Effects of Exhaust Gas Temperature and Fuel Composition on Particulate Emission from Spark Ignition Engines Robert E. Sampson and George S. Springer' Department of Mechanical Engineering, The University of Michigan, Ann Arbor, Mich. 48104
An experimental apparatus was built to collect particles directly from the exhaust of a spark ignition engine mounted on a dynamometer test stand. 'Three different sampling systems, consisting of filters and an Andersen stack sampler, were constructed. The performances of these sampling systems were evaluated, and the conditions necessary for obtaining representative samples of the exhaust stream were established. The total weight and the size distributions of particles emitted from a 1970 Chevrolet 350-in.3 V-8 engine were measured over a wide range of exhaust gas temperatures (85-820°F). The experiments were performed with both leaded and unleaded fuels at a steady engine speed and constant road load condition. The weight of the particles deposited along the inside wall of the exhaust system was also determined. The results of previous investigations indicate that several parameters contribute to the weight, size, and composition of particulates emitted from spark ignition engines-e.g., summary given by Springer (1972). These parameters include fuel composition, engine design and maintenance, operating conditions, and exhaust system (geometry, pressure, temperature). Although the effects of some of these parameters have been studied in detail, a systematic study has not yet been made of the influence of exhaust gas temperature on the weight and size of particulates. The major objective of this investigation was, therefore, to determine the effects of exhaust gas temperature on the total amount and size distribution of particulates formed in the exhaust of spark ignition engines. A further goal was to evaluate these effects using both leaded and unleaded fuels. The two parameters, temperature and fuel lead content were selected for study because of their significance in the particulate formation process. It is recognized, however, that besides temperature and fuel, other parameters may also influence particulate emissions. Nevertheless, to focus attention on their effects, only these two parameters were varied during the experiments. All other parameters were held constant a t values representative of an automobile cruising a t 55 mph. The measurements were performed with sampling systems suitable for collecting particles a t temperatures up to 850°F and for determining the total weight and size distribution of the collected particles.
tests was maintained constant a t 14.6. The engine was coupled to an eddy current dynamometer. The engine was lubricated by Standard Permalube 1OW30 oil, which was changed every 50 hr of operation and whenever the fuel was changed. Both leaded (Standard Regular) and unleaded (Indolene Clear) fuels were used in the experiments. The physical properties of the fuels are given by Sampson and Springer (1972). Exhaust System. Pressure and velocity fluctuations in the exhaust system were minimized by combining the exhaust from all eight cylinders through two standard exhaust manifolds and a crossover pipe. To further reduce these fluctuations, a 2 f t long, 1 ft diam cylindrical surge tank was attached to the crossover pipe (Figure 1). The surge tank was insulated to maintain a temperature drop along it similar to that normally obtained in a muffler. A muffler was not included in the exhaust system. Following the surge tank, the exhaust was discharged into a tunnel vent directly, through a bypass filter, or through a simulated exhaust system. The discharge systems were constructed of 2-in. black iron pipes. Valve GI controlled the flow to the tunnel vent, Valve GOthrough a ,142-mm diam Type A Gelman glass fiber filter (F1, Figure 1) located 32 in. from the surge tank. Flow to the simulated exhaust system was controlled by Valve G3. This exhaust system was mounted horizontally and consisted of three 2-ft-long pipes in series, followed by a 90" pipe elbow, a 3-in.-long pipe, another 90" pipe elbow, and then three more 2-ft-long pipes in parallel with the first three. To collect deposits on the wall, inserts made of 0.003 in. thick, 6.5 in. long, and 6.0 in. wide stainless steel shim stock were fitted tightly into each end of the 2-ft-long pipe sections. The temperature of the entire exhaust system could be regulated by heating tapes and by two fans. The temperatures of the exhaust pipes were measured by thermocouples cemented into the walls of the pipes. The exhaust gas temperature was measured by thermocouples inserted
Flaiiblc Tubing
Experimental Apparatus
The experimental apparatus consisted of three major components-the engine, the exhaust system, and the sampling system (Figure 1). Engine. The engine was a 350-in.3 displacement, 255 hp ( a t 4800 rpm) 1970 Chevrolet V-8 production engine with a 9 : l compression ratio. The air-fuel ratio during the
TO TUNNEL VENT G I , GZ,GJ GATE VALVES SAMPLING PORTS APPROXIMATELY ! F T
0
1
To whom correspondence should be addressed.
Figure 1. Schematic of exhaust system Volume 7, Number 1, January 1973
55
,
9
1
1 1 lki i R A
E
0 250 0 125
60
050
0 2 5 0 0 150
6 0
050
60
050
0 2 5 0 0210 4
Figure 2.
Probe dimensions
Both Systems I and I1 were surrounded with beaded heaters and ceramic fiber insulation, and could reach temperatures up to 850°F. The temperatures were monitored with thermocouples located as indicated in Figure 3. In System I11 a filter was connected to the probe through a y,-in. i.d., 5-in.-long stainless steel tube (Figure 3c), equipped with a stainless steel metering valve. Ambient air could be mixed with the gas sample. The airflow rate was measured by a wet test meter a t the inlet of the air line. All filters used in Systems 1-111 were 47-mm diam Gelman Type A glass fiber filters, which are 98% efficient in removing particles of diameters larger than 0.05 p . These filters (and also filter F1 in Figure 1) were mounted in specially designed filter holders capable of withstanding temperatures up to 850°F. After each sampling system the exhaust gas temperature was reduced to 70°F by a water-cooled heat exchanger. The flow rate was then measured by a wet test meter. Flow through the sampling system was provided by a vacuum pump.
, ,
30
~ 0 3 7 5 ~ 0 3 3455 ~ 0 5 0 0 0425
55
1
60 I 0 0
60
I50
A L L UNITS IN INCHES
directly into the middle of the exhaust stream. Sampling ports were also provided along the entire exhaust system. Sampling System. The sampling system consisted of a probe, a particle collection system, a flow measurement apparatus, and a vacuum pump. The gas sample containing the particles was drawn from the exhaust stream through a stainless steel sampling probe (Figure 2). Three different particle collection systems were employed (Figure 3). With System I the total weight and size distribution of the particles were measured a t temperatures corresponding to the gas temperature in the exhaust stream. With Systems I1 and I11 the total weight of particles was determined a t temperatures below the gas stream temperature. In System I1 the sample was cooled by water. In System I11 the sample was mixed with ambient air, thereby reducing the sample temperature to levels below those obtainable in System 11. System I consisted of a Model 50-000 high-temperature Andersen stack sampler (with inconel gaskets and stainless steel separating plates) in parallel with a bypass filter (Figure 3a). A filter was also placed immediately behind the sampler. The flow was controlled by two 3/~-in.stainless steel gate valves. In System 11, a 10 in. long, yl6-in. i.d., stainless steel tube was placed between two filters (Figure 3b). This tube was surrounded by a stainless steel tube. Water was circulated between the two tubes, and the temperature of the gas was adjusted by controlling the water flow rate.
E x p e r i m e n t a l Procedure
After installation, the engine was operated for 50 hr a t various speeds as recommended by the manufacturer. During this time the exhaust was discharged directly into the tunnel vent. All experiments were performed a t a steady engine speed (1800 rpm) and constant load (24.5 bhp), corresponding to a 1970 full-sized Chevrolet cruising a t 55 mph under road load conditions. Prior to undertaking the experiments, a series of tests was conducted to ensure that the sample drawn from the exhaust stream was representative of the conditions in the stream itself. In these tests, leaded fuel was used, and the particles were collected utilizing Sampling System 11. Particulate concentrations were measured a t 30, 35, 40, 45, 60, 90, and 120 min after starting the engine. No change was observed in the results of these measurements, suggesting that a 30-min warmup period was adequate. Therefore, before collecting a sample the engine was always warmed up for 30-40 min. When switching fuels, particle concentrations were measured after operating the engine for 24, 32, 24, 36, and 40 hr. No appreciable change was noted in the concentrations measured a t these times. All the data reported were taken after the 32-hr period. When sampling directly from the exhaust stream, it is of primary importance that the sample be withdrawn isokinetically (Watson, 1954; Habibi, 1970). This can be FROM PROBE
FROM PROBE
FROM PROBE I
I
fI INSULATION
il
HEATER
FLOW
ANDERSEN SAMPLER FILTER I F 3 )
t TO HEAT EXCHANGER
0 ) SYSTEM
Figure 3.
Schematic of particle collection
systems 56
Environmental Science & Technology
i
1 TO HEAT EXCHANGER
I
b l SYSTEM
II
TO HEAT EXCYANGER
c ) SYSTEM
IU
@ - VALVE I
- THERMOCOUPLE
Table I . Effect of Surge Tank on Pressure Fluctuation and Particle Concentration in the Exhaust Stream
Pressure fluctuations at the location of the surge tank (in Hg) With surge tank
Above mean Below mean
Without surge tank
Static
Stagnation
Static
Stagnation
0.15
0 30
1.20
1.80
0.15 0 30 120 Particle concn. ma/ft3
1.20
With surqe t a n k
Av
Without surqe tank
0.31 4 0.49
0.394 0.473 0.446 0.358
0.402
0.417
achieved only in steady flow, which does not exist in an automobile exhaust system. T o determine the effects of velocity fluctuations on the concentration of particulates in the sample, the exhaust stream (in the absence of the surge tank) was sampled 100 in. downstream from the exhaust port. The particle concentrations thus measured ranged from 0.39 to 0.47 mg/ft3 with the average value of 0.41 mg/ft3. The particle concentration was also determined by the filter collecting particles from the entire exhaust stream (F1 in Figure 1) and was found to be 0.42 mg/ft3. The good agreement between the results of these two independent measurements indicates that, by this point, fluctuations were damped sufficiently so as not to affect the sample appreciably. Although fluctuations did not appear to be a significant factor, a surge tank was placed in the exhaust system to further minimize their effects. I t was found that the surge tank reduced the fluctuations in static and stagnation pressures, but did not influence significantly the particle concentration (Table I). Since variations in the sampling flow rate may also affect the particle samples, particulate concentrations were measured a t various sampling flow rates. The results revealed that near the flow rates required for isokinetic sampling even 200-300% variations in sampling flow rate did not cause more than a 10% difference in the measured particle concentrations. This experimental observation is in agreement with the theoretical results of Watson (1954). In all subsequent tests, samples were collected a t isokinetic flow rates calculated as outlined by Sampson and Springer (1972). T o determine the effect of the probe diameter on the sample, particulate concentrations were measured using 0 NATURAL CONVECT10
soot
FORCED CONVECTION FIT TO DATA
A
-
TEMPERATURE
five different size probes (Figure 2). The results did not show any systematic variation in particle concentration with probe diameter. In all subsequent tests, a Y,-in. 0.d. probe was used, The distribution of particles across the exhaust stream was assessed by placing the probe at different locations across the pipe (Sampson and Springer, 1972). The maximum difference in particle concentration across the exhaust stream was 4%. Therefare, in all subsequent measurements, the probe was placed in the center of the stream. The time required to collect a representative sample on the filters was also evaluated, and was found to be 10 min. T e s t Procedures
Prior to collection of samples, the temperatures of the exhaust gas and the wall of the simulated exhaust system were recorded along its entire length under steady-state conditions (Figure 4). During these measurements, the exhaust system was either exposed to room air and hence cooled by natural convection only, or it was cooled by forced convection created by fans blowing longitudinally along the exhaust system. The latter was used to simulate conditions that might occur with a moving automobile. Before each experiment the engine was warmed up for 30-40 min while discharging through the tunnel vent. During this time, the samplers were brought to the desired temperature, and the simulated exhaust system was heated to the temperature it would have attained had the exhaust gas been flowing through it. Once steady-state temperature was reached, the main gas stream was switched to the simulated exhaust system. Sampling duration was 2 hr for the Andersen Sampler, and 10 min for the filters. Using System I, the exhaust system was sampled a t six locations approximately 24 in. apart along the simulated exhaust system. The entire sampling system was kept a t the temperature of the gas stream a t the location of the sampling probe. With System 11, samples were collected only a t the beginning of the simulated exhaust system. The temperatures of the filters (F4, Fg in Figure 3 b ) were adjusted to different values in the range of 450-820°F. With System 111, samples were collected a t the end of the simulated exhaust system. The ratio of the air-to-exhaust sample flow rate was 8 to 1. The particle concentration in ambient air was found to be 1.2 x 10-4 mg/ft3. This is less than 0.1% of the particle concentration in the exhaust stream. Deposition on the wall of the simulated exhaust system was studied by measuring the weight of the particles deposited on the inserts. Before the experiments, the filters were kept for 24 hr in a constant-temperature desiccator, and were then weighed. After completion of a test the filters were weighed, returned to the disiccator, and weighed again every 24 hr until there was no change in the weight. The Andersen Sampler plates and the inserts in the exhaust system were handled in the same manner. Results
500
WALL TEMPERATURE
I
O
50
'
"
'
1
(
1
1
1
(
1
1
1
1
100 150 200 DISTANCE ALONG EXHAUST SYSTEM (INCHES)
Temperature variation of the exhaust gas and the inside wall of the exhaust system Figure 4.
Particle Concentration. The particle concentration as a function of exhaust gas temperature is shown in Figure 5 . The results are presented in terms of mg/ft3 and gram/ mile for both leaded and unleaded fuel. The data were obtained using Sampling Systems I and I1 above 450"F, and Sampling System I11 a t 85°F. The concentrations given in Figure 5 are nearly the same as would be emitted from the exhaust system since only a small percentage of particles are deposited on the wall (see below). Volume 7, Number
1, January 1973
57
Table II. Particle Emission During Steady and Cyclic Operating Conditions Emissiona Engine
Speed
Leaded fuel
Steady speeds, emission in gram. mile Various 1954-57 models 50 m p h 0.085-0.295 1970 Chevroiet, V - 8 60 mph 0.03-0.16 1970 Chevrolet, V - 8 55 rnph 0.06 Steady speeds, emission in mg/ft3 Various 1961-62 models 25-60 m p h -0.93 55 mph 0.85-0.94 1970 Chevrolet, V-8 Cyclic operating conditions, emission in gram, mile Various 1966-70 models 7-Mode federa1 cycle 0.149-0.189 1966 "Popular Model" V - 8 Simulated consumer 0.4 test cond. 1969 Ford, V - 8 7-Mode federa1 cycle 0.28-1.20
Unleaded fuel
-
...
0.0033-0.0167 0.01
Ratio
Investigator
...
McKee and McMahon (1960) Moran and Manary (1970) Present study
10: 1
-6:l
0.1 5-0.16
-6:l
Mueiler et ai. (1962) Present study
0.069-0.23
-2:l
Ter Haar et al. (1972)
...
...
0.15
Habibi (1970)
2.66:l
0.07-0.55
Ninomiyaetal. (1970)
-2:1-4:1
a i r some instances emission rates given in different units (gram mile or mg ft31 could not be converted to the same units.
When leaded fuel is used. the exhaust gas temperature has a significant effect on the concentration of particles. At 700"F, the particle concentration changes abruptly, increasing markedly with decreasing temperature. This is likely caused by condensation of PbClBr vapors in the exhaust stream. The concentration remains constant between 675" and 450°F, but increases again wheri the temperature drops from 450" to 85°F. This increase is probably due to the condensation of heavy hydrocarbons in the gas stream. Kote that under normal operating conditions exhaust gases are unlikely to be cooled much below 85'F. Therefore, the concentrations obtained at this temperature are reasonably close to the maximum values that would be emitted in practice from an auto exhaust. When using unleaded fuel, the temperature seems to have little effect on particle concentration at temperatures above 450°F. This suggests that few particles were formed in the exhaust system. Consequently, most particles present in the exhaust stream must have originated in the combustion chamber. Although a chemical analysis of the particles was not performed, earlier studies (Macfarlane et al., 1964; Schalla and Hibbard, 1957) indicated that such particles are composed mostly of carbon. An increase in particle concentration was observed between the temperatures of 450" and 85"F, probably due to condensation of heavy hydrocarbons. As indicated before. the base fuels used in these tests were different. Therefore, the differences in particulate concentrations with leaded and unleaded fuels may have been caused not only by the differences in the fuel lead content but also by differences in aromatic content. The presence of organic matter in exhaust particles b a s demonstrated by McKee and McMahon (1960), Begeman (1964), and Moran and Manary ( 1970). It is noted that Moran et al. (1971) also investigated the dependence of particulate concentration on temperature in the exhaust system of a spark ignition engine. However, they did not present data in the range 100-600°F and, therefore, their results cannot be directly compared with those obtained in the present study. Figure 5 clearly indicates the effects of exhaust gas temperature and fuel composition on particle concentration. These results, however, are only for a specific engine operating a t a constant speed and a t road load conditions. The effects of engine speed and load on the results were not evaluated. These effects can be assessed by comparing the present results with those obtained by other investigators, 58
Environmental Science & Technology
--
Comparisons can be performed only with the data taken at 85°F since none of the previous investigators sampled at higher temperatures. The comparisons, given in Table 11, reveal that for both leaded and unleaded fuels, the particle concentrations are nearly the same for different engines and for different speeds, provided the engine speed remained constant. This suggests that the results in Figure 5 are representative of steady operating conditions. During cyclic operating conditions, the particle concentrations are generally higher than during steady-state operation, but the ratio of particle concentration with leaded and unleaded fuel is considerably lower than with steady speeds (Table 11). Size Distribution. Size distributions of particles, as measured by the Andersen Sampler, are presented in Figure 6 and Table 111. To demonstrate the effects of fuel composition and exhaust gas temperature, only representative results are shown. Data obtained at additional temperatures were found to be scattered around these curves in a random manner. Similar to the particle concentration, the size distribution was unaffected by the temperature in the ranges of 450-675°F and 450-820°F for leaded and unleaded fuel, respectively. For leaded fuel, there was a change in the particle size distribution at 700°F which is about the same temperature at which a change in particle concentration was noted. The weight of the smaller particles (particles less than 1 diam) is much higher at lower temperature (500°F) than at the higher temperature (820°F) (see Table 111). I
'
I
1
l
l
I
I n
0 LEADED FUEL
t
A UNLEADED FUEL FIT TO DATA
-
\ \ \
I
'--
Figure 5. Particle concentration in the exhaust stream
Table Ill. Representative Particle Size Distributions Wtofparticles, r n g / f t 3 Unleaded fuei 500'F
1.0 0.4 0.4 0.5 0.8 0.7 0.7 0.4 3.4
Leaded fuei 500'F
0.80 0.38 0.34 0.63 0.70 0.33 0.70 0.70 51.7
x
"
102
Leaded fuel S2O'F
0.45 0.45 0.50 0.35 0.35 0.20 0.25 0.25
1
/ I 50
Figure
7.
L M
Lumulative weignr
01
250
particles aeposltea along m e
in-
20.5
The particle size distributions are significantly different for leaded and unleaded fuel. For leaded fuel, approximately 88 wt % of the particles are smaller than 0.35 and 98 wt % smaller than 10.0 p. For unleaded fuel, approximately 40 wt % of the particles are smaller than 0.35 p and 88 wt % are smaller than 10.0 p (Figure 6). The weight of small particles (smaller than 0.35 p ) is much higher with leaded than with unleaded fuel. However, the weight of large particles (larger than 0.35 p ) is approximately the same for both leaded and unleaded fuel (Table
B
m).
The foregoing results imply that lead compounds exhausted as particulate matter are mostly in particles smaller than 0.35 p . Similar results were obtained by other investigators (Habihi e t al., 1970; Hirschler e t al., 1957; Hirschler and Gilbert, 1964; Lee et al., 1971; Mueller et al., 1962; Ninomiya e t al., 1970) under both cyclic and steady-state operations using different engines. Generally, it has been found that cyclic operations yield larger particles than steady operating conditions. This is felt to be caused by agglomeration of particles a t the wall, and subsequent reentrainment during acceleration. Surface Deposition. Particle deposition along the inside surface of the exhaust system was measured using both leaded and unleaded fuels. Deposition is caused by the transport of particles from the gas stream to the wall and by the formation of particles directly on the wall. The former is caused by several mechanisms, including Brownian diffusion, electrostatic, gravitational, thermal, and inertial forces. The latter is due to condensation on the wall, which was generally -150°F cooler than the gas stream. Although the deposition rate is not necessarily uniform along the exhaust system, the cumulative weight of particles deposited along the exhaust system increases uni-
Figure 8. Scanning electron microscope pictures of the deposition on the inside wall of the exhaust System A-Leaded fuel. 75 in. from exhaust manifold; 0-leaded fuel, 220 in. from exhaust manifold: C-unleaded fuel. 75 in. from exhaust manifold; 0-Unleaded fuel, 220 in. from exhaust manifold; height of the letters c o r m Sponds to approximately 1 p
formly (Figure 7). In comparing Figure 7 to Figure 5, it is seen that the weight deposited up to any point in the system is less than 10% of the particles contained in the stream a t that point. The differences in the sizes of the deposited particles using leaded and unleaded fuels were investigated by looking a t the deposits utilizing a scanning electron microscope. Photographs of the deposits a t the beginning and a t the end of the exhaust system demonstrate two significant effects: considerably more particles are deposited a t the beginning of the exhaust system with leaded fuel than with unleaded fuel, and with leaded fuel many more small particles are deposited than large ones (Figure 8). It must be recognized, however, that in a n actual exhaust system, particles accumulate on the walls and flake off pericdically. Acknowledgment The authors thank J. T. Ganley for his help in the experiments. Literature Cited
Figure 6. Representative particle size distributions
Begeman, C. R., in "Vehicle Emissions," Society of Automotive Engineers, Technical Progress Series 6, New York, N.Y., 1964, pp 163-74.
Habibi, K., Enuiron. Sei. Teehnol., & 239-53 (1970). Volume 7,Number 1, January 1973 59
Habibi, K., Jacobs, E. S., Kunz, W. G., Pastell, D. L., “Characterization and Control of Gaseous and Particulate Exhaust Emissions from Vehicles,” presented a t the Air Pollut. Contr. Ass., West Coast Section, Fifth Technical Meeting, San Francisco. Calif. Reoort available from E . I. DuPont DeNemours and Co., Wilmiigton, Del., October 1970. Hirschler, D. A , , Gilbert, L. F., Lamb, F. W., Kiebylski, L. M., Ind. Eng. Chern., 49, 1131-42 (1957). Hirschler, D. A., Gilbert, L. F., Arch. Enuiron. Health. 8 . 297-313 (1964). Lee, R. E . . Patterson, R. K., Crider, W. L., Wagman, J., Atrnos. Environ , 5,225-37 (1971). Macfarlane, J. J., Holderness, F. H . , Whitcher, F. S. E., Cornbust Flame, 8,215-27 (1964). McKee, H . C., McMahon, W. A . , J. Air Pollut Contr. Ass., 10, 456-62 (1960). Moran, J. B., Manary, 0. J., “Effect of Fuel Additives on the Chemical and Physical Characteristics of Particulate Emissions in Automotive Exhaust,” Interim Reut. to the Nat. Air Pollut. Contr. Ass., submitted by the Dow. Chemical Co.. Midland, Mich., July 1970. Moran, J. B., Manary, 0. J., Fay, R. R., Baldwin, M. J., “Development of Partic;late Emission Control Techniques for Spark Ignition Engines, Final Rept. to the Environmental Protection Agency, submitted by The Dow Chemical Co., Midland, Mich., July 1971. ~
~~
~
~~
~~~~
~~~
Mueller, P. K., Helwig, H. L., Alcocer, A. E . , Gorg, W. K., Jones, E . E., Symposium on Air Pollution Measurement Methods, American Societv for Testing and Materials, Special Tech? Publ. No. 352, pp60-77 (1962).Ninomiva, J . S., Bergman, W., Simpson, B. H., “Automotive Particulate Emissions,” presented 2 the Second International Clean Air Congress, Int. Union of Air Pollut. Prevention Ass., Washington, D.C. report available from Automotive Emissions Office, Ford Motor Co., Dearborn, Mich., December 1970. Sampson, R. E., Springer, G. S., “Effects of Temperature and Fuel Lead Content on Particulate Formation in Spark Ignition Engine Exhaust,” Fluid Dynamics Laboratory, Publication No. 72-1. DeDartment of Mechanical Engineerine. The Universitv of Michiga;, Ann Arbor, Mich., 1972. Schalla, R. L., Hibbard, R. R., NACA Report No. 1300, Chap. IX, pp 346-455 (1957). Springer, G. S., “Engine Emissions,” G. S. Springer and D. J. Patterson, Eds., Chap. V, Plenum Press, New York, N.Y., 1972. Ter Haar, G. L., Lenane, D. L., H u , J . N., Brandt, M., J . Air Pollut. Contr. Ass., 22,39-46 (1972). Watson, H. H., Ind. Hyg. Quart., 3 , 2 1 4 (1954). I
Y
Received for review May 19, 1972. Accepted October 17, 1972. This work u a s supported by the Environmental Protection Agency under Grant No. AP-01012-01.
NOTES
Dependence of Hi-Vol Measurementson Airflow Rate Arnold L. Cohen Office of Research and Monitoring, National Environmental Research Center, Environmental Protection Agency, Cincinnati, Ohio 45268
Comparisons of particulate concentration data obtained from Hi-Vol air pollution samplers, operating a t different airflow rates, can lead to errors. A difference of only 1% was found between hi-vol samplers operated a t 50 and 60 cfm; however, hi-vols operated a t 40 and 60 cfm exhibited a 3.8% difference. Loss of submicron aerosol particles through the filter pores a t the higher face velocities account for the discrepancies. The test data reported here point out the necessity to control the conditions under which air samples are collected in the various localities. H
The high volume (hi-vol) air sampler is widely used by air pollution control agencies and industry in the U.S., Canada, Japan, and other countries to determine the concentration of total suspended particulate matter. The sampler is operated for a 24-hr period by drawing air a t the rate of 40-60 cfm through an 8 x 10-in. glass fiber filter capable of removing nearly 100% of all particulates 0.3 in diam or greater (Pate and Tabor, 1962). Although most hi-vol samplers in use are of approximately the same configuration-i.e., gabled roof design, plywood or aluminum construction, comparable air inlet area-the samplers are not operated in a uniform fashion. Air pollution control agencies may sample a t an initial flow rate that can vary from 35 cfm to 65 cfm. Since the particle capture velocity a t the sampler air inlet is a direct function of airflow rate, the question arises as to the comparability of particulate concentration data collected from samplers not operating at the same flow rate. 60
Environmental Science & Technology
A study was conducted a t the Environmental Toxicology Laboratory of the National Environmental Research Center, Cincinnati, to determine the effect of airflow rate on particulate concentration measurements made with hivol samplers. Four gabled roof hi-vol samplers (GMW Model 2000) were operated on the third-floor roof of the Laidlaw Avenue Laboratory in conjunction with a National Air Surveillance Network (NASN) cascade impactor sampler (Lee and Flesch, 1969) for determining the size distribution of suspended particulate matter. Samples were collected from June 17 to July 22, 1971. The hi-vols were operated for 24-hr sampling periods a t flow rates that varied from 30 to 60 cfm. The flow rates on each sampler were changed periodically to eliminate possible effects inherent in an individual sampler. The initial flow rate was set on each hi-vol using a variable transformer, and, at the end of the 24-hr sampling period, the flow rate was again determined with a calibrated rotameter according to prescribed EPA procedures (U.S. EPA, 1971). The average flow rate for the 24-hr period was determined from the initial and final readings. In general, the decrease in flow rate owing to particulate loading on the filters did not exceed 3 cfm. The results of these tests are presented in Table I. The data indicate a trend toward lower concentration measurements with increasing airflow rate. The particulate concentrations over the entire seven-day sampling period varied from approximately 100 to 200 wg/m3, thereby providing a wide aerosol range for evaluation. This range is in general agreement with NASN results for 1970 where the total suspended particulate concentration varied from 45