EPA/600/R-14/308
                                        February 2014
  Environmental Technology
  Verification Report
Sunset Laboratory Model 4 OC-EC Field Analyzer
                   Prepared by

                  Baireiie
                The Business of Innovation

             Under a cooperative agreement with

                U.S. Environmental Protection Agency

-------
Environmental Technology Verification
                  Report

   ETV Advanced Monitoring Systems Center

Sunset Laboratory Model 4 OC-EC Field Analyzer
                     By

                  Kenneth Cowen
                  Thomas Kelly
                  Amy Dindal
                    Battelle
                Columbus, Ohio 43201

-------
                                            Notice
The U.S. Environmental Protection Agency, through its Office of Research and Development, partially
funded and collaborated in the research described herein. This report has been subjected to the Agency's
peer and administrative review.  Any opinions expressed in this report are those of the author(s) and do
not necessarily reflect the views of the Agency,  therefore, no official endorsement should be inferred. Any
mention of trade names or commercial products does not constitute endorsement or recommendation for
use.

-------
                                       Foreword

The EPA is charged by Congress with protecting the nation's air, water, and land resources.
Under a mandate of national environmental laws, the Agency strives to formulate and implement
actions leading to a compatible balance between human activities and the ability of natural
systems to support and nurture life. To meet this mandate, the EPA's Office of Research and
Development provides data and science support that can be used to solve environmental
problems and to build the scientific knowledge base needed to manage our ecological resources
wisely, to understand how pollutants affect our health, and to prevent or reduce environmental
risks.

The Environmental Technology Verification (ETV) Program has been established by the EPA to
verify the performance characteristics of innovative environmental technology across all media
and to report this objective information to permitters, buyers, and users of the technology, thus
substantially accelerating the entrance of new environmental technologies into the marketplace.
Verification organizations oversee and report verification activities based on testing and quality
assurance protocols  developed with input from major stakeholders and customer groups
associated with the technology area. ETV consists of six environmental technology centers.
Information about each of these centers can be found on the Internet at http://www.epa.gov/etv/.

Effective verifications of monitoring technologies are needed to assess environmental quality
and to supply cost and performance data to select the most appropriate technology for that
assessment. Under a cooperative agreement, Battelle has received EPA funding to plan,
coordinate, and conduct such verification tests for "Advanced Monitoring Systems for Air,
Water, and Soil" and report the results to the community at large. Information concerning this
specific environmental technology area can be found on the Internet at
http://www.epa.gov/etv/centers/centerl.html.

-------
                                   Acknowledgments

The authors wish to acknowledge the support of all those who helped plan and conduct the
verification test, analyze the data, and prepare this report.  Specifically the authors thank Andrea
Polidori of the South Coast Air Quality Management District, Joann Rice of EPA Office of Air
Quality Planning and Standards, and Zachary Willenberg, Battelle Quality Assurance Officer, for
their review of this verification report.
                                           in

-------
                                      Contents


Notice	i

Foreword	ii

Acknowledgments	iii

List of Figures	vi

List of Tables	vii

List of Abbreviations	viii

Chapter 1 Background	1

Chapter 2 Technology Description	2

Chapters Test Design and Procedures	5
  3.1 Introduction	5
  3.2 Test Procedures	7
  3.3 Field Site	8

Chapter 4 Quality Assurance/Quality Control	11
  4.1 Amendments/Deviations	11
  4.2 Reference Method	11
      4.2.1 Reference Method Sampling	12
      4.2.2 Reference Method Analysis	17
  4.3 Audits	21
      4.3.1 Performance Evaluation Audit	21
      4.3.2 Technical Systems Audit	23
      4.3.3 Data Quality Audit	24
  4.4 QA/QC Reporting	24
  4.5 Data Review	24

Chapters Statistical Methods	25
  5.1 Comparability	25
  5.2 Correlation	26
  5.3 Precision	26
  5.4 Data Completeness	26
  5.5 Operational Factors	26

Chapter 6 Test Results	27
  6.1 Comparability	30
      6.1.1 Regression Analysis	37
      6.1.2 Relative Percent Difference Analysis	41
                                          IV

-------
  6.2 Correlation	42
  6.4 Data Completeness	47
  6.5 Operational Factors	48
       6.5.1 Routine Maintenance	48
       6.5.2 Consumables/Waste Generation	48
       6.5.3 Ease of Use	48

Chapter 7  Performance Summary	49

-------
                                   List of Figures

Figure 2-1. Model 4 OC-EC field analyzer	4
Figure 3-1. Illustration of measurements of carbonaceous paniculate matter	6
Figure 3-2. Aerial photograph of test site and surrounding area	9
Figure 3-3. RAAS reference sampler installed at sampling site	9
Figure 3-4. BGIPQ200 reference samplers installed at sampling site	10
Figure 4-1. Duplicate Reference Method EC results from TOR analysis	13
Figure 4-2. Duplicate Reference Method EC results from TOT analysis	14
Figure 4-3. Differences between the duplicate reference method EC results from
            TOR analysis	15
Figure 4-4. Differences between the duplicate reference method EC results from
            TOR analysis	15
Figure 4-5. Regression of reference method results from TOR analysis by sampler type	16
Figure 4-6. Scatter plot of reference method results from TOT analysis by sampler type	16
Figure 6-1.  Measured 2-hour average thermal EC concentration from the duplicate
            Model 4 OC-EC analyzers during testing	27
Figure 6-2.  Measured 2-hour average corrected thermal  EC concentration from the
            duplicate Model 4 OC-EC  analyzers during  testing	28
Figure 6-3.  Measured 2-hour average optical EC concentration from the duplicate
            Model 4 OC-EC analyzers during testing	29
Figure 6-4. Calculated differences between the hourly average EC concentration
            from the duplicate Model Model 4 OC-EC analyzers during testing
            (Difference = SN 3218-3219 - SN 90)	29
Figure 6-5. Comparison of uncorrected thermal EC results from the
            Model 4 OC-EC analyzers and the mean reference method TOR EC concentrations.
            	31
Figure 6-6. Comparison of corrected thermal EC results from the Model 4 OC-EC
            analyzers and the  mean reference method TOR EC concentrations	32
Figure 6-7. Comparison of uncorrected thermal EC results from the Model 4 OC-EC
            analyzers and the  mean reference method TOT EC concentrations	33
Figure 6-8. Comparison of corrected thermal EC results from the Model 4 OC-EC
            analyzers and the  mean reference method TOT EC concentrations	34
Figure 6-9. Comparison of optical EC results from the Model 4 OC-EC analyzers
            and the mean reference method TOR EC concentrations	35
Figure 6-10. Comparison of optical EC results from the Model 4 OC-EC analyzers
            and the mean reference method TOT EC concentrations	36
Figure 6-11. Scatter plot of uncorrected Model 4 EC 12-hour thermal averages
            against mean reference method TOR results	37
Figure 6-12. Scatter plot of corrected Model 4 EC 12-hour thermal averages against mean
            reference method  TORresults	38
Figure 6-13. Scatter plot of uncorrected Model 4 EC 12-hour thermal averages
            against mean reference method TOT results	38
Figure 6-14. Scatter plot of corrected Model 4 EC 12-hour thermal averages against
            mean reference method TOT results	39
                                          VI

-------
Figure 6-15. Scatter plot of Model 4 EC 12-hour optical averages against mean reference
            method TOR results	40
Figure 6-16. Scatter plot of corrected Model 4 EC 12-hour optical averages against mean
            reference method TOT EC results	40
Figure 6-16. Regression of uncorrected Model 4 2-hour thermal EC measurements	44
Figure 6-17. Regression of corrected Model 4 2-hour thermal EC measurements	45
Figure 6-18. Regression of Model 4 2-hour optical EC measurements	45
Figure 6-19. Regression of uncorrected Model 4 12-hour thermal EC averages	46
Figure 6-20. Regression of corrected Model 4 12-hour thermal EC averages	46
Figure 6-21. Regression of Model 4 12-hour optical EC averages	47


                                   List of Tables

Table 4-1. Summary of Reference Method Field Blank Results	17
Table 4-2. Auto-Calibration Results of DRIEC/OC Analyzer	18
Table 4-3. Full Calibration Results used for DRI Analyzers	19
Table 4-4. Results of Periodic Calibration Checks	19
Table 4-5. Temperature Calibration Results	20
Table 4-6. System Blank Results	20
Table 4-7. Laboratory Blank Results	21
Table 4-8. Replicate EC Analysis Results	22
Table 4-9. Summary of Flow Rate PE Audit	23
Table 4-10.  Summary of Temperature and Pressure PE Audit	23
Table 6-1. Summary regression results of the Model 4 OC-EC analyzers and the
           reference method	41
Table 6-2. Summary of Regression Results of the Model 4 OC-EC Analyzers and the
           Reference Method, Excluding Apparent Outlier Data Points	41
Table 6-3. Summary of Relative Percent Difference between the Model 4 OC-EC Analyzers
           and the TOR Reference Method Results	42
Table 6-4. Summary of regression results of the Model 4 OC-EC analyzers and the
           reference method	43
Table 6-5. Summary of Calculated Precision Results of the Model 4 OC-EC Analyzers	43
Table 6-6. Summary of Unit-to-Unit Regression Results of the Model 4 OC-EC Analyzers	44
Table 6-7. Summary of data completeness for the Model 4 OC-EC analyzers	47
Table 6-8. Summary of maintenance performed on Model 4  OC-EC analyzers	48
Table 7-1. Summary of Verification Test Results for the Model 4 OC-EC	49
                                          Vll

-------
                                    List of Abbreviations
AMS        Advanced Monitoring Systems Center
BC          black carbon
CC>2         carbon dioxide
csv          comma separated variable
CV          coeffi ci ent of vari ati on
DL          detection limit
DQO        data quality objective
DRI         Desert Research Institute
EC          elemental carbon
EPA         U.S. Environmental Protection Agency
ETV         Environmental Technology Verification
FID         flame ionization detector
He          helium
IMPROVE   Interagency Monitoring of PROtected Visual Environments
KHP         potassium hydrogen phthalate
LAC         light absorbing carbon
LPM         liters per minute
MnC>2        mangenese oxide
Hg/m3        micrograms per cubic meter
|j,L          microliter
NDIR        non-dispersive infrared
NIST        National Institute of Science and Technology
OC          organic carbon
PE          performance evaluation
PM          particulate matter
ppm         parts per million
QA          quality assurance
QC          quality control
QMP        quality management plan
r2            coefficient of determination
RISC        reduced instruction set computer
RPD         relative percent difference
TOR         thermal optical reflectance
TOT         thermal optical transmittance
TSA         technical systems audit
                                         Vlll

-------
                                    Chapter 1
                                   Background


The U.S. Environmental Protection Agency (EPA) supports the Environmental Technology
Verification (ETV) Program to facilitate the deployment of innovative environmental
technologies through performance verification and dissemination of information. The goal of
the ETV Program is to further environmental protection by accelerating the acceptance and
use of improved and cost-effective technologies. ETV seeks to achieve this goal by providing
high-quality, peer-reviewed data on technology performance to those involved in the design,
distribution, financing, permitting, purchase, and use of environmental technologies.

ETV works in partnership with recognized testing organizations; with stakeholder groups
consisting of buyers, vendor organizations, and permitters; and with the full participation of
individual technology developers. The program evaluates the performance of innovative
technologies by developing test plans that are  responsive to the needs of stakeholders,
conducting field or laboratory tests (as appropriate), collecting and analyzing data, and
preparing peer-reviewed reports. All  evaluations are conducted in  accordance with rigorous
quality assurance and quality control (QA/QC) protocols to ensure that data of known and
adequate quality are generated and that the results are defensible.

The EPA's National Risk Management Research Laboratory and its verification organization
partner, Battelle, operate the Advanced Monitoring Systems (AMS) Center under ETV. The
AMS Center recently evaluated the performance of the Sunset Laboratory Model 4 OC-EC
Field Analyzer at an ambient air monitoring site in Columbus, Ohio.  Black carbon (BC)
monitors were identified as a priority technology category for verification through the AMS
Center stakeholder process.

-------
                                    Chapter 2
                             Technology Description

The objective of the ETV AMS Center is to verify the performance characteristics of
environmental monitoring technologies for air, water, and soil. This report provides results
for the verification testing of the Sunset Laboratory Model 4 OC-EC Field Analyzer. The
following is a description of the Model 4 OC-EC analyzer, based on information provided by
the vendor. The information provided below was not verified in this test.

Sunset Laboratory's semi-continuous Model 4 OC-EC Field Analyzer has been developed as
a field deployable alternative to integrated filter collection with subsequent laboratory
analysis. This instrument can provide time-resolved analyses of organic carbon (OC) and
elemental carbon (EC) in airborne particulate matter (PM) on a semi-continuous basis. A
quartz filter disc is mounted in the oven within the instrument, and samples are collected for
the desired time period.  Once the collection is complete, the oven is purged with helium, and
a stepped-temperature ramp based on the NIOSH 5040 reference method1 increases the oven
temperature to 840 °C, thermally desorbing organic compounds and pyrolysis products into a
manganese dioxide (MnO2) oxidizing oven. As the carbon fragments flow through the MnO2
oven, they are quantitatively converted to carbon dioxide (CO2) gas. The CO2 is swept out of
the oxidizing oven with the helium stream and measured directly by a self-contained non-
dispersive infrared (NDIR) detector system. The analyzer oven is cooled to 550 °C then a
second temperature ramp is initiated in an oxidizing (O2 in helium)  gas stream and  any
elemental carbon is oxidized off the filter and into the oxidizing oven and NDIR. The EC is
then detected in the same manner as the OC.

The Sunset Laboratory thermal/optical/transmittance (TOT) method uses the high light
absorbance characteristic of EC to correct for the pyrolysis-induced error. This is done by
incorporating a tuned diode laser (red, 660 nm), focused through the sample chamber such
that the laser beam passes through the mounted filter in the sample oven. Initial
transmission of the modulated laser beam is recorded. As the oven ramp proceeds, the laser
transmission is monitored continuously by the data system.  Any  charring of the OC results
in a decrease in transmission of the laser.  After the initial temperature ramp, when  the
helium carrier is switched to a He/O2 mixture, all of the EC is oxidized off and the laser
transmission returns to the background level. When the resulting NDIR data are reviewed
with an overlay of the laser absorbance, the point in the temperature ramp at which the laser
transmission equals the initial laser transmission is the split point. Any EC detected before
this point was formed pyrolytically by charring of the OC. This carbon  is subtracted from

-------
the EC area observed during the oxidizing phase of the analysis and is assigned as OC.  The
primary assumption for this correction is that the particulate bound EC and the pyrolytically
formed EC either have the same absorption coefficient, or else the pyrolytically formed EC
will oxidize earlier. Three characteristic components of the semi-continuous Model 4 OC-
EC analyzer are important in the strength of the analysis.  The first of these is the optical
detection and correction for EC. Elemental carbon is naturally present in many of these
samples from some combustion source such as a diesel exhaust. This black material is a very
strong absorber of light, and almost always the only absorber in the red light region. In
addition to this EC in the sample, EC can be formed from some charring of the OC fraction
of the sample as it is pyrolyzed during the initial temperature ramp. This can begin occurring
as low as 300 °C depending on the organic components on the filter. This charring of OC
could result in an artificially low measurement of the OC and a higher than actual
measurement for the original EC if no correction was made.

The second component of the Model 4 OC-EC analyzer is the use of a sensitive, linear
detector for this measurement.  Historically, this has been a flame ionization detector (FID).
Recently Sunset Laboratory has introduced its own self-contained flow-through NDIR
system.  It is a RISC processor controlled pseudo-dual beam instrument with temperature
stabilized source and  a linearized signal output. The vendor indicates that calibrations can be
used for up to a year.  NDIR eliminates the requirement for air or hydrogen at the field site.

The third important component of the Model 4 OC-EC analyzer is the incorporation of a
fixed volume loop used to inject an external standard at the end of every analysis. The
resulting external standard data are incorporated into every data package and used along with
the known carbon concentration in the loop to calculate the analytical results.

The Sunset Laboratory semi-continuous instruments monitor the sample deposition by
measuring the change in laser transmittance over the sample period. This change in
transmittance can be directly correlated to the deposition  of elemental carbon.  This  optically
derived elemental carbon  or "OptEC" is computed using  a proprietary equation to
compensate for the non-linear  effects of the highly scattering effects of the quartz filter. It is
assumed that  at the beginning of each sample collection period that there is no EC. The
absorbance of the sample  is determined from the transmittance where the initial laser is 100%
T and the final laser/initial laser as the end of the sample  collection is the final percent
transmittance. Absorbance is then calculated from the %T.

           OptEC (ugEC/cm2) = absA2 * X2 + abs*  Xi + X0

Multiplication by the deposit area (cm2) and division by the sample volume (M3) yields the
results in ugOptEC/M3.

-------
Figure 2-1. Model 4 OC-EC field analyzer

-------
                                     Chapter 3
                           Test Design and Procedures
3.1 Introduction
The ETV Program's AMS Center conducts third-party performance testing of commercially
available technologies that detect or monitor natural species or contaminants in air, water,
and soil.  Stakeholder committees of buyers and users of such technologies recommend
technology categories, and technologies within those categories, as priorities for testing.
Among the technology categories recommended for testing are "black carbon" monitors.
Because of the nature of BC, this technology category includes monitors for both BC and
EC.  Two stakeholders were selected to serve as peer reviewers for the quality assurance
project plan (QAPP)2 and this verification report. The responsibilities of verification test
stakeholders/peer reviewers included:

       •     Participate in technical panel discussions (when available) to provide input to
            the test design;
            Review and provide input to the QAPP; and
       •     Review and provide input to the verification report/verification statement.

The QAPP and this verification report were reviewed by experts in the fields related to black
carbon monitors. The following experts provided peer review:

       •     Andrea Polidori, South Coast Air Quality Management District
            Joann Rice, EPA.

The purpose of this verification test was to generate performance data on BC monitors so
organizations and users interested in installing and operating these systems can make
informed decisions about their use. Black carbon is a term that is commonly used to  describe
strongly light absorbing carbon (LAC), which is thought to play a significant role in  global
climate change through direct absorption of light, interaction with clouds, and by reducing
the reflectivity of snow and ice.  BC is formed from the incomplete combustion of fossil
fuels, biofuels, and biomass and can be emitted from both anthropogenic and natural sources.
It is a primary component of soot and has been linked to adverse health effects and visibility
reduction. Consequently, there is a great deal of interest in monitoring BC in the atmosphere.
However, differences in measurement techniques result in measurements that are
operationally defined and characterize the paniculate matter based on either its light
absorbing properties (leading to determination of BC) or its refractory properties (leading to

-------
determination of EC), as illustrated in Figure 3-1. In this figure, the use of the subscript a
denotes that the measurements are technique specific and result in estimations of BC or EC
that are "apparent" based on the technique being used.  The methods used to determine EC
are termed thermal-optical in Figure 3-1 because they involve conversion of particulate
carbon to gaseous form under varying temperatures and controlled atmospheres while the
parti culate sample is monitored by either transmission or reflection of light.


This verification test was conducted according to procedures specified in the peer-reviewed
ETV Quality Assurance Project Plan (QAPP)for Verification of Black Carbon Monitors.2
Deviations from this QAPP are described in Section 4.1 of this report.
      Light-Absorption Classification

        More
   light-absorbing
               Light-
              absorbing
               carbon
               (LAC)
                          Black
                          carbon
Brown  *
carbon
 (BrC)
        Less
    light-absorbing
                         Thermal-Optical Classification

                                     More refractory
                            Elemental
                             carbon
                                                      Organic
                                                      carbon
                                                       (OCJ
                                     Less refractory

                      * Measurement technique-specific split point
        Figure 3-1. Illustration of measurements of carbonaceous particulate matter.
                                 (Source: U.S. EPA)3

The performance of the Model 4 OC-EC analyzer was verified by evaluating the following
parameters:

    •  Comparability with collocated reference method results
    •  Correlation with collocated reference method results
    •  Precision between duplicate units
    •  Data completeness
    •  Operational factors such as ease of use, maintenance and data output needs, power
       and other consumables use, and operational costs.

-------
3.2 Test Procedures
The test was conducted at the Battelle Columbus Operations Special Support Site (BCS3)
located at 2555 International St., in Columbus, OH. For this test, duplicate Model 4 OC-EC
monitors continuously sampled ambient air for approximately 33 days, during which period
filter samples were collected on thirty days to use as a basis of comparison for the analyzers
being tested.  Specifically, during each test period duplicate integrated filter samples were
collected over successive  12-hour periods each day (except as noted below) using
commercially available air sampling equipment.  The duplicate reference samples were
collected from 7:00 am to approximately 6:50 pm and from 7:00 pm to approximately 6:50
am daily. From April 5 through April 25, a single four-channel Anderson Model RAAS-400
speciation sampler was used to collect the duplicate reference filter samples using separate
channels  of the sampler.  Samples were not collected during one of the planned 12-hour
sampling periods (beginning on the evening of April 16) because of adverse weather
conditions. On April 25, a failure in the electronics of the RAAS sampler rendered it
inoperable and the lack of replacement parts resulted in the RAAS being inoperable for the
remainder of the test period. Duplicate BGI Model PQ200 samplers were immediately
available and were used in replacement of the RAAS for the collection of the reference
samples from April 26 to May 7.

The reference samples were collected on pre-cleaned quartz fiber filters at a nominal flow
rate of 16.7 LPM with both the RAAS and BGI samplers.  The change in samplers was not
expected to result in any differences in the measured EC concentrations since the samples
were collected at the sample flow rate and with similar PM2 5 size selective inlets.  Actual
differences that may have been observed are likely the result of differences in the calibrated
flow rates of the samplers. Over the 3 3-day field period, a total of 118 filter samples were
successfully collected (i.e., duplicate samples during 59 of the 60 12-hour sampling periods).

The reference samples were analyzed for EC by Desert Research Institute (DRI) using the
DRI Model 2001 EC-OC analyzers implementing the Interagency Monitoring of PROtected
Visual Environments (IMPROVE) thermal/optical reflectance (TOR) method, which
monitors the filter sample by means of optical reflectance.4 DRI also reported EC results
from the IMPROVE method with the filter monitored by thermal/optical transmission
(TOT).4 Results from the Model 4 OC-EC analyzers were compared to these filter sample
results to assess the comparability of the Model 4 OC-EC analyzer results to the filter sample
results.

The precision of the Model 4 OC-EC analyzers was determined from comparisons of paired
data from the duplicate units. Other performance parameters such as data completeness,
maintenance requirements, ease of use,  and operational costs were assessed from
observations by the Battelle field testing staff.  This test was not intended to simulate long-
term (e.g., multi-year) performance of BC monitors. As such, performance and maintenance
issues associated with long-term use of the Model 4 OC-EC analyzers are not addressed in
this report.

Note that in this report the filter samples will be referred to as "reference  samples."
However, it should be noted that the IMPROVE method is not a true "Reference Method" in

-------
that it is not recognized as an absolute standard. Nonetheless, it is used within the
IMPROVE network as the standard method for EC analysis. Thus the method was used in
this test as an analytical technique used for comparison to the BC monitors.  Other
thermal/optical reference methods such as the NIOSH 5040 method may result in different
results.

For the verification test, duplicate Model 4 OC-EC analyzers were installed inside an
environmentally controlled instrument trailer at the BCS3 ambient air monitoring site in
Columbus,  OH. The two Model 4 OC-EC units were installed by the vendor and were
operated continuously over the 33-day testing period. Trained Battelle testing staff
performed periodic maintenance activities on the analyzers.  These activities were
documented and are reported in Section 6.5 of this report. The two analyzers collected
ambient air through a common inlet protected by a rain guard.  The RAAS-400 and the
duplicate PQ200 samplers were installed on the platform such that each inlet was more than
one meter from the inlet. The sample air flow passed through an organic vapor denuder, and
then was split in two to pass through separate PM2.s cyclone separators before entering  the
analyzers. The analyzers were programmed on a two hour measurement cycle which
included 108 minutes of sample collection and  12 minutes of thermal sample analysis using
the NIOSH 5040 analysis profile. The cycles were timed to allow synchronization with the
12-hour reference method sampling periods. These two-hour cycle thermal measurements
for OC and EC were reported by the two Model 4 OC-EC analyzers corrected to standard
temperature and pressure (25 °C and  1 atmosphere), along with minute-by-minute optical EC
measurements made by the two analyzers.
3.3 Field Site
Figure 3-2 shows an aerial photograph of the BCS3 facility test site (red marker "A") and the
surrounding area.  The test site is located approximately /^ mile north of a rail yard and in the
vicinity of multiple industrial and shipping facilities which result in frequent truck traffic past
the site.  The site also receives regionally transported air pollution due to its location on the
western side of the Columbus metropolitan area.  An environmentally controlled mobile
laboratory was installed at the site to serve as a shelter for the Model 4 OC-EC analyzers and
as work space for the testing staff.  Figures 3-3 and 3-4 show the sampling trailer with RAAS
sampler and BGI samplers, respectively, installed on a platform next to the trailer.

-------
Figure 3-2. Aerial photograph of test site and surrounding area.
Inlet
 Figure 3-3.  RAAS reference sampler installed at sampling site.

-------
Figure 3-4. BGI PQ200 reference samplers installed at sampling site.
                              10

-------
                                    Chapter 4
                      Quality Assurance/Quality Control
QA/QC procedures and all verification testing were performed in accordance with the QAPP
for this verification test1 and the quality management plan (QMP) for the AMS Center4
except where noted below.  QA/QC procedures and results are described below. Other than
an initial flow check during installation of the Model 4 OC-EC analyzers no additional
QA/QC activities were performed on the Model 4 OC-EC analyzers. Maintenance activities
performed on the Model 4 OC-EC analyzers is included in Section 6.5.1.

4.1 Amendments/Deviations

Two deviations to the test/QA plan were prepared, approved, and retained in the test
documentation. The deviations established the following modification to the test/QA plan
and the test procedures:

   •   The RAAS speciation sampler was replaced with duplicate BGIPQ200 samplers
       approximately halfway through the sampling period because of a failure in the
       RAAS sampler. This change did not adversely affect the data quality since the new
       samplers functionally performed the same as the RAAS samplers. See Section 4.2.1
       for a comparison of the results from the two sampler types.
   •   Routine flow checks were not performed as specified in the test plan for a portion of
       the test period.  This deviation is not expected to adversely affect the results since
       subsequent leak checks met the acceptance criteria and no systemic bias was observed
       in the reference results.
4.2 Reference Method

The following sections describe the QA/QC procedures employed in the collection and
analysis of reference samples. Only the results for the EC analyses are presented since OC
results are not used to evaluate the performance of the Model 4 OC-EC Analyzers.
                                         11

-------
4.2.1 Reference Method Sampling
This verification test included a comparison of Model 4 OC-EC Analyzers results to those of
the reference measurements described in Section 3.2. Figures 4-1 and 4-2, respectively, show
the results of the TOR and TOT measurements of the duplicate reference samples collected
during the testing period.  In these figures, one of the two channels used in the RAAS for the
collection of the reference samples was designated as the "primary sample" and the other
channel was designated as the "collocated sample" to show potential biases between the two
channels.

Similarly, one of the BGI  samplers was designated as the "primary sample" and the other as
the "collocated sample." Figures 4-3 and 4-4 show the corresponding differences between the
primary and collocated reference sample measurements and indicate that there is no clear
systematic bias observed in either the TOR or TOT results for either the RAAS sampler or
the duplicate BGI samplers.

During the verification test the mean of the EC measurements for the duplicate reference
samples ranged from <1.5 to 32.9 micrograms per filter (|j,g/filter) which corresponds to
airborne concentrations ranging from <0.13 to 2.7 micrograms per cubic meter (|j,g/m3) based
on the sample volumes of-12 m3.  According to DRI's documentation, at concentrations of
greater than 10 times the method detection limit (10 x MDL -1.5 |j,g/filter) the expected
precision between duplicate samples is -  10%.3  In general, the majority of the reference
samples had EC concentrations that were  below 10 times the MDL, consequently the percent
difference between the duplicate samples  was typically greater than 10%. Reference results
that were reported as less than the MDL were assigned a value of l/2 MDL for this report.
Over all results with both  reference sampler types, the calculated percent differences (i.e., the
difference between the two duplicate results divided by their mean) ranged from -98% to
109%, with an average of 3%. The EC concentrations measured from the filters collected
with the RAAS sampler ranged from <0.13 to 1.1 ng/m3, with a mean concentration of 0.35
± 0.28 ng/m3. The average percent difference between the duplicate samples collected with
the RAAS was 21% ± 71%. The EC concentrations measured from the filters collected with
the BGI PQ200 samplers ranged from <0.13 to 2.7 ng/m3,  with a mean concentration of 0.64
± 0.62 ng/m3. The average percent difference between the duplicate samples collected with
the BGI samplers was 1% ± 50%.

Figure 4-5 shows a scatter plot of the TOR results for the reference method samples
indicating in which sampler type used to collect the filters  (e.g.,  RAAS and BGI PQ200).
Figure 4-6 presents a  similar plot for the TOT analyses.  These figures show that the TOT
results exhibit a slope closer to 1.0, an intercept  closer to zero, and a higher r2 value, relative
to the TOR results.

During testing a total  of 12 reference method field blank samples were collected,
representing 10% of the total reference method samples. The field blanks were installed in
the filter cassettes and loaded into the reference method samplers without drawing air
through the filters.  Table  4-1 presents a summary of the field blank results including the
results for the EC measurements for both  the TOR and TOT analyses by the DRI Model 2001
analyzer.  Table 4-1 shows that in all cases the measured EC on the reference method blank
filters was below the detection limit of 1.5 |j,g/filter.

                                          12

-------
CO


"B)


C
o
"•p
re
o
o
o
LLI
•a
o
0)
o
c
0)
a:
2 -
1 -
   0
• Primary Sample-RAAS

0 Primary Sample-BGI

• Collocated Sample-RAAS

E Collocated Sample-BGI
              I


      CO
                    CO


                    CNJ
                            CO

                            o5
CO


CD
CNI
CO
                                           Date

                Figure 4-1. Duplicate Reference Method EC results from TOR analysis.
                                         13

-------
CO



1
C
o
"•p
re
 0)
a:
2 -
 o
 c
 o
o

O
LLI


 O 1 -
 0)
 o
 c
   0
• Primary Sample-RAAS


0 Primary Sample-BGI


• Collocated Sample-RAAS


a Collocated Sample-BGI
                                       AJA
                                      llu
                                                                             3J3l
                                                        CO



                                                        CD
                                                        CNI

                                                        ^
                                          Date


                Figure 4-2. Duplicate Reference Method EC results from TOT analysis.
                                                                     CO



                                                                     £?
                                                                     i?5
                                         14

-------
       o
      is
       o
      o
      o
      £
u. j -

0.3 -
0.1 -
4/:
-0.1 -
_n R -




|





Jl
rl* ' ill
nsl





4/12/13
• Difference-RAAS
• Difference-BGI


I










1 I
I il. n
J/19/T3 I 4/26/1 1

                                              Date


Figure 4-3. Differences between the duplicate reference method EC results from TOR analysis
          0.5
       g
       is
       o
       o
       o
       o
       LU

-------
         a.
         ns
         w
         u
         •s
         o
         .o
         "o
         o
         c
         o
         •s
         o
         o
         o
         o
         o
   2 -
                        • RAAS
                        • BGI PQ200
                                      BGI PQ200
                                    y = 0.899x+ 0.041
                                       R2 = 0.920
                                      N = 21
                                                            RAAS
                                            y = 0.706x + 0.067
                                               R2 = 0.608
                                                N = 38
                                    1                  2                 3
                       EC Concentration - Primary Sample (jig/m3)

Figure 4-5.  Regression of reference method results from TOR analysis by sampler type.
        a.
        ns
        w
        o
        o
        o
        +J
        ns
        O
        o
        o
        LU
2 -
            1 -
                   »RAAS
                   • BGI PQ200
                                 BGI PQ200
                               y = 0.957x +0.008
                                  R2 = 0.946
                                   N = 21
                                           RAAS
                                       y = 0.785x+0.033
                                          R2 = 0.679
                                          N = 38
              0123
                     EC Concentration - Primary Sample (p,g/m3)

Figure 4-6. Scatter plot of reference method results from TOT analysis by sampler type.
                                        16

-------
Additionally, flow rate checks of the reference samplers were performed at least every three
days during the sampling period to ensure that the reference samplers were operating within
5% of their nominal flow rate.

            Table 4-1. Summary of Reference Method Field Blank Results
Filter ID
BTOQ001
BTOQ002
BTOQ011
BTOQ055
BTOQ072
BTOQ073
BTOQ074
BTOQ075
BTOQ090
BTOQ097
BTOQ109
BTOQ110
EC (ng/filter)
TOR
0.00
0.00
0.05
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.10

TOT
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
4.2.2 Reference Method Analysis

Routine calibrations of the DRI Model 2001 carbon analyzers used to analyze the reference
samples were performed at the beginning and end of each day by injecting known volumes of
either CH4 or CC>2 with nominally the same amount of carbon (approximately 21.4 |j,g) and
comparing the resulting OC3 and EC1  measurements4. The acceptable level is ±5%
difference between peaks injected during the OC3 and EC1 temperature step.  Table 4-2
presents a summary of these routine calibrations for each of the carbon analyzers (identified
as uniquely numbered CA units below) and the calculated percent differences between
carbon measurements from the OC3 and EC1 measurements for each calibration.
Exceedences of the acceptance criterion require recalibration of the analyzer.
                                        17

-------
             Table 4-2. Auto-Calibration Results of DRIEC/OC Analyzer.

Date
4/23/2013
5/17/2013
5/18/2013
5/20/2013
5/21/2013


4/23/2013
5/17/2013
5/18/2013
5/20/2013
5/21/2013


4/23/2013
5/17/2013
5/18/2013
5/20/2013
5/21/2013
OC3 (jig)
CA#6
21.80



19.46
CA#7
20.90

22.75


CA#8
22.35
20.10
20.31
20.27

CA#9
20.22
22.33
22.37
22.29
22.24
CA#10
21.67



21.15
CA#11
21.08
21.89
21.87

21.85
CA#12
21.06
21.10
21.17
21.01
21.00
CA#13
22.59




CA#16
17.67
22.50
22.37
22.39
22.41
EC1 (ng)
CA#6
21.11



20.10
CA#7
20.53

22.31


CA#8
22.02
20.26
20.09
20.25

CA#9
20.78
22.04
22.00
21.95
22.00
CA#10
21.29



21.70
CA#11
20.91
21.27
21.17

21.24
CA#12
20.94
21.00
20.82
20.94
20.92
CA#13
21.40




CA#16
19.48
21.43
21.40
21.46
21.52
%Diff
CA#6
3.3%



-3.2%
CA#7
1.8%

2.0%


CA#8
1.5%
-0.8%
1.1%
0.1%

CA#9
-2.7%
1.3%
1.7%
1.5%
1.1%
CA#10
1.8%



-2.5%
CA#11
0.8%
2.9%
3.3%

2.9%
CA#12
0.6%
0.5%
1.7%
0.3%
0.4%
CA#13
5.6%




CA#16
-9.3%
5.0%
4.6%
4.3%
4.1%
 CA#16 was taken offline on 4/23/2013 for FID and laser issues.
fMissing values indicate the instruments were offline during the day.

Full instrument calibrations are performed semiannually or after major maintenance or
repairs and are used to establish the calibration slope used in converting CC>2 detector counts
to ug of carbon. Four types of standards are used for full instrument calibration: 5% nominal
methane (CFLi) in helium (He), 5% nominal CC>2 in He, potassium hydrogen phthalate (KHP),
and sucrose. Instrument calibration involves spiking pre-fired quartz punches at four different
levels from 5.0 to 20.0 microliters (ul) of the 1,800 parts per million (ppm) KHP and sucrose
solutions and injecting CH4 and CC>2 gases at four levels from 200 to 1,000 uL. The
calibration slopes derived from the two gases and the KHP- and sucrose-spiked filter punches
are averaged together to yield a single calibration slope for a given analyzer. This slope
represents the response of the entire analyzer to generic carbon compounds and includes the
efficiencies of the oxidation and methanator zones and the sensitivity of the FID.  Table 4-3
presents a summary of the recent full calibrations of the analyzers used to analyze the
reference samples. Note that the treatment of the calibration data ensures that the data passes
through the origin, so no intercept is presented. The calculated slopes are compared to
previous calibration results and  should be with 10% of previous calibrations if no major
changes to the instrument have been made.  If the differences in the slope exceed 10%, the
calibration must be repeated to verify the results.
                                          18

-------
Table 4-3. Full Calibration Results used for DRI Analyzers
Analyzer
CA#6
CA#7
CA#7
CA#7
CA#8
CA#8
CA#9
CA#10
CA#10
CA#10
CA#11
CA#12
CA#13
CA#16
Date
4/3/2013
5/7/2013
4/26/2013
10/11/2012
5/13/2013
12/7/2012
4/18/2013
5/21/2013
5/4/2013
3/29/2013
3/14/2013
4/1/2013
11/19/2012
3/10/2013
Slope
21.482
19.436
22.000
20.513
20.094
20.483
22.040
21.457
23.228
20.502
21.320
20.904
20.818
22.685
r2
0.991
0.976
0.996
0.986
0.981
0.987
0.988
0.982
0.996
0.991
0.987
0.989
0.985
0.987
The instrument calibration was verified several times a week using sucrose and KHP
standards near the midpoint of the calibration curve (18 ugC). Table 4-4 presents a summary
of the calibration checks performed.  In all cases the agreement between the measured and
standard concentration was within the ±1 ugC acceptance criterion.

                   Table 4-4. Results of Periodic Calibration Checks
Date
4/23/2013
5/17/2013
5/20/2013
5/21/2013
CA#6 CA#8
18.718 17.729
17.433
17.505
17.739
CA#9
18.441
17.783
17.447
17.523
CA#10
17.450


17.223
CA#11
18.119
18.149
18.439
17.332
CA#12
17.876
17.688
17.692
17.402
CA#13 CA#16
18.071 17.761
17.749 18.104
18.595
18.147
       fMissing values indicate the instruments were offline during the day.

Temperature calibrations are performed at least semiannually on all instruments to verify
that the sample temperature is as accurate as possible. Quick-drying temperature-indicating
liquids of different melting points, Tempilaq°G (Tempil, Inc., South Plainfield, NJ, USA),
were used as temperature indicators. Temperature indicators of 121, 184, 253, 510, 704, and
816 °C were chosen for the IMPROVE_A protocol temperature calibration. The accuracy of
Tempilaq°G is certified within 1% of its designated temperature and is traceable to the
National Institute of Standards and Technology (NIST). Table 4-5 shows the results of the
most recent temperature calibrations of the analyzers used to analyze the reference samples.
In all cases the linear relationship between the thermocouple and standard Tempilaq°G values
met the r2 > 0.99 acceptance criterion.
                                         19

-------
                     Table 4-5. Temperature Calibration Results

CA#6
CA#6
CA#7
CA#7
CA#8
CA#8
CA#9
CA#9
CA#10
CA#11
CA#12
CA#13
CA#13
CA#16
Date
11/26/2012
5/16/2013
4/29/2013
6/3/2013
12/7/2012
5/13/2013
4/5/2013
4/22/2013
2/19/2013
3/15/2013
2/25/2013
11/19/2012
5/21/2013
3/11/2013
Slope
1.016
1.027
1.041
1.01
1.018
1.037
1.012
1.026
1.015
1.019
1.02
1.012
0.995
1.012
Intercept
8.3
19.0
8.9
3.0
5.8
7.6
8.7
1.8
8.0
11.7
7.1
9.7
12.7
11.2
r2
0.999
0.999
0.991
1.000
1.000
0.997
0.999
0.999
0.999
0.999
1.000
0.999
0.999
1.000
System blanks were performed once a week without filter punches in the analyzer to
determine the instrument baseline. Calculated carbon concentrations from the system blank
should not be more than 0.2 ug carbon. Table 4-6 presents a summary of the EC system
blank results for the analyzers used to analyze the reference samples. Table 4-6 shows that
the great majority of the EC system blanks showed no detectable carbon, and all EC blanks
easily met the 0.2 ugC acceptance criterion.
                           Table 4-6. System Blank Results
EC (ug)

4/28/2013
5/5/2013
5/12/2013
5/19/2013
5/26/2013
CA#6
0.000
0.000
0.002

0.004
CA#8
0.000
0.000

0.000
0.000
CA#9
0.003
0.000

0.000
0.000
CA#10


0.068

0.000
CA#11
0.000
0.000
0.000
0.000
0.000
CA#12
0.000

0.000
0.000
0.000
CA#13
0.000
0.000
0.000


CA#16

0.000
0.000
0.000

 System blanks on 5/12/2013 were high due to instruments being idle for the weekend and subsequent
laboratory blank checks indicated normal conditions.
Laboratory blanks were performed daily to check for system contamination and evaluate
laser response. If total carbon exceeded 0.2 ugC, values were voided and additional
laboratory blanks were run after performing the oven bake procedure until the system is clean
(i.e., OC < 0.2 ug C/cm2 and no EC). Analyzers exceeding the limit for laser drift,
reflectance, transmittance, total carbon, and calibration peak area after three laboratory blank
                                          20

-------
runs must be taken offline for maintenance. Table 4-7 presents the results of the laboratory
blanks for the analyzers used to analyze the reference samples. One of the analyzers (CA#7)
repeatedly failed the blank check and was taken offline on 5/18/2013 for maintenance.

                        Table 4-7. Laboratory Blank Results
EC (ng)

4/23/13
5/17/13
5/18/13
5/20/13
5/21/13
CA#6
0.000


0.063
0.000
CA#7


0.017
0.005
0.000
CA#8
0.000
0.000
0.000
0.000
0.011
CA#9
0.
0.
0.
0.
0.
.000
.000
.000
.000
.000
CA#10
0.000
0.000

0.000
0.012
CA#11
0.
0.

0.
0.
.000
.000

.000
.000
CA#12
0.000
0.000
0.000
0.003
0.021
CA#13 CA#16
0.000 0
0.000 0
0
0
0
.000
.000
.000
.000
.005
       CA#7 was taken offline for 5/18/2013.

Replicates of analyzed samples were performed at the rate of one per group often samples.
The replicate was selected randomly and run immediately after each group often was
completed. The random analyzer for the replicate was identified using a chart created in
Microsoft Excel using the random number generator, which results in replicate analysis on
the same and different analyzers. The ug/cm2 values for EC were compared with the original
run for both the TOR and TOT analysis. Precision was determined from replicate
measurements as the average fractional difference between original and replicate analysis
concentrations.  Concentration uncertainty is the fractional precision times sample
concentration. If sample concentration times fractional precision is zero, then the detection
limit is used as concentration uncertainty. Table 4-8 shows the results of the replicate
analyses. The results of the replicate analyses ranged from 0% to 52% for the TOR analysis
and from 0% to 36% for the TOT analysis. In general, the percent difference exceeded the
goal of 15% for the majority of the duplicate analyses indicating a lower degree of data
quality than desired.
4.3 Audits

Three types of audits were performed during the verification test: a performance evaluation
(PE) audit of the reference method sampling, a technical systems audit (TSA) of the
verification test performance, and a data quality audit. Audit procedures are described further
below.

4.3.1 Performance Evaluation Audit
A PE audit of the RAAS reference method sampler was performed by measuring the sample
flow rate through the two channels used for collection of the reference samples.  The flow
rate through each channel was measured using a NIST-traceable flow transfer standard
(BIOS DryCal, Serial No. 103777). After installation of the BGI PQ200 samplers, the flow
rates of those samplers were audited using a BGI DeltaCal calibrator (Serial No. 001255).
The results of those checks are summarized in Table 4-9, and indicate that the sampler flow
rates were well within the target ±5% tolerance of the nominal 16.7 L/min flow rate.
                                         21

-------

Table 4-8. Replicate EC Analysis
Results
Runl
Filter ID
BTOQ037
BTOQ041
BTOQ107
BTOQ114
BTOQ133
BTOQ086
BTOQ055
BTOQ057
BTOQ081
BTOQ009
BTOQ078
BTOQ095
TOR
6.97
7.77
2.55
3.81
4.37
4.33
0.00
5.22
8.21
3.44
4.76
8.68
TOT
4.55
6.06
1.89
3.00
2.43
2.52
0.00
4.17
6.09
1.99
3.30
6.91
Run 2
BTOQ037
BTOQ041
BTOQ107
BTOQ114
BTOQ133
BTOQ086
BTOQ055
BTOQ057
BTOQ081
BTOQ009
BTOQ078
BTOQ095
5.85
6.40
1.50
4.38
3.89
3.47
0.00
3.96
6.96
3.27
6.16
7.50
3.76
5.21
1.46
3.93
2.50
1.76
0.00
3.23
4.43
2.03
3.81
6.08
Percentage Diff.
BTOQ037
BTOQ041
BTOQ107
BTOQ114
BTOQ133
BTOQ086
BTOQ055
BTOQ057
BTOQ081
BTOQ009
BTOQ078
BTOQ095
17.5%
19.3%
52.1%
14.0%
11.8%
22.1%
0.0%
27.3%
16.5%
5.2%
25.6%
14.7%
19.0%
15.1%
25.8%
26.9%
3.1%
35.5%
0.0%
25.3%
31.6%
2.1%
14.2%
12.7%
22

-------
                     Table 4-9. Summary of Flow Rate PE Audit
Date
4/5/13
4/26/13
Reference Sampler
RAAS - Channel 1
RAAS - Channel 4
BGI - Sampler 1
BGI - Sampler 2
Measured
Flow
(L/min)
16.63
16.67
16.82
16.83
Difference from
Nominal
-0.6%
0.0%
0.9%
1.0%
Additionally, the temperature and pressure sensors of the reference samplers were audited
using NIST-traceable transfer standards. A summary of those audit results are shown in
Table 4-10 and indicate that both the temperature and pressure sensors in the reference
samplers were within the acceptance criteria for the verification test (i.e., ± 2°C for
temperature, and ± 5 mmHg for pressure).

            Table 4-10. Summary of Temperature and Pressure PE Audit
Date
4/5/13
4/8/13
4/26/13
Reference
Sampler
RAAS
RAAS
BGI - Sampler 1
BGI - Sampler 2
Sampler
Temp. (°C)
15.3
12.6
13.0
Audit Temp.
15.6
12.1
12.8
Sampler
Pressure
(mmHg)
736
749
748
Audit
Pressure
(mmHg)
736
749
749
4.3.2 Technical Systems Audit
A Battelle QA Officer performed one ISA as part of this verification test. The TSA was
performed at the BCS3 site in Columbus, OH. The TSA focused on observation of the
reference method sampling and field QA/QC procedures in preparation for the field test.  The
purpose of the audit was to ensure that the verification test was being performed in
accordance with the AMS Center QMP, and the test/QA plan for this verification test. In the
audit, the Battelle QA Officer observed the reference method sampling and sample recovery,
compared the actual test procedures being performed to those specified or referenced the
test/QA plan,  reviewed data acquisition and handling procedures, inspected documentation of
reference sample chain of custody, performance of flow, pressure, and temperature PE audits,
and reviewed test record books.  One finding and five observations were noted requiring two
deviations.  The first deviation pertained to leak checks not occurring at the recommended
frequency after each flow check. The VTC started performing leak checks after each flow
check.  It is not expected that the failure to conduct the leak checks had a substantial impact
on the results  since subsequent leak checks passed the acceptance criteria and there were no
systemic biases between the reference results prior to conducting regular leak checks. The
second deviation was to address the change from the Anderson RAAS to the two BGI PQ
200 samplers. While  not a finding, this deviation was necessary to document the change in
                                         23

-------
samplers and did not impact the quality of results. The remaining observations noted were
minor and did not impact the quality of results.

4.3.3 Data Quality Audit
At least 10% of the data acquired during the verification test were audited. Battelle's Quality
Manager traced the data from the acquisition, through reduction and statistical analysis, to
final reporting, to ensure the integrity of the reported results. All calculations performed on
the data undergoing the audit were checked and QC results verified. Corrections from the
other vendor report were made to the Magee data prior to QA review,  resulting in minor
comments. A few data issues were noted in the data quality audit, with minimal to no effect
on the overall quality of the verification results.

4.4 QA/QC Reporting

Each audit was documented in  accordance with Sections 3.3.4 and 3.3.5 of the QMP for the
ETV AMS Center.6  The results of the audits were submitted to the EPA.

4.5 Data Review

All data received from DRI for the reference measurements underwent 100% review and
validation by Battelle technical staff before being used for any statistical calculations.  This
review included a review of the data files containing the measured EC values from the
individual thermal steps for each filter and tracing of the calculated total EC  measurements
for both the TOR and TOT methods.  The Model 4 OC-EC Analyzer data were reviewed to
ensure that the minute by minute data were appropriately averaged into hourly and 24-hour
values. Based on review of Model 4  OC-EC Analyzer data files and operator logs, a small
number of 2 hour measurements were missing because of instrument maintenance activities.
Those data are detailed in Section 6.5.1. All reference data were found to be valid and were
included in the data  analysis.

Records generated in the verification test received a one-over-one review (e.g., review by
staff not involved in the generation of the record, but with at least the  same technical
expertise as the person generating the record) before these records were used to calculate,
evaluate, or report verification results. Data were reviewed by a Battelle technical staff
member involved in the verification test. The person performing the review added his/her
initials and the date to a hard copy of the record being reviewed.
                                         24

-------
                                    Chapter 5
                               Statistical Methods
The statistical methods used to evaluate the quantitative performance factors listed in Section
3.1 are presented in this chapter.
5.1 Comparability

The Model 4 OC-EC analyzers were evaluated for comparability in two ways.  Firstly,
comparability was determined from a linear least squares regression analysis of the measured
EC concentrations from the Model 4 analyzers against the corresponding mean EC results
from the duplicate reference samples.  Separate analyses were performed for the thermal EC
and optical EC results of the Model 4 analyzers. For comparison to the reference results,
average concentrations from each of the analyzers were determined separately for each of the
12-hour sampling periods, by averaging the monitor's individual 2-hour results over the
corresponding sampling period. The analyzers that were tested were set such that the
measurement times were synchronized with the reference measurements to the extent
possible. The 12-hour EC averages from the two Model 4 OC-EC analyzers were plotted
against the mean of the corresponding duplicate reference method EC measurements.  The
slope and intercept of these plots were determined from a linear regression analysis and
reported independently for each of the monitors.


Additionally, comparability was determined in terms of the relative percent difference (RPD)
between the mean value of the reference measurements and the results from each Model 4
OC-EC  analyzer tested. The RPD was calculated using Equation 1:
                                                    .100                         (1)
where:              Ct is the average EC concentration measured by the analyzer during
                    the ith reference sampling period, and
                                         25

-------
                     C(?ef}i  is the mean of the duplicate reference method EC
                    concentrations for the /'th reference sampling period.
5.2 Correlation
The degree of correlation of the results from each Model 4 OC-EC analyzer to the reference
method results was determined based on the coefficient of determination (r2) of the linear
regression performed to assess comparability (Section 5.1).  Correlation was determined
separately for each unit of each analyzer undergoing testing, and relative to the results from
the reference method.
5.3 Precision
Precision (P) was determined based on a comparison of paired measurements from the
duplicate analyzers being tested.  For this assessment of precision, the P between the paired
measurements from the duplicate analyzers was calculated using Equation 2:
where C(l), and C(2), are the EC concentrations measured by the first and second of the two
duplicate analyzers. Precision was calculated for the duplicate analyzers for each reference
sampling period, and the overall mean precision was also calculated. For this calculation,
measurement data below the vendor's stated instrumental detection limit was excluded from
the analysis.

5.4 Data Completeness
Data completeness was assessed in two ways, based on the overall data return achieved by
each analyzer during the testing period. First, for each of the analyzers data completeness
was calculated as the total hours of apparently valid data reported by the monitor divided by
the maximum total possible hours  of monitoring data in the entire field period. Also, for
each analyzer data completeness was calculated as the percentage of 12-hour reference
method sampling periods in which the analyzer provided at least 9 hours of valid data (75%).
The causes of any substantial incompleteness of data return were established from operator
observations or vendor records, and noted in the discussion of data completeness results.

5.5 Operational Factors
Operational factors such as maintenance needs, data output, consumables used, ease of use,
repair requirements, etc., were evaluated based  on observations recorded by Battelle staff,
and explained by the vendor as needed. A laboratory record book was maintained at the test
site, and was used to enter daily observations on these factors.  Examples of information
recorded in the record book include the daily status of units, maintenance performed, and
observations recorded during the installation and removal of the units.
                                          26

-------
                                     Chapter 6
                                    Test Results
Figure 6-1 shows the results of the 2-hour thermal EC measurements from the duplicate
Model 4 OC-EC analyzers for the verification testing period. Investigation of the raw data
files indicated that beginning on 4/28/13 at 15:00 (indicated by the vertical dashed line) and
continuing through the end of the test period, an anomalous spike in the optical reference
signal occurred in the raw data for one of the two duplicate analyzers being tested (RT3219).
This spike occurred consistently in the time segment between 238 and 258 seconds of the
thermal analysis.  This reference signal is measured at a wavelength where no CC>2
absorption occurs and serves as a baseline for correction of the measurement channel signal.
The recurring spike in the reference channel resulted in a positive bias in the measured
thermal EC concentrations for that analyzer only, which  can be observed in Figure 6-1 as a
divergence of the traces from the two analyzers over the  last 9 days of the test.
            5
            4/5/13
4/12/13
4/19/13
4/26/13
5/3/13
                                            Date
 Figure 6-1.  Measured 2-hour average thermal EC concentration from the duplicate Model 4
                            OC-EC analyzers during testing.
                                         27

-------
Prior to that time period, the 2-hour EC results from the two analyzers were more closely in
agreement.

The cause of this anomaly in the RT3219 reference signal could not be identified or
reproduced once the analyzers were returned to the vendor facility.  To correct this anomaly,
the vendor replaced the values in the raw reference channel data file with the average of the
values recorded from 234 to 237 seconds (i.e., just prior to the spike) in the same file. This
correction was done only for the one analyzer that exhibited the anomaly (RT3219).  Both
the raw and corrected values for the thermal EC measurements are presented in this report
along with the optical EC measurements recorded by the analyzer. Figures 6-2 and 6-3  show
the corrected thermal EC data and the optical EC data for the two analyzers.  Figure 6-2
shows that the correction applied to the RT3219 analyzer results in improved agreement of
the 2-hr thermal EC traces from the two analyzers for the last 9 days of the testing period.
          3 -
      o
      •&
      S
      4-1
      C
      01
      o
      o
      o
      O
2 -
          1  -
                                          -RT3218

                                          RT3219 (Corrected)
          0
          4/5/13
             4/12/13        4/19/13        4/26/13          5/3/13
                                 Date
 Figure 6-2.  Measured 2-hour average corrected thermal EC concentration from the duplicate
                        Model 4 OC-EC analyzers during testing.

The calculated differences (expressed in terms of RT3218 - RT3219) are presented in Figure
6-4, for both the raw and the corrected data.  The raw data show a sharp divergence between
the two analyzers after 4/28/13.
                                         28

-------
         u
         c
         o
         o

         o
             3 -
         I   2
         IS
                                              RT3218


                                              RT3219
             0

             4/5/13
                           4/12/13
                                         4/1 9/1 3
                                                       4/26/13
                                                                      5/3/13
                                               Date
  Figure 6-3.  Measured 2-hour average optical EC concentration from the duplicate Model 4

                              OC-EC analyzers during testing.
                1.0
            
-------
6.1 Comparability
The comparability of the Model 4 OC-EC analyzers with the reference method was
determined in two ways. Firstly, comparability was determined from a linear least squares
regression analysis of the EC concentrations measured by the Model 4 OC-EC units and by
the reference methods as described in Section 5.1.1. Also, comparability was determined
from the RPD of the Model 4 OC-EC data and the mean of the reference method data for
each sampling period, as described in Section 5.1.2. For these calculations, the 12-hour
results for the Model 4 analyzers were calculated as the averages of the 2-hour readings
collected during the respective reference method periods. Comparability was determined
independently for both the raw thermal data and the corrected thermal data from the Model 4
OC-EC analyzers as well as for the optical EC measurements of the Model 4 OC-EC
analyzers. In each case the comparability was calculated with respect to both the TOR and
TOT reference method results. The results of these analyses are presented below.

Figures 6-5 shows a time series plots of the mean TOR reference method EC results and the
corresponding 12-hour average uncorrected Model 4 OC-EC thermal EC results. Figure 6-6
shows the same time series plot, but using the corrected thermal EC data from the Model 4
unit RT3219.  Similarly, Figures 6-7 and 6-8 show the time series plots of the mean  TOT
reference method results with the corresponding 12-hour average uncorrected and the
corrected Model 4 OC-EC thermal EC results. Figures 6-9 and 6-10 show time series plots
of the TOR and TOT reference method results with the corresponding 12-hour average
Model 4 OC-EC optical EC results. The dates shown correspond to the start of the respective
sampling periods, with the first sampling period beginning on April 5 at 7:00 pm and ending
on April 6 at 6:50 am.
                                         30

-------
          to

           E
           c
           g
           +->

           SH
           +->
           c
           0
           o
           c
           o
           o

           O
           HI
              2 -
1 -
                          RT3218


                          RT3219


                          Reference Method
              0

              4/5/13
                  4/12/13
4/19/13

      Date
4/26/13
5/3/13
Figure 6-5. Comparison of uncorrected thermal EC results from the Model 4 OC-EC analyzers and the mean reference method TOR EC

                                                 concentrations.
                                                      31

-------
         to
         I2
         re
         +->
         £
         0
         U
         £
         O
         o
         o  1
         HI
                          RT3218
                          RT3219 (Corrected)
                          Reference Method
            0
             4/5/13
   ill
4/12/13
4/19/13
      Date
4/26/13
5/3/13
Figure 6-6. Comparison of corrected thermal EC results from the Model 4 OC-EC analyzers and the mean reference method TOR EC
                                              concentrations.
                                                    32

-------
o -
M"*
1
c
.0
Concentr
IN
O
GO
0
.c
"0 1
0
u
1
"0
K.
0 -
4/f




• RT3218
• RT3219


• Reference Method















I
5/13









I




I




ill
4/12/13








I,



I



Lull




till



I



1





1

III
4/19/13 4/26/13 5/3/13
Date
Figure 6-7. Comparison of uncorrected thermal EC results from the Model 4 OC-EC analyzers and the mean reference method TOT EC
                                                       concentrations.
                                                             33

-------
         to




         1

          £
         .0
         '+->

         |2


          0
          u
          c
          o
         o

         O
         m

         •o
          o

         +•>
          0  1
         s

          0
          u
          c

          Si

         £
          0
             0

             4/5/13
RT3218


RT3219 (Corrected)


Reference Method
                                  .
                                   ..
    4/12/13
4/19/13

      Date
4/26/13
5/3/13
Figure 6-8. Comparison of corrected thermal EC results from the Model 4 OC-EC analyzers and the mean reference method TOT EC

                                                concentrations.
                                                     34

-------
          to

          1
          £
          O
           £
           0
           U
           £
           O
          o
          o
2
1
                         RT3218
                         RT3219
                         Reference Method
              0
              4/5/13
i~~i~~i~"i~~i~~r
     4/12/13
                                  •
                                  Liil
                                  4/19/13
                                       Date
4/26/13
5/3/13
Figure 6-9. Comparison of optical EC results from the Model 4 OC-EC analyzers and the mean reference method TOR EC concentrations.
                                                    35

-------
           to




           1

           £
           .0
           '+->

           |2


           0
           u
           c
           o
           o

           O
           m

           •o
           o

           +•>
           0  1
           s

           0
           u
           c

           Si

           £
           0
RT3218


RT3219


Reference Method
              0

              4/5/13
                     III
Lit  III
                      LUl
   4/12/13
               4/19/13

                    Date
4/26/13
i~ ~i~ ~i~ ~i~ ~i~ ~i
   5/3/13
Figure 6-10. Comparison of optical EC results from the Model 4 OC-EC analyzers and the mean reference method TOT EC concentrations.
                                                   36

-------
6.1.1  Regression A nalysis
Figures 6-11 and 6-12 show linear regressions of the corrected and uncorrected Model 4 OC-
EC 12-hour thermal EC averages, respectively, with the corresponding TOR reference
method results. These figures show that there is a negative bias (i.e., slope <  1) of the Model
4 OC-EC results relative to the reference method TOR results, as well as a positive intercept
of-0.3  ng/m3 for each of the  regression lines. Comparison of the regression  analyses shows
that use of the corrected data (Figure 6-11) substantially changes the slope and intercept of
the RT3219 data. Also indicated in Figure 6-11 is a data point that appears to be an outlier
from the rest of the data. These data resulted from measurements on April 26, 2013, and can
be seen in Figures 6-5 to 6-10. This data point is clearly apparent in each of the regression
plots (Figures 6-11 through 6-16) and consistently falls below the regression lines in each
plot. A linear relationship appears to exist below 1.5 ng/rn3 and that additional measurements
at higher concentrations are needed to confirm a linear relationship at higher concentrations.

Figures 6-13 and 6-14 show linear regressions of the corrected and uncorrected Model 4 OC-
EC 12-hour averages, respectively, with the corresponding TOT reference method results.
The slopes of the regression lines in these figures are greater than unity indicating positive
bias of the Model 4 OC-EC results relative to the TOT reference results. As with the TOR
results, the intercepts of the regression lines are -0.3 |J,g/m3.
                                             RT3218 = 0.797x + 0.280
                                                    R2 = 0.854
                                             RT3219 = 0.903x
                                                    R2 = 0.783
              o
             '+*
              o
             +*
              c
o
o
o
LU
TJ
O
                                    1                2
                            Reference Method EC Concentration (|ig/m3)

 Figure 6-11. Scatter plot of uncorrected Model 4 EC 12-hour thermal averages against mean
                             reference method TOR results.
                                          37

-------
                        *RT3218

                        • RT 3219 (Corrected)
                                              RT3218 = 0.797x + 0.280
                                                    R2 = 0.854

                                          RT3219 (Corr.) = 0.819x + 0.290
                                                    R2 = 0.865
                  0123
                            Reference Method EC Concentration (jig/m3)

 Figure 6-12. Scatter plot of corrected Model 4 EC 12-hour thermal averages against mean
                             reference method TOR results.
            -  2
             o
             o
             o
             o
            o
            o
            LU
                        *RT3218

                        • RT3219
RT3218 = 1.057X +0.293
      R2 = 0.876

RT3219 = 1.215x +0.355
      R2 = 0.827
                                    1                 2                 3
                            Reference Method EC Concentration (iig/m3)

Figure 6-13. Scatter plot of uncorrected Model 4 EC 12-hour thermal averages against mean
                             reference method TOT results.
                                          38

-------
                 + RT3218

                 • RT 3219 (Corrected)
                                    RT3218 = 1.057x + 0.293
                                           R2 = 0.876

                                  RT3219 (Corr.) = 1.080x + 0.306
                                           R2 = 0.878
           0123
                    Reference Method EC Concentration (ng/m3)
   Figure 6-14.  Scatter plot of corrected Model 4 EC 12-hour thermal averages against mean
                             reference method TOT results.

Figures 6-15 and 6-16 show linear regressions of the optical Model 4 OC-EC 12-hour
averages with the corresponding TOR and TOT reference method results, respectively. The
slopes of the regression lines in these figures are less than unity indicating negative bias of
the Model 4 OC-EC optical results relative to the reference results. The intercepts of the
regression lines are -0.1 |J,g/m3.

Table 6-1 presents a summary of the regression results for the corrected and uncorrected
thermal EC results, and the optical results from the Model 4 OC-EC analyzers relative to
both the TOR and the TOT reference method EC results. The uncertainties (one standard
deviation) of the calculated slopes and intercepts are included parenthetically. These
regression results include the apparent outlier data points noted above at the highest observed
concentration with each Model 4 analyzer.  Table 6-2 presents revised regression results
excluding those data points. In both Table 6-1 and 6-2, bolded entries indicate slopes or
intercepts whose 95% confidence interval does not include 1.0 or 0.0, respectively.
                                          39

-------
                                             RT3218 = 0.656x + 0.134

                                                    R2 = 0.878
                                             RT3219 = 0.701x +0.140

                                                    R2 = 0.882
o
'+*

2
+*

o
o
c
O
o

o
LU
"5
TJ
O
                 0123

                          Reference Method EC Concentration (jig/m3)


Figure 6-15. Scatter plot of Model 4 EC 12-hour optical averages against mean reference

                                method TOR results.
                         *RT3218


                         • RT3219
                                            RT3218 = 0.874x + 0.143

                                                  R2 = 0.910



                                            RT3219 = 0.934x + 0.150

                                                  R2 = 0.914
                 0123

                          Reference Method EC Concentration (ng/m3)


Figure 6-16. Scatter plot of corrected Model 4 EC 12-hour optical averages against mean

                         reference method TOT EC results.
                                        40

-------
    Table 6-1. Summary regression results of the Model 4 OC-EC analyzers and the
                                 reference method
Analyzer - Mode
RT3218- Thermal
RT32 19 -Thermal
RT3219 (Corrected) - Thermal
RT3218- Optical
RT32 19 -Optical
TOR
Slope
0.797 (0.044)
0.903 (0.063)
0.819 (0.043)
0.656 (0.034)
0.701 (0.034)
Intercept
(Hg/m3)
0.280 (0.028)
0.346 (0.040)
0.290 (0.027)
0.134 (0.021)
0.140 (0.022)
TOT
Slope
1.057(0.053)
1.215 (0.073)
1.080(0.053)
0.874 (0.036)
0.934 (0.038)
Intercept
(Hg/m3)
0.293 (0.025)
0.355 (0.035)
0.306 (0.025)
0.143 (0.017)
0.150 (0.018)
 Table 6-2. Summary of Regression Results of the Model 4 OC-EC Analyzers and the
             Reference Method, Excluding Apparent Outlier Data Points
Analyzer - Mode
RT3218- Thermal
RT32 19 -Thermal
RT3219 (Corrected) - Thermal
RT3218- Optical
RT32 19 -Optical
TOR
Slope
0.982 (0.046)
1.131(0.072)
0.936 (0.053)
0.783 (0.036)
0.838 (0.037)
Intercept
(Hg/m3)
0.211 (0.025)
0.261 (0.038)
0.247 (0.028)
0.087 (0.019)
0.090 (0.020)
TOT
Slope
1.239 (0.057)
1.455 (0.081)
1.180 (0.066)
0.999 (0.039)
1.068(0.040)
Intercept
(Hg/m3)
0.245 (0.023)
0.292 (0.033)
0.280 (0.027)
0.111 (0.016)
0.115 (0.016)
Table 6-1 shows that with the outlier data points included, the regression slopes for both the
thermal (corrected) and optical EC from the Model 4 analyzers were closer to 1.0 in
comparison to the TOT data than in comparison to the TOR data.  The regression intercepts
were closely similar in the TOR and TOT regressions. Table 6-2 shows that the same was
true of the Model 4 optical data when the outlier data points were excluded. However, the
regression slopes for the Model 4 thermal (corrected) EC data were closer to 1.0 with the
TOR data when the outlier data were excluded.
6.1.2 Relative Percent Difference A nalysis
Table 6-3 presents a summary of the calculated RPD between the 12-hour averages for the
corrected and uncorrected thermal EC, and the optical EC results from the Model 4 OC-EC
analyzers relative to the TOR and the  TOT reference method results, respectively. For these
calculations, reference method results below twice the method detection limit (i.e., below
0.26 ng/rn3) were excluded.  For perfect agreement between the Aethalometers and the
reference method results, the RPD would be zero. Table 6-3 shows that the thermal EC data
from the two Model 4 OC-EC analyzers exhibited similar RPD values relative to the
reference data, when the RT3219 corrected data were considered.  The RPD values were
                                        41

-------
lower in comparison to the TOR data than to the TOT data. The Model 4 optical EC data
showed a strong trend in the opposite direction, exhibiting RPD values of less than 4%
relative to the TOR data and approximately 20 to 30% relative to the TOT data. In general,
the thermal EC measurements from the Model 4 OC-EC analyzers were greater than those
from the reference method resulting in positive RPD values. It should be noted that only
about two thirds of the TOR reference method results and fewer than half the TOT reference
method results were above twice the detection limit.
   Table 6-3.  Summary of Relative Percent Difference between the Model 4 OC-EC
             Analyzers and the TOR and TOT Reference Method Results
Analyzer - Mode
RT3218- Thermal (All)
RT3218 - Thermal (RAAS)
RT3218- Thermal (BGI)
RT32 19 -Thermal (All)
RT3219 - Thermal (RAAS)
RT32 19 -Thermal (BGI)
RT3219 (Corrected) - Thermal (All)
RT3219 (Corrected) - Thermal (RAAS)
RT3219 (Corrected) - Thermal (BGI)
RT3218- Optical (All)
RT3218 - Optical (RAAS)
RT3218- Optical (BGI)
RT32 19 -Optical (All)
RT3219 - Optical (RAAS)
RT32 19 -Optical (BGI)
RPD
TOR
45.3%
51.6%
36.6%
66.2%
54.2%
83.4%
45.7%
60.1%
25.0%
-3.8%
-1.1%
-7.3%
2.0%
4.9%
-2.0%
TOT
80.6%
98.4%
56.4%
101.4%
95.7%
109.2%
77.6%
100.7%
46.1%
23.6%
33.8%
9.6%
31.1%
41.4%
17.2%
6.2 Correlation
Table 6-4 presents a summary of the coefficient of determination (r2) values for the corrected
and uncorrected thermal, and the optical results from the Model 4 OC-EC analyzers relative
to both the TOR and the TOT reference method results. The correlation results were
calculated including all of the data and also with the apparent outliers shown in Figure 6-10
removed.

Table 6-4 shows that r2 values for both the thermal EC and optical EC were always slightly
higher relative to the reference TOT data than to the corresponding reference TOR data. In
addition, the r2 values observed with the optical EC data were always higher than those for
                                        42

-------
the thermal EC data in comparison to the same TOR or TOT data. Finally, the exclusion of
the outlying data increased the r2 values in the comparisons listed in Table 6-4 in all but one
case: the corrected data from Model 4 unit RT3219 showed lower r2 values when the outlier
data were excluded than when those data were included.
  Table 6-4. Summary of regression results of the Model 4 OC-EC analyzers and the
                                  reference method

Analyzer/Mode
RT3218- Thermal
RT32 19 -Thermal
RT3219 (Corrected) - Thermal
RT3218- Optical
RT32 19 -Optical
r2 (All data)
TOR
0.854
0.783
0.865
0.878
0.882
TOT
0.876
0.827
0.878
0.910
0.914
r2 (Outlier removed)
TOR
0.890
0.816
0.847
0.896
0.902
TOT
0.896
0.852
0.850
0.920
0.926
6.3 Precision

Table 6-5 presents a summary of the calculated unit-to-unit precision results for the corrected
and uncorrected thermal EC, and the optical EC results from the duplicate Model 4 OC-EC
analyzers. For this calculation, measurement data below the vendor's stated instrumental
detection limit of 0.5 ng/m3 was excluded. The total number of paired measurements in
which the readings from both analyzers exceeded the detection limit is included
parenthetically for each calculation.

Table 6-5 shows that correction of the Model 4 unit RT3219 data as described in the
introduction to this chapter resulted in a substantial improvement in the calculated inter-unit
precision of the two Model 4 analyzers. Precision values for the corrected thermal EC results
were 12.4% for the 2-hr data and 9.7% for the 12-hr data.  The precision of the optical EC
results was better than for the thermal EC results, with both the 2-hr and 12-hr data showing
precision better than 7%.

 Table 6-5. Summary of Calculated Precision Results of the Model 4 OC-EC Analyzers


Uncorrected
Corrected
Thermal
2-hour
19.9% (N=168)
12.4% (N=157)
12-hour
14.3% (N=38)
9.7% (N=36)
Optical
2-hour
5.7%(N=91)
~
12-hour
6.3%(N=16)
~
Figures 6-16 through 6-18 show regressions of the 2-hour data for the corrected and
uncorrected thermal EC measurements and the 2-hour optical measurements for the duplicate
Model 4 analyzers, and Figures 6-19 through 6-21 show scatter plots of the corresponding
                                         43

-------
12-hour averages. Table 6-6 summarizes the regression statistics from the Model 4 unit-to-
unit comparison plots shown in Figures 6-16 through 6-21.  The slope, intercept, and r2
values for the Model 4 thermal EC comparisons for both the 2-hr and 12-hr data were all
improved by correction of the unit RT3219 data. The regressions using the corrected data
show slopes at or near 1.0, intercepts below 0.04 |ig/m3, and r2 values exceeding 0.91. The
regressions of the Model 4 optical EC data show slopes slightly  above  1.0, near-zero
intercepts, and r2 values of 0.999, substantially higher than the r2 values of the thermal EC
comparisons.

    Table 6-6. Summary of Unit-to-Unit Regression Results of the Model 4 OC-EC
                                      Analyzers
Mode
Thermal
Thermal -
Corrected
Optical
2-hour
Slope
1.075(0.021)
1.000(0.016)
1.057(0.002)
Intercept
(Hg/m3)
0.064 (0.017)
0.024(0.013)
0.002(0.001)
r2
0.880
0.912
0.999
12-hour
Slope
1.104(0.055)
0.980 (0.036)
1.065(0.004)
Intercept
(Hg/m3)
0.044 (0.040)
0.035 (0.027)
-0.001 (0.002)
r2
0.872
0.925
0.999
              o
              I
              o
              o
              o
              LU
              O>
              CM
              CO
                  4 -
             £   3
2 -
                        y = 1.075x +0.064
                           R2 = 0.880

                            N = 372
                               1234
                                 RT3218 - EC Concentration (jig/m3)
      Figure 6-16. Regression of uncorrected Model 4 2-hour thermal EC measurements.
                                          44

-------
B)
       3   4
        o
        0)
        u
        o
       o
       o
       LU
       "(5

        5
       o>
       CM
       CO
                  y=1.000x + 0.024
                      R2 = 0.912
                          1234
                            RT3218 - Thermal EC Concentration (|j.g/m3)
Figure 6-17.  Regression of corrected Model 4 2-hour thermal EC measurements.
        o
       is
        0)
        o
        o
       o
       o
       LU
        Q.
       O
        i
       O>
       CM
       CO
       OL
                        3 -
                2 -
                        1  -
                         y = 1.057x +0.002
                            R2 = 0.999
                             N = 372
                            RT3218 - Optical EC Concentration (jig/m3)
     Figure 6-18. Regression of Model 4 2-hour optical EC measurements.
                                     45

-------
      i
      o
       0)
       o
       o
      o
      o
      LU
      O>
      CM
      CO
y= 1.104X +0.044
   R2 = 0.872
   N=62
             0123
                          RT3218 - Thermal EC Concentration (ng/m3)

Figure 6-19. Regression of uncorrected Model 4 12-hour thermal EC averages
           3
      i
      o
      I   2
      I
      o
      o
      o
      LU
      CM
      CO
y = 0.980x + 0.035
   R2 = 0.925
     N=62
             0123
                          RT3218 - Thermal EC Concentration (ng/m3)

 Figure 6-20.  Regression of corrected Model 4 12-hour thermal EC averages.
                                    46

-------
                         y = 1.065x-0.001
                            R2 = 0.999

                             N=62
                                      1                  2
                                RT3218 - Optical EC Concentration (jig/m3)
              Figure 6-21. Regression of Model 4 12-hour optical EC averages.
6.4 Data Completeness

Table 6-7 presents a summary of the data completeness for the duplicate Model 4 analyzers
during the testing period.  Each of the analyzers recorded a total of 372 valid measurements
out of 378 measurement periods during the verification test. The six periods of missing data
were all associated with periods when the analyzers were offline to conduct routine
maintenance as indicated in Section 6.5.1. The single missing 12-hour average for each of
the analyzers was also the result of the analyzers being offline for maintenance.

      Table 6-7.  Summary of data completeness for the Model 4 OC-EC analyzers
Analyzer
RT3218
RT3219
2-hour
Total
Periods
378
378
Valid
Measure
ments
372
372
%
Complete
98%
98%
12-hour
Total
Periods
63
63
Valid
Measure
ments
62
62
%
Complete
98%
98%
                                         47

-------
6.5 Operational Factors

This section addresses the maintenance, consumables, waste generation, ease of use, and
other factors relevant to automated field operation of the Model 4 OC-EC analyzer.

6.5.1 Routine Maintenance
Table 6-8 shows the maintenance activities that were performed on the two Model 4 OC-EC
analyzers during the verification test.

     Table 6-8. Summary of maintenance performed on Model 4 OC-EC analyzers
Date
4/3/1 3 a
4/15/13
4/25/13
5/3/13
Maintenance
Filters changed - both analyzers
Filters changed - both analyzers
Filters changed - both analyzers
Filters changed - both analyzers
Approximate
time
-15 minutes
-15 minutes
-15 minutes
-15 minutes
Data loss
2 hours (one
measurement cycle)
2 hours (one
measurement cycle)
6 hours (three
measurement cycles)b
2 hours (one
measurement cycle)
a Filter change occurred after installation of the Model 4 analyzers but before reference sampling
  began.
b The analyzers were remotely stopped, but the filters were not changed until ~4 hours later.

6.5.2 Consumables/Waste Generation
During the verification test, the Model 4 OC-EC analyzers required the use of various
compressed gases including ultra-high purity helium,  5% CFLt in helium, and 10% oxygen
(O2) in helium.  These gases were supplied to the analyzers from individual cylinders (size
AQ) that were returned to the supplier with a substantial amount of the gases remaining. The
Model 4 OC-EC analyzers also required the replacement of quartz fiber filters (two punches
per analyzer) approximately weekly.

6.5.3 Ease of Use
Installation of two Model 4 units with inlets was completed in approximately 4 hours. After
installation, the units were allowed to  operate overnight and were calibrated the following
morning using sucrose standard solutions. The calibration of both units was completed in
approximately 3 to 4 hours. Routine operation required no effort other than brief daily
instrument checks and approximately weekly data downloads. The data was processed using
the vendor software to generate comma separated variable (csv) data files.
                                         48

-------
                                   Chapter 7
                            Performance Summary

Table 7-1 presents a summary of the results of the verification of the Model 4 OC-EC
analyzers during this verification test.

         Table 7-1. Summary of Verification Test Results for the Model 4 OC-EC
Comparability-
Regression
analysis
comparison to
reference samples


Analyzer/Mode
RT3218 Thermal
RT32 19 Thermal
RT32 19 Corrected
Thermal
RT3218 Optical
RT32 19 Optical
Comparability- Calculation
between Model 4 OC-EC re
reference method results
ofRPD
suits and
Correlation - Regression analysis
comparison to reference samples
Precision - Compar
results from duplic
Model 4 OC-EC
Analyzers
ison of
ate

Uncorrected
Corrected
TOR
Slope
0.797 (0.044)
0.903 (0.063)
0.819 (0.043)
0.656 (0.034)
0.701 (0.034)
Intercept
0.280 (0.028)
0.346 (0.040)
0.290 (0.027)
0.134(0.021)
0.140(0.022)
Analyzer/Mode
RT32 18 Thermal
RT32 19 Thermal
RT3219 Corrected Thermal
RT3218 Optical
RT3219 Optical
Analyzer/Mode
RT32 18 Thermal
RT32 19 Thermal
RT3219 Corrected Thermal
RT3218 Optical
RT32 19 Optical
TOT
Slope
1.057 (0.053)
1.215 (0.073)
1.080 (0.053)
0.874 (0.036)
0.934 (0.038)
Intercept
0.293 (0.025)
0.355 (0.035)
0.306 (0.025)
0.143 (0.017)
0.150(0.018)
RPD
TOR
45.4%
66.2%
45.7%
-3.7%
2.1%
TOT
80.6%
101.4%
77.6%
23.6%
31.2%
r2
TOR
0.854
0.783
0.865
0.878
0.882
TOT
0.876
0.827
0.878
0.910
0.914
RPD (# of Observations)
Thermal
2-hour
19.9%
(N=168)
12.4%
(N=157)
12-hour
14.3%
(N=38)
9.7%
(N=36)
Optical
2-hour
5.7%
(N=91)
~
12-hour
6.3%
(N=16)
~
                                       49

-------
 Table 7-1 (continued). Summary of Verification Test Results for the Model 4 PC-EC
Precision - Regression
analysis of results from
duplicate monitoring
systems
Data Completeness
Maintenance
Consum ables/waste
generated
Mode
Thermal
Uncorrected
Thermal
Corrected
Optical
Analyzer
RT3218
RT3219
Period
2-hour
12-hour
2-hour
12-hour
2-hour
12-hour
Period
2-hour
12-hour
2-hour
12-hour
Slope
1.075(0.021)
1.104(0.055)
1.000(0.016)
0.980 (0.036)
1.057(0.002)
1.065(0.004)
Total Periods
378
63
378
63
Intercept
0.064 (0.017)
0.044 (0.040)
0.024 (0.013)
0.035 (0.027)
0.002 (0.001)
-0.001 (0.002)
Valid
Measurements
372
62
372
62
r2
0.880
0.872
0.912
0.925
0.999
0.999
%
Complete
98%
98%
98%
98%
• Routine maintenance consisted of replacing filters approximately
weekly.
• Three different compressed gas cylinders required to operate the units
• Internal filters replaced weekly.
Ease of use
Installation of two Model 4 units with inlets completed in ~4 hours.
Calibration of units completed in ~3-4 hours, after allowing the units
to operate overnight.
Routine operation required no effort other than brief daily instrument
checks and approximately weekly data downloads.
Data processed using vendor software to generate csv data files.	
                                            50

-------
                                   Chapter 8
                                   References

1.  Birch, M.E., Gary, R.A., 1996. Elemental carbon-based method for monitoring
   occupational exposures to particulate diesel exhaust. Aerosol Science Technology 25,
   221D241.

2.  Battelle, Quality Assurance Project Plan for Verification of Black Carbon Monitors:
   Version 1, Battelle, Columbus, Ohio, April 12, 2013.

3.  Report to Congress on Black Carbon, EPA-450/R-12-001, U.S. Environmental Protection
   Agency, March 2012, available at http://www.epa.gov/blackcarbon/.

4.  Desert Research Institute, DRI Model 2001  Thermal/Optical Carbon Analysis
   (TOR/TOT) of Aerosol Filter Samples - Method IMPROVE_A, DRI SOP#2-216r3,
   prepared by DRI, Reno, NV, October 22, 2012.

5.  U.S. EPA, Environmental Technology Verification Program Quality Management Plan,
   EPA Report No: EPA 600/R-08/009, U.S. Environmental Protection Agency, Cincinnati,
   Ohio, January 2008.

6.  Battelle,  Quality Management Plan for the ETV Advanced Monitoring Systems Center,
   Version 8.0, U.S. EPA Environmental Technology Verification Program, prepared by
   Battelle, Columbus, Ohio, April 5, 2011.
                                        51

-------