[Federal Register Volume 63, Number 24 (Thursday, February 5, 1998)]
[Rules and Regulations]
[Pages 6032-6037]
From the Federal Register Online via the Government Publishing Office [www.gpo.gov]
[FR Doc No: 98-2878]



[[Page 6031]]

_______________________________________________________________________

Part VI





Environmental Protection Agency





_______________________________________________________________________



40 CFR Part 50



National Ambient Air Quality Standards for Particulate Matter; Final 
Rule

  Federal Register / Vol. 63, No. 24 / Thursday, February 5, 1998 / 
Rules and Regulations  

[[Page 6032]]



ENVIRONMENTAL PROTECTION AGENCY

40 CFR Part 50

[AD-FRL-5961-6]


National Ambient Air Quality Standards for Particulate Matter

AGENCY: Environmental Protection Agency (EPA).

ACTION: Final rule.

-----------------------------------------------------------------------

SUMMARY: On July 18, 1997, EPA announced a supplemental comment period 
for the limited purpose of taking comments on certain field and 
laboratory test results associated with the development of the 
reference method (Appendix L of 40 CFR Part 50) for measuring particles 
with an aerodynamic diameter less than or equal to a nominal 2.5 
micrometers (PM2.5) in the ambient air. In the announcement, 
EPA indicated that upon the close of the comment period it would decide 
whether any further action would be appropriate. Having carefully 
assessed the comments received, EPA has determined that no further 
action is necessary.

ADDRESSES: The comments received during the supplemental comment period 
and EPA's responses to those comments have been entered into Docket No. 
A-95-54. The docket is available for public inspection in the Central 
Docket Section of the U.S. Environmental Protection Agency, South 
Conference Center, Rm. 4, 401 M St., SW., Washington, DC 20460. The 
docket may be inspected between 8 a.m. and 3 p.m., Monday through 
Friday, except legal holidays, and a reasonable fee may be charged for 
copying.

FOR FURTHER INFORMATION CONTACT: John H. Haines, MD-15, Air Quality 
Strategies and Standards Division, Office of Air Quality Planning and 
Standards, Environmental Protection Agency, Research Triangle Park, NC 
27711, telephone: (919) 541-5533, email: [email protected] or 
Neil H. Frank, MD-14, Emissions, Monitoring and Analysis Division, 
Office of Air Quality Planning and Standards, Environmental Protection 
Agency, Research Triangle Park, NC 27711, telephone: (919) 541-5560, 
email: [email protected].

SUPPLEMENTARY INFORMATION: On July 18, 1997, EPA published (62 FR 
38652) a final rule revising the national ambient air quality standards 
for particulate matter. In Unit VI.B. (Appendix L--New Reference Method 
for PM2.5) of the preamble to the final rule, EPA concluded 
that the proposed design and performance specifications for the 
reference sampler, with modifications described in the final rule, 
would achieve the design objectives set forth in the proposal. 
Accordingly, EPA adopted the sampler and other method requirements 
specified in the revised Appendix L as the reference method for 
measuring PM2.5 in the ambient air. As discussed in the 
preamble to the final rule, a series of field tests were performed 
using prototype samplers manufactured in accordance with the proposed 
design and performance specifications. The results of these field tests 
confirmed that the prototype samplers performed in accordance with 
design expectations. Operational experience gained through these field 
tests did, however, identify the need for minor modifications as 
discussed in the preamble to the final rule. As explained in that 
preamble, EPA made other modifications to the proposed design and 
performance specifications in response to public comment. As part of 
this process, EPA performed laboratory tests to ensure that the 
modifications achieved their intended objectives. While the results of 
the field and laboratory tests were largely confirmatory in nature and 
did not indicate a need to alter the basic design and performance 
specifications, they did identify areas that needed further refinement. 
Given that these tests were performed, by necessity, during and after 
the close of the public comment period and because the results were not 
available for placement in the docket until late in the rulemaking 
process, the preamble to the final rule announced that a supplemental 
comment period would be afforded for the limited purpose of taking 
comments on these field and laboratory test results. The following 
documents present the results of the field and laboratory tests and 
associated analyses that EPA considered, as discussed in Unit VI.B. of 
the preamble to the final rule, in making minor modifications or other 
refinements to the proposed reference method for measuring 
PM2.5 in the ambient air. The documents are:
    1. Adaptation of the Low-Flowrate, PM10, Dichotomous 
Sampler Inlet to Fine Particle Collection.
    2. Filter Temperature Specification Report.
    3. Flow Rate Specification Report.
    4. Laboratory and Field Evaluation of FRM Sampler Report.
    5. Prototype PM2.5 Federal Reference Method Field 
Studies Report.
    In a separate document published on July 18, 1997 (62 FR 38762), 
EPA announced a supplemental comment period for the limited purpose of 
taking public comment on the five documents specified above. The 
document emphasized that comments received on the reference method for 
PM2.5 that went beyond the scope of the five documents would 
not be considered. The EPA also indicated in the document that upon the 
close of the supplemental comment period, it would consider the 
comments received and then decide whether any further action was 
appropriate. In response to the July 18, 1997 document, EPA received 
comments from three organizations. The EPA has conducted a careful 
assessment of the comments and has concluded that they raise no issues 
not considered prior to promulgation of Appendix L or addressed in the 
quality assurance guidelines to be presented in Section 2.12 of the 
Quality Assurance Manual for Air Pollution Measurement Systems. 
Accordingly, EPA has concluded that no additional rulemaking action is 
necessary as a result of the comments received during the supplemental 
comment period. A summary of the significant issues raised by the 
commenters and EPA's responses has been entered in Docket No. A-95-54 
and is reproduced as Appendix A to this document.

Appendix A--Responses to Significant Comments on Field and Laboratory 
Test Results Regarding Federal Reference Method for Measuring 
PM2.5 in the Ambient Air, Docket No. A-95-54, October 1997

Summary

    On July 18, 1997 (62 FR 38762), EPA announced a supplemental 
comment period for the limited purpose of taking public comment on the 
results of various laboratory and field tests and associated analyses 
involving the new Federal Reference Method for measuring 
PM2.5 in the ambient air (Appendix L of 40 CFR part 50). The 
new Federal Reference Method (FRM) was adopted on July 18, 1997 (62 FR 
38652) in conjunction with new national ambient air quality standards 
for PM2.5 (40 CFR 50.7). During the supplemental comment 
period announced on July 18, three organizations submitted comments.
    The EPA has reviewed the comments received and has concluded that 
none of them presents issues that were not previously considered in the 
development of the FRM for PM2.5, or that have not been 
addressed in the specific quality assurance guidelines to be presented 
in Section 2.12 of the Quality Assurance Manual for Air Pollution 
Measurement Systems. Accordingly, it is unnecessary to take further 
rulemaking action or to postpone

[[Page 6033]]

implementation of the Federal Reference Method for PM2.5 as 
a result of any of the comments.
    Significant comments raised in each commenter's letter are 
summarized below, together with EPA's responses.
    Item VI-D-04 Author: EPRI.
    Comment: FRM sampler provides biased results due to known losses of 
volatile and semi-volatile aerosol components.
    Response: The FRM sampler was never intended to collect and measure 
all semi-volatile aerosol components. The sampler was designed to 
closely approximate the measurements obtained by the type of samplers 
used in the health studies that served as the basis for the 
PM2.5 standards. Moreover, the new monitoring regulations 
require supplemental monitoring at a 50-site national speciation 
network in which volatile and semi-volatile aerosol components will be 
measured, thus providing a more complete characterization of the 
ambient aerosol.

    Item VI-D-05 Author: American Petroleum Institute.
    Comment: Efficacy of the rain shroud has not been demonstrated 
regarding minimizing rain or snow intrusion.
    Response: The EPA has been evaluating three identical prototype 
inlets which meet the dimensional specifications of the new 
PM2.5 FRM inlet. In these field tests conducted at Research 
Triangle Park, NC, three prototype FRM samplers containing the 
prototype inlets were collocated with six prototype FRM samplers 
containing the older style PM10 inlet (as proposed for the 
PM2.5 reference method sampler on December 13, 1996). 
Although relatively few significant rain events occurred in the area 
during this time period, inspection of the samplers appeared to 
indicate that the new inlet design was more effective at minimizing 
rain intrusion than the older design.
    The performance of the prototype inlets was also evaluated under 
artificial conditions designed to simulate periods of heavy rainfall. 
For these tests, two identical prototype reference method samplers were 
collocated outdoors such that their inlets were at the same elevation 
but positioned approximately 0.7 m apart horizontally. One of the two 
samplers used the prototype new PM2.5 inlet design while the 
other sampler used the older PM10 inlet design. An 
oscillating type sprinkler was then used to expose the two samplers to 
conditions of accelerated rainfall. The sprinkler nozzle was oriented 
to provide equal coverage to the two inlets and adjusted so the angle 
of incidence continuously varied between 0 deg. and 90 deg. relative to 
the inlet. A rain gauge was positioned between the two samplers and 
used to measure the quantity of simulated rainfall to which the 
samplers were exposed. Over a 2-day time period, eight discrete tests 
were conducted, each having a duration of 3 hours. At the completion of 
each test, the sprinkler was turned off, the rain gauge measurement was 
noted, and the water volume was measured in each of the sampler's 
collection jars. Prior to the next test, the rain gauge and collection 
jars were emptied, and the inlet locations were alternated between 
samplers in order to minimize any positional effects or flow system 
effects on the test results.
    Results of these simulated rainfall tests are summarized in Table 
1. The simulated rainfall during each 3-hour time period ranged between 
3.5 inches and 7 inches with a mean value of 4.75 inches. Inspection of 
Table 1 reveals that the older style PM10 inlet collected a 
range of 80 ml to 450 ml of water during each rain event. As expected, 
observations during the simulated tests indicated that rain intrusion 
into the inlet was maximum when rain impinged at an angle normal to the 
face of the sampler's insect screen. This phenomenon is typically 
observed in the field during periods of rain accompanied by elevated 
horizontal wind speeds. In contrast to the older PM10 inlet, 
no water droplets were observed to collect inside the prototype 
PM2.5 inlet during any of the eight replicate tests. During 
the entire testing totaling 38 inches of simulated rainfall, the new 
PM2.5 inlet collected no water while the older 
PM10 inlet collected over 1600 ml of water. Although these 
simulated rainfall tests cannot exactly simulate all the conditions 
that the samplers might encounter in the field, these results indicate 
that the new PM2.5 inlet design was much more effective at 
minimizing rain intrusion than the older, original PM10 
design.

Table 1.--Results of Simulated Rainfall Tests for PM2.5 Inlet Evaluation
------------------------------------------------------------------------
                                                  Volume of water in    
                                  Simulated       collection jar (ml)   
           Test No.               rainfall   ---------------------------
                                  (inches)     PM10  inlet  PM2.5  inlet
------------------------------------------------------------------------
1.............................  4.5.........  100.........  0           
2.............................  4.5.........  220.........  0           
3.............................  4.0.........  80..........  0           
4.............................  4.5.........  200.........  0           
5.............................  5.0.........  450.........  0           
6.............................  5.0.........  80..........  0           
7.............................  3.5.........  80..........  0           
8.............................  7.0.........  420.........  0           
                                Mean =......  Mean =......  Mean =      
                                4.75 in.....  204 ml......  0 ml        
------------------------------------------------------------------------

    Comment: Filter temperature overheats measured in February do not 
adequately represent those which might be measured in summer.
    Response: Evaluation of prototype FRM at RTP, NC after February 
indicated that overheats of 3 deg. C were occasionally observed but 
5 deg. C overheats were not observed even on days when radiant fluxes 
at the sampling site exceeded 1200 W/m\2\.
    Comment: The 6/30/97 McElroy/Frank memorandum provides a tabular 
summary of FRM PM2.5 precision measurements used to revise 
upward the method detection limit (MDL) specification from 1 
g/m3 to 2 g/m3. Detailed 
analysis is difficult since individual data are not provided or cited. 
However, inserting the reported mean daily precisions into the 
definition of MDL (and assuming that blank means=0) yields minimum MDLs 
of 2.3 g/m3 for Denver and RTP locations and 3.7 
g/m3 for Azusa, values that differ from those 
reported in the table where Denver = 2 g/m3, RTP = 
3 g/m3, Azusa = 2 g/m3.
    Response: The change in estimated method detection limit from 1 
g/m3 to 2 g/m3 was due to 
information gained through field use of prototype samplers since the 
regulation was initially proposed. As specified originally in the 
December 13, 1996 proposal, the detection limit of the PM2.5 
mass concentration measurement ``* * * is determined primarily by the 
repeatability (precision) of filter blanks * * *.'' At the time the 
regulation was proposed, field data had not yet been collected to 
determine the variability of field blanks. For this reason, laboratory 
blanks were used to provide a preliminary estimate of the method's 
precision. Once prototype samplers became available, specialized field 
studies conducted in Denver, Azusa, and RTP provided a data base upon 
which to provide actual estimates of the method's detection limit. The 
final regulation as promulgated on July 18, 1997 updated the 
preliminary estimate and modified the text to indicate that field 
blanks were used for estimating the method detection limit. In 
particular, Section 3.1 was modified to read, ``The

[[Page 6034]]

lower detection limit of the mass concentration measurement range is 
estimated to be approximately 2 g/m3, based on 
noted mass changes in field blanks * * *.'' Thus, the use of actual 
field data in conjunction with a minor modification in the MDL's 
definition accounted for the revision in the method detection limit.
    The commenter apparently misinterpreted the precision table 
included in the docket (reproduced in Table 2 below). The values 
reported in the last column of the table refer to the precision of 
measured PM2.5 concentrations and have no relationship with 
measured precision of field blanks. This apparent misinterpretation led 
to the commenter's conclusion that the original method detection limit 
calculations were in error. The enclosed Table 3 below presents actual 
data from the three field sites relating to the observed mass changes 
in the field blanks. As indicated in the final column of Table 3, the 
method detection limits determined at Denver, Azusa, and RTP were 2 
g/m3, 2 g/m3, and 3 g/
m3, respectively. This actual field information was the 
basis for the July 18, 1997 text which stated that the method detection 
limit ``* * * is estimated to be approximately 2 g/
m3.''

                                                Table 2.--Summary of Precision Tests at 3 Separate Sites                                                
                                           [Method Detection Limit (Field Blanks) = |Mean| + 10 * (Std. Dev.)]                                          
--------------------------------------------------------------------------------------------------------------------------------------------------------
                                                                                                                                 Method      Mean daily 
                                                                                                                 Mean PM2.5     detection     precision 
              Site                       Dates          No. days   Prototype samplers  PM2.5 range (g/     conc.         limit      (std. dev.)
                                                                       evaluated                m\3\)           (g/  (g/  (g/
                                                                                                                    m\3\)         m\3\)         m\3\)   
--------------------------------------------------------------------------------------------------------------------------------------------------------
DENVER, CO......................  Dec. 10-22.........         10  6 Graseby-Andersen.  1.4 to 20.6............         10.9             2          0.23 
AZUSA, CA.......................  March 25-April 10,           9  6 Graseby-Andersen.  6.0 to 32.1............         18.6             2          0.37 
                                   1997.                                                                                                                
RTP, NC.........................  April 4-30, 1997...         13  3 R&P..............  7.2 to 18.5............         11.7             3          0.23 
--------------------------------------------------------------------------------------------------------------------------------------------------------


                                             Table 3.--Calculated Method Detection Limit at 3 Separate Sites                                            
                                            [Method Detection Limit (Field Blanks) = Mean + 10 * (Std. Dev.)]                                           
--------------------------------------------------------------------------------------------------------------------------------------------------------
                                                                                                                                Standard                
                                                                                                      Total        Mean of    deviation of     Method   
                                                                                       Number of    number of    daily field   daily field    detection 
                      Site                                       Dates                  sampling      field        blanks        blanks         limit   
                                                                                          days        blanks    (g/  (g/  (g/
                                                                                                                    m\3\)         m\3\)         m\3\)   
--------------------------------------------------------------------------------------------------------------------------------------------------------
Denver, Co......................................  Dec. 10-22, 1996..................           10           30         -.010          0.19             2
Azusa, CA.......................................  March 25--April 10, 1997..........            8           24          0.18          0.22             2
RTP, NC.........................................  April 4-30, 1997..................            8           24          0.52          0.27             3
--------------------------------------------------------------------------------------------------------------------------------------------------------

    Comment: The 25 C limit should be termed ``post-acquisition'' 
rather than ``post-sampling.''
    Response: This is a good suggestion, and this terminology will be 
employed in Section 2.12 of the Quality Assurance Handbook for Air 
Pollution Measurement Systems.
    Comment: The 9/96 G. H. Achtelik report offers at best a lower 
bound estimate of filter volatiles loss.
    Response: Studies are currently being performed in Riverside, CA to 
further characterize the effects of volatile losses. In addition, EPA 
requires a 50-site chemical speciation network in which volatile and 
semi-volatile aerosol components will be measured.
    Comment: Midnight to midnight sampling may provide different 
measured concentrations than noon to noon sampling due to water of 
crystallization effects.
    Response: It was necessary to maintain the midnight to midnight 
sampling for PM2.5 to be consistent with the sampling 
schedules for other particulate measurements and to not unduly 
constrain the work schedules of site operators. However, if such 
effects are suspected, operators are encouraged to re-weigh filters 
after additional conditioning (beyond the minimum 24 hours).
    Comment: A number of lingering problems were identified in the 
field tests.
    Response: One of the purposes of these field tests was to develop 
preventative maintenance guidelines for routine operation of these 
samplers. None of these problems was unexpected, and each will be 
addressed in Section 2.12 of the Quality Assurance Handbook for Air 
Pollution Measurement Systems. Note also that these tests were 
performed using prototype and not production model PM2.5 
samplers.
    Comment: A field calibration protocol should be developed to test 
the performance of the inlets.
    Response: While the intent of the comment is understood, the 
recommended calibration protocol would be cumbersome, time consuming, 
and not precise enough to measure any realistic changes in fractionator 
performance.
    Comment: Poor correlation achieved by the Tucson site technician 
might indicate the samplers are not user-friendly and/or require 
special field personnel.
    Response: It should be noted that all of these studies were 
performed using prototype samplers that were operated using procedures 
that were at that time still under development. Taking this under 
consideration, the intramethod and intermethod results from all the 
other studies could have been interpreted as being closer than 
expected. The lower intramethod precision observed at the Tucson site 
can no doubt be attributed to a combination of contributing factors. As 
noted in the EPA staff report, ``* * * the Tucson study was operated by 
a site technician as additional and unassisted duties to his normal 
work load * * *.'' Of equal importance is the fact that the mean 
concentration at the Tucson site was appreciably lower than at any of 
the other five sampling sites. At low ambient concentrations, the 
effect of

[[Page 6035]]

sample handling, conditioning, and weighing uncertainties becomes much 
more important than at higher concentrations. It is reasonable to 
expect, therefore, that higher intrasampler variability would be 
observed at the Tucson site than at the other sampling sites. An 
assertive quality assurance program will be included within the 
implementation of the national monitoring network.
    Specialized tests were conducted in Azusa, CA to determine if local 
site personnel would experience significantly more variability with the 
prototype FRM samplers than would be experienced by specially trained 
researchers. First, aerosol researchers conducted 6 days of 22-hour 
sampling using six identical PM2.5 samplers. Mean precision 
in PM2.5 concentrations was measured to be 0.4 g/
m3. Using the same procedures, site operators from the South 
Coast Air Quality Management District then conducted their own 
precision tests with the same samplers. Mean precision in 
PM2.5 concentrations was also measured to be 0.4 g/
m3. Incidentally, this measured intrasampler variability was 
appreciably less than the 2 g/m3 maximum value 
allowed by the regulations.

    Item VI-D-06 Author: National Cotton Council of America.
    Comment: Based on impactor theory developed by Ranz and Wong, 
Parnell et al contend that the impactor cutpoint is actually 2.74 
m rather than the 2.5 m design value.
    Response: There are basically two problems associated with the 
Parnell et al approach. First, although the 1952 Ranz and Wong research 
led to important insights regarding impactor theory, it was an early 
work which could not properly account for the effects of complex 
impactor design parameters such as jet-to-plate distance, throat 
length, and fluid Reynolds number. Only the development of 
sophisticated numerical analysis techniques in conjunction with the 
advent of high speed computers allowed detailed analysis of fluid flow 
fields and of particle trajectories within the flow fields. In 
particular, important advances in our understanding of inertial 
impactors were made by Marple (1970) and Marple and Liu (1975). It was 
upon these improved design guidelines that the EPA prototype WINS was 
developed. Based on this well-accepted inertial impactor theory, one 
would predict a cutpoint of 2.44 m aerodynamic diameter for 
the WINS impactor rather than the 2.74 m value predicted by 
the simplistic approach of Ranz and Wong.
    The second problem associated with the Parnell et al. approach is 
that impactor theory can never be used to reliably predict an actual 
impactor's performance. Despite advances since the Ranz and Wong work, 
conventional impactor theory only provides starting guidelines upon 
which to base impactor design. In reality, a number of factors can 
affect a given impactor's performance including actual component 
dimensions, flow rate, particle bounce, particle re-entrainment, wall 
losses, and electrostatic effects. If one is interested in determining 
an impactor's actual performance, therefore, the impactor must be 
calibrated in the laboratory under carefully controlled conditions 
using primary calibration aerosols. The novel geometry of the WINS 
impactor reinforced the need for laboratory calibration to determine 
its actual performance. As described in ``Modification and Evaluations 
of the WINS Impactor,'' the experimentally determined cutpoint of the 
WINS impactor was measured to be approximately 2.48 m 
aerodynamic diameter at standard temperature and pressure conditions.
    References: Marple V.A. and Willeke K. (1976) Impactor design. 
Atmos. Envir. 10:891-896.
    Marple V. A. and Liu B.Y.H. (1975) On fluid flow and aerosol 
impaction in inertial impactors. J. Coll. & Interface Sci. 53:31-34.
    Comment: PM from agricultural operations has different 
characteristics than that used in the laboratory calibration. Actual 
performance of the WINS may be different in the field.
    Response: Laboratory tests showed that there was no difference in 
collection between liquid and solid aerosols. Fractionation of the 
aerosol using its aerodynamic properties automatically accounts for the 
particle's physical size, shape, and density.
    Comment: The data presented in ``Flow Rate Specification Report'' 
seems to indicate that flow rate errors in FRM prototype samplers are 
not random but systematically understate the actual flow rates. As a 
consequence, the sampled particles actually have a higher momentum than 
the FRM measurements imply, adversely affecting the interpretation of 
the penetration curves.
    Response: It is important to understand that no flow control system 
is inherently accurate and that all systems require periodic 
calibration. There are several factors which affect the flow rate 
accuracy of any individual FRM sampler. Because automatic volumetric 
flow control involves separate measurements of several key parameters 
(e.g., ambient temperature, ambient pressure, etc.), any inaccuracies 
in their actual measurements will naturally result in inaccuracies in 
flow control. Although these parameters are typically calibrated at the 
same time as the initial flow calibration, any drift in their response 
since the time of calibration will naturally result in variations in 
flow control. For example, if pressure transducer circuitry is not 
properly compensated for temperature, significant reductions in ambient 
temperature can result in directional biases in ambient pressure 
measurements. These pressure measurement biases can, in turn, naturally 
result in directional biases in flow control.
    Because collocated, identical instruments are typically calibrated 
in the field using the same flow transfer standard, it is reasonable to 
expect that any directional bias in the transfer standard's calibration 
will also result in biases among the group of collocated samplers in 
the same direction as that of the transfer standard. Thus, if the flow 
transfer standard and NIST traceable audit device do not agree exactly, 
we tend to observe directional differences in flow response among a set 
of samplers. In the case of the sample flow data provided in the 
docket, the actual flow rates measured by the NIST traceable flow 
standard were always higher (mean value = 0.9 percent higher) than the 
flow value indicated by the instruments. Actual flow rates are 
positively biased, therefore, which accounts for the percent error 
direction used in reporting the flow audit results.
    Regardless of one's individual choice of bias direction, the effect 
of the flow bias can be predicted with respect to magnitude and 
direction. These effects can be conveniently grouped into aspiration 
and particle transport effects, effects of flow bias on fractionator 
performance, and effects of flow bias on calculated PM2.5 
concentrations. These factors are considered separately below.
    Aspiration and Particle Transport Effects: Although major biases in 
sampler flow rate can adversely effect the sampler's inlet aspiration, 
minor flow rate biases should have negligible effects on the inlet's 
ability to withdraw representative aerosol samples from the ambient air 
and transport the aspirated aerosol efficiently throughout the sampling 
system. The FRM specifications for flow rate control were designed to 
ensure that large errors in flow control would be identified during 
sampling and that appropriate action (i.e., sampler shutdown and/or 
warning flags) would be automatically taken.

[[Page 6036]]

    Effects on Fractionator Performance: Similar to the effect of flow 
rate bias on the sampler's aspiration performance, minor flow rate 
biases should have negligible effects on the sampler's ability to 
accurately fractionate an aspirated aerosol. For small variations in 
flow rate (such that the jet Reynolds number is not significantly 
altered), the fractionator's cutpoint is inversely proportional to the 
square root of the volumetric flow rate. For the EPA WINS impactor 
which possesses a cutpoint of 2.48 m at 16.67 L/min., for 
example, a 2 percent increase in flow rate would result in only a 1 
percent decrease in cutpoint to 2.46 m. Similarly, a 2 percent 
decrease in flow rate would result in only a 1 percent increase in 
cutpoint to 2.50 m. Moreover, these 1 percent predicted 
changes in fractionator cutpoint would result in an even smaller bias 
in collected PM2.5 mass concentration. Since the expected 
mass collected is a function of both the fractionation curve and the 
mass size distribution of the aerosol to which it is exposed, numerical 
sensitivity analysis has been performed on three idealized ambient 
distributions. Assumed parameters for the distribution are identical to 
those used in 40 CFR part 53 Table F-3 for coarse, ``typical,'' and 
fine ambient aerosol distributions. Since only the cutpoint of the 
fractionator curve can be expected to change at low flow rate biases, 
the predicted fractionation curve can numerically integrate with each 
of the ambient distributions to calculate the expected measured mass 
concentration as a function of flow rate bias.
    Results presented in the table below indicate that a maximum bias 
in expected mass concentration of approximately 0.6 percent would be 
associated with flow biases of 2 percent. Note that higher flow rates 
result in lower fractionator cutpoints, which results in lower mass 
gains than would normally occur.

------------------------------------------------------------------------
                                       Expected bias in measured mass   
                                   concentration solely as a function of
                                       flow-induced cutpoint changes    
                                  --------------------------------------
           Distribution              -2% flow     0% flow      +2% flow 
                                       bias         bias         bias   
                                    (Dp50=2.46   (Dp50=2.48   (Dp50=2.50
                                   m)  m)  m)
                                    (percent)    (percent)    (percent) 
------------------------------------------------------------------------
Coarse...........................       +0.5            0         -0.6  
``Typical''......................       +0.2            0         -0.2  
Fine.............................       +0.2            0         -0.2  
------------------------------------------------------------------------

    Effects on Calculated PM2.5 Mass Concentration: As 
discussed above, the effects of flow biases on inlet aspiration 
performance and fractionator cutpoint are essentially negligible. The 
primary effect of flow rate biases on PM2.5 measurements 
concerns the calculation of PM2.5 concentration from the 
measured mass gain of the filter divided by the volume of air sampled 
as reported by the sampler. Because the FRM samplers are designed to 
continuously adjust volumetric flow rate to the design setpoint flow 
rate of 16.67 actual L/min., the sampled air volume reported by the 
instrument is typically very close to the design flow rate times the 
sampling duration. If, for example, the flow rate reported by the 
sampler was in fact low by 2 percent, the sampler would have sampled, 
fractionated, and collected a fine particulate mass which was 
approximately 2 percent higher than it should have been. Since the 
calculated PM2.5 concentration is simply the measured mass 
divided by the indicated sampled air volume, the calculated 
PM2.5 concentration would be positively biased by 
approximately 2 percent. Note that the effects of flow biases on 
fractionator performance and collected aerosol mass are in opposite 
directions, thus partially offsetting each other.
    Comment: The fractionator used in the FRM should be evaluated in 
the laboratory after collecting appreciable quantities of polydisperse 
particles on the impaction plate.
    Response: These sensitivity tests were in fact conducted in the 
laboratory and described in ``Modification and Evaluation of the WINS 
Impactor.'' The WINS impactor was exposed to laboratory generated 
polydisperse Arizona test dust for three 24-hour periods where the mean 
dust concentration was measured to be 330 g/m3. 
After each 24-hour collection period, the performance of the loaded 
substrate was evaluated in the laboratory using primary calibration 
aerosols. Results showed that the fractionator could be exposed to 
ambient aerosol concentrations averaging 330 g/m\3\ for 6 
consecutive days before a 5 percent bias in measured PM2.5 
concentration would be expected.
    Comment: Favorable results of collocated field tests should not 
imply that the samplers are accurately measuring PM2.5 
values, only that similar samplers produce similar results. To verify 
accuracy, the six samplers should be simultaneously tested in the 
laboratory using a known and typical aerosol as described in the 
previous comment.
    Response: Because the size and volatility of particles comprising 
fine ambient particulates vary over a wide range of environmental and 
sampling conditions, the accuracy of PM2.5 measurements 
cannot be defined in an absolute sense. Instead, EPA defines 
PM2.5 sampler accuracy based on how well the sampler meets 
all design, construction, and operational specifications set forth for 
samplers approved for determining compliance with the PM2.5 
regulations. In particular, field accuracy can be defined by the level 
of agreement between a given PM2.5 sampler and a collocated 
PM2.5 reference audit sampler operating simultaneously. In 
the case of collocated prototype FRM samplers, favorable agreement 
among the samplers implies that adequate control is being exercised 
over uncertainties associated with the sampler's construction, 
calibration, setup, and operation.
    Laboratory calibration of size selective components requires 
accurate generation and measurement of primary aerosol standards under 
very carefully controlled conditions. Simultaneous calibration of six 
identical samplers under these conditions would be impractical. To 
ensure that production samplers accurately meet the required 
specifications, the samplers must be manufactured in an ISO-9001 
registered facility, and the facility must be maintained in compliance 
with all applicable ISO 9001 requirements. The manufacturer must also 
conduct specific tests and submit supporting evidence to EPA 
demonstrating conformance to critical component specifications such as 
materials, dimensions, tolerances,

[[Page 6037]]

and surface finishes. In conjunction with final assembly and inspection 
requirements, field tests are used to demonstrate that the samplers 
meet required performance specifications.

List of Subjects in 40 CFR Part 50

    Environmental protection, Air pollution control, Carbon monoxide, 
Lead, Nitrogen dioxide, Ozone, Particulate matter, Sulfur oxides.

    Authority: Secs. 109 and 301(a), Clean Air Act, as amended (42 
U.S.C. 7409, 7601(a)).

    Dated: January 29, 1998.
Carol M. Browner,
Administrator.
[FR Doc. 98-2878 Filed 2-4-98; 8:45 am]
BILLING CODE 6560-50-P