Environ. Sci. Technol. 2005, 39, 9424-9430
Cold Temperature PM Emissions Measurement: Method Evaluation and Application to Light Duty Vehicles J Y R K I R I S T I M A¨ K I , † J O R M A K E S K I N E N , * , † ANNELE VIRTANEN,† M A T T I M A R I C Q , ‡ A N D P A¨ I V I A A K K O § Aerosol Physics Laboratory, Institute of Physics, Tampere University of Technology, P.O. Box 692, FIN-33101 Tampere, Finland, Research and Advanced Engineering, Ford Motor Company, P.O. Box 2053, MD 3083, Dearborn, Michigan 48121, and VTT Processes, Energy and Environment, Engines and Vehicles, P.O. Box 1601, FIN-02044 VTT, Finland
This work examines the methodology to sample and measure the number and size of motor vehicle particulate emissions (PM) at subambient temperatures. The study has two principal objectives. The first is to address the following question: which aspects of the particle sampling, dilution, and size measurement process must be made at the vehicle test temperature to obtain an accurate representation of the PM emissions? The second is to perform a preliminary overview of how subambient temperature operation affects PM emissions from the major classes of current model light duty vehicles. The principal findings are the following: (1) The temperature of the particle size instruments, test cell versus room temperature, has little effect on the measurements. (2) Once the engine has warmed, solid particle (soot) mode emissions in the cold test cell are similar to those at room temperature. The first finding simplifies cold temperature emissions testing because it allows particle sizing instruments to be placed outside the cold test cell and operated at room temperature. The latter is consistent with the expectation that solid particles are formed in the engine and are therefore relatively unaffected by ambient conditions after engine warm-up. Use of cold dilution air in the roomtemperature test cell increases the number and size of nuclei particles; however, the effect of dilution air temperature was inconclusive in the cold test cell.
Introduction That gaseous emissions increase at cold ambient temperatures has been known for over 30 years (1, 2). Since then, a relatively large decrease in the gas-phase emissions has been achieved by the use of catalytic converters and engine control systems. At present the question of how temperature affects particulate matter (PM) emissions has become important. Epidemiological associations between ambient * Corresponding author phone: +358-3-31152676; fax +358-331152600; e-mail:
[email protected]. † Tampere University of Technology. ‡ Ford Motor Company. § VTT Processes. 9424
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 24, 2005
particulate matter levels and potential adverse health effects are driving an increased scrutiny of particle emissions sources. To limit PM levels, many parts of the world, including Europe, the United States, and Japan, are lowering motor vehicle tailpipe PM standards significantly, to the point where conventional measurement methodologies may no longer be reliable (3). At these new low levels even small details in the vehicle testing and PM sampling protocols may play significant roles in the measurement results. The ambient temperature is certainly one factor that could affect PM emissions, the conventional wisdom being that lower temperatures lead to higher PM emissions. As a result of the more stringent emissions standards, there is an increased interest within the emissions research community in the issues of vehicle exhaust dilution and PM sampling. Historically, these issues have focused on the solid particle mode of soot particles, with its accompanying adsorbed hydrocarbons and sulfate, that is typical of diesel emissions. Several authors have reported data showing that, in addition to this solid particle mode, vehicle PM emissions are often accompanied by a nucleation mode of particles in the 5-30 nm range. This mode is very sensitive to small changes in the dilution parameters and sampling system, e.g., residence time and dilution air temperature (3-5), dilution ratio (6), storage-release effects in the transfer lines (7-10), and dilution air humidity level (9, 11, 12). The current PM emissions standards are mass based, whereby the solid particle mode generally dominates. However, with efforts to reduce PM emissions, such as the application of diesel particulate filters, several authors (6, 13, 14) have reported under some vehicle operating conditions an increase in the nanoparticle (dp < 50 nm) number count when the mass of the soot mode decreases. The combination of tighter emissions standards and the relatively larger role of the nucleation particles leads to the need for a reliable method to measure the nucleation mode. In the EU-funded project Particulates [characterization of exhaust particulate emissions from road vehicles; Growth program, http://vergina.eng.auth.gr/mech/particulates/] a dilution system reported by Ntziachristos et al. (15) was developed for this purpose. Mathis et al. (9) used the Particulates dilution scheme to study the combination of dilution ratio and dilution temperature necessary for reproducible nucleation mode particle formation. To date, the vast majority of motor vehicle PM emissions studies have been carried out at room temperature (∼25 °C). Although there is general agreement that decreasing the dilution air temperature will increase the nucleation mode, there is little information on what effect the ambient temperature has on either the nucleation mode or soot mode tailpipe emissions. Aakko and Nylund (16) studied the particle emissions with different fuels and engine technologies at subzero temperatures. Medium duty diesel engine data indicated large, 10-fold, increases in PM emissions at subzero temperatures. However, the temperature effects observed for light duty vehicles were considerably smaller. Mathis et al. (17) found that conventional diesel vehicles showed almost no dependence on test cell temperature. On the other hand, gasoline and diesel particulate filter equipped diesel vehicles showed increased PM at low temperature. In addition to the potential effect of temperature on vehicle emissions, several issues increase the complexity of the dilution and sampling at subzero temperatures. Below zero, the high water content of exhaust gas requires careful precautions. If cooling of the exhaust occurs in the sampling line instead of during dilution, there is a risk of water freezing 10.1021/es050578e CCC: $30.25
2005 American Chemical Society Published on Web 11/10/2005
TABLE 1. Technical Specifications of the Cars Measureda diesel 1 (pretest) model year engine displacement (L) power (kW) torque (Nm/rpm) emissions certification a
1999 1.9 85 EURO 2
diesel 2
diesel 3
gasoline 1
gasoline 2
2002 1.4 50 160/2000 EURO 3
2002 2.0 66 205/1900 EURO 3
2002 1.6 MPI 79 148/3750 EURO 3
2003 1.8 GDI 90 174/3750 EURO 3
MPI ) multiport fuel injection, GDI ) gasoline direct injection.
in the sample lines. Semivolatile PM may to some degree evaporate if the temperature rises after their formation, potentially requiring the control of temperature gradients along the sample line. The question thus arises as to whether the measurement instruments themselves must be located at the vehicle test temperature to fulfill this requirement. In previous cold temperature studies (16, 17) this question has not been addressed. In the present article we examine the effect of temperature on three aspects of PM measurement: (1) operation of the measurement instruments; (2) dilution of the exhaust; (3) emissions from the vehicles. The temperature range extends from -7 to 25 °C. Results are reported from five light duty vehicles, three diesel, one port fuel injection gasoline (MPI), and one gasoline direct injection (GDI).
Experimental Methods The experimental tests were carried out at the Technical Research Centre of Finland (VTT) in Espoo, Finland. The vehicles were run on a Froude Consine (1.0 m) chassis dynamometer enclosed in a climate-controlled test cell that can maintain its temperature within (0.5 °C during soak and within (1.5 °C during vehicle operation. A Pierburg 12.5 WT constant volume sampler (CVS) was used for filter-based gravimetric measurements of PM mass. The test vehicles, all passenger cars, are described in Table 1. Diesel 1 was used for preliminary testing and to investigate the effect of temperature on the particle measurement instruments. Two common-rail diesel vehicles, 2 and 3, and gasoline vehicles 1 and 2 were used to examine how the test cell temperature and the dilution air temperature affect the characteristics of the PM emissions. The two diesel vehicles are both of Euro 3 technology but from different manufacturers. Although vehicles with more stringent emissions classifications might have lower PM emissions, the effect of temperature on their PM emissions is not expected to differ qualitatively from the test vehicles used here, with the exception of those vehicles that incorporate a diesel particulate filter. Regulated emissions, CO, HC, NOx, and CO2, were measured with Pierburg AMA 2000 gas analyzers. The vehicles were procured from a rental agency, determined to be properly functioning, and, except for the fuel, tested as received. The sulfur levels of the diesel fuel and gasoline were 300 and 40 ppm, respectively. The relatively high fuel sulfur level was chosen because it has been found to enhance nucleation (10, 11, 14). It was thought that the nucleation mode would be more sensitive than solid particles to temperature and, thereby, provide a more stringent test of the measurement system. Chemical analyses of the diesel and gasoline fuels used in this study are presented in the Supporting Information. The sampling system was similar to what has been described by Ntziachristos et al. (15), except that we used a sample probe prior to the porous tube type diluter (18). A schematic of the dilution system is presented in the Supporting Information. A residence time chamber was used after the primary dilution to provide time for particle growth.
Secondary dilution was performed with an ejector pump diluter after the residence time chamber. The primary dilution ratio was determined using NOx analyzers for diesel vehicles and CO2 analyzers for gasoline cars. The primary dilution ratio in the steady-state tests varied between 9 and 16 with an average of 12.3 and standard deviation (STD) of 8.7%. The secondary ejector diluter provides a nominal dilution ratio of 7, but this was not monitored during the tests. The dilution air temperature was controlled with a heater and heat exchanger. The heater was used to provide warm dilution air when testing at subzero cell temperatures, and the heat exchanger was used to provide cooled dilution air at normal room temperatures. A 3-way valve allowed selection of the desired dilution temperature. The test matrix included two test cell temperatures of +22 °C (labeled “warm”) and -7 °C (labeled “cold”), as well as two primary dilution air temperatures of +20 and -3 °C, also labeled “warm” and “cold”. The nominal temperature set points, as well as the range of temperatures experienced during testing, are collected in the Supporting Information. PM emissions were measured with an electrical lowpressure impactor (ELPI) (19, 20) and a scanning mobility particle sizer (SMPS) (21). First, steady-state tests were run with diesel 1 to ascertain whether it is necessary to maintain the particle instruments at the test cell temperature, i.e., locate them inside the test cell. The goal was to ascertain if nanoparticle evaporation is significant enough to alter the measured number size distribution when the instruments are kept at room temperature for a “cold” test. This was accomplished by comparing simultaneous measurements from two (ELPI-SMPS) instrument pairs. One pair was located outside, and the other inside, the cold cell, with the exception that the condensation particle counter (CPC) for the SMPS of the “cold” pair was located outside the test cell so that the CPC could operate within its temperature specification. The classifier portion of the “cold” SMPS remained inside the test cell. The data from the “cold” versus “warm” instrument tests reveal that instrument temperature is not an important factor in PM emissions measurement, at least with respect to ELPI and SMPS measurements of diesel exhaust (see below). Therefore, the subsequent investigation into the importance of dilution air temperature and test cell temperature was conducted with “warm” instruments. These tests were also carried out with two ELPI-SMPS instrument pairs. Both pairs were located after the secondary dilution. One pair measured particles that first passed through a thermodenuder. This device removes semivolatile material, and these data are referred to as “solid” particle results. Those after denuder data are corrected for the size-dependent particle losses through the thermodenuder, which average ∼30%. The solid particle losses are experimentally determined for this unit at the 250 °C temperature in the 20-250 nm size range. The other instrument pair measured the total particulate aerosol, and the data are referred to as “total” number of particles. The particles removed by the thermodenuder are referred as “semivolatile” particles. An example is presented in the Supporting Information. Prior to each day’s testing, nebulized VOL. 39, NO. 24, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
9425
dioctyl sebacate (DOS) particles were used to ensure that both ELPI-SMPS pairs were operating within their specifications. The steady-state tests were run at 50 km/h at a load of 1.7-4.1 kW, at 80 km/h at load of 6.4-9.7 kW, and at 80 km/h at 15 kW. Transient PM emissions were recorded over the European ECE15 + EUDC drive cycle. The diesel cars were also driven over the ARTEMIS cycle to obtain reference information. Because the SMPS must be scanned over particle size, it was not used for transient PM measurements. Instead, the instrument pair consisted of a CPC and an ELPI, which are capable of real-time measurement. The CPC model used counted particles larger than 3 nm in diameter. The after denuder CPC is also corrected for ∼30% particle loss in the thermodenuder though the losses below 20 nm are not well characterized. Two repeat sets of measurements were carried out at each test cell temperature. ECE15 + EUDC cycles were run once at the beginning of each day followed by the steady-state measurements. The error bars presented in the figures are either the standard deviation of the measurements or 10% bars depending on which one is larger. For diesels the standard deviations of total and solid particle counts in warm steady-state measurements were below 14% in normal and below 30% at the 15 kW load points; exact values can be found in Supporting Information. The standard deviations are comparable to what has been reported previously (15) with similar dilution system for intralaboratory repeatability (10%) of accumulation and (50%) for nucleation mode particles. With the gasoline vehicles there were stability problems. The large standard deviations for the MPI vehicle’s steady-state tests are due to low particle counts and the relative instability of the particle emissions. For GDI there were problems with the speed/load behavior of the engine, which caused instability during the 80 km/h normal load measurement. The difficulties at 80 km/h 15 kW load were similar to those for the MPI vehicle.
Results: Instrument Temperature How the instrument temperature influences PM measurement can be examined by comparing data recorded simultaneously from a single test vehicle by two sets of instruments. This is illustrated in Figure 1 for two different combinations of cell and dilution air temperature. The upper plot compares data recorded by the “warm” SMPS-ELPI instrument pair (set 1) to the corresponding “cold” instrument data (set 2) when the vehicle (diesel 1) is run at 80 km/h and a load of 9.7 kW in a “warm” cell and the exhaust is diluted with “warm” air, i.e., a control experiment with both instrument pairs at room temperature. The lower graph makes the analogous comparison for the test vehicle run in a “cold” cell and the exhaust diluted with “cold” air. At this speed/load point the particle size distributions are bimodal, indicating the presence of both a solid particle mode (centered at ∼70 nm) and a nucleation mode (at 20-25 nm). A similar bimodal shape is found at both test cell/dilution air temperature combinations. In each case the instrument temperature exhibits a negligible influence on the measured particle size distributions, including the nucleation mode, which one might expect to be sensitive to temperature. In light of this finding, the remainder of PM emissions measurements were performed with the instruments located outside the test cell at room temperature.
Results: Dilution Temperature The effect of primary dilution air temperature on the measured PM size distributions was investigated at both the “warm” and “cold” test cell temperatures. At operating points 9426
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 24, 2005
FIGURE 1. Comparison of PM size distributions measured simultaneously using “cold” versus “warm” instrument sets: (A) vehicle run with “warm cell” and “warm dilution air”; (B) vehicle run with “cold cell” and “cold dilution air”. Type: diesel 1, 80 km/h, 9.7 kW load. where a nucleation mode is not present the dilution air temperature has no noticeable effect on the size distribution. When present, the nucleation mode peak diameter lies between 20 and 40 nm and partially obscures the solid particle mode. Use of a thermodenuder to evaporate and remove the semivolatile nucleation mode particles provides a view of the solid particle mode, which is centered at about 50-80 nm (Figure S2, Supporting Information). At the “warm” cell temperature the use of “cold” dilution air enhances the nucleation mode from diesels 2 and 3, both in terms of number and volume concentration (Figure 2). The volume concentration is calculated by assuming spherical particles. Although solid particles are not spherical, it is believed that in both cases this approach is biased equally enough to allow comparison of the results. The volume of semivolatile particles roughly doubles with the “cold” dilution air, whereas the solid mode remains unchanged within experimental reproducibility. The analogous investigation into the effect of dilution air temperature proved inconclusive for the “cold” test cell (Figure 2). Cold dilution air leads to a small, ∼10%, increase in the nucleation mode for diesel 2, whereas for diesel 3 it leads to an apparent 50% decrease in nuclei mode volume. Possibly this counterintuitive result is caused by test-to-test variability coupled with a weak dependence on temperature. At 80 km/h and high load there is already a great deal of nucleation; thus, dilution temperature plays a smaller role than when nucleation is near threshold. In this case variability in vehicle operation, which is likely exacerbated at low temperature, could exceed the effect of dilution temperature and lead to the inconclusive observations. In contrast to the ambiguity exhibited by the nucleation mode, the solid mode particles again appear to be insensitive to the dilution air temperature.
FIGURE 2. Comparisons of the total number (upper graph) and volume (lower graph) of solid and semivolatile particles measured at the four combinations of “warm” and “cold” dilution air in a “warm” versus “cold” test cell. The cell temperature is indicated on the horizontal axis, and the dilution air temperature is inscribed on each bar.
Results: Effect of Temperature on Vehicle PM Emissions The comparisons in this section focus on the differences in vehicle PM emissions when operated at “warm” versus “cold” ambient temperature. Unless the test cell and dilution air temperatures are individually specified, “warm” refers to measurements made in a “warm” test cell using “warm” dilution air and “cold” refers to data taken in a “cold” test cell and the exhaust treated with “cold” dilution air. For reference, the Supporting Information reports the regulated emissions recorded during the transient tests. Steady-State Tests. Number distributions from the three diesel vehicles tested at steady state are much alike. The solid particle mode is located around 54-110 nm in peak diameter, has a geometric standard deviation (GSD) of 1.71.9, and has a total concentration of (6-30) × 106 particles/ cm3. However there are differences in the extent of the nucleation mode. All 3 of the diesel test vehicles exhibit only a single PM mode at 50 km/h. When the speed is increased to 80 km/h, the solid particle mode increases (Figure 3a) and diesel 1 develops nucleation mode emissions (Figure 1). At high load all three diesel vehicles exhibit an extensive nucleation mode. This picture is qualitatively the same for “warm” and “cold” vehicle operation. The peak of the solid particle mode remains in the range 54-80 nm. Vehicle operation at -7 °C does not introduce a nucleation mode where one was not present at 25 °C. Nor does “cold” operation lead to large increases in particle emissions. In fact there is generally a 30% or so decrease in diesels’ solid particle emissions as the temperature is lowered (Figure 3a). The reason for this remains unclear. The magnitude of the change is of the same order as test-to-test variability, but the consistent decrease points to a systematic effect. Without additional studies into sampling and dilution, possible effects such as changes in thermophoretic losses or dilution ratio cannot be ruled out.
In “warm” operation, the PM emissions of the MPI gasoline vehicle are more than 2 orders of magnitude lower than the solid particle mode from the diesel vehicles. Unlike with the diesel vehicles, there is no sudden increase in particle number concentration at 80 km/h and high load from a nucleation mode. Under “cold” conditions the PM emissions show a large increase on a relative basis, roughly 10-fold compared to “warm” operation except at 15 kW load. However, on an absolute basis the PM emissions remain a factor of 10 below those from the diesel vehicles. The majority of particles fall below 100 nm in diameter. The low particle concentrations (
