United States
                     Environmental Protection
                     Agency
                                    Environmental Monitoring
                                    Systems Laboratory
                                    Las Vegas,  NV 89193-3478
                     Research and Development
                                    EPA/600/S4-90/032   Mar.  1991
iSrEPA       Project  Summary

                     Evaluation  of a  Remote
                     Sensor for  Mobile  Source
                     CO  Emissions
                     Donald H. Stedman and Gary A. Bishop
  Carbon  monoxide  (CO)  emission
measurements of thousands of vehicles
per day are possible  with a recently
evaluated remote sensor developed at
the University of Denver. Funded by
the Environmental Monitoring Systems
Laboratory - Las Vegas (EMSL-LV) In-
novative Research Program, the evalu-
ation has demonstrated the compara-
bility of volume concentration mea-
surements made by this method with
traditional emission monitoring Instru-
mentation.  Measurements are made
unobtrusively as vehicles pass through
an infrared light beam  directed across
one traffic lane about 25 centimeters
above the pavement. A video camera
records the vehicle registration num-
ber of each vehicle as its CO emissions
are measured so that characteristics of
Individual vehicles and  vehicle fleet
categories can be associated with each
measurement. Determining appropriate
applications and monitoring protocols
for this technology is the second phase
of this Innovative  Research Project.
Similar remote sensing technology for
monitoring mobile hydrocarbon and ni-
trogen  oxide emissions can be devel-
oped to address the urban ozone non-
attainment problem.
  This project summary was developed
by EPA's Environmental Monitoring
Systems Laboratory, Las Vegas, NV, to
announce key findings of the research
project that Is fully documented In a
separate report of the  same title (see
Project Report ordering Information at
back).
                                                        Introduction
                                                          This project  report describes an EPA-
                                                        sponsored evaluation of a device designed
                                                        to measure the carbon monoxide (CO)
                                                        emissions from a passing  automobile by
                                                        means of remote sensing.  The University
                                                        of Denver developed the device with sup-
                                                        port from the Colorado Office of Energy
                                                        Conservation as an  approach to promote
                                                        improved vehicle fuel economy. High CO
                                                        emissions are indicative of incomplete fuel
                                                        combustion caused  by out-of-adjustment
                                                        or defective fuel and emission systems.
                                                        Surveillance for CO emissions will identify
                                                        vehicles with potential fuel  efficiency ben-
                                                        efits if repaired. As a result of this original
                                                        purpose, the CO emissions monitoring
                                                        system is called FEAT, for  Fuel Efficiency
                                                        Automobile Testing.
                                                          The FEAT, shown schematically in Fig-
                                                        ure  1, consists of three basic  units: an
                                                        infrared (IR) light source, an IR detector
                                                        and a personal computer. IR absorption is
                                                        used to determine the amounts of CO and
                                                        CO, emitted by a passing automobile. The
                                                        IR fight source, located on one  side of a
                                                        roadway, sends a collimated beam into a
                                                        gas  filter radiometer equipped  with two
                                                        liquid-nitrogen-cooled indium antimonide
                                                        photovoltaic detectors. A 4.3 micron filter
                                                        isolates the CO region before one of the
                                                        two detectors. The resulting beam passes
                                                        through a rotating gas filter wheel, half of
                                                        which contains a CO and H2 mixture and
                                                        the  other half  N2.  The rotating  wheel
                                                        modulates the signal and provides both a
                                                        reference channel and a CO data channel.
                                                          The system is installed across a single-
                                                        lane highway with the IR beam located at
                                                                     Printed on Recycled Paper

-------
                        Carbon  Monoxide  Remote  Sensing
                            Computer.
                                     Calibration
Figure 1. Schematic diagram.of the University of Denver CO remote sensor.
the height of most exhaust pipes (about
25  centimeters above the pavement).
When a vehicle enters the optical path a
drop in reference  voltage  signals the
vehicle's presence. Span voltages from
each of the three signal channels (CO,
CO2, and reference) are acquired before
the vehicle enters the beam, and zero
correction voltages for each channel are
acquired while the vehicle  is completely
blocking the  beam. As the vehicle exits
the beam, a one half second voltage ver-
sus time trace from  each of the three
channels is obtained.
  The voltage values obtained as a ve-
hicle exits the beam are corrected for the
ambient measurements obtained in front
of the vehicle. Voltages are converted to
path-averaged concentrations with a linear
calibration equation. Calibration constants
are determined periodically while in use
by measurements of certified compressed
gas mixtures to allow path-averaged CO
and CO.J concentrations to be determined.
The concentrations of the CO and CO2 in
the exhaust plume change jointly as they
rapidly disperse in the turbulent wake be-
hind the vehicle. The computer calculates
a ratio of CO to CO2 that corresponds to
the slope of a best fit line for a scatter plot
of these two gases. The confidence limits
of the slope must be better than  20% for
the software to accept the  measurement
as valid.
  Using the combustion reaction equa-
tions, the ratio of CO to COa is converted
to exhaust  percent CO. The  results are
stored  on floppy disk and  printed on  a
video  tape  together with a freeze-frame
video image of the rear of the vehicle at a
rate as fast  as one vehicle per second. At
some  locations emissions measurements
of over one thousand vehicles per hour
have been made.
  The  remainder of this document sum-
marizes the evaluation of the  FEAT.  It  is
organized in four  sections  which corre-
spond  to the objectives of the  project.
These objectives are:

   1. to evaluate the theory used to con-
     vert the CO to CO  ratio to percent
     CO and to grams CO per gallon of
     fuel;
   2. to validate the theory using existing
     data  from laboratory-grade  gas
     monitors;
   3. to inter-compare FEAT and labora-
     tory-grade emissions measurements
     of vehicles on a dynamometer; and
   4. to discuss potential applications for
     remote sensing of vehicle emissions.


Theory
  The  FEAT depends on the carbon and
oxygen balances of the combustion reac-
tion equations  to  convert the measured
CO to CO2 ratio into percent CO and grams
CO per gallon of fuel burned. This entails
certain  assumptions concerning  the
chemical formula  for the fuel  and  the
presence  of  unburned  hydrocarbons.
Complete derivation of the theory used by
FEAT and an analysis of the sensitivity of
the FEAT results to the required assump^
tions  is  the  subject of  this  part of  the
evaluation project.
  The conversion equation from the mea-
sured ratio to  exhaust  percent CO  de-
pends on  a knowledge of the  carbon to
hydrogen ratio in the fuel. However the
error  introduced by assuming a typical
value is very small, particularly for  low
emission vehicles  (less  than  4% CO).
About 70% of the vehicles measured by
FEAT have percent CO emissions of  less
than  1% CO.  Even considering a very
high CO emitting  vehicle with the most
extreme cases of fuel carbon to hydrogen
ratios, 1:0 for pure carbon (e.g., coal)  and
1:4 for methane, the FEAT reported  per-
cent CO of 8.8 would be  26% low for the
coal-fired vehicle and 33% high for the
methane vehicle.
  When FEAT readings are converted to
mass emissions in  grams CO per gallon
of fuel, hydrocarbon emissions may either
be neglected, or corrected using an aver-
age hydrocarbon emissions factor. In either
case the corrections are small, particularly
for the low emission vehicles. For most
vehicles (i.e., about 95%) the uncorrected

-------
 grams CO per gallon values are within 6%
 of the corrected values.
   Presently under development is a hy-
 drocarbon  remote  sensing monitoring
 channel for FEAT which will eliminate this
 small source of inaccuracy when reporting
 grams CO per gallon of fuel. This hydro-
 carbon  channel will also  enable remote
 sensing to report hydrocarbon exhaust
 emissions directly,

 Theory Validation
   Ratios of CO to CO2 from  conventional
 exhaust system probe measurements can
 be directly compared to FEAT  measure-
 ments  of  the  ratio.  However,  a  direct
 comparison of percent CO  can not be
 done without correcting the  probe value
 for dilution air. Dilution of vehicle exhaust
 by air which is not part of the combustion
 process (e.g.,  as introduced by an air
 pump) results in a lower CO concentration
 at the probe. FEAT exhaust measurements
 are unaffected by dilution air. Probe read-
 ings are  adjusted upwards  to  take  into
 account any dilution caused by the excess
 air by using the probe exhaust oxygen
 (O2) measurement as an indicator. Unless
 stated otherwise, all exhaust probe results
 used in  comparisons have been adjusted
 with the oxygen correction for dilution air.
  A data  set obtained from the Chrysler
 Corporation allowed a FEAT-independent
 evaluation of the  FEAT theory. Measure-
 ments included real-time CO,  CO2, and O2
 vehicle  emissions concentrations before
 and after an exhaust system catalyst. The
 malfunctioning vehicle was operated on a
 dynamometer for the test. Catalyst  effi-
 ciency was near 100% for low percent CO
 emissions but dropped to about 50% when
 presented with high percent CO emissions.
 The same equations that FEAT uses for
 calculation of percent CO where applied
 to the ratio of CO to CO2 calculated from
 the Chrysler data.
   Figure  2 shows the  resulting FEAT
 theory calculated percent CO (the lines)
 compared to the Chrysler measured per-
 cent CO (the boxes and X's). The agree-
 ment is excellent, except for a slight in-
 ability of  the FEAT theory  to respond
 completely to the highest CO peaks, which
 may be due  to  lags in laboratory instru-
 mentation. The  FEAT theory applies as
 well after the catalyst as before it, indicat-
 ing that  the catalyst follows the same
 combustion reaction equation as the en-
 gine.

 Dynamometer Testing
  An  interface was required which would
 enable the  FEAT unit  to  measure  the
 emissions from a vehicle on a dynamom-
 eter since the normal FEAT mode of op-
 eration requires a passing vehicle to trigger
 its measurement cycle. The interface con-
 sisted of a system for compressing raw
 exhaust into  a two liter stainless steel
 canister from  which there were two exits.
 At the bottom of  the canister a small  leak
 eliminated the water which had  condensed,
 and maintained ventilation of  the tubing.
At the side of the canister a short tube led
through a computer controlled solenoid
valve to a ring of holes normally used as
 the intake for calibration gas to the FEAT
 system. A small rotating  blade was used
 to simulate the presence of a vehicle and
 to trigger the release of exhaust into the
 beam. Since the blade rotated every two
 seconds,  a reading of the exhaust was
 available  every two seconds when the
 FEAT was  operated in this  mode. The
 hardware and  software  were otherwise
 identical to those used by the FEAT dur-
 ing normal on-road operation.
   Comparison  testing was performed at
 the  Environmental  Testing Corporation
 (ETC),  a private automobile emissions
 testing facility that specializes in high alti-
 tude certification tests for several automo-
 bile manufacturers.  ETC is equipped with
 dynamometer facilities, constant volume
 sampling systems, and a full  suite of ex-
 haust gas analytical instrumentation.
  Tests were conducted on November 1,
 1989 and November 15, 1989. Single blind
 and double blind test protocols were em-
 ployed to compare FEAT to laboratory CO
 measurements. The  majority of the tests
 were  conducted in  a steady-state mode
 where a vehicle on  a dynamometer was
 allowed to stabilize at a specific speed for
 a specific load prior to calling for the mea-
 sured values from  the FEAT and  ETC
 monitors. In  one test, the emissions from
 a cold vehicle were  monitored as the ve-
 hicle warmed up, thus allowing the chang-
 ing emissions  from the  period before,
during, and after catalyst warm-up to be
 monitored. In that test, O2 data from ETC
was  unavailable to allow dilution correc-
tion to their values.
                             Comparison   Between   FEAT   Equation
                                                 and   Vehicle   Data
                                      10
                       G  Chrysler angina  out
                      	FEAT prediction
     T	 T' r~	1	•"•••IT	T
    20     30     40     50     60
         TIME  (sec.  )
                                                                                           80
                               •+•  Chyrsler tailpipe out

                                    • FEAT  prediction
Figure 2.  Chrysler data on a malfunctioning vehichle. Upper line is the FEA T equation prediction from the engine emissions. Points are the oxygen corrected
         data. Lower line and points are the equivalent data from the tailpipe.                                          wxyyeuwreciea

-------
                                       Measurement    Comparison
                                                                        D Open comparison
                                                                        4- Blind  comparison
                                                                               Fit
                                                        ETC   % C O


Figure 3.  Uncorrected FEAT readings against ETC readings. Blind comparisons conducted 11/1/89 and 11/15/89. Open comparison data obtained on 11/
         15/89 is dilution corrected. All data not corrected for calibration cylinder discrepancies.
  Figure 3 is a scatter graph of the FEAT-
and ETC-measured percent CO along with
a best fit line for the steady  state tests.
The slope of the line is 0.94, the intercept
0.011% CO and the R2 is 0.99. The FEAT
data are reported as raw data. If the cali-
bration  data are used from the day  on
which the measurements were taken, the
FEAT readings would be increased by a
factor of approximately 1.04. For the non-
steady-state test the vehicle started at CO
levels too high for ETC to read, then as
the engine and catalyst warmed  up the
emissions decreased to very low  values.
Comparison  for this test also showed
agreement between the  two approaches.
  The agreement between the FEAT and
laboratory-grade instrument indicates that
the FEAT correctly measures the instan-
taneous emissions  of on-road vehicles.
Precision was determined to be better than
one third of one percent CO for a range of
measured  emissions up to greater than
ten percent CO.

Possible Applications
   FEAT has been  shown to make fast
and accurate  measurements of  the in-
stantaneous CO emissions from vehicles
as they pass by the instrument.  Broadly
speaking, this technology  has application
for characterization of CO emissions from
individual vehicles  or characterization of
fleet CO emissions by aggregation of indi-
vidual measurements.
  The value of individual vehicle monitor-
ing is the identification of the small fraction
of the vehicles that are responsible for the
majority  of  the  mobile emissions  (i.e.,
roughly 20% of the vehicles produce 60%
of the emissions). Fleet emissions charac-
terization might produce improved emis-
sions input for air quality predictive models,
or more  quickly identify a subset of the
fleet with systematic failures that can be
addressed through actions such as recall
or tampering investigations. However, to
fully realize  specific applications requires
additional research. This section discusses
some of the applications of the technology
and the associated issues which must be
addressed. Vehicle emissions are variable
and depend upon many factors including
engine/control system  condition,  vehicle
load, driving mode (e.g., rate of accelera-
tion/ deceleration, etc.), and engine and
catalyst temperature.  The traditional ap-
proaches for emissions monitoring attempt
to minimize the variations by  controlling
the  influential factors.  The  Federal Test
Procedure  (FTP), which is used to certify
new vehicle emissions, does this by mak-
ing  integrated  emissions  measurements
on vehicles operated  on  a dynamometer
under very  carefully controlled conditions
during a prescribed operating and vehicle
load cycle.  The FTP  emission measure-
ments are on a mass basis as opposed to
volume  concentration measurements.
Though considerably  less well controlled
than the FTP, most Inspection and Main-
tenance  (I & M) programs  use steady-
state tests conducted on warmed engines
under idle and fast idle, no-load conditions.
I  & M emissions are measured on a vol-
ume concentration basis. The present new
vehicle  emission .control standards and
the maintenance requirements for the on-
road fleet are designed to be responsive
to these two test procedures.
   The remote sensor has less control over,
or capability to  monitor the factors which
influence  emissions than the traditional
approaches. In  addition, it makes a "snap
shot"  type of measurement which  could
catch a vehicle during an anomalous point
in its  emission  cycle. Judicious selection
of the monitoring location will allow some
control over load as influenced by road
grade and position with  respect to road
conditions (i.e., highway  entrance or exit
ramps) and traffic  signs (i.e.  stop signs).
Use of radar (recently integrated into the
FEAT data, system) will  permit vehicle
driving mode to be monitored.
   The remote  sensor  has two  powerful
advantages over the  traditional ap-
proaches.  It measures emissions of ve-
hicles as they are in the real-world, under
load in which a traditional I & M test does
not. A single system can  make thousands
of measurements  per day  in a cost effi-
cient  way.
   Fleet and individual vehicle applications
have  different monitoring requirements. For
fleet monitoring applications, the large data
sets that can be easily gathered make the

-------
aggregated  results immune  from errors
resulting from emission variability, though
still subject to any bias resulting from non-
representative emissions conditions. If bias
cannot be sufficiently minimized or deter-
mined, it could prevent intercomparisons
between emissions  in different parts  of
the country.  However, bias  would  have
little effect in the intercomparison in rela-
tive terms of various segments of the same
fleet (e.g., vehicle age, make and model,
etc.) monitored at a single  location  and
time. Individual vehicle monitoring is sub-
ject to both emissions bias and variability.
It should be possible to minimize variabil-
ity by selection  of monitoring  sites that
increase the chances of a vehicle being in
a steady-state  mode  of  operation.  Simi-
larly,  bias  may be controlled  by picking
monitoring  locations that will result in rep-
resentative emissions. Multiple measure-
ments of the same vehicle is another ap-
proach to decrease problems  caused by
variability.
  There  are  a  number of possible goals
that could  be established as  part of an
individual vehicle  monitoring  program.
Obvious gross emitting vehicles could be
required to improve emissions through re-
pair or adjustment; yehicles with high test
results could be subject to additional tests;
or obviously clean vehicles could be given
an exemption from periodic I & M tests.
Research is required to establish the re-
mote sensor measurement value that cor-
responds to the action level desired.

Notice
  The  information in this document has
been funded wholly or in part by the United
States  Environmental Protection Agency
under Cooperative Agreement CR-815778-
01-0 to the University of  Denver.  It has
been subject to the  agency's  peer  and
administrative review, and it has been ap-
proved for publication as  an EPA docu-
ment. Mention of trade names or commer-
cial products does not constitute endorse-
ment or recommendation for use.

-------

-------

-------
   DonaldH.Stedman and'GaryA. Bishop are with the University of Denver, Denver,
     CO 80208.
   Marc L Pttchford is the EPA Project Officer (see below).
   The complete report, entitled "Evaluation of a Remote Sensor for Mobile Source CO
     Emissions," (OrderNo. PB91-148320/AS; Cost: $17.00, subject to change) will
     be available only from:
          National Technical Information Service
          5285 Port Royal Road
          Springfield, VA 22161
          Telephone: 703-487-4650
   The EPA Project Officer can be contacted at:
          Environmental Monitoring Systems Laboratory
          U.S. Environmental Protection Agency
          Las Vegas, NV 89193-3478
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati, OH 45268
BULK RATE
POSTAGE & FEES PAID
EPA PERMIT NO. G-35
  Official Business
  Penalty for Private Use $300
  EPA/600/S4-90/032

-------