Environ. Sci. Technol. 2007, 41, 6074-6083
Regulated and Non-Regulated Emissions from In-Use Diesel-Electric Switching Locomotives ANIKET A. SAWANT, ABHILASH NIGAM, J. WAYNE MILLER, KENT C. JOHNSON, AND DAVID R. COCKER, III* College of EngineeringsCenter for Environmental Research and Technology (CE-CERT), and Department of Chemical and Environmental Engineering, Bourns College of Engineering, University of California, Riverside, California 92521
Diesel-electric locomotives are vital to the operation of freight railroads in the United States, and emissions from this source category have generated interest in recent years. They are also gaining attention as an important emission source under the larger set of nonroad sources, both from a regulated emissions and health effects standpoint. The present work analyzes regulated (NOx, PM, THC, CO) and non-regulated emissions from three in-use dieselelectric switching locomotives using standardized sampling and analytical techniques. The engines tested in this work were from 1950, 1960, and 1970 and showed a range of NOx and PM emissions. In general, non-regulated gaseous emissions showed a sharp increase as engines shifted from non-idle to idle operating modes. This is interesting from an emissions perspective since activity data shows that these locomotives spend around 60% of their time idling. In terms of polycyclicaromatic hydrocarbon (PAH) contributions, the dominance of naphthalene and its derivatives over the total PAH emissions was apparent, similar to observations for on-road diesel tractors. Among nonnaphthalenic species, it was observed that lower molecular weight PAHs and n-alkanes dominated their respective compound classes. Regulated emissions from a newer technology engine used in a back-up generator (BUG) application were also compared against the present engines; it was determined that use of the newer engine may lower NOx and PM emissions by up to 30%. Another area of interest to regulators is better estimation of the marine engine inventory for port operations. Toward that end, a comparison of emissions from these engines with engine manufacturer data and the newer technology BUG engine was also performed for a marine duty cycle, another application where these engines are used typically with little modifications.
Introduction Transportation of freight by railroad is a critical component of the global economy. Freight railroads in the United States account for about 42% of the intercity freight market on a per ton-mile basis (1). Freight railroads may be typically * Corresponding author phone: (951) 781-5695; fax: (951) 7815790; e-mail:
[email protected]. 6074
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 17, 2007
classified into two broad categories: line-haul and switching & terminal (S&T). Line-haul railroads provide point-to-point freight transportation over length scales of the order of 1001000 miles, while S&T railroads provide logistical support to line-haul railroads within a specified area, such as a port. It has been reported that emissions resulting from S&T-type railroad operation, especially with respect to NOx, are significant for the Port of Los Angeles (2). The diesel-electric locomotive is virtually the exclusive form of propulsion for freight railroads in the United States. It is also a significant contributor to mobile source emissions in the United States, contributing 11% and 4% of the mobile source emissions inventory for NOx and PM10, respectively (3), up from 4% and 1% respectively for the preceding decade (4). Briefly, the operation of a diesel-electric locomotive is as follows (5, 6). The diesel engine drives an alternator that generates electric current, which is used to drive traction motors located within the locomotive wheel assemblies. This drive system decouples the diesel engine from the wheels and allows the engine to be operated under steady-state conditions at fixed percentages of rated power, called “notches”. Line-haul and switching locomotives differ markedly in their modes of operation, with the former spending significantly greater proportions of time in the highest notch position than the latter. Overall locomotive emission factors are calculated by weighting the notch-specific emissions by numerical factors that reflect the time spent in each notch. Table 1 lists engine speed, power, and weighting factor for each notch position for the switching locomotives tested in this study. Current locomotive emission standards in the United States follow a “tier” system (3) listed in Table 2 and are applicable to both new and remanufactured locomotives. Regulations for nonroad sources have not kept pace with on-road regulations. Thus, although nonroad diesel engines comprise a small percentage of all heavy-duty diesel sources combined, their contribution to the total emissions inventory due to heavy-duty diesel engines is out of proportion to their numerical population. Further, emissions from locomotives were found to contribute a significant percentage of the NOx and PM emission inventory for nonroad sources, at 30% and 12%, respectively (3). Regulatory agencies in the United States, most notably the U.S. Environmental Protection Agency (U.S. EPA) and the California Air Resources Board (CARB), have started to address the relative lack of regulation on nonroad sources in greater depth. For example, the U.S. EPA issued regulations governing locomotive emissions in 1998 (3), followed by a proposal in 2004 to move to stricter standards by 2007 along with switching locomotives to fuel having a sulfur level comparable to the current on-highway federal standard of 500 ppm (7). A review of the scientific literature suggests that the knowledge base of locomotive emissions is less comprehensive than that of on-road diesel emissions. In an early study, Bryant and Tennyson (8) reported emission factors for NO, NO2, CO, THC, and aldehydes from 4 different locomotives. The engines were also run on a simulated “road duty” cycle, and it was found that CO was the only component with emissions lower than the prevailing California standards for on-highway diesel engines. Researchers from the Southwest Research Institute (San Antonio, TX) have also performed much work in this area. For example, Fritz and Cataldi (5) measured regulated emissions from two locomotive engines from different manufacturers on a test stand, and compared their values with the prevailing U.S. EPA AP-42 emission factors (9). Their results with respect to influence of fuel sulfur level were inconclusive, as has been reported 10.1021/es061672d CCC: $37.00
2007 American Chemical Society Published on Web 07/24/2007
TABLE 1. Percentage of Maximum Engine Power, Maximum Engine Speed, and Emissions Weighting Factor for Each Tested Notch Position throttle position
percent of full loada
maximum engine speed (567/645b; rpm)
EPA weighting factora
idle 1 2 3 4 5 6 7 8
0.0 4.5 11.5 23.5 35.0 48.5 64.0 85.0 100
283/323 283/323 374/398 454/483 523/575 614/653 683/738 763/830 843/908
59.8 12.4 12.3 5.8 3.6 3.6 1.5 0.2 0.8
a EPA Regulatory Support Document, 1998 (3). b Displacement per cylinder in cubic inches, EMD Locomotive Service Manual.
TABLE 2. Emissions Standards for Switching Locomotives (U.S. EPA, 1998 (3)) emission standard (g bhp-1 hr-1) Tier (year of manufacture)
HC
CO
NOx
PM
Tier 0 (1973-2001) Tier 1 (2002-2004) Tier 2 (2005-)
2.10 1.20 0.60
8.0 2.5 2.4
14.0 11.0 8.1
0.72 0.54 0.24
elsewhere (3). However, a later study (10) showed significant (16-39%) reductions in PM emissions with 50 ppm S fuel relative to a high-sulfur (4760 ppm S) fuel. The reduction of NOx emissions from locomotives has also received attention. Fritz et al. (11) investigated the impact of injection timing on NOx emissions from 13 locomotives in three different configurations. They found that a 4° retardation in injection timing led to a 25% reduction in NOx emissions, albeit at the cost of somewhat higher smoke opacity readings. Other studies have also reported NOx emissions advantages from locomotive engines designed to run on liquefied natural gas (12, 13) and engines equipped with NOx control technologies such as selective catalytic reduction (13). Popp et al. (6) have also discussed the feasibility of using a remote sensing system to measure emissions from in-use locomotives, and found the NOx emission factor to compare favorably with data from Southwest Research Institute for a similar engine. In the United States as in other parts of the world, regulations cover emissions from four species (or agglomerations of species) emitted by diesel engines, namely, oxides of nitrogen (NOx), particulate matter (PM), total hydrocarbons (THC), and carbon monoxide (CO). Of these, THC and PM are not individual compounds, but rather complex mixtures of potentially several thousand compounds grouped together. THCs are segregated into groups of compounds in terms of both number of carbon atoms (e.g., C6, C7, etc.) and subgroups based on the presence of heteroatoms, typically oxygen (e.g., carbonyls, acids, alcohols, etc.). Compounds of interest from a human health standpoint include 1,3-butadiene, the so-called BTEX (benzene, toluene, ethylbenzene, xylenes) compounds, and carbonyl compounds (formaldehyde, acetaldehyde, acrolein). In the particle domain, PM is typically classified into elemental carbon (EC), organic carbon (OC), trace elements, and other inorganic compounds. The OC fraction contains several thousand compounds; among the most important of these from a human health perspective are the polycyclic aromatic hydrocarbons (PAHs), including naphthalene, benzo[a]pyrene, and others. Additionally, the concentration of the OC fraction of PM is a strong function of ambient temperature
and humidity, leading to partitioning of the “semivolatile” compounds between the gas and particle phases. While not regulated at the source(s), the U.S. EPA and CARB classify several compounds found in diesel exhaust as “hazardous air pollutants” (HAPs) and/or “mobile source air toxics” (MSATs), including the specific compounds named above. Various groups like Rogge et al. (14), Schauer et al. (15), and Miguel et al. (16) have investigated the concentrations of PAHs and/or n-alkanes from on-road light-, medium-, and heavy-duty diesel vehicles. Note that in these cases, it is primarily on-road engines that have been the subject of investigation. More specifically, there appears to be very little information in the literature on speciated gas- and particlephase emissions from nonroad sources. A study to investigate emissions from three typical switching locomotives used in the Port of Los Angeles (POLA) was conducted in the Spring of 2004. The present paper reports the regulated and non-regulated emissions from these locomotives, the first study of its kind on a 2007 CFRcompliant test system, and discusses the implications from a nationwide emission inventory as well as local air quality basis. Although the present paper discusses emissions from three diesel engines in the context of switching locomotive applications, it is to be noted that engines of this size and power range are essentially identical to those marketed by heavy-duty diesel original equipment manufacturers (OEMs) for diverse applications including marine propulsion (“Category 2” marine diesel engines, 5 e [volume per cylinder] e 20 L (17)), oil rig drilling (both land-based and offshore), and stationary power generation (both land-based and marine applications) (18). For example, the Electro-Motive Division (EMD) 12-645 and 16-645 engine families, both tested in this study (see “Experimental Section”) are used in harbor craft (19) and as emergency standby sets for critical installations such as hospitals and nuclear power plants (18). Therefore, it is not unreasonable to extend the relevance of the results set forth in this paper to this broader spectrum of nonlocomotive applications as well. This extended relevance is of interest, especially in the context of increased regulatory interest in all nonroad sources, beyond locomotives alone (17). The section “Implications for Other Transportation Sources” discusses this in greater detail.
Experimental Section Sampling System. The sampling system used in this study was the UCR/CE-CERT Heavy-Duty Mobile Emissions Laboratory (MEL), using a constant volume sampling system (CVS) for total exhaust capture. This indigenously developed system has been described in detail elsewhere (20, 21). A custombuilt stainless steel fitting was used to mate one stack at a time on each locomotive to the raw exhaust inlet on the MEL. Inside the MEL, the raw exhaust is mixed with filtered ambient air for primary dilution. Ports located 10 tunnel diameters downstream of the mixing point connect through heated lines to provide sample to real-time analyzers for gas-phase species including NOx, CO, THC, and CO2. A CFR 2007-compliant secondary dilution system (21) is used to collect samples for non-regulated gas- and particle-phase emissions, including PM2.5 collected on 47 mm, 2.0-µm pore size Teflon PTFE membrane filters (Pall, East Hills, NY). For further details refer to the Supporting Information. Test Matrix and Cycle. The test matrix for this study was based on fleet composition of the locomotive supplier, the largest switching locomotive operator in the POLA. Briefly, the test matrix includes locomotives that differ in engine family, cylinder count, output power, and exhaust plumbing (i.e., number of stacks). Two of the engines had undergone significant modifications in displacement to accommodate low-NOx fuel injectors. The fleet consisted of engines all manufactured by Electro-Motive Division (EMD), formerly VOL. 41, NO. 17, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
6075
TABLE 3. Details of Locomotives Tested locomotive model
no. of axles
engine mfr.
engine
engine model year
horse-power
turbo-charged?
no. of stacks
no. of stacks tested
SW-1200 SD-18 SD-20
4 6 6
EMD EMD EMD
12-567-Ca 16-567-D3a 16-645-CE
1956 1964 1979
1200 1800 2000
no no (de-turbo’d) no
2 2 4
2 1 1
a These 567 in3 engines have been rebuilt to a displacement of 645 in3 to accommodate low NO injectors that are compatible with 645-series x engines but not with the older 567-series engines.
FIGURE 1. NOx, PM, THC, and CO emissions for one of the locomotives tested on a per unit work (g bhp-1 hr-1) basis. Note: Idle values divided by 5 for clarity. of General Motors Corporation. EMD controls about 70% of the market for locomotive engines as a whole (4) and thus may be considered a suitable surrogate for in-use locomotive engines. The engines chosen belonged to two EMD engine series, the 567 and 645 (the numbers refer to the displacement per cylinder in cubic inches). Table 3 lists details of the locomotive test matrix. One factor not apparent from the table is that the older 567 locomotives were rebuilt using new 645-compatible components and low NOx fuel injector and effectively converted into 645 engines through piston, liner, and head modifications. All engines were tested at 9 steady-state notch positions: idle and notches 1 through 8 in descending order of power (i.e., notch 8 first). Notches 7 and 8 for locomotive SD-18 were not tested because of backpressure considerations. The power generated by the locomotive during this testing was routed through a load bank provided by the locomotive owner.
Results and Discussion Emissions by Notch Position. Figure 1 shows the general trend of NOx, PM, THC, and CO emissions for one of the locomotives tested on a per unit work (g bhp-1 hr-1) basis. All values are corrected for engine backpressure that was observed at the higher operating notches. Figure S1a-d (Supporting Information) show the emissions on a per unit work (g bhp-1 hr-1) basis for NOx, PM, THC, and CO for all the locomotives tested, uncorrected for back-pressure. We observe that, in general, the per unit work emission factors are highest for the idle notch, with a significant drop-off with increasing notch position. This trend is especially pronounced in the case of THC and CO, where the idle values are divided by 5 for clarity on the chart. The primary reason for this is the low power usage when the engine is idling. It should be noted that although the EPA regards idle as a state with zero 6076
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 17, 2007
percent engine load (see Table 1), we observed this not to be the case for the locomotives tested; the load bank registered an idling power usage in the region of 5-6 bhp for all engines tested, primarily due to auxiliary power requirements. When the emissions are viewed on a per kilogram of CO2 emitted [g (kg CO2)-1] basis (refer to SI Figure S2a-d, uncorrected for backpressure; Figure S3a-d, corrected for backpressure) this trend largely disappears, with the exception of total hydrocarbons. This is because the energy output by notch increases nonlinearly relative to increases in CO2 output for all locomotive stacks studied. Stack-to-Stack Variability. In general, locomotive engines typically have multiple stacks for routing of exhaust. Typically, one stack may serve a bank of 4 or 6 cylinders; it is not uncommon for a 16-cylinder engine to have 4 stacks. The method specified by the U.S. EPA for the testing of locomotive emissions (22, 23) specifies that emissions from multiplestack engines be tested by ducting the stacks together, or alternatively, proportional sampling is permitted from each stack individually provided that the CO2 concentrations in each exhaust stream are within 5% of each other at notch 8. In the present study, one of the locomotives (SW-1200; 1956) was tested for all 8 notches at both its stacks with total exhaust capture. Differences in g hr-1 CO2 emissions were found to be less than 4% absolute between the two stacks, making this engine “eligible” for testing on an individual stack basis. Therefore the assumption that different stacks on the same engine, each producing similar quantities of CO2, should produce similar levels of other species appears to be a reasonable one. The present data show that this is not necessarily the case. Revisiting Figures 1 and S1 (in the SI), we observe that differences in emissions for the regulated species are in many cases greater than the 4% difference observed for CO2, with
FIGURE 2. Weighted emission factors for locomotives from present study compared with manufacturer’s data. Note: NOx values divided by 10 for clarity.
FIGURE 3. Comparison of EPA (3) switch cycle data (hollow shapes) with manufacturer’s data (solid circle), and present study (solid triangles). the most prominent differences being observed for THC and CO during idling. In addition, load bank data indicates that the power differences between the two stacks, although typically less than 2%, do not follow the same trends as the CO2 differences. In other words, the variability observed between the two stacks of the same engine is of the order one might observe between two distinct engines. Comments on Back-Pressure. The CVS system used for this work was designed to measure full exhaust capture from engines up to 550 kW. For this testing at the highest notches, backpressure rapidly increased to about 16 in. of water, which is well below the specified maximum backpressure limit of 30 in. of water (29) (Table S17). For most of the cases the backpressure was well below 10 in. of water, having values above that for notches 7 and 8 only. According to Kim et al. (27) and Stamatelos et al. (28) increasing backpressure on an engine increases fuel consumption and effects emissions. For our experiments NOx remained almost same and PM values increased slightly as can be seen from uncorrected NOx and PM emission charts in the Supporting Information. These tests were originally focused on emissions from the locomotives within the local communities where the notch position rarely exceeds Notch 4. Notch 6 is the notch where backpressure issues were first encountered. Notches 6, 7,
and 8 contribute only 2.5% of the weighing factor for locomotives as per Regulatory Support Document (EPA). Therefore, impacts of the higher notches on overall emissions are small. Throughout this paper, regulated emissions from notches 6-8 are corrected for by assuming an emissions plateau in per CO2 emissions as provided by engine manufacturers and literature data. There is insufficient data available from the literature to provide similar corrections for non-regulated species and they are therefore presented as measured. The maximum correction encountered for this data on overall weighted emission rates was
