the transmission configuration (either 3- or 4-speed), neither for NOx (Figure 5) nor for PM (Figure 13). The most plausible explanation to this difference is drawn from Table 1 that shows the sample sizes for various model years of 6V-92TA and Series 50 engine models, along with the corresponding transmission configurations of test buses powered by these engines. All Series 50 engines tested in buses with 3-speed and 4-speed transmissions were newer than the 1993 year model, most of them (16 tests) being 1994 model years in buses with 3-speed transmissions. By contrast, most emission testing of the 6V-92TA engine models has been performed on vehicles with 3-speed transmissions (82 buses, as compared to 51 buses with 4-speed transmissions), but only 23% of these tests (19 vehicles) included model years later than 1990. On the other hand, about 90% of the 6V-92TA engines tested in vehicles with 4-speed transmissions (46 out of 51 buses) were relatively new, of the 1991, 1992, and 1993 model years. Quick inspection of the histograms displayed in Figures 14 and 15 provides further insight into the reasons, and the meanings, of the numerical results brought up in Table 3 as well as in Figures 12 and 13. Although the median of PM emissions from 6V-92TA engines in buses with 3-speed transmissions is slightly smaller than the one corresponding to 4-speed transmissions (0.92 vs 0.93), their highly skewed distribution in Figure 14 explains both 1762
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 9, 2001
FIGURE 16. Variation trends of average PM emissions over a decade of model years for DDC engines in transit buses. their higher averages and wider confidence intervals displayed in Figure 12. The unusually high level of scatter shown in Figure 13 for the Series 50 engine in buses with 4-speed transmissions is caused by only a few PM measurements obtained in the unusually high range of 1.75-2 g/mile, as presented in Figure 15. The expected reductions in exhaust emissions with the new, Series 50, engine as well as with improvements from one model year to another are much more pronounced and conclusive for the PM than for the NOx. This conclusion is evident not only from a brief comparison between the equivalent Figures 8 and 16 but also by examining the tabular results presented in Tables 2 and 4 for NOx and Tables 3 and 5 for PM emissions. The general trend of continuously decreasing emission levels following the 1990 model year is consistent for the 6V-92TA engines in regard with both the NOx and the PM data, but it is reversed for NOx emissions from the Series 50 engines. Similarly to the trends of NOx emissions illustrated in Figure 8, it appears that the modern, Series 50 engine, starts operating with its 1993 model at levels of PM emissions that are almost identical with those generated by its predecessor, the 6V-92TA engine in the same model year, the last of its production. Unlike the NOx emissions, however, the measured levels of PM emissions from 1993 to 1996 models of the Series 50 engine never exceeded the lowest emission levels from 6V-92TA engines, as measured for its last, 1993 model year. In addition to the above observations about average values, it is important to notice that the lower levels of data scatter observed in newer engines, both for NOx and PM, along with more symmetric distributions, lead to narrower ranges of 95% confidence limits even for relatively small sample sizes. The exception to this trend seems to originate, again, from the NOx levels measured on Series 50 engines, mainly those of the 1994 model year.
Emissions of CO and CO2 Oxides of carbon are measured in exhaust emission tests, since CO2 provides an indicator of fuel consumption, while CO is known to be toxic and is a species responsible for nonattainment of the CO National Ambient of Air Quality Standards in several urban environments. The results available for these gases in the database generated by WVU’s transportable laboratories demonstrate significant superiority of the new, Series 50, engine over its predecessor, the popular 6V-92TA. This assertion is evident from the frequency distributions displayed, respectively, in Figures 17 and 18, for exhaust measurements of CO and CO2. The results indicate not only a significant reduction, of about 57% for CO and
TABLE 5. Descriptive Statistics of PM Data (g/mile) from DDC Engines of Various Model Years model 6V-92TA models
Series 50 models
year
count
mean
median
std dev
coeff variat
95% conf
1980 1981 1982 1983 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996
4 4 5 11 5 12 17 73 37 88 73 136 65
2.77 5.58 1.59 2.63 0.47 0.83 0.52 1.56 1.30 1.52 1.13 0.81 0.58
2.76 5.67 1.62 1.97 0.38 0.73 0.66 1.16 1.08 1.66 1.10 0.85 0.71
0.18 0.49 0.22 1.37 0.15 0.25 0.44 1.22 0.82 1.16 0.41 0.39 0.31
0.06 0.09 0.14 0.52 0.31 0.30 0.85 0.78 0.63 0.76 0.37 0.48 0.53
0.28 0.79 0.27 0.92 0.18 0.16 0.23 0.28 0.27 0.24 0.10 0.07 0.08
FIGURE 17. Frequency distributions of CO emissions from main models of DDC engines, tested through CBD cycles of transit buses.
count
mean
median
std dev
coeff variat
95% conf
20 99 65 54
0.34 0.26 0.42 0.28
0.18 0.19 0.34 0.27
0.37 0.20 0.34 0.07
1.09 0.77 0.81 0.25
0.17 0.04 0.08 0.02
CO emissions from Series 50 engines is almost 3 times lower than the value corresponding to 6V-92TA engines (0.27 as compared to 0.77), despite the fact that the sample size of the latter is about 2.5 times larger. Statistical analysis of test results shows that test vehicles with 4-speed transmissions yielded, on average, CO emissions lower by about 45% than those with 3-speed transmissions, while engines fueled by D1 yielded CO emissions that were lower, on average, by about 33.5% than those fueled by D2 diesel. However, the average emissions of carbon dioxide were higher by about 24% for 6V-92TA engines in buses with 4-speed transmissions, than from the same engine models in buses with 3-speed transmissions. However, such variations in emission levels may be closely associated with changes in the engine model year and tradeoffs between emissions and efficiency. An investigation of year-to-year variations of the average levels of CO and CO2 test data reveals that the CO emissions display, as expected, a visible trend of reductions with newer model years. A different trend, however, is displayed by the CO2 emissions, that appear to have increased in the last model years of the 6V-92TA engines, without any significant changes with the model year in the case of the Series 50 engines.
Trends of Hydrocarbon (HC) Emissions
FIGURE 18. Frequency distributions of CO2 emissions from main models of DDC engines, tested through CBD cycles of transit buses. 18% for CO2, in the average emissions from Series 50 model engines, as compared with the 6V-92TA models, but also less variability of the measured emissions data. The coefficients of variation are almost twice higher for the 6V-92TA than for the Series 50 data, whereas the “95% confidence level” for
A rather dramatic improvement in HC emissions has been achieved with the replacement of the 6V-92TA engines by its successor, the Series 50 model, as it is clearly illustrated by the results shown in Figure 19. The average level of such emissions is almost eight times lower for the Series 50 engine, while the range of their 95% confidence interval is narrower by about 2.5 than that corresponding to 6V-92TA models. This improvement may be attributed partly to the 4-stroke operation and partly to improved fuel injection hardware. One may notice, however, that the shape of HC data distribution for the 6V-92TA models is more symmetric than the patterns displayed by the other types of regulated emissions for the same test samples. Such enhanced consistency of HC data is an indication of the fact that measurements of HC may be more robust than other emission results due to uncontrollable variations of testing and hardware conditions. Despite the above conclusive distinction between the two engine models, no evident trend can be identified, for either one of them, regarding the variations of HC emissions with the model year of the engine. Contrary to what one might expect, no noticeable improvement seems to be associated with the testing of newer engines, either of the 6V-92TA or the Series 50 models. The available VOL. 35, NO. 9, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1763
FIGURE 19. Frequency distributions of HC emissions from main models of DDC engines, tested in transit buses simulating the CBD driving cycle. test data indicate a visible trend of increasing HC emissions for the 6V-92TA engines from the 1990 to the 1991 and then to the 1992 models, whereas a similar rising trend, but significantly less pronounced, is observed for the Series 50 engines from the 1993 to the 1994 and then to the 1995 model years. However, it should be remembered that series 50 engine HC emissions are very low, often close to background levels.
Acknowledgments The authors are grateful to the staff of the WVU Transportable Laboratories for obtaining the data used in this study, and to the U.S. Department of Energy, Office of Transportation Technologies, and the National Renewable Energy Laboratory, for support of most of the previous bus emissions characterization programs.
Literature Cited (1) Heywood, J. B. Internal Combustion Engine Fundamentals; McGraw-Hill Book Co.: New York, 1988.
1764
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 9, 2001
(2) Clark, N. N.; Nine, R. D.; Daley, J. J.; Atkinson, C. M.; Tennant, C. J.; Lyons, D. W.; Peerenboom, W. H.; Suski, V. A. Heavy Duty Truck Emissions: Driving Routes and NO/NO2 Ratios; 8th CRC On-Road Vehicle Emissions Workshop, San Diego, CA, April 1998. (3) Bagley, S. T.; Baumgard, K. J.; Gratz, L. D.; Johnson, J. H.; Leddy, D. G.; Characterization of Fuel and After-treatment Device Effects on Diesel Emissions; Health Effects Institute Research Report, No. 76; September 1996. (4) Health Effects Institute. Diesel Emissions and Lung Cancer: Epidemiology and Quantitative Risk Assessment; A Special Report of the Institute’s Diesel Epidemiology Expert Panel; June 1999. (5) Graboski, M. S.; Yanowitz, J.; McCormick, R. L. In-Use Emissions From Heavy Duty Vehicles Operating in the Colorado Northern Front Range Area; 8th CRC On-Road Vehicle Emissions Workshop, San Diego, CA, April 1998. (6) Yanowitz, J.; McCormick, R. L.; Graboski, M. S. Environ. Sci. Technol. 2000, 34(5), 729-740. (7) Clark, N. N.; Gautam, M.; Lyons, D. W.; Bata, R. M.; Wang, W.; Norton, P.; Chandler, K. Natural Gas and Diesel Bus Emissions: Review and Recent Data, SAE Paper 973203; 1997. (8) Clark, N. N.; Gautam, M.; Boyce, J.; Wang, W.; Lyons, D. Emissions Performance of Natural Gas and Diesel Fueled School Buses with Cummins 8.3 L Engines; Spring Technical Conference, ASME Internal Combustion Engine Division, Paper 99-ICE-76, Proceedings ICE Vol. 32-2. (9) Gautam, M.; Clark, N.; Lyons, D.; Long, T. Jr.; Howell, A.; Loth, J.; Palmer, G. M.; Wang, W. G.; Bata, R. Design Overview of a Heavy Duty Mobile Vehicle Emissions Testing Laboratory, ASME DE - Vol. 40, Advanced Automotive Technologies; ASME Winter Annual Meeting, Atlanta, GA, Dec 1-6, 1991; pp 199-207. (10) Lyons, D.; Bata, R.; Wang, W. G.; Clark, N.; Palmer, G. M.; Gautam, M.; Howell, A.; Loth, J.; Long, T., Jr. Design and Construction of a Transportable Heavy Duty Vehicle Emission Test Laboratory, ISATA Paper No. 920450; 25th International Symposium on Automotive Technology and Automation (ISATA), Florence, Italy, June 1-5, 1992. (11) Cheremisinoff, N. P. Practical Statistics for Engineers and Scientists; Technomic Publishing Company: Lancaster, PA, 1987. (12) Kiemele, M. J.; Schmidt, S. R. Basic Statistics: Tools for Continuous Improvement; Air Academy Press: Colorado Springs, CO, 1990.
Received for review June 26, 2000. Revised manuscript received January 22, 2001. Accepted February 12, 2001. ES001416F
