Brake Wear Particulate Matter Emissions - Environmental Science


Brake Wear Particulate Matter Emissions - Environmental Science...

1 downloads 114 Views 76KB Size

Research Brake Wear Particulate Matter Emissions BHAGWAN D. GARG, STEVEN H. CADLE,* PATRICIA A. MULAWA, AND PETER J. GROBLICKI General Motors R&D Center, MD 480-106-269, Warren, Michigan 48090-9055 CHRIS LAROO† Diversified Services, Inc., 26211 Central Park Blvd., Suite 600, Southfield, Michigan 48076 GRAHAM A. PARR AEROTEK Laboratory Services, 33097 Schoolcroft Rd., Livonla, Michigan 48150

Current particulate matter (PM) emission factor models estimate brake wear particulate matter emission rates using data derived from asbestos brakes. However, most brake pads are now produced from nonasbestos materials. Little work has been performed on emissions from brakes using these materials. Therefore, a brake wear study was performed using seven brake pad formulations that were in high volume use in 1998. Included were semimetallic brakes, brakes using potassium titanate fibers, and brakes using aramid fibers. Brakes were tested on a brake dynamometer under four wear conditions. On average, 35% of the brake pad mass loss was emitted as airborne PM. The observed wear rates correspond to vehicle emission rates of 5.1-14.1 mg/mi. On average, 86 and 63% of the airborne PM was smaller than 10 µm in diameter (PM10) or 2.5 µm in diameter (PM2.5), respectively. The large number of particles observed in some wear tests was attributed to condensation, a process that is highly dependent on dilution condition. Analysis of airborne PM showed very few inhalable fibers. On average, 18% of the airborne PM was carbonaceous material. Elemental analysis indicated that metallic species together with silicon, phosphorus, sulfur, and chlorine accounted for most of the remaining mass. Estimates of brake wear PM10 and PM2.5 emission rates from light-duty vehicles are made from brake dynanometer wear tests.

Introduction The U.S. EPA provides the PART5 model to the states for use in constructing the mobile source PM inventory. California uses the EMFAC7g model, which is similar to the PART5 model. These models calculate the on-road vehicle PM emissions from exhaust, brakes, and tires as well as fugitive dust. The PART5 model uses PM10 emission factors for lightduty gasoline vehicles of 13, 13, 8, and 35 mg/mi for exhaust, * Corresponding author phone: (810)986-1603; fax: (810)986-1910; e-mail: [email protected]. † Present address: Engine Measurement Division, Horiba Instruments Inc., 900 Hines Dr., Ann Arbor, MI 48108. 10.1021/es001108h CCC: $19.00 Published on Web 09/16/2000

 2000 American Chemical Society

brake, tire, and fugitive emissions, respectively, for the year 2000. The tire and brake wear emission rates are not varied by vehicle model year. While several studies have recently examined the exhaust PM emission rates from in-use vehicles (1-3), we are not aware of any recent studies of brake wear emissions. The 13 mg/mi brake wear emission factor is based on work by Cha et al. (4), who determined the mass emission rate, particle size distribution, and fiber emission rate from asbestos brakes. At the time of the study, brake pads were predominately made with asbestos. Since then, asbestos use in brakes has greatly diminished, having been replaced by a variety of other materials. Thus, the emission rate of fine particles from current brake pads and linings is unknown. This study determined mass emission rates from seven brake pads currently in high-volume use on new vehicles. It also examined particle size and composition. Fiber emissions are a special concern when dealing with brakes since all brake pad formulations use at least one fibrous material, including steel fibers, aramid fibers, and potassium titanate fibers. Therefore, fibers were measured as well. In addition, brake wear has been cited as a possible source of copper found in urban runoff. This runoff is believed to be a significant source of copper pollution in the San Francisco Bay area (5). Emission rates of copper and other elements were determined.

Experimental Section The selection of brake pads and linings for test is confounded by the large variety of compositions on the market. Filip et al. (6) analyzed the composition of 82 brake pads available in the American market. The pads are produced from four main components: binders, fibers, fillers, and modifiers. Various modified phenol-formaldehyde resins are used as the binders. Fibers were classified as metallic, mineral, ceramic, aramid, or potassium titanate. Fillers tend to be low-cost materials such as barium and antimony sulfate, kaolinite clays, magnesium and chromium oxides, metal powders, etc. Some fillers fulfill functions other than simply occupying space. Graphite is a major modifier used to influence friction. For this study, we chose to test brake pads in high-volume use in the 1998 production year. For the sake of simplicity, we will refer to both pads (disc brakes) and shoes (drum brakes) as pads, since pads constitute the major portion of the brakes in use on current production vehicles. Brake inventory data for General Motors was collected from a number of sources including automotive journals and internal contacts. We chose seven brake pads that represent 88% of the brakes used by GM in 1998 production. The selected pads are listed in Table 1 along with information on the brake type, the amount of friction material in the pad, and the inertial test weight. Semi-quantitative data on the composition of these pads gathered from material data safety sheets are available (7). All of the pads used a phenolic resin binder. However, the type of fiber used varies, as shown in Table 1. All tested brake pads contained barium sulfate along with other fillers such as calcium, zirconium, antimony, magnesium, and zinc compounds. Most of the brake pads include the addition of carbonaceous material such as carbon black, rubber, and cashew dust. For the pads selected, the ratio of car to truck brakes is 1.13. Five of the pads were from front brakes and six were disc-type brakes. Front brakes were emphasized since they typically handle 70% of the braking load. Note that the mass of the brake pads varies with vehicle size, from a minimum of 105 g for the small vehicles to 380 g for the larger trucks. VOL. 34, NO. 21, 2000 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

4463

TABLE 1. Tested Brake Pads id

vehicle

location

type

wt friction material (g)

inertia wt

type

1 2 3 4 5 6 7

LeSabre/Bonneville Saturn Malibu/Cutlass Regal/Intrigue/Grand Prix C/K pickups C/K pickups Blazer/Jimmy/Bravada

front front front rear front rear front

disc disc disc disc disc drum disc

230 105 105 215 380 350 360

1984 1376 1671 592 1616 1434 1577

semimetallic (steel fiber) potassium titanate, aramid, and copper fiber aramid, mineral, and copper fiber aramid fiber semimetallic, (steel fiber) semimetallic, (steel fiber) potassium titanate, aramid, and copper fiber

Brake Testing. Brakes were tested on a brake dynamometer (Link Friction Test Machine, model D) located at the GM Milford Proving Grounds. This dynamometer has a 60 hp DC drive capable of accelerating the brake and wheel assembly to speeds of 130 km/h. The inertial load can be varied from 120 to 2000 kg. Testing is done with production brake and wheel assemblies. Braking is controlled with a hydraulic system capable of generating a maximum pressure of 320 kPa (2000 psi). The system is computer operated. Parameters monitored during a test include speed, deceleration, pressure, torque, brake temperature, cooling air speed, and air temperature. Brakes were tested using engineering wear test procedure BSL-035, which is similar to JASO-C427-75. This procedure tests brakes repeatedly at four temperatures, nominally 100, 200, 300 and 400 °C. The temperature is the brake surface temperature as measured by a thermocouple inserted in a hole in the rotor. After each brake application, the temperature rises. The next brake application starts as soon as the brake cools to its target temperature. The temperature rise is roughly 25 °C during each brake application. Braking was done against an inertial mass representative of the vehicle for which it was designed. All braking events were done from 50 to 0 km/h at a deceleration rate of 2.94 m/s2 (0.3 g). Application of the brakes is designed to simulate a regular, on-road braking event. The brake temperature is controlled by adjusting the airflow past the brakes and by adjusting the interval between brake events. In some cases, it was necessary to increase the initial wheel speed from 50 to 65 km/h in order to obtain the 400 °C test temperature. All tests were started with new brake pads. Brakes were burnished (preconditioned) per BSL-032 at 100 °C for 200 stops. Brakes were then removed, weighed, and reinstalled. Any accumulated brake dust was removed at this time as well. The burnish runs were followed by the 100 °C test sequence, which consisted of 1000 stops. The entire 1000 stop test sequence varied from 7 to 13 h depending on the brake lining being tested. On average, the first 20 stops were done with no airflow past the brakes to rapidly get them to the test temperature. At higher test temperatures, this was increased to as many as 80 stops. At the end of the test, the brake pads were removed and weighed. Brake dust was removed from the test apparatus, and the brakes were reinstalled. The brake dust was collected in Ziplok bags and retained for elemental analysis. Since collection of brake dust was not quantitative, the brake dust was not weighed. Subsequently, the 1000 stop 200 °C test sequence was run, followed by the 500 stop 300 and 400 °C tests. Total elapsed time for these tests varied from 1.5 to 13 h. The brake assembly was enclosed in a 14 in. × 18 in. × 36 in. chamber. The inlet to the chamber was covered with a HEPA high-efficiency particle filter. Airflow was controlled by a computer-controlled blower located downstream. The exit from the chamber was through a 0.55 m diameter duct. Sampling probes were inserted into the horizontal section of ductwork 6.7 m from the brake assembly. Airflow through the system averaged 80.4, 30.0, 13.5, and 11.9 m3/min for the 100, 200, 300, and 400 °C tests, respectively. The 80.4 m3/min 4464

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 34, NO. 21, 2000

flow rate corresponds to an average velocity of 5.72 m/s. At this flow rate, the residence time between the brake assembly and the sampling point was approximately 1 s. Particle Sampling. Filter samples for mass and elemental analysis were collected on 47 mm diameter, 2.0-µm pore size Gelman Teflon filters. Nuclepore 47 mm diameter, 0.8µm pore size polycarbonate filters were also used to collect particles for elemental analysis. Pallflex 47 mm diameter Tissue Quartz 2500 QAT-UP filters were used to collect samples for carbon analysis. These filters were prefired at 900 °C in air for 3 h to remove carbon. Quartz filters were collected on the 100 and 300 °C tests while Nuclepore filters were collected on the 200 and 400 °C tests. BGI 47-mm aluminum filter holders were used. The inlet to these filter holders was modified to accept a 9 cm long, 35 mm i.d. tube that was used to improve the uniformity of the deposit. The upstream side of this tube tapered down to the 1/4 in. OD diameter tube that was used as the sample probe. Sample probes were inserted close to the centerline of the ductwork and had a 90° bend so they faced into the airflow. Particle samples were also collected on mixed cellulose ester filters (MCE) with a 0.8-µm pore size purchased from SKC. This filter is used for fiber analysis based on NIOSH fiber counting method 7400. The 47-mm filters were all operated at a flow rate of 18.4 L/min. The MCE filters were initially collected at a flow rate of 4.6 L/min; however, this was decreased to 1.1 L/min for some samples to avoid overloading. Particle size distributions were obtained with a micro orifice uniform deposit impactor (MOUDI). The impactor was operated with five stages and a final filter. The particle size cuts for the stages were >18, 10, 2.5, 1.0, and 0.1 µm. Uncoated aluminum foil was utilized as the collection surface. It should be recognized that impactors are subject to errors due to particle bounce. This problem tends to reduce the measured particle size. The final filter was a 37 mm diameter, 2.0-µm pore size Gelman Teflon filter. The sample flow rate was 30 L/min. Particle number distributions were measured with a Dekati electrical low-pressure impactor (ELPI). This instrument charges the aerosol and passes it through a 13-stage impactor (stage cut points from 0.03 to 10 µm). The current deposited on each stage is measured once per second, providing a near real-time size distribution. However, three factors combined to make the acquisition and analysis of valid ELPI data difficult. The ELPI data were collected before an upgrade became available that added shielded insulators and new software. This meant that the deposition of fine particles to the large particle stages greatly limited the accuracy of large particle measurements. Great care had to be taken in the zeroing of the instrument. The automatic zeroing cycle was usually inadequate. In addition, negative signals were observed on the largest particle stages when bursts of particles entered the instrument. These factors combined to reduce the amount of valid data collected. Continuous particle counts were made with a TSI electrical aerosol analyzer (EAA). The analyzer was operated continuously at a setting that measures all particles greater than 0.01 µm in aerodynamic diameter. This analyzer was designed

TABLE 2. Brake Wear Mass Emission Rates brake id

temp (°C)

mass (mg/stop)

airborne (mg/stop)

% airborne

brake id

temp (°C)

mass (mg/stop)

airborne (mg/stop)

% airborne

1 1 1 1 2 2 2 2 3 3 3 3 4 4

100 200 300 400 100 200 300 400 100 200 300 400 100 200

4.50 11.79 6.09 22.33 1.29 2.90 9.81 30.57 1.75 2.21 4.86 27.06 0.27 0.63

1.66 3.20 1.58 2.15 0.50 0.88 1.47 7.12 0.48 0.30 0.39 2.99 0.11 0.17

0.37 0.27 0.26 0.10 0.39 0.30 0.15 0.23 0.28 0.14 0.08 0.11 0.41 0.26

4 4 5 5 5 5 6 6 6 6 7 7 7 7

300 400 100 200 300 400 100 200 300 400 100 200 300 400

1.40 12.09 2.27 2.12 4.13 5.65 2.48 7.72 30.14 131.0 1.76 5.18 10.97 15.43

0.17 3.23 0.90 0.86 1.22 0.82 0.29 0.89 2.23 10.81 0.84 0.97 1.74 3.01

0.12 0.27 0.40 0.41 0.30 0.14 0.12 0.12 0.07 0.08 0.48 0.19 0.16 0.19

for atmospheric studies and would have been saturated by the number of particles present in the 300 and 400 °C tests. To avoid this problem, cartridge filters with different size “pinholes” were used to provide an additional 2-37-fold dilution, as needed. Particle counts made with the EAA in this manner are complicated by the fact that the EAA response is not identical for all particle sizes. When operated at a single size setting, as in this study, an integrated response factor must be used that reflects the size distribution of the particles being measured. We used an integrated response factor of 1.24 × 105 particles mL-1 V-1. This response factor is considered to be accurate to no more than a factor of 2. The airflow in the duct was run at 143 m3/min for 10-15 min before the start of each test to remove any easily reentrained particles deposited during the preceding test. Sampling commenced when the brakes reached their operating temperature and stopped immediately after the end of the test sequence. Tunnel blanks and media blanks were collected and retained for analysis. Measurement of Gases. A continuous hydrocarbon (HC) flame ionization detector (FID) analyzer was operated during this study. During preliminary tests, CO and CO2 nondispersive infrared analyzers were operated as well. The brake test apparatus was located in an active vehicle test area and drew its air from the room environment. Thus, background concentrations of these species were variable, making accurate measurements of these species difficult. It was judged that an increase of 2.5 ppm CO or CO2 could have been detected if it had occurred regularly with each application of the brakes. This concentration was equivalent to ∼3 mg/stop as compared to wear rates of ∼10 mg/stop. At this sensitivity these gases were not detected. Sample Analysis. Filters and MOUDI foils were equilibrated in a constant temperature (64 °F) and humidity (RH ) 56%) balance room for a minimum of 24 h before initial and final weightings on a Cahn microbalance. MCE filters were analyzed for fibers using NIOSH Method 7400. Filters were cleared using acetone vapor and mounted on a slide using triacetin for refractive index matching. The MCE filters were analyzed optically using a Leitz Laborlux K microscope with phase contrast condenser and objective lens. The microscope was equipped with a video camera that was connected to a frame grabber to allow digital examination of the image. A computer image analysis system was used to make a quantitative data file of length and width for objects that were identified as possible fibers with the microscope. The minimum fiber length detectable using this method was approximately 0.5 µm. A subset of the samples was also sent to Clayton Laboratory Services in Livonia, MI, for confirmatory fiber counting. Quartz filters were sent to the Desert Research Institute for the determination of organic carbon (OC) and elemental

carbon (EC) by the thermal optical reflectance (TOR) method (8). Teflon filters, Nuclepore filters, and brake dust samples were sent to Pixe Analytical Laboratories for elemental analysis. PIXE (proton-induced X-ray emission) analysis is able to detect up to 72 elements with atomic numbers greater than 10.

Results Table 2 gives the brake wear rates as a function of test temperature. Wear rates are given as the mass lost per stop averaged over the entire test, which includes the warm-up stops. As expected, the mass lost per stop tends to increase as the brake temperature goes up. The lowest wear rate at 100-300 °C occurred for brake lining 4, which is a rear disc brake. This is due, in part, to the fact that rear brakes carry less of the braking load than front brakes. On the other hand, brake 6 is a rear drum brake and has the highest mass wear rate at 400 °C. A variety of factors contribute to the differences in wear rates including brake composition and the inertia weight being stopped. Also given in Table 2 is the average mg/stop of airborne particulate matter and the percent of the total wear detected as airborne particulate matter. For all seven brake pads, the percent airborne PM for the 100 °C tests is higher than for the 400 °C tests, averaging 35 ( 0.12 and 0.16 ( 0.07%, respectively. The averages for 200 and 300 °C were 0.24 ( 0.10 and 0.16 ( 0.08%. While the higher percentage of airborne PM for the lower temperature tests may be due to differences in wear processes, it could also be influenced by the 6.7-fold reduction in airflow for the higher temperature tests. Reductions in airflow could influence the results both by allowing more time for particles to deposit in the ductwork and by reducing the likelihood of the particles being entrained in the first place. To put these data into perspective for emissions inventories, it will be necessary to rely on engineering estimates of in-use brake lining wear rates. Front brakes are expected to last 35 000 mi under normal usage while rear brakes are expected to last 70 000 mi. At the end of life, 80% of the friction material will have been worn off. On this basis, the total brake wear for a small car would be approximately 18 mg/mi. For a large car it would be approximately 28 mg/mi, while for a large pickup truck it would be approximately 47 mg/mi. Normal in-use brake service generates brake temperatures of 100-200 °C as reported by Sasaki (9). The average airborne fractions found in this study at 100 and 200 °C were 35 and 24%. Therefore, we will assume 30% of the wear is airborne PM. On this basis, the airborne PM from small cars to large pickup trucks would range from 5.1 to 14.1 mg/mi. Particle Size Distribution. Figure 1 shows MOUDI particle size distributions for four of the 100 °C tests. The distributions for brakes 1, 4, and 7 are dominated by the largest and smallest VOL. 34, NO. 21, 2000 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

4465

FIGURE 1. Particle mass size distributions.

TABLE 3. Particle Size Determined with the MOUDI Impactor temp (°C)

MMDa (µm)

% mass