A ew Method for the Real-Time Measurement of Diesel Aerosol

Transcription

1 A ew Method for the Real-Time Measurement of Diesel Aerosol Contract Final Report Department of Health and Human Services Centers for Disease Control National Institute for Occupational Safety and Health (NIOSH) Grant 1 R01 OH /1/2005 to 7/31/2008 Prof. David B. Kittelson, Ph.D. Principal Investigator kitte001@umn.edu Phone: Co-investigators: Winthrop F. Watts, Jr., Ph.D. Jason P. Johnson Adam C. Ragatz University of Minnesota Department of Mechanical Engineering 111 Church St. SE Room 1100 ME Minneapolis, MN, /12/2010

2 TABLE OF CO TE TS LIST OF TABLES... 3 LIST OF FIGURES... 4 LIST OF TERMS AND ABBREVIATIONS... 5 ACKNOWLDEGEMENTS... 6 ABSTRACT... 6 SIGNIFICANT FINDINGS... 8 TRANSLATION OF FINDINGS... 9 IMPACT SCIENTIFIC REPORT A. Specific Aims B. Accomplishments C. Introduction D. Methods and Materials Engine Test Conditions, Fuels, Aftertreatment Devices Aerosol Instrumentation Particle Sizing Particle Counting Surface Area Monitors Photometers Filter Samples Catalytic Stripper Sampling System Quality Assurance E. Results and Discussion Initial Engine Tests Evaluation of Emission Control Devices Catalyzed Diesel Particulate Filter (CDPF Diesel Oxidation Catalyst (DOC) Instrument Comparison Underground Mine Study Portable Catalytic Stripper Additional Research F. Summary G. References PUBLICATIONS

3 LIST OF TABLES Table 1. Instrument and position in the sampling system Table 2. CDPF filtration efficiencies Table 3. CDPF SMPS, photometer and filter measurements for the 1400 RPM 250 N-m condition Table 4. Filter measurements for ECD-1 and B50 fuel with and without the DOC Table 5. SMPS measurements for ECD-1 and B50 fuel with and without the DOC Table 6. AM510, DustTrak and aethalometer measurements for ECD-1 and B50 fuel with and without the DOC Table 7. Percent reduction or increase calculated for the DOC with filter data Table 8. Percent reduction or increase calculated for the DOC with SMPS data Table 9. Percent reduction or increase calculated for the DOC with AM510, DustTrak and aethalometer data Table 10. Correlation coefficients for the ECD-1 fuel without the DOC Table 11. Correlation coefficients for the ECD-1 fuel with the DOC Table 12. Correlation coefficients for the B50 fuel without the DOC Table 13. Correlation coefficients for the B50 fuel with the DOC Table 14. Summary correlation coefficients Table 15. Summary of samples collected during the mine study Table 16. MSA gravimetric samples collected at each mine sampling location Table 17. AM510 photometer samples collected at each mine sampling location Table 18. EC-OC < 1.0 µm filter samples collected at each mine sampling location Table 19. Ratios of reference samples to AM510 photometer samples Table 20. SMPS data for number, length, surface area and volume concentrations for catalytic strippers and penetration fractions Table 21. Summary statistics for the portable and rack-mounted PAS Table 22. AM 510 photometer response at two different flow rates Table 23. NO 2 concentrations upstream and downstream of the DOC

4 LIST OF FIGURES Figure 1. Typical Diesel number, surface area and mass weighted size distributions Figure 2. Volatile nature of nuclei mode particles (Source Kittelson, et al., Figure 3. Weighted average SMPS size distribution for the four test condition Figure 4. Residence time chamber, units are in inches Figure 5. Filter holder and flow meters used to sample DPM after primary dilution Figure 6. Filter critical orifice flow control manifold used to control filter flow rate Figure 7. Sample distribution manifold Figure 8. Schematic of the sampling system Figure 9. Unusual regular spike of particles from the Deere engine Figure 10. Broken cylinder ring Figure 11. Response with and without CS, dilution ratio approximately 200: Figure 12. SMPS size distribution with and without the CS Figure 13. SMPS number size distributions for the 1400 RPM 250 N-m condition with ECD-1 fuel with and without the CDPF and CS Figure 14. DOC evaluation at 1400 rpm 50 N-m for ECD-1 and B50 fuels Figure 15. DOC evaluation at 1400 rpm 100 N-m for ECD-1 and B50 fuels Figure 16. DOC evaluation at 1400 rpm 250 N-m for ECD-1 and B50 fuels Figure 17. DOC evaluation at 1400 rpm 450 N-m for ECD-1 and B50 fuels Figure 18. EC and OC averages with and without the DOC for the two fuels Figure 19. SMPS number and volume size distributions for ECD-1 fuel with and without the DOC Figure 20. Suite of instruments setup in the crusher operator s control room Figure 21. Schematic of portable instrument Figure 22. Drawing of the CS portion of the portable instrument with photo Figure 23. Power consumption for the portable CS Figure 24. Measured and predicted particle penetration through the lab and mini-cs Figure 25. SMPS size dependent particle losses in the mini-cs and lab CS Figure 26. Volume size distributions of the lab and mini-cs Figure 27. Stripchart for 1400 rpm 100 N-m condition up and downstream of the DOC 52 Figure 28. Stripchart for 1400 rpm 450 N-m condition up and downstream of the DOC 53 4

5 LIST OF TERMS A D ABBREVIATIO S B50 50% soymethyl ester and 50% ultra-low sulfur Diesel fuel NO Nitric oxide B99 99% soymethyl ester and 1% ultra-low sulfur Diesel fuel NO 2 Nitrogen dioxide BaO Barium oxide NOx Oxides of nitrogen BaSO 4 Barium sulfate NSAM Nanoparticle surface area monitor BC Black carbon O 2 Oxygen BP British Petroleum OC Organic carbon CARB California Air Resources Board PAHs Polycylic aromatic hydrocarbons CDPF Catalyzed Diesel particulate filter PAS Photoelectric aerosol sensor CO 2 Carbon dioxide PE Photoelectic CPC Condensation particle counter PM Particulate matter CRC Coordinating Research Council QAUP Pre-fired quartz ultra pure filter CS Catalytic stripper R 2 Correlation coefficient DC Diffusion charger SMPS Scanning mobility particle sizer DOC Diesel oxidation catalyst SOF Soluble organic fractioin DOS Dioctyl sebacate SO 2 Sulfur dioxide DPF Diesel particulate filter SO 3 Sulfur trioxide DPM Diesel particulate matter TC Total carbon DR Dilution ratio TL Transfer line EAD Electical aerosol detector ULSD Ultra-low sulfur Diesel fuel EC Elemental carbon UV Ultraviolet ECD-1 Ultra-low sulfur Diesel fuel provided by BP VW Volkswagon ECU Electronic control unit EEPS Engine exhaust particle sizer EPA Environmental Protection Agency H 2 O Water H 2 SO 4 Sulfuric acid JM Johnson Matthey MSHA Mine Safety and Health Administration NIOSH National Institute for Occupational Safety and Health 5

6 ACK OWLDEGEME TS This research was made possible by a grant from the National Institute for Occupational Health and Safety (NIOSH). It would not have been possible without the collaboration of Vulcan Materials Co., TSI, Inc. and Johnson Matthey, Plc. Vulcan Materials hosted the two week mine study and we thank all the miners and managers at the Central Quarry in Lexington, KY for their cooperation and assistance. In particular we thank Mr. Kelly Bailey, Cooperate Director of Industrial Hygiene and Health Services. Without his assistance and support the mine study would not have been possible. TSI, Inc. loaned the University many aerosol instruments and provided technical support for this study. In particular we want to thank Dr. Beau Farmer, Chief Technical Officer, and Mr. Brian Osmondson and Mr. Gregory Olson, Marketing Managers, for their support. Johnson Matthey (JM) provided numerous catalysts and technical support for this project. We want to thank Dr. Martyn Twigg, Chief Scientist, for his continued support of research conducted in the University of Minnesota s Power and Propulsion Laboratory. We would also like to thank the NIOSH Pittsburgh Research Laboratory and Dr. James Noll for analyzing the elemental and organic carbon samples collected during the study. Finally we would like to thank John Deere for providing the engine and technical support. ABSTRACT Diesel exhaust is a complex mixture of gases and Diesel Particulate Matter (DPM), and in metal and non-metal underground mines the Mine Safety and Health Administration (MSHA) regulates DPM concentrations using a time-weighted-averaged, 8 hr, full-shift, permissible exposure limit. The limit is 160 µg/m 3 of total carbon (TC) and National Institute for Occupational Safety and Health (NIOSH) standard method 5040 is used to determine the amount of elemental (EC), organic carbon (OC) and TC in the sample. This sampling and analysis method does not provide real-time data, nor does it provide information on particle volatility, particle size, or aerosol number concentration. These factors are known to be affected by emission control devices, engine duty cycle, fuel and lubrication oil composition and other factors. The goals of the project were: Evaluate the DPM control efficiency of selected catalyzed emission control devices in the laboratory using traditional and non-traditional measures. Non-traditional measures include particle size, particle volatility, aerosol number and volume concentrations. Traditional measures include mass, elemental carbon (EC) and organic carbon (OC). Evaluate and recommend procedures for the use of a low cost instrument package for the routine test cell evaluation of Diesel engines equipped with and without emission control devices. Develop a portable CS for use with portable aerosol instruments used in underground, non-gassy mines to obtain real-time data on the physical and chemical characteristics of DPM to which miners are exposed. 6

7 Control of temperature, residence time, flow, and dilution air quality are critical for repeatable, real-time measurement of nanoparticles, and the minimization of artifacts on filters collected for gravimetric analysis. This is particularly important for evaluating DPM emissions from 2007 compliant engines conforming to the 0.01 g/bhp-hr U.S. Environmental Protection Agency (EPA) standard with or without aftertreatment devices. This standard is roughly a 90% reduction from the previous level corresponding to an approximate 100 µg mass gain on the filter during an EPA certification test. A catalyzed Diesel particulate filter (CDPF) consisting of a catalyzed metallic prefilter and a catalyzed filter was evaluated. The CDPF particulate matter removal efficiency as measured by number, surface area or volume was > 99.9 %, and filters used to collect mass and EC/OC had no visible deposits. We compared the gravimetric based filter measurements used to determine the mass concentration to estimates of mass derived from the scanning mobility particle sizer (SMPS) to determine the amount of filter artifact. Filter artifact is defined as the weight gain not attributable to the collection of suspended particles. Our estimates show that in this case the standard TX40 filter overestimated mass by roughly a factor of 45, which is in general agreement with what is reported in the literature. The significance of this finding is that reliance upon filter based measurements to determine mass emissions from a low emitting engine with CDPF can lead to significant mass measurement error and misinterpretation of results. Extensive tests were conducted using a Diesel oxidation catalyst (DOC) and the findings were mixed. These devices have little impact on EC, reduce OC and may increase or decrease nitrogen dioxide (NO 2 ) and sulfate concentration, depending upon operating conditions. The DOC reduced the OC concentration when the catalyst light off temperature was achieved allowing oxidation of volatile organic material. On the other hand, oxidation of sulfur dioxide (SO 2 ) at higher operating temperatures to sulfates increased the mass concentration. The same was true for the formation of NO 2 under some circumstances. The nuances of catalyst storage and release are not well understood and prediction of what a specific catalyst will do under specific conditions is problematical without laboratory testing. Use of DOCs on Diesel equipment used in occupational settings such as an underground mine is advisable only when sufficient information is available to evaluate the potential consequences of using these devices on mine air quality. Laboratory measurements from a suite of portable, real-time aerosol instruments correlated well (correlation coefficient (R 2 ) > 0.95) with time weighted average DPM mass and EC concentrations. The relationship between the instruments and the time weighted averaged measurements (and each other) was affected primarily by the amount of volatile material available to form nanoparticles (< 30 nm in diameter). The amount of volatile material and the nanoparticle concentrations were affected by the engine condition (load and speed), fuel type, dilution condition, and presence of a catalyzed aftertreatment device. Real-time instruments can estimate the solid particulate matter mass attributable to EC in the laboratory. However, the user must be aware of how the physical and chemical aerosol characteristics can impact the estimate. In field tests conducted at an underground limestone mine the TSI AM510 portable photometer (equipped with a Dorr Oliver cyclone and 1.0 µm impactor) qualitatively tracked time weighted average mass and EC-OC measurements. The 7

8 correlation coefficient (R 2 ) between the TC and the calibration factor adjusted photometer measurements was The main issues holding back the use of a photometer for real-time estimation of DPM are the removal of non-dpm associated particulate matter associated with mining activity from the aerosol stream and calibration of the photometer to mine specific aerosol. The photometer calibration for DPM must be re-evaluated when other aerosols like drill oil mist or welding fume are present in the sampling area. The main issues holding back the use of a photometer for real-time estimation of DPM is the removal of non-dpm associated particulate matter (PM) from the aerosol stream and calibration of the photometer to mine aerosol. Work remains to reduce the interfering dust to improve the specificity of the measurement. The mini-catalytic stripper (CS) designed and built for this project is battery operated and was used in combination with the photoelectric aerosol sensor (PAS) to provide information on the volatile and non-volatile fractions. The CS could also be used with other portable instruments such as the diffusion charger (DC) or photometer. The prototype was operated for over 4 hrs using two lithium ion batteries typical of those used in laptop computers. Further design modifications will package the mini-catalytic stripper with one or more portable instruments or sensors in an instrument that is smaller than the size of a shoebox. The EC to OC ratio is altered by the type of fuel, engine operating condition, and by the use of a catalyzed aftertreatment device such as a DOC. The DOC operates just like a CS in that it removes volatile organic material that is available to nucleate and form nanoparticles. The fact that the EC to OC ratio varies makes it problematical to attempt to establish a consistent relationship between TC and EC. SIG IFICA T FI DI GS The major findings from the project include: The two stage dilution tunnel built for the project is an enhanced version of previous dilution systems. It adds a 2.3 µm cut cyclone, improves the temperature and flow control for the residence time portion of the tunnel, and relies on ultra-clean dilution air for dilution. Control of temperature, residence time, flow, and dilution air quality are critical for repeatable, real-time measurement of nanoparticles in the laboratory, and the minimization of artifacts on filters collected for gravimetric analysis. This is particularly important for evaluating DPM emissions from 2007 compliant engines conforming to the 0.01 g/bhp-hr EPA standard with or without aftertreatment devices. This standard is roughly a 90% reduction from the previous level corresponding to an approximate 100 µg mass gain on the filter during an EPA certification test. A catalyzed Diesel particulate filter (CDPF) consisting of a catalyzed metallic prefilter and a catalyzed filter was evaluated. The CDPF particulate matter removal efficiency as measured by number, surface area or volume was > 99.9 %, and filters used to collect mass and EC/OC had no visible deposits. We compared the gravimetric based filter measurements used to determine the mass concentration to estimates of mass derived from the scanning mobility particle sizer (SMPS) to determine the amount of filter artifact, defined as the weight gain not attributable to the collection of suspended particles. Our estimates show that in this case the filter 8

9 overestimated mass by roughly a factor of 45, which is in general agreement with what is reported in the literature. Extensive tests were conducted using a Diesel oxidation catalyst (DOC) and the findings were mixed. In general, DOCs have little impact on EC, reduce OC and may increase or decrease nitrogen dioxide (NO 2 ) and sulfate concentration, depending upon operating conditions. The nuances of catalyst storage and release are not well understood and prediction of what a specific catalyst will do under specific conditions is problematical without laboratory testing. Estimates of DPM concentration measured by portable, real-time aerosol instruments compared well (R 2 > 0.95) with time-weighted average DPM mass concentration and EC concentrations. The relationship between the instruments and the time-weighted average measurements (and each other) is affected primarily by the amount of volatile material available to form nanoparticles (< 30 nm in diameter). The amount of volatile material and the nanoparticle concentrations is affected by the engine condition (load and speed), fuel type, dilution condition, and presence of a catalyzed aftertreatment device. The TSI AM 510 portable photometer (equipped with a Dorr Oliver cyclone and 1.0 µm impactor) qualitatively tracked time weighted average mass and EC and TC measurements made in an underground limestone mine. The calibration factor TC relationship had an R 2 of The main issue holding back the use of a photometer for real-time estimation of DPM is the removal of non-dpm associated particulate matter from the aerosol stream. We believe that a newly introduced TSI DustTrak II monitor combined with a recently developed inlet that removes particles > 0.45 µm will be successful for tracking DPM concentrations underground. The mini-catalytic stripper designed and built for this project is battery operated and was used in combination with the PAS to provide information on the volatile and non-volatile fractions. It could also be used with other portable instruments such as the DC or photometer. The prototype was operated for over 4 hrs using two lithium ion batteries typical of those used in laptop computers. We believe further design modifications can package the mini-catalytic stripper with one or more portable instruments or sensors (i.e. PAS, DC, CPC and photometer) in an instrument that is smaller than the size of a shoebox. The EC to OC ratio is altered by the type of fuel, engine operating condition, and by the use of a catalyzed aftertreatment device such as a DOC. The DOC operates just like a catalytic stripper in that it removes volatile organic material that is available to nucleate and form nanoparticles. The fact that the EC:OC ratio varies makes it problematical to attempt to establish a consistent relationship between TC and EC. TRA SLATIO OF FI DI GS The emphasis of this project was to develop a sampling method for Diesel aerosol that can be used in the underground mine environment to provide real-time data for the management of mine ventilation, diesel usage, and mine air quality. A suite of portable real-time instruments was evaluated in the laboratory using a Deere engine fueled with two types of Diesel fuel (ultra-low sulfur Diesel and a 50% blend of soy methyl ester with ultra-low sulfur Diesel B50). Evaluations were done with and without emission control devices. Laboratory results showed that several portable instruments have the 9

10 potential to monitor Diesel aerosol in real-time. Among them are the portable TSI AM510 or DustTrak photometers, and the EcoChem Analytics photoelectric aerosol sensor. However, each is subject to errors caused by volatile material in the exhaust that may form nanoparticles. By combining these instruments with a portable catalytic stripper instrument performance and the specificity of the measurement are improved. For these instruments to be used successfully in a dusty mine environment a pre-selector to remove mine dust from the sample is needed. This was demonstrated in the field study conducted at an underground limestone mine where the TSI AM510 photometer qualitatively tracked the total and elemental carbon concentration. A more robust preselector would improve the DPM estimate. Our laboratory evaluations of the CDPF and DOC focused on the DOC because the amount particulate matter emitted from the CDPF was to low to be meaningful using gravimetric or EC/OC methods. However, we were still able to evaluate particulate matter emissions from the CDPF using our suite of real-time aerosol instruments. Laboratory tests showed that the ratio of EC to OC varied depending upon the type of fuel, engine operating condition, and whether an aftertreatment device such as a DOC was used. The DOC acts like a catalytic stripper in that it removes volatile organic material that is available to nucleate and form nanoparticles. An EC to OC ratio that varies makes it problematical to attempt to establish a mathematical relationship between TC and EC, and makes accurate prediction of the EC concentration from TC data difficult. The EPA 2007 particulate matter emission standard for heavy-duty on-road trucks is 0.01 g/bhp-hr. This corresponds to a 100 µg mass gain on the filter during a certification test. During an evaluation of a catalyzed Diesel particulate filter we compared the gravimetric based filter measurements used to determine the mass concentration to estimates of mass derived from the scanning mobility particle sizer (SMPS) to determine the amount of filter artifact, defined as the weight gain not attributable to the collection of solid particles. Our estimates show that in this case the filter overestimated mass by roughly a factor of 45, which is in general agreement with what is reported in the literature. Control of temperature, residence time, flow, and dilution air quality is critical for the repeatable, real-time measurement of nanoparticles, and the minimization of artifacts on filters collected for gravimetric analysis. IMPACT Underground mine operators must comply with the MSHA DPM standard in underground metal and nonmetal mines and do not have access to real-time data on DPM levels. Rather they rely upon collection of filter samples and EC and OC analysis to determine DPM concentrations. Improvements in mine design, ventilation, Diesel usage and operator work practices are made without the benefit of real-time DPM data to evaluate changes made in the mine environment. A real-time DPM instrument or sniffer will provide mine operators with valuable information to evaluate changes and assist in maintaining compliance with MSHA standards. We have demonstrated that several commercially available, portable instruments have the potential to make these measurements, but the presence of volatile material that may form nanoparticles, and the presence of mine dust can interfere with accuracy and specificity of the measurements. We demonstrated that a mini-catalytic stripper designed to remove volatile material and 10

11 used with these instruments improve the accuracy and specificity of the DPM measurement. If used with these instruments in the underground mine environment this combination has the potential to collect real-time DPM data that could be used to evaluate practices designed to minimize miner exposure to DPM. Usage of this package of instrumentation in other occupational or environmental situations where Diesel exhaust is a concern is likely to be much easier due to the lower dust concentrations. Our research was a collaborative effort involving Vulcan Materials, TSI, Inc., and Johnson Matthey. Each company has an interest in bringing the concept of a portable, real-time DPM monitor to reality. Vulcan Materials will use such an instrument to monitor DPM levels in underground limestone mines where the control of DPM is a high priority. TSI, Inc. will benefit by bringing to market a mini-catalytic stripper that can be used with a number of their instruments to provide added information on the volatile nature of the aerosol. Johnson Matthey benefits by providing the catalyst core used in the catalytic stripper, and by gaining fundamental knowledge on the performance of their catalysts. Ultimately the success of the project hinges on bringing a product to the market place that can be used by industrial hygienists and air pollution specialists. It is important to note that with the inclusion of black carbon (BC) as a global warming pollutant developing inexpensive ways to monitor BC atmospheric concentrations becomes an important priority. Our research has the potential to impact this environmental research area. We plan to continue our work with TSI and Johnson Matthey to develop the minicatalytic stripper. Future work at the University will be focused on downsizing the stripper, characterizing performance including particle losses, and minimizing power consumption. In the future we hope to evaluate our instrument package in an underground mine and have already obtained verbal agreement from Vulcan Materials to host such a study. This evaluation would be combined with an evaluation of an exhaust soot sensor developed jointly by the University of Minnesota and Honeywell with funding from the Department of Energy and the California Air Resources Board (CARB). 11

12 SCIE TIFIC REPORT A. Specific Aims Diesel exhaust is a complex mixture of gases and Diesel Particulate Matter (DPM), and in metal and non-metal underground mines the Mine Safety and Health Administration (MSHA) regulates DPM concentrations using a time-weighted-averaged, 8 hr, full-shift, permissible exposure limit. The limit is 160 µg/m 3 of total carbon (TC), which was upheld by the U.S. Court of Appeals for the District of Columbia (U.S. Court of Appeals, 2007). Recent publications by MSHA have provided guidance on sampling and error factor calculations (MSHA 2008a, b). NIOSH method 5040 is used to determine elemental carbon (EC), organic carbon (OC) and TC exposure but this method does not provide data in real-time, nor does it provide information on particle volatility, particle size, or aerosol number concentration. These factors are known to be affected by catalyzed emission control devices, fuel, lube oil composition and other factors. The original goals of this three year project were: Evaluate the DPM control efficiency of selected catalyzed emission control devices in the laboratory using traditional and non-traditional measures. Non-traditional measures include particle size, particle volatility, aerosol number and volume concentrations. Traditional measures include mass, elemental carbon (EC) and organic carbon (OC). Evaluate and recommend procedures for the use of a low cost instrument package for the routine test cell evaluation of Diesel engines equipped with and without emission control devices. Develop a portable CS for use with portable aerosol instruments used in underground, non-gassy mines to obtain real-time data on the physical and chemical characteristics of DPM to which miners are exposed. B. Accomplishments The accomplishments of the project include: Installed a 2005 Deere 4045H, 4-cylinder, 4.5-L, 129-kW (at 2400 rpm) engine in the laboratory. The engine is turbocharged, aftercooled with common rail fuel injection, and has EPA tier 2 approval for off-highway applications. Assembled a suite of real-time instruments including the TSI hand-held DustTrak and AM 510 photometers for mass measurement, the TSI Electrical Aerosol Detector (EAD) and Nanoparticle Surface Area Monitor (NSAM) and the Matter Engineering portable and rack-mounted Diffusion Chargers (DC) for surface area measurement, the EcoChem Analytics portable and rack mounted Photoelectric Aerosol Sensor (PAS) for BC measurement, the TSI 3007, 3775, 3025A Condensation Particle Counters (CPC) for particle number measurement, two TSI Scanning Mobility Particle Sizers (SMPS) and the TSI Engine Exhaust Particle Sizer (EEPS) for number, mass, volume, and surface area size distribution measurement. 12

13 Designed, constructed and evaluated a temperature controlled, two-stage minidilution tunnel that uses ultra-clean dilution air. Detected an engine fault by observing unusual particle number concentrations. Evaluated the performance of the instruments at four engine conditions (1400 rpm and 50, 100, 250 and 450 N-m) with the engine fueled with Ultra Low Sulfur Diesel (ULSD) containing < 7 ppm S or biodiesel fuel containing 50% soy methyl ester. Compared the response of the real-time instruments to time weighted averaged gravimetric mass and EC/OC samples. Evaluated the performance of the instruments listed above with the two fuels and a Diesel Oxidation Catalyst (DOC) installed in the exhaust of the Deere engine. Conducted a limited evaluation of a catalyzed Diesel particulate filter (CDPF) on the Deere engine fueled with ULSD fuel. Conducted a 2-week, in-mine evaluation of the performance of selected instrumentation at Vulcan Materials Central Limestone Quarry. Designed and evaluated three mini-catalytic strippers. Designed and evaluated a prototype portable instrument consisting of a mini-catalytic stripper (CS) and PAS. This instrument runs on battery power for more than 4 hrs. The prototype instrument was shown to representatives of TSI, Inc a major aerosol instrument manufacturer, and further development is ongoing. C. Introduction Figure 1 shows the physical relationship between Diesel nucleation, accumulation and coarse modes for three weighted size distributions (number, surface area and mass) (Whitby, et al., 1976; Kittelson, 1998), and an alveolar plus tracheo-bronchial deposition curve (ICRP, 1994). The curves have a lognormal, trimodal form and the concentration in any size range is proportional to the area under the corresponding curve in that range. The nucleation mode (3-30 nm) typically contains < 10% of the particle mass but > 90% of the particle number. Nucleation mode particles are usually composed of nearly allvolatile material as illustrated in Figure 2 (Kittelson et al., 2002, 2005), and discussed below. Most of the DPM mass is found in the accumulation mode (roughly nm size particles), and is composed of carbonaceous agglomerates and adsorbed materials. The coarse mode consists of particles larger than 500 nm in diameter and contains 5-20% of the mass. These large particles are formed by reentrainment of DPM, which has been deposited on cylinder and exhaust system surfaces, and crankcase fumes are also in this range. All three modes are better defined by their formation mechanisms than by rigid size boundaries. 13

14 ormalized Concentration (Ctotal -1 )dc dlogdp Nuclei Mode Deposition Accumulation Mode Coarse Mode Deposition ,000 10,000 Diameter (nm) Number Surface Mass Deposition (Alveolar + Tracheo-Bronchial, ICRP 1994) Figure 1. Typical Diesel number, surface area and mass weighted size distributions shown with tracheo-bronchial plus alveolar deposition (Source Whitby and Cantrell, 1976; Kittelson, 1998) 1.0E E+08 Mode 1 Mode 1, CS Mode 4 Mode 4, CS Corrected for dilution ratio Catalytic stripper (CS) operated at 300 ºC d /dlogdp, part/cm 3 1.0E E E E+04 Mode rpm, 1400 N-m Mode rpm, 350 N-m 1.0E+03 Caterpillar 3176, C-12, 6-cylinder, 12 L, turbocharged and aftercooled engine with high pressure, electronically controlled, direct fuel injection. The engine was fuel with ultralow sulfur fuel containing 50-ppm sulfur Dp, nm Figure 2. Volatile nature of nuclei mode particles (Source Kittelson, et al., 2005) 14

15 The chemical nature of DPM is complex. It is composed of volatile material found in the nucleation mode (Sakurai, et al., 2003a,b; Ziemann, et al., 2002; Kittelson, et al., 2005, 2006), and carbonaceous agglomerates found in the accumulation mode. Nucleation mode particles are formed as exhaust dilutes and cools allowing volatile material to pass from the gas to particle phase by two paths: nucleation to form new particles and adsorption or condensation on existing particles. Nucleation is favored if there is little carbonaceous surface area on which to adsorb. Nucleation is a highly nonlinear process so small changes in the ratio of volatile precursors to carbonaceous agglomerates during dilution will influence the concentration of nanoparticles (Abdul- Khalek, et al., 1999; Khalek, et al., 2000, Kittelson, et al., 2002). Carbonaceous agglomerates have large surface areas on which volatiles can condense or adsorb. The ratio of carbonaceous agglomerates to volatile precursor species changes when catalyzed aftertreatment devices are used such as flow through oxidation catalysts or catalyzed filters. Oxidation catalysts are typically used to reduce carbon monoxide and volatile hydrocarbons while filters reduce particulate matter. Catalyzed filters may also reduce the hydrocarbon fraction, but the degree of reduction depends upon the type of catalyst. In addition, filters reduce the solid carbonaceous fraction in the exhaust. Depending upon the catalysts used both devices can dramatically increase the conversion of SO 2 to SO 3 and H 2 SO 4. Volatile nucleation mode nanoparticles form during dilution by heterogeneous or homogeneous nucleation, while simultaneous removal by coagulation is taking place. In the ambient atmosphere, coagulation with existing accumulation mode particles occurs in a matter of minutes while in a mine this process will take place more quickly. In the mine atmosphere, the concentration of accumulation mode particles (dust and DPM) is higher, resulting in more rapid scavenging of newly formed particles as well as their precursors. The U.S. Bureau collaborated with the University of Minnesota to develop a size selective DPM sampler for use in underground mines (Rubow, et al., 1990a,b, McCartney, et al., 1992; Cantrell, et al., 1990, 1993). The research showed that the submicron aerosol in coal mines is nearly all contributed by DPM while the coarse mode is composed primarily of coal and rock dust. In metal and nonmetal mines about 20% of the Diesel aerosol contributes to the > 0.8 µm fraction (Cantrell, et al., 1990). The information derived from these studies is useful because commonly used real-time aerosol instruments generally respond to specific size fractions such as PM10 or PM2.5, but cannot distinguish between different types of aerosol such as dust, oil mist or DPM. Pre-treating mine aerosol with a cyclone and/or an impactor removes larger non-dpm. Further treating the sampled aerosol by passage through a CS will remove volatile material leaving primarily solid carbonaceous agglomerates and any dust that was not removed by the cyclone/impactor. Although some DPM is removed in this process we believe the resulting sample will yield a more sensitive and specific response for DPM when measured by a portable instrument such as PAS, photometer or surface area monitor. The PAS is particularly interesting because it responds to photoemitting substances on the surface of aerosol particles. Ultraviolet (UV) irradiation of the sampled aerosol leads to the emission of photoelectrons from surface material that readily undergoes photoemission (Burtscher, 1992). The remaining positively charged aerosol 15

16 particles are separated from the electrons and collected on a filter connected to an electrometer. The measured current is a function of the UV irradiation wavelength and intensity, the total available surface area and the photoemission properties of the surface materials. Commercially available instruments usually use a wavelength of 222 nm. The PAS strongly responds to Diesel accumulation mode particles (Matter, et al., 1999). This is particularly true when surface bound PAHs, having a high photoelectric (PE) yield, are present on the particles. However, a moderate response is still present even when DPM found in the accumulation mode consists primarily of EC, which has a lower PE yield (Baltensperger, et al., 2001; Siegmann, et al. 2000). For Diesel accumulation mode particles, the PAS signal correlates with the accumulation mode surface area concentration (Dahmann, et al., 2000). In mixed urban aerosols, on the other hand, where the PAH surface concentration (for example from spark ignition engines) may be much higher, the high PE yield of the PAH dominates the PAS response (Bukowiecki, et al., 2002). Therefore, PAS signals can only be quantitatively interpreted in a certain context. Nevertheless, PAS signal observations are useful for monitoring of relative changes using the measured electrometer response. In we conducted a pilot study for the NIOSH Pittsburgh Research Laboratory to evaluate the response of a suite of portable, near real-time aerosol instruments to Diesel exhaust aerosol with and without a CS. We determined instrument response as a function of particle size and volatility (Kittelson, et al., 2005) as illustrated in Figure 2. We showed that the response of the PAS was strongly influenced by the physical and chemical nature of Diesel aerosol confirming work done by Maisels, et al. (2002) and Jung, et al. (2005). The presence of a large, predominantly volatile nuclei mode particles < 30 nm in diameter, and/or the presence of volatile material on the surface of the solid carbonaceous agglomerates in the accumulation mode suppressed the PAS response. Removal of the volatile material by passage of the aerosol through the CS enhanced the response, and improved correlations between the PAS, the Diffusion Charger (DC), and the Scanning Mobility Particle Sizer (SMPS). The instruments included in this study were selected based upon low cost, portability, and the ability to provide near real-time data that would be useful in evaluating DPM emissions in the engine laboratory, and an underground mine. These findings are of significance for underground mining. Potentially useful, battery-powered, portable, near real-time aerosol monitors have been tried underground (Bugarski, et al., 2004; Watts, 2004). In the laboratory, the PAS with the CS provided a better estimate of the solid carbonaceous fraction most likely associated with EC. D. Methods and Materials Engine Test Conditions, Fuels, Aftertreatment Devices The engine used for all laboratory testing was a model year 2005, Deere 4045H, 4-cylinder, 4.5-L, 129-kW (at 2400 rpm) provided by Deere at no cost. This engine is turbocharged, aftercooled with common rail fuel injection, and has EPA tier 2 approval for off-highway applications. Exhaust gas recirculation was not used. All tests were conducted at steady-state conditions at 1400 rpm. Loads were set at 50, 100, 250, or 450 N-m. These conditions were selected to provide the instruments a variety of aerosol conditions based upon the number size distribution mainly nucleation 16

17 mode (1400 rpm, 50 N-m), bimodal distribution (1400 rpm, 100 N-m), and two conditions with a large accumulation mode (1400 rpm, 250 and 450 N-m). Figure 3 shows the weighted average of all size distributions collected at these conditions with ultralow sulfur (ECD-1) Diesel fuel containing 6 ppm S with no aftertreatment device in the exhaust. The standard deviations are shown. The weighted average is calculated as the sum of the measurements divided by their variances of the mean, divided by the sum of the reciprocals of the variances of the mean. Two fuels were used during the project. ECD-1 fuel is an ultra-low sulfur fuel containing 6 ppm S provided by BP. We blended ECD-1 with a soy methyl ester (B99) fuel purchased in MN to make B50 with a 3 ppm fuel S content. A complete chemical analysis from BP is available for the ECD-1 fuel and a partial analysis of the BP-50 was done by Southwest Research Institute. The rationale for evaluating two fuels is that biofuel blends are known to decrease exhaust EC and mass, and a number of mines are using B50 or higher blends to reduce these concentrations underground. The limestone mine that hosted the in-mine study was using B50 at the time of the study, and was in the process of increasing to B99. Donaldson provided two aftertreatment devices a Diesel oxidation catalyst (DOC) and a catalyzed Diesel particulate filter (CDPF). Far more tests were done with the DOC than the CDPF because of the high efficiency of DPM removal by the CDPF. To provide meaningful samples for gravimetric and EC analysis, sampling times with the CDPF in place would have had to be very long (hrs), making the evaluation far more costly, and subject to significantly more artifact in the filter based measurements. 4.0E E E+07 ECD-1 fuel, no DOC, no CS Weighted average SMPS scans Standard deviation 1400 RPM 50 Nm 1400 RPM 100 Nm 1400 RPM 250 Nm 1400 RPM 450 Nm d /dlogdp, part/cm 3 2.5E E E E E E Dp, nm Figure 3. Weighted average SMPS size distribution for the four test condition with standard deviation 17

18 Aerosol Instrumentation Particle Sizing Scanning Mobility Particle Sizer (SMPS): During the project a TSI 3071A Electrostatic Classifier with TSI 3025A or 3010 CPC was used to determine the particle number size distribution. The 3071A was eventually replaced by a new TSI 3080 in The SMPS sizes particles by electrical mobility using the CPC to count the selected particles. It was configured to cover the size range of 8 to 300 nm over a period of 90 s with a 30 s interval between scans. From the number distribution we calculate the surface area, active surface area, volume and mass weighted size distributions, and the total number, surface area, active surface area, and volume concentrations. The SMPS active surface is calculated using expressions given by (Pandis, et al., 1991; Kasper, et al., 2000, 2001) assuming spherical particles and that the species charging the particles is hydrated proton which has a mean free path of 14.5 nm (Pui, et al., 1988). Active surface is an important metric because the photoelectric aerosol sensor and diffusion charger signals are proportional to active surface. Volume concentrations are calculated assuming spherical particles. Engine Exhaust Particle Spectrometer (EEPS ): The TSI EEPS sizes particles from 5.6 to 560 nm and has a 1 s response time. The EEPS has been described in detail elsewhere (Johnson, et al., 2004). The EEPS classifies particle size using electrical mobility, counting particles with a series of electrometers to determine the size distribution. The EEPS fast response time makes it excellent for measurements made during transient engine operating conditions. Particle Counting Condensation Particle Counters (CPC): We used numerous TSI CPCs during the project that covered a range of particle size from about 3 nm to 1,000 nm in diameter. The principle of operation is condensation of a liquid such as butanol, isopropyl alcohol or water on particles to grow them to an optically detectable size by an optical particle counter. We also used a portable, battery-powered TSI 3007 CPC. The 3007 has an upper concentration limit of 100,000 part/cm 3 and a 95% response time of < 9 s, and particle size detection limit of 10 nm. Surface Area Monitors Electrical Aerosol Detector (EAD): The TSI 3070A EAD was used to measure the total aerosol length concentration (mm/cm 3 ). Aerosol length can be thought of as number concentration times average diameter, or simply diameter to the first power weighting. This measurement falls between the number concentration and the surface area concentration (diameter to the second power weighting). It is well suited for particles in the range of 10 to 1,000 nm. Values are provided in real time and are not biased according to chemical composition. Nanoparticle Surface Area Monitor (NSAM): The TSI 3550 NSAM is based on the EAD. It measures the human lung-deposited surface area of particles (reported as 18

19 µm 2 /cm 3 ) corresponding to either the tracheobronchial or alveolar regions of the lung. The dynamic range of the instrument is 0 to 10,000 µm 2 /cm 3. It is sensitive to particles as small as 10 nm, and data are collected every second. Diffusion Charger (DC): The DC measures the total active surface area of particulate matter. Positively charged ions are produced by a glow discharge, forming in the neighborhood of a very thin wire. These ions attach themselves to the sampled aerosol stream with a known probability. The charged aerosol particles are then collected on a filter. The electric current flowing from the filter to ground potential is measured and is proportional to the number of ions attached to the particles. For particles in the free molecular range, the attachment is proportional to the surface area of the particles, but is independent of the composition of the particles (Adachi, et al., 1985). Siegmann, et al. (1999) contend that the DC measures the so-called active surface in the size range from slightly above ten to a few hundred nm. The active surface is the effective surface area available for mass transfer in a kinetically limited situation and should be appropriate for describing the gas to particle mass transfer taking place in a diluting exhaust plume. Portable and rack mounted instruments were used. Photoelectric Aerosol Sensor (PAS): The PAS responds to photoemitting substances on the surface of aerosol particles. Ultraviolet irradiation of the sampled aerosol leads to the emission of photoelectrons from surface material that readily undergoes photoemission (Burtscher, 1992). The remaining positively charged aerosol particles are separated from the electrons and collected on a filter connected to an electrometer. The measured current is a function of the UV irradiation wavelength and intensity, the total available surface and active surface area and the photoemission properties of the surface materials. Photometers DustTrak and AM 510: These instruments are portable, battery-powered, laser photometers with real-time mass concentration readout and data logging capability. The monitor is designed to provide exposure assessment by measuring particle concentrations corresponding to respirable size, PM10, PM2.5 or PM1.0 size fractions. The signal is proportional to the total light scattered by the aerosol sample. Filter Samples Filter Samples: Filter samples were collected for gravimetric and EC/OC analysis after the first stage of dilution where the filter face temperature is < 47 ºC. Our procedure was to collect one TX40 filter for gravimetric analysis and three pre-combusted quartz fiber (QAUP) filters in separately loaded 37 mm cassettes for the EC/OC analysis by NIOSH method TX40 filters are one of two types of filters used for gravimetric analysis of engine particulate matter samples specified by the EPA (Teflon filters are the other), and QAUP filters are specified in the NIOSH 5040 method. We collected one dynamic blank for every three EC/OC samples by placing two QAUP filters in series so that the top filter collects the particulate matter and adsorbed organics while the second filter collects adsorb organic material, which is used as an OC correction factor (Hering, et al., 1990). Spacers were used in the cassettes to provide even deposition and flow was 19

20 controlled as described below at 4, 8 or 12 L/min. Flows were monitored and recorded using TSI flow meters. The NIOSH Pittsburgh Research Laboratory did all the EC/OC analysis and the TX40 filters were pre- and post-weighed in a temperature and humidity controlled weigh room using a Cahn microbalance. Aethalometer: The Aethalometer provides a real-time readout of the concentration of BC. The Aethalometer uses continuous filtration and optical measurement method to give a continuous readout of real-time data. A wide body of published research shows that the Aethalometer BC measurement is closely proportional to filter-based EC measurements. The Aethalometer performs the optical analysis and data readout on the spot. More information is found elsewhere (Magee Scientific, 2005). The instrument works by measuring the attenuation of a light beam passing through a sample collected on a filter tape. Catalytic Stripper Catalytic Stripper (CS): A rack mounted CS was used to remove sulfur compounds and the soluble organic fraction (SOF) by passing the diluted Diesel exhaust over two heated catalysts. The CS was built at the University (Abdul-Khalek, et al., 1995; Abdul-Khalek, 1996; Stenitzer, 2003), and contained two ceramic substrates provided by Johnson Matthey. One catalyst, referred to as the S-Trap, was designed to remove sulfur compounds by absorption; while the second, called the Oxicat, was designed to remove volatile hydrocarbons. All components in the exhaust gas containing sulfur oxides are caught by the S-Trap and are stored as BaSO 4 in the S-Trap as shown below. H 2 SO 4 H 2 O + SO 2 SO 2 + ½ O 2 SO 3 BaO + SO 3 BaSO 4 When the S-Trap exceeds its capacity for BaSO 4 storage, it is regenerated by heating at 450 to 500 ºC in nitrogen. Our S-trap is very oversized and has been operated for years in the laboratory without regeneration. The Oxicat removes the volatile hydrocarbons in the exhaust gas stream, which consists mainly of unburned fuel and lubrication oil. The Oxicat is a flow through ceramic substrate coated with 75 g/ft 3 of Pt. In normal operation the CS is heated to 300 ºC and the aerosol flow rate through the CS is 10 L/min. The reaction shown below summarizes what occurs. Further details of the CS are found elsewhere (Stenitzer, 2003). SOF + xo 2 yco 2 + zh 2 O Particle losses in the CS would be expected to impact performance. Due to the small size of Diesel aerosol the principal mechanisms by which particles are lost in the CS are thermophoresis and diffusion. Previous work documented particle losses at a variety of operating conditions including the 10 L/min and 300 C conditions used in this study using sodium chloride aerosol and a differential mobility analyzer. Penetration is approximately 70 % for particles above 100 nm where most of the mass is found, but 20

21 decreases to about 25% at 10 nm due to diffusion losses. Further details are available elsewhere (Stenitzer, 2003). Stenitzer (2003) determined the performance of the CS for removing volatile particles. At 300 ºC and 10 L/min, essentially 100% of sulfate and lubricating oil particles in the 15 to 200 nm diameter range are removed. It was also shown that restructuring of the carbonaceous agglomerates does not occur even at very high temperatures (Higgins, et al., 2003). During the project three mini-catalytic strippers were built using catalyzed ceramic or metallic substrates provided by Johnson Matthey or Miratech. Performance of these strippers were compared to the rack mounted CS used in the laboratory. Sampling System Measurements were made using a two stage ejector dilutor system designed and built for the project. This system is similar to those described previously (Abdul-Khalek, et al., 1999), and extensively evaluated during the Coordinating Research Council (CRC) E-43 project (Kittelson, et al., 2002, 2006). Exhaust enters the system through an 8 cm long, cm diameter stainless sampling probe immersed in the exhaust flow, and passes through a short section (25 cm) of stainless steel tubing referred to as the transfer line (TL). To prevent particle losses the TL is insulated. Long, poorly insulated TLs with low flow rates result in a substantial loss of nanoparticle precursors (Wei, et al., 2001a,b). An Air-Vac TD 110-air ejector pump with a critical flow orifice provided the first stage of dilution. The system was designed to give a primary dilution ratio (DR) ranging from 5 to 20:1 with a regulated dilution air flow rate of 100 L/min. It is critical that the orifice not be allowed to plug with soot as this will increase the DR and impact the resulting measurements. Ultra-clean compressed air was provided by a Donaldson pressure swing adsorption system was used to filter, dry and clean the compressed air used for dilution (Swanson, et al., 2007). Following primary dilution the sample passed through a URG cyclone (URG ES) that has a 50% cutpoint of 2.3 µm at a flow rate of 100 L/min and allows 99% of particles < 1.0 µm to enter the sampling system. Diluted aerosol then passed into an insulated residence time chamber shown in Figure 4. The chamber was designed for a residence time of 1 s at 100 L/min. Pressure and temperatures were measured in the chamber as well as in the exhaust and transfer line. Surrounding the residence time chamber were cm (0.25 in outside diameter) copper coils that carried cold water (approximately 1 ºC water) at a flow rate of approximately 10 L/min. Water was recirculated through an ice chest containing water and ice. This system maintained the residence time chamber temperature between 15 and 24 ºC. Temperature in the chamber was measured using a type K thermocouple located in the center of the chamber. After the residence time chamber secondary dilution took place using another TD 110 air ejector. The nozzles at the right end of the chamber shown in Figure 4 were also used to direct sample air into a 4 position, cassette filter holder that was used to collect 37 mm filter samples for gravimetric or EC/OC analysis. Spacers were inserted into the cassettes to ensure even DPM distribution for EC analysis. Flow through the filters was measured and recorded using TSI L/min flow meters as shown in Figure 5. Flow through the filters was controlled by a 12 place, critical orifice manifold shown 21

22 in Figure 6. The flow rates were 4, 8 or 12 L/min, and it was possible to collect four samples at one time at each flow rate. The flow rate used for sampling was determined by the expected filter loading time. Figure 4. Residence time chamber, units are in inches Figure 5. Filter holder and flow meters used to sample DPM after primary dilution The primary DR was determined by measuring the raw exhaust NO concentration and the diluted sample NO concentration using NO gas analyzers and heated sample lines. The analyzers were Rosemount 880A or CAI 600 series analyzers. The secondary dilution ratio was monitored using a separate high sensitivity EcoPhysics NO analyzer. If additional dilution was required for the 3025A CPC then a leaky filter dilutor was used and the concentration of DOS particles with and without the dilutor was used to determine the leaky filter DR. 22

23 Figure 6. Filter critical orifice flow control manifold used to control filter flow rate to 4, 8, or 12 L/min After secondary dilution the sample stream was distributed using a manifold as shown in Figure 7. This was the first of two such manifolds used in the sampling system. The first manifold distributed diluted exhaust to instruments used without the CS while the second manifold distributed diluted exhaust to instruments used with and without the CS. Flow through the CS was continuously recorded and maintained at 10 L/min. A bypass flow passed through the CS when instruments received sample directly from manifold #1 preventing the CS from overheating and to maintain a steady-state temperature. CS temperature was maintained at 315 C + 3 ºC. All tubing going to the instrumentation or filters was cm (0.25 in) or cm (0.375 in) conductive tubing to minimize particle losses. The stripper used for this portion of the work (Stenitzer, 2003) was a modification of Abdul-Khalek s (1995) design and is capable of handling flows up to 10 L/min. Table 1 lists the instrumentation used in the project and the location within the sampling system and Figure 8 shows a schematic view of the entire sampling system. The partial flow dilution system used in this research is far less expensive to build than the full-flow constant volume tunnels used for EPA engine certification that typically cost a million dollars or more. Partial flow systems, such as this, mimic realworld exhaust dilution and allow for the formation of nanoparticles during dilution and cooling (Abdul-Khalek, et al., 1999, Kittelson, et al., 2002, 2006). Performance of this type of dilution system was characterized at Cummins and Caterpillar laboratories during the E-43 project and detailed procedural recommendations for sampling are found in the E-43 report (Kittelson, et al., 2002). 23

24 Table 1. Instrument and position in the sampling system Instrument Measures Manifold # 1 Manifold # 2 w/wo CS SMPS Size distribution Y EEPS Size distribution Y NSAM Surface area Y EAD Particle length Y Aethalometer Black carbon Y 3775 CPC Particle number Y PAS-rack Photoemission Y DC-rack Surface area Y 3007 CPC Particle number Y DustTrak Particle mass Y AM 510 Particle mass Y PAS-portable Photoemission Y DC-portable Surface area Y SMPS Scanning mobility particle sizer, EEPS Engine exhaust particle sizer, NSAM nanoparticle surface area monitor, EAD, Electrical aerosol detector, CPC condenstation particle counter, PAS Photoemission aerosol sensor, DC Diffusion charger, AM 510 photometer, DustTrak photometer Figure 7. Sample distribution manifold, two of these were used 24

25 Figure 8. Schematic of the sampling system Quality Assurance Our quality assurance procedures were developed during the Coordinating Research Council E-43 Diesel project, and during the DOE funded spark ignition project (Kittelson, et al., 2002, 2003). Details of the procedures developed during the E-43 project are available elsewhere (Ayala, et al., 2003). A brief description is provided below. A leak test was performed each day using an absolute filter connected to the inlet of manifold #1. This was followed by evaluating the response of the aerosol instruments to a polydisperse, 100 ppm dioctyl sebacate (DOS) aerosol that produced a mode at about 50 nm. DOS was diluted in isopropanol. The DOS aerosol was generated by a Collison atomizer and was dried by two charcoal-filled diffusion driers and diluted with dried, filtered compressed air using ejector pump dilutors. The leaky filter dilution ratio was estimated using this aerosol. The gas analyzers were calibrated every six months using a five-point calibration procedure that follows the manufacturers guidelines. The analyzers were zeroed and spanned daily using National Institute of Standards traceable gases. An important issue, which has been largely ignored, is the quality of the compressed air used for dilution and cooling of the exhaust samples prior to making aerosol measurements. This is especially important when measuring low aerosol number and mass concentrations. We used a Donaldson high purity filtration system (ALD ) designed to provide high purity compressed air. This system dries the air stream, 25

26 removes particles and organic material by sequential passage of the dilution air stream through driers, filters and scrubbers. We believe that failure to use such a system can lead to the misinterpretation of results due to artifact formation, especially when taking exhaust filter samples from engines meeting EPA 2007 and later standards. E. Results and Discussion Initial Engine Tests While mapping the performance of the Deere engine we noticed an unexpected, regular (approximately 5 min interval), repeatable pulse of nanoparticles at a 1400 rpm, 100 N-m (a light load condition) as illustrated by the condensation particle counter (CPC) data in Figure 9. This pulse had not been previously observed with an older version of a similar Deere engine. As a consequence a great deal of effort was expended to determine if the pulse was an artifact of sampling or due to the engine. 3.0E E+03 CPC PAS A CPC, part/cm 3 2.0E E E PAS, fa 5.0E Deere engine, 1400 rpm, 100 Nm; not corrected for dilution ratio, 8/1/ E+00 22:00 22:07 22:14 22:21 22:28 Time 0 Figure 9. Unusual regular spike of particles from the Deere engine While mapping the engine we were also setting up and evaluating our dilution system and instrumentation, and this instrumentation (described in more detail later) assisted in diagnosing the problem. As shown in Figure 9, the PAS response decreased when the number of particles measured by the CPC increased suggesting that the particle pulse was composed of volatile nanoparticles probably from lubricating oil. There are a number of locations within the engine that can contribute to excess oil consumption. At Deere s suggestion we measured the flow of crankcase fumes and found the flow varied inversely with CPC count. This suggested a variation in the leakage rate past the piston 26

27 rings. Next, we used the Deere engine software package to access the ECU to turn off injection to individual cylinders. It was found that the problem continued with cylinders 1, 3 and 4 turned off individually, but went away when cylinder 2 was turned off. The engine was disassembled, and the problem was traced to a broken piston ring in cylinder 2 shown in Figure 10. Oil was collecting and was periodically released into the exhaust forming high concentrations of nanoparticles during exhaust cooling and dilution. This fault would not have been detected by filter sampling, opacity measurements, or other routine diagnostic procedure, because the amount of mass emitted would have been below the level of detection. Further, engine performance was not substantially degraded. The ring was replaced and the problem disappeared. Figure 10. Broken cylinder ring (photo provided by Dana Corporation, Perfect Circle Division, manufacturer of the piston ring) After the engine was repaired we repeated the test to determine the impact of the repair. During the test we observed the response of three portable instruments (3007 CPC, PAS, and AM510 photometer) with and without the CS. In Figure 11, the regular pulse of particles shown in Figure 9 has disappeared, and the CPC and AM 510 responses decrease after the CS while the PAS response increases. A 30% decrease in concentration is expected due to solid particulate matter losses within the CS, and additional reductions in concentrations are due to the removal of adsorbed volatile material, and volatile nanoparticles as shown by the CPC. Evidence that the increase in PAS response is due to the removal of volatile nanoparticles is given by the SMPS size distributions in Figure 12 measured over the same time period. The SMPS is not a near real-time instrument, taking about 90 s to obtain one size distribution, thus each period in Figure 11 is about 6 min long, time enough for 3 SMPS scans to be collected or a total of 24 scans, 12 with and 12 without the CS. In Figure 12 three number size distributions are shown; no CS, with CS and with CS corrected for particle losses. The decrease in the nanoparticle tail (left edge) of the SMPS size distributions shrinks by more than an order of magnitude even after the correction for particle losses. This is due mainly to the removal of volatile nanoparticles. The accumulation mode (centered between nm) shrinks about 40%, but after correction for particle losses this reduction is about 10%. The sharp drop-off on the nanoparticle tail at the lower limit of SMPS detection, about 8 nm was due to instrument behavior (that was later corrected) while the increasing trend before the drop off suggests that there is a nucleation mode below 8 nm. The figure shows DR corrected data, which in this case was 246 to 1. DR impacts the magnitude of emissions not the shape of the size distribution. 27

28 6.0E CPC AM 510 PAS E CPC, part/cm 3 4.0E E E+03 No CS CS No CS CS No CS CS No CS CS PAS, fa; AM510, µg/m E+03 Deere engine 1400 rpm, 100 N-m; catalytic stripper at 300 ºC Not corrected for dilution ratio, 1/11/ E+00 14:30 14:37 14:44 14:51 14:58 15:06 15:13 Time Figure 11. Response with and without CS, dilution ratio approximately 200: E E+07 Deere engine 1400 rpm, 100 N-m; Samples collected with and without the CS at 300 ºC Corrected for dilution ratio, 1/11/2007 No catalytic stripper Catalytic stripper corrected for particle losses d /dlogdp, part/cm 3 8.0E E E E+06 Catalytic stripper 0.0E Dp, nm Figure 12. SMPS size distribution with and without the CS 28

29 Evaluation of Emission Control Devices Catalyzed Diesel Particulate Filter (CDPF) A Catalyzed Diesel Particulate Filter (CDPF) consisting of a catalyzed metallic prefilter and a catalyzed filter was provided by Donaldson for evaluation. Limited tests were conducted because of the very high filtration efficiency shown in Table 2 and illustrated in Figure 13 where the SMPS size distributions are shown with and without the CDPF and with and without the CS for the 1400 rpm, 250 N-m condition and ECD-1 fuel corrected for dilution ratio, and CS losses (Stenitzer, 2003). On average the primary dilution ratio applied to the filter samples was 14:1 and the total dilution ratio was 256:1. Table 3 provides the measured concentration and standard deviation for the gravimetric, EC/OC, SMPS, AM510, and DustTrak data. The CDPF SMPS particulate matter removal efficiency as measured by number, surface area or volume was > 99.9%, and as a result after one hour of sampling the filters used to collect mass and EC/OC had no visible deposits. At 1400 rpm and 250 N-m the exhaust aerosol is composed primarily of carbonaceous material as illustrated by the lack of effect on the number size distribution when the sample is passed through the CS. The impact of filter artifacts is also illustrated by data in Table 3. To obtain an estimate for the artifact, the SMPS volume is multiplied by 1 g/cm 3 and then the ratios of filter and SMPS mass with and without the CDPF are compared. The ratio without the CDPF is 1.53 and with the CDPF it is For a given sampling configuration and test conditions the ratio would not be expected to be very different upstream and downstream of the CDPF. This suggests that the filter artifact is causing the mass downstream of the CDPF to be grossly over estimated, in this case by a factor of roughly 45. This estimate is in general agreement with other estimates found in the literature (Chase, et al., 2004; Khalek, I. A., 2007; Noll, et al., 2008; Swanson, et al., 2009). The photometer data shown in Table 2 also show lower filtration efficiencies. It should be remembered that these photometers were calibrated at the factory using Arizona road dust. Recalibration with DPM would be required for a more accurate assessment. Table 2. CDPF filtration efficiencies determined by the different measures for the 1400 RPM 250 N-m condition with ECD-1 fuel Metric CDPF efficiency, pct SD, pct SMPS number, part/cm SMPS surface area, µm 2 /cm SMPS volume, µm 3 /cm Mass, mg/m AM510, mg/m DustTrak, mg/m

30 1.0E E+07 ECD-1 fuel, with and without CDPF and CS, weighted average SMPS, STDEV Corrected for DR and CS losses d /dlogdp, part/cm 3 1.0E E E E RPM 250 Nm 1400 RPM 250 Nm CS 1400 RPM 250 Nm CDPF 1400 RPM 250 Nm CDPF CS 1.0E E Dp, nm Figure 13. SMPS number size distributions for the 1400 RPM 250 N-m condition with ECD-1 fuel with and without the CDPF and CS. Weighted average corrected for DR and CS losses with standard deviations Table 3. CDPF SMPS, photometer and filter measurements for the 1400 RPM 250 N-m condition. Weighted averages and standard deviations with ECD-1 fuel Condition Mass EC OC TC Avg, mg/m 3 SD, mg/m 3 Avg, mg/m 3 SD, mg/m 3 Avg, mg/m 3 SD, mg/m 3 Avg, mg/m 3 SD, mg/m 3 ECD ECD-1 + CDPF N/A N/A N/A Condition SMPS number SMPS surface area SMPS volume Avg, part/cm 3 SD, part/cm 3 Avg, µm 2 /cm 3 SD, µm 2 /cm 3 Avg, µm 3 /cm 3 SD, µm 3 /cm 3 ECD E E E E E E+02 ECD-1 + CDPF 6.80E E E E E E+00 Condition AM 510 DustTrak Avg, mg/m 3 SD, mg/m 3 Avg, mg/m 3 SD, mg/m 3 ECD ECD-1 + CDPF Diesel Oxidation Catalyst (DOC) The DOC was also provided by Donaldson. It was evaluated at four engine conditions with two fuels; ECD-1 and B50. Thus, there are two comparisons of interest ECD-1 versus B50 with and without the DOC installed in the exhaust. The ECD-1 fuel was evaluated in 2007 and 2008 and the B50 fuel was evaluated only in Data presented are a weighted average of all samples. 30

31 Figures show SMPS number size distributions for the two fuels with and without the DOC and or the CS. The left hand chart shows results for the ECD-1 fuel and the right hand chart shows the B50 fuel results. The scales are identical to allow easier comparison, and standard deviations are not included to allow easier viewing of the distributions. Figure 14 shows results for the 1400 rpm 50 N-m condition. The average exhaust temperature entering the DOC at this condition was ºC for ECD-1 and ºC for B50. At this temperature the DOC is not fully active as indicated by the larger reductions in the nucleation mode achieved when the sample passes through the CS. The B50 fuel produced a larger nucleation mode, which is consistent with the fact that the B50 fuel produces less carbonaceous material and more volatile organics than Diesel fuel. This point is illustrated further with filter data presented later. 8.0E+07 ECD-1 fuel, with and without DOC or CS, weighted average SMPS, CS corrected for losses 7.0E+07 d /dlogd p, particles/cm 3 6.0E E E E RPM 50 Nm 1400 RPM 50 Nm DOC 1400 RPM 50 Nm CS 1400 RPM 50 Nm DOC CS 2.0E E E Dp, nm 8.0E+07 Figure 14. DOC evaluation at 1400 rpm 50 N-m for ECD-1 (left) and B50 (right) fuels Figure 15 shows results for the 1400 rpm 100 N-m condition. The average 6.0E+07 exhaust temperature entering the DOC at this condition was ºC for ECD-1 and ºC for B50. At this temperature the DOC 5.0E+07 is fully active as indicated by the large reductions across the size range when the sample passes through DOC, and no 4.0E+07 additional reduction when the sample passes through the CS. The B50 fuel again produced a larger nucleation mode than the ECD-1 3.0E+07 fuel. d /dlogdp, particles/cm 3 7.0E E+07 B50 fuel, with and without DOC or CS, SMPS weighted averages, CS corrected for losses 1400 RPM 50 Nm 1400 RPM 50 Nm DOC 1400 RPM 50 Nm CS 1400 RPM 50 Nm DOC CS 1.0E E Dp, nm 31

32 8.0E+07 ECD-1 fuel, with and without DOC or CS, weighted average SMPS, CS corrected for losses 8.0E+07 B50 fuel, with and without DOC or CS, SMPS weighted averages, CS corrected for losses d /dlogdp, particles/cm 3 7.0E E E E E RPM 100 Nm 1400 RPM 100 Nm DOC 1400 RPM 100 Nm CS 1400 RPM 100 Nm DOC CS d /dlogdp, particles/cm 3 7.0E E E E E RPM 100 Nm 1400 RPM 100 Nm DOC 1400 RPM 100 Nm CS 1400 RPM 100 Nm DOC CS 2.0E E E E E d /dlogd p, particles/cm 3 Dp, nm 0.0E Figure 15. DOC evaluation at 1400 rpm 100 N-m for ECD-1 (left) and B50 (right) fuels Figure 16 shows results for the 1400 rpm 250 N-m condition. The average exhaust temperature entering the DOC at this condition was ºC for ECD-1 and ºC for B50. At this temperature the DOC is fully active as indicated by the reductions across the size range achieved when the sample passes through DOC, and no additional reduction when the sample passes through the CS. At this condition the B50 fuel produced volatile particles across the entire size range. The volatile particles were removed by the DOC. 8.0E E E E E E E+07 ECD-1 fuel, with and without DOC or CS, weighted average SMPS, CS corrected for losses 1400 RPM 250 Nm 1400 RPM 250 Nm DOC 1400 RPM 250 Nm CS 1400 RPM 250 Nm DOC CS d /dlogdp, particles/cm 3 8.0E E E E E E E+07 Dp, nm B50 fuel, with and without DOC or CS, SMPS weighted averages, CS corrected for losses 1400 RPM 250 Nm 1400 RPM 250 Nm DOC 1400 RPM 250 Nm CS 1400 RPM 250 Nm DOC CS 1.0E E E Dp, nm 0.0E Dp, nm Figure 16. DOC evaluation at 1400 rpm 250 N-m for ECD-1 (left) and B50 (right) fuels Figure 17 shows results for the 1400 rpm 450 N-m condition. The average exhaust temperature entering the DOC at this condition was ºC for ECD-1 and ºC for B50. There is essentially no difference between the two fuels or between the DOC or CS conditions suggesting that nearly all of the emitted particulate matter is carbonaceous with little volatile material. The standard deviation for the DOC temperature is greater for the ECD-1 fuel tests than the B50 tests. The ECD-1 fuel was 32

33 tested on multiple days/months in both 2007 and 2008 and the temperature variation reflects changes in laboratory/ambient temperatures. 8.0E+07 ECD-1 fuel, with and without DOC or CS, weighted average SMPS, CS corrected for losses 8.0E+07 B50 fuel, with and without DOC or CS, SMPS weighted averages, CS corrected for losses d /dlogdp, particles/cm 3 7.0E E E E E RPM 450 Nm 1400 RPM 450 Nm DOC 1400 RPM 450 Nm CS 1400 RPM 450 Nm DOC CS d /dlogdp, particles/cm 3 7.0E E E E E RPM 450 Nm 1400 RPM 450 Nm DOC 1400 RPM 450 Nm CS 1400 RPM 450 Nm DOC CS 2.0E E E E E Dp, nm 0.0E Figure 17. DOC evaluation at 1400 rpm 450 N-m for ECD-1 (left) and B50 (right) fuels Tables 4-9 summarize data for the DOC evaluation. Table 4 provides direct averages and standard deviations for the filter data and Tables 5-6 provide weighted averages and standard deviations for SMPS, photometer and aethalometer measurements and Tables 7-9 provided the corresponding percent reduction/increases measured by these instruments. A negative percentage in Tables 7-9 indicates an increase after the DOC. Table 4. Filter measurements for ECD-1 and B50 fuel with and without the DOC. Direct averages and standard deviations DOC Condition Mass EC OC TC Avg, mg/m 3 STDEV, mg/m 3 Avg, mg/m 3 STDEV, mg/m 3 Avg, mg/m 3 STDEV, mg/m 3 Avg, mg/m 3 STDEV, mg/m 3 ECD-1 None 1400 RPM 50 Nm None 1400 RPM 100 Nm None 1400 RPM 250 Nm None 1400 RPM 450 Nm DOC 1400 RPM 50 Nm DOC 1400 RPM 100 Nm DOC 1400 RPM 250 Nm DOC 1400 RPM 450 Nm B50 None 1400 RPM 50 Nm None 1400 RPM 100 Nm None 1400 RPM 250 Nm None 1400 RPM 450 Nm DOC 1400 RPM 50 Nm DOC 1400 RPM 100 Nm DOC 1400 RPM 250 Nm DOC 1400 RPM 450 Nm Dp, nm 33

34 Table 5. SMPS measurements for ECD-1 and B50 fuel with and without the DOC. Weighted averages and standard deviations DOC Condition SMPS number SMPS surface area SMPS volume Avg, part/cm 3 STDEV, part/cm 3 Avg, µm 2 /cm 3 STDEV, µm 2 /cm 3 Avg, µm 3 /cm 3 STDEV, µm 3 /cm 3 None 1400 RPM 50 Nm 1.68E E+06 ECD E E E E+01 None 1400 RPM 100 Nm 1.31E E E E E E+01 None 1400 RPM 250 Nm 1.72E E E E E E+02 None 1400 RPM 450 Nm 1.76E E E E E E+02 DOC 1400 RPM 50 Nm 6.40E E E E E E+01 DOC 1400 RPM 100 Nm 9.83E E E E E E+01 DOC 1400 RPM 250 Nm 1.49E E E E E E+02 DOC 1400 RPM 450 Nm 1.95E E E E E E+02 B50 None 1400 RPM 50 Nm 3.63E E E E E E+01 None 1400 RPM 100 Nm 3.21E E E E E E+02 None 1400 RPM 250 Nm 2.03E E E E E E+02 None 1400 RPM 450 Nm 1.45E E E E E E+02 DOC 1400 RPM 50 Nm 6.19E E E E E E+01 DOC 1400 RPM 100 Nm 1.41E E E E E E+01 DOC 1400 RPM 250 Nm 1.43E E E E E E+02 DOC 1400 RPM 450 Nm 1.48E E E E E E+02 Table 6. AM510, DustTrak and aethalometer measurements for ECD-1 and B50 fuel with and without the DOC. Weighted averages and standard deviations DOC Aftertreatment AM 510 DustTrak Aethalometer Avg, mg/m 3 STDEV, mg/m 3 Avg, mg/m 3 STDEV, mg/m 3 Avg, mg/m 3 STDEV, mg/m 3 None 1400 RPM 50 Nm ECD None 1400 RPM 100 Nm None 1400 RPM 250 Nm None 1400 RPM 450 Nm DOC 1400 RPM 50 Nm DOC 1400 RPM 100 Nm DOC 1400 RPM 250 Nm DOC 1400 RPM 450 Nm B50 None 1400 RPM 50 Nm None 1400 RPM 100 Nm None 1400 RPM 250 Nm None 1400 RPM 450 Nm DOC 1400 RPM 50 Nm DOC 1400 RPM 100 Nm DOC 1400 RPM 250 Nm DOC 1400 RPM 450 Nm Table 7. Percent reduction or increase calculated for the DOC with filter data Cond ition Percent reduction or in crease Mass EC OC TC ECD RPM 50 Nm RPM 100 Nm RPM 250 Nm RPM 450 Nm B P RPM 50 Nm RPM 100 Nm RPM 250 Nm RPM 450 Nm Negative m ean s p ercent increase 34

35 Table 8. Percent reduction or increase calculated for the DOC with SMPS data Condition Percent reduction or increase Number Surface Volume ECD RPM 50 Nm RPM 100 Nm RPM 250 Nm RPM 450 Nm B RPM 50 Nm RPM 100 Nm RPM 250 Nm RPM 450 Nm Negative means percent increase Table 9. Percent reduction or increase calculated for the DOC with AM510, DustTrak and aethalometer data Condition Percent reduction or increase AM 510 DustTrak Aethalometer 1400 RPM 50 Nm ECD RPM 100 Nm RPM 250 Nm RPM 450 Nm B RPM 50 Nm RPM 100 Nm RPM 250 Nm RPM 450 Nm Negative means percent increase The evaluation of the DOC shows that for specific test conditions the mass, and in some cases the EC concentration increases. Such is the case with the ECD-1 fuel at the 250 and 450 N-m test conditions as shown in Figure 18 where charts are shown for the EC and OC data from Table 4 for the two fuels with and without the DOC. The standard deviations shown in the chart are for TC. It is known that sulfates are formed as gaseous SO 2 is oxidized to SO 3 leading to the formation of H 2 SO 4 and other sulfate containing compounds. However, ECD-1 fuel contains 6 ppm S and it would not be expected that sulfate formation would account for the entire observed increase. Increases are shown for both mass (8.7%) and EC (27.5%) at the 1400 RPM 250 N-m condition and for EC (23.5%) at the 1400 RPM 450 N-m test condition. Decreases in mass are shown for the 1400 RPM 50 and 100 N-m test conditions. This is expected because the DOC reduces the volatile material for the two conditions where a large proportion of the mass is in the nucleation mode. OC is also reduced (28.7%) at the 250 N-m condition but volatile material composes a much smaller fraction of the mass at this condition. The increase in EC concentration at the 250 and 450 N-m test conditions are unexpected results, and might be due to an increase in backpressure caused by the DOC. 35

36 There is another possible explanation for the unexpected increase in EC after the DOC. The dilution system relies on Air-Vac TD 110-air ejector pumps with critical flow orifices for primary and secondary dilution. At high load conditions the engine emits more DPM, and a fraction of the DPM collects on the critical flow orifice during primary dilution. If the rate of collection is greater for exhaust not passing through the DOC than it is for exhaust that passes through the DOC than a sampling error is introduced into the mass and EC/OC measurements. This is likely the case as exhaust passing through the DOC will have less organic material and will be less sticky, and will therefore be less likely to collect on the orifice. The end result is the introduction of a bias in the mass and EC/OC estimates. The solution to the problem is to design a simple nozzle instead of a sharp edge orifice that is less prone to collection of large DPM. Evidence supporting this explanation is shown in Figure 19 where the ECD-1 number size distribution with and with out the DOC previously shown in Figure 17 are shown again along with the volume size distributions. Note that for both particle number and volume larger particles are passing through the DOC in the 100+ nm range. This is consistent with the idea that drier DPM particles passed through the sharp edged orifice EC OC ECD-1 vs. B50 o DOC EC OC ECD-1 vs. B50 With DOC C oncentration, m g/m Standard deviation for TC is shown C oncentration, m g/m Standard deviation for TC is shown 4 ECD-1 B50 ECD-1 B ECD-1 B50 ECD-1 B RPM 50 N-m 1400 RPM 100 N-m 1400 RPM 250 N-m 1400 RPM 450 N-m 2 0 ECD-1 B50 ECD-1 B50 ECD-1 B50 ECD-1 B RPM 50 N-m 1400 RPM 100 N-m 1400 RPM 250 N-m 1400 RPM 450 N-m Figure 18. EC and OC averages with and without the DOC for the two fuels at the four test conditions. Standard deviations are for the TC average concentrations 36

37 2.5E E+07 ECD-1 fuel, with and without DOC weighted average SMPS 1400 RPM 450 Nm 1400 RPM 450 Nm DOC 1400 RPM 450 Nm 1400 RPM 450 Nm DOC 5.0E E+04 d /dlogdp, particles/cm 3 1.5E E E E+04 dv/dlogdp, µm 3 /cm 3 5.0E E E Dp, nm 0.0E+00 Figure 19. SMPS number and volume size distributions for ECD-1 fuel with and without the DOC Instrument Comparison The engine test conditions were selected to provide Diesel aerosol with and without a nucleation mode. The presence of volatile nanoparticles in the nucleation mode impacts instrument performance. The addition of a DOC and B50 fuel to the test matrix resulted in further variation in the test aerosol and increased the size of the correlation matrix. A correlation matrix was used to determine the correlation between the real-time and filter-based measurements. Tables 10 through 13 summarize the correlation coefficients (R 2 ) for the cases when R 2 for EC was greater than The order is based upon the EC R 2, from highest to lowest. Correlations are shown for rackmounted/portable instruments broken down by fuel either with or without the DOC. Instruments used with the CS are highlighted in gray. Only correlations between the instruments and mass, EC, OC and TC are shown since filter measurements are used for determining compliance with occupational health standards. Correlations for the EEPS and SMPS are not shown because these instruments are laboratory instruments that are not normally used in underground mines. Units shown in these tables are for the instruments and filter measurements not for the R 2, which are dimension less values. Tables 10 and 11 summarize the R 2 for the ECD-1 fuel with and without the DOC. OC has the lowest overall R 2 values for the ECD-1 fuel. 37

38 Table 10. Correlation coefficients for the ECD-1 fuel without the DOC Instrument Mass mg/m 3 EC mg/m 3 OC mg/m 3 TC mg/m 3 DustTrak mg/m PAS fa CS DustTrak mg/m AM510 mg/m CS PAS fa CS AM510 mg/m Aethalometer µg/m Gray denotes instrument was used with the catalytic stripper in line. Table 11. Correlation coefficients for the ECD-1 fuel with the DOC Instrument Mass mg/m 3 EC mg/m 3 OC mg/m 3 TC mg/m 3 DustTrak mg/m CS DustTrak mg/m Portable PAS fa AM510 mg/m CS AM510 mg/m PAS fa CS Portable PAS fa CS PAS fa NSAM Alveolar µm 2 /cm Gray denotes instrument was used with the catalytic stripper in line. Table 12 and 13 summarize the R 2 for the B50 fuel with and without the DOC. The B50 correlations for OC without the DOC are poor, but improve when the DOC is installed, and are about the same as ECD-1 fuel R 2. The elimination of OC produced by B50 fuel is a function of catalyst, operating conditions including catalyst temperature, OC chemistry and flow rates. The CS and the DOC are similar but not identical. The CS operates at a constant temperature of about 316 ºC while the DOC operating temperature depends upon engine condition ranging from 160 to 500 ºC. Further compounding the problem is the instrument response to exhaust aerosol containing particles coated with an OC layer. ECD-1 fuel in general produces less OC and more EC especially at the highest load conditions (Table 4) than B50. There is more carbonaceous surface area for the adsorption of volatile precursor material that can form nanoparticles, thus the ECD-1 nucleation mode fuel is smaller than the B50 nucleation mode (Figures 14-16). 38

39 Table 12. Correlation coefficients for the B50 fuel without the DOC Instrument Mass mg/m 3 EC mg/m 3 OC mg/m 3 TC mg/m 3 DustTrak mg/m CS PAS fa CS Portable PAS fa AM510 mg/m CS DustTrak mg/m Aethalometer µg/m CS AM510 mg/m Portable DC µm 2 /cm CS Portable DC µm 2 /cm Gray denotes instrument was used with the catalytic stripper in line. Table 13. Correlation coefficients for the B50 fuel with the DOC Instrument Mass mg/m 3 EC mg/m 3 OC mg/m 3 TC mg/m 3 AM510 mg/m DustTrak mg/m CS AM510 mg/m CS DustTrak mg/m PAS fa Portable PAS fa CS PAS fa CS DC µm 2 /cm NSAM Alveolar µm 2 /cm CS Portable DC µm 2 /cm EAD mm/cm Gray denotes instrument was used with the catalytic stripper in line. Table 14 summarizes the R 2 obtained for all data regardless of test condition, and for all data collected for each fuel regardless of whether a DOC was present. The PAS, DustTrak and AM510 photometers are the only instruments that have an R 2 for EC that is greater than 0.90 for both fuels with and without the DOC and no CS. 39

40 Table 14. Summary correlation coefficients Both fuels with and without DOC Instrument Mass mg/m 3 EC mg/m 3 OC mg/m 3 TC mg/m 3 CS PAS fa PAS fa DustTrak mg/m AM510 mg/m All ECD-1 with and without DOC Instrument Mass mg/m 3 EC mg/m 3 OC mg/m 3 TC mg/m 3 DustTrak mg/m CS DustTrak mg/m AM510 mg/m PAS fa CS AM510 mg/m CS PAS fa Portable PAS fa All B50 with and without DOC Instrument Mass mg/m 3 EC mg/m 3 OC mg/m 3 TC mg/m 3 AM510 mg/m DustTrak mg/m CS AM510 mg/m CS PAS fa CS Portable PAS fa CS DustTrak mg/m Underground Mine Study Vulcan Materials hosted the underground mine study at the Central Quarry from July 22, to August 2, The Central mine is a three level, room-and-pillar underground limestone mine located in Lexington, Kentucky. The mine operates two 10- hour shifts per day (production and maintenance), 5 days a week, and a single 8-hr shift on Saturday. The underground mine is accessed by decline roadways. Limestone is drilled and blasted and the shot rock is loaded onto haul trucks and transported to the underground primary and secondary crushing plants where it is crushed and sized. The material is then conveyed to the surface crushing and screening plants for further processing and stockpiling. Caterpillar Diesel front-end loaders and haul trucks load and transport the limestone. Diesel roof bolters, scalers, oilers, water trucks, pickup trucks and other Diesel equipment are used throughout the mine. During the period of the study B50 fuel was used, and fuel analysis by Southwest Research showed that the fuel contained 3 ppm S. The objective of the study was to evaluate the ability of the TSI SIDEPAK AM510 Personal Aerosol Monitor to estimate DPM concentrations. The AM510 was selected because of the overall high R 2 measured in the laboratory (Tables 10-14), the low cost of the instrument compared to the other instruments, and because we were able to obtain 10 AM510 samplers for use in the study. The AM510 photometers were equipped with Dorr-Oliver cyclones and 1.0 µm Bureau of Mines impactors. The 40

41 combination of the time-weighted average mass concentrations obtained from the AM510 and the impactor substrate provided an estimate of the time-weighted average respirable dust concentration. Results were compared to gravimetric filter samples and EC-OC filter samples. The gravimetric samples were also collected using Dorr-Oliver cyclones and Bureau of Mines impactors. The EC-OC samples were collected using SKC cassettes with Dorr-Oliver cyclones and 1.0 µm SKC impactors. Three EC-OC samples were collected for every gravimetric sample. Sample locations were selected after visiting the mine and talking with mine personnel. The sampling locations provided a range of concentrations, a mixture of Diesel aerosol and limestone dust, and were considered safe sampling locations. The locations included the crusher operator s air conditioned control room on level three, the maintenance bay on level two, the main ramp from level 2 to level 3, and a haul truck operating on level 3. The baskets on the haul truck were located outside the operator s cabin near the entry ladder. The crusher operator s control room had the lowest concentrations and the haul truck had the overall highest concentrations because of the high concentrations of dust to be expected during loading. The suite of samplers used at each location is shown in Figure 20. Figure 20. Suite of instruments setup in the crusher operator s control room. From left to right: Gravimetric sampler, EC-OC sampler, AM510, EC-OC sampler, 2 AM510s and an EC-OC sampler. The first EC-OC sampler with blue tape was the sample from which the backup filter was analyzed as a control. The wooden sticks shown in each basket align the Dorr-Oliver inlets in the same direction. At each of the sampling locations two baskets of instruments were used. The baskets contained three photometers as described previously, one gravimetric sampler and three EC-OC samplers. The tenth photometer was placed in a basket either at the ramp or the maintenance location. The gravimetric sample was collected using the sizeselective method developed by the U. S. Bureau of Mines (McCartney et al., 1992). In that method a Dorr-Oliver cyclone, Bureau of Mines impactor and an MSA filter cassette (model ) are held in a Mine Safety Appliance (MSA) model personal respirable dust sample holder. An MSA ELF pump calibrated for 1.7 L/min provided 41

42 sample flow. The MSA filters were pre- and post-weighed using a Cahn microbalance located in a temperature and humidity controlled room after a 4-hr equilibration period. Both laboratory and field blank filters were included as gravimetric controls. The EC-OC samples were collected using Dorr-Oliver cyclones and SKC cassettes containing the 1.0 µm built in impactor (catalog No ). Each SKC cassette contains a backup high purity quartz filter and one of every three backup filters was analyzed to determine an EC-OC artifact correction. EC-OC analysis was conducted by the NIOSH Pittsburgh Research Laboratory using NIOSH method An MSA ELF pump calibrated for 1.7 L/min provided sample flow. Figure 4 shows the sampling array. During the first week the sampling locations included crusher operator control room, main ramp and haul truck. During the second week the maintenance bay sampling site was used rather than the haul truck because of the high dust concentrations encountered on the haul truck and the desire not to overload the impactors with limestone dust. The main ramp samples were collected at a height of 1.5 m beside the ramp. The maintenance bay samplers were located on a parts cabinet located near a rib at a height of about 1.5 m. Only the main ramp and crusher operator room locations were sampled on Saturday. Sampling lasted about 6 hr each weekday and 3 hrs on Saturday. The duration of sampling was established by estimating the amount of collected mass, and working with the miners schedule to ensure work completion by the end of the work shift. Instrument and MSA ELF pump flow calibrations were done underground either at the beginning or at the end of each day using a bubble meter (Sensidyne Gilibrator) that was factory calibrated prior to the beginning of the study. All instruments were calibrated to 1.7 L/min with the cyclone and impactor in-line and contained in a calibration jar. TSI modified the AM510 to allow the pumps to operate at a flow rate of 1.7 L/min, normally the AM510 operates at 1.0 L/min. The AM510 was also zeroed each day. Table 15 is a summary of the samples that were collected at each location with a description of data resulting from each sample. The size selective sampler with MSA filter cassette is identical to the Bureau of Mines sampler, and provides estimates of < 1.0 µm, > 1.0 µm, and respirable dust mass. By pre- and post-weighing the greased impactor substrate used in the 1.0 µm impactor an estimate of the > 1.0 µm, respirable mass is obtained. The SKC DPM sampler provides < 1.0 µm estimates of EC, OC and TC. The AM510 photometers provide an estimate of the < 1.0 µm mass concentration using light scattering. These data are averaged and compared to the time-weighted average MSA filter data. All data were recorded over the same sampling periods. Table 15. Summary of samples collected during the mine study Sampler Respirable Dust, mg/m 3 EC, mg/m 3 OC, mg/m 3 TC, mg/m 3 < 1.0 µm > 1.0µm Total < 1.0 µm < 1.0 µm < 1.0 µm Size selective MSA Yes a Yes b Yes SKC DPM Yes d Yes d Yes AM 510 Yes c Yes b Yes a Determined by gravimetric analysis of MSA filter b Determined by gravimetric analysis of impactor substrate c Determined by photometer d Determined by NIOSH method 5040 for elemental carbon 42

43 Tables summarize the results for the gravimetric, EC-OC-TC, and AM510 samples collected at the mine. In the tables the number of samples refers to the number of time-weighted averaged samples used to calculate the mean and standard deviation. In this case each AM510 photometer provided one time-weighted averaged estimate each day even though the instruments were providing data at a rate of once per second. Table 19 shows average ratios obtained from the AM510, MSA and EC-OC samples. These values could be used to establish alternative calibration factors for the AM510 under the assumption that the gravimetric and EC-OC samples act as reference samples. The AM510 is factory calibrated to the respirable fraction of standard ISO , A1 Test Dust (Arizona test dust 1-10 µm in size). The calibration factor established at the factory is 1.0. A new calibration is established by dividing the average obtained from the reference sampler by the AM510 average. To illustrate the calculation TC is used as the reference sample since the current MSHA standard is based on TC. The ratio is obtained by dividing mg/m 3 (crusher TC) from Table 18 by mg/m 3 (crusher AM510 < 1.0 µm) from table 17. The ratios for the crusher, ramp, and haul truck locations are relatively close together ranging from to while the maintenance area location is lower at This suggests that a different aerosol mixture is present in the maintenance bay, which is consistent with the welding activity that takes place in that area. It is important to recognize that presence of welding fumes, drill oil mist or other aerosol in specific areas of the mine can impact instrument performance and calibration. At this mine the main sources of aerosol were mining activity generating limestone dust and diesel vehicles generating DPM. In areas where drill oil mist or welding fumes are present separate calibration factors should be established. Table 16. MSA gravimetric samples collected at each mine sampling location. Averages and standard deviations Location N samples < 1.0 µm, mg/m 3 > 1.0 µm, mg/m 3 Resp Conc, mg/m 3 Avg Std Avg Std Avg Std Crusher Vehicle Ramp Maintenance Table 17. AM510 photometer samples collected at each mine sampling location. Averages and standard deviations Location Number of samples < 1.0 µm, mg/m 3 > 1.0 µm, mg/m 3 Resp Conc, mg/m 3 Avg Std Avg Std Avg Std Crusher Vehicle Ramp Maintenance

44 Table 18. EC-OC < 1.0 µm filter samples collected at each mine sampling location. Averages and standard deviations Location Number of samples EC OC TC % µg/m 3 µg/m 3 µg/m 3 TC a EC Avg Std Avg Std Avg Std % > TLV Crusher Vehicle Ramp Maintenance Table 19. Ratios of reference samples to AM510 photometer samples Location Respirable dust ratios < 1.0 µm < 1.0 µm MSA/AM510 RD > 1.0 µm impactor TC/AM510 EC/AM510 MSA/AM510 Crusher Vehicle Ramp Maintenance Portable Catalytic Stripper A portable or mini-cs enhances the capability of portable instruments to distinguish between volatile and non-volatile particulate matter in near real-time, and in the case of the PAS and possibly other instruments enhances instrument response in the presence of volatile nucleation mode particles. As discussed previously, the University has used a CS in the laboratory since the early 1990s (Abdul-Khalek, et al., 1995; Abdul- Khalek, 1996; Stenitzer, 2003). Under the direction of Prof. Kittelson a design team composed of senior mechanical engineering students was given the task of designing, building and evaluating a portable CS for use with portable instruments. A detailed design report is found elsewhere (Feldkamp, et al., 2008). Figure 21 shows a schematic of the portable instrument that includes an inlet with pre-selector, heater and catalyst, temperature controller and thermocouple, heat exchanger, portable instrument (in this case a portable PAS), flow controller, pump and power supply. Figure 22 shows a drawing of the CS portion of the portable instrument, and the catalyzed stainless steel core provided by Miratech. The Miratech cores were used in the senior design team s mini-cs. Johnson Matthey provided catalyzed ceramic cores that are being evaluated for use in a second generation mini-cs. 44

45 Figure 21. Schematic of portable instrument Figure 22. Drawing of the CS portion of the portable instrument with photo of the Miratech stainless steel substrate used in the catalyst section The sample aerosol is heated by a star wound heater, shown in Figure 22 that acts as both a heater and a diffuser of the sample flow upstream of the catalyst. The purpose of the heater is to heat the flow and catalyst to the desired temperature of 300 ºC, so that volatile hydrocarbons can be removed by oxidation by the CS. The catalyst is surrounded by a mat material to prevent leakage around the catalytic core, and cased in a stainless steel tube that is wrapped with insulation. On either side of the tube are ceramic gaskets that prevent conductive heat loss. The heater is seated in the upstream ceramic tube and coated with magnesium oxide to inhibit particle formation due to heating of the metallic material. On either end of the assembly, there are stainless cones to allow for a flow transition for expanding from a small tube and then converging to a smaller tube on the exit. The design of the upstream cone is important to prevent channeling, thus ensuring uniform flow through all cells of the catalyst. The design does not include the S-trap for sulfate removal that is included in the laboratory version of the CS. Key design parameters for the portable CS are power consumption, particle losses, and volatile material removal efficiency. Each was evaluated by the design team and the results are discussed briefly below. The designed flow rate for the mini-cs is 1 L/min. The lab CS has a flow rate of 10 L/min. 45