EPA-600/2-76-169
June 1976
DEVELOPMENT OF A PROPORTIONAL SAMPLER
FOR AUTOMOBILE EXHAUST EMISSIONS TESTING
By
Harold J. Hasklns
Industrial and Environmental Products Operation
Aeronutronic Division
Aeronutronic Ford Corporation
Ford Road, Newport Beach, California 92663
Contract No. 68-02-1755
Project Officer
Peter Gabele
Environmental Sciences Research Laboratory
Research Triangle Park, North Carolina 27711
LIBRARY
U. S. FNVI.ROf.'f.'V ' PROTECTS
-------
DISCLAIMER
This report has been reviewed by the Environmental Sciences Research
Laboratory, U.S. Environmental Protection Agency, and approved for pub-
lication. Aproval does not signify that the contents necessarily reflect
the views and policies of the U.S. Environmental Protecion Agency, nor
does mention of trade names or commercial products constitute endorse-
ment or recommendation for use.
11
-------
ABSTRACT
This report documents the successful development of a proportional sampler
for measurement of automobile exhaust mass emissions. The device dynami-
cally maintains a sample mass flow at a constant, known fraction of the
tailpipe exhaust mass flow. Neither the sample nor the exhaust is diluted
in any manner. Sampling proportionality is maintained through all modes
of the EPA urban driving schedule.
The proportional sampler major components are an exhaust flowmeter, sample
flow control valve, electronic signal processor, sample conditioning sys-
tem, and an exhaust heat exchanger. The flowmeter is a vortex-shedding
type. The sample valve is a flapper valve operated in an on/off pulsed
mode. The signal processor controls the valve pulse frequency to maintain
sample mass proportionality.
Extensive development testing led to use of a water-to-exhaust heat
exchanger upstream of the flowmeter. This provided acceptable flowmeter
accuracy and satisfactory overall performance for the device, which was
then delivered to EPA's Triangle Park facility. The equipment design
and test results are discussed in the report.
This report was submitted in fulfillment of Contract Number 68-02-1755,
by Aeronutronic Ford Corporation, under the sponsorship of the Environ-
mental Protection Agency. Work was completed as of December 1975.
111
-------
CONTENTS
Page
Abstract iii
Figures vi
Tables viii
Abbreviations and Symbols ix
Acknowledgments xiv
1. Introduction 1
2. Summary 3
3. Equipment Design 8
4. Test Results 57
References 81
Appendices 82
v
-------
FIGURES
No. Page
1 Proportional Sampler Fluid System Schematic 15
2 Proportional Sampler Electrical System Diagram 20
3 Proportional Sampler Console (Front View) 22
4 Control Panel and Electronics Assembly 23
5 Console Interior (Right Side View) 24
6 Console Interior (Rear View, With Heat Exchanger
Removed) 25
7 Console Interior (Left Side View, With Heat Exchanger
Removed) 26
8 Sample Flow Control Valve Schematic Diagram 33
9 Flapper Valve Cross-Section 34
10 Signal Processor Simplified Block Diagram 42
11 Exhaust Cooling Load Estimates 51
12 Cooling Capacitor for Open-Cycle Water-Exhaust Head
Exchanger 52
13 Exhaust Heat Exchanger Schematic Diagram 53
14 Heat Exchanger Coil Assembly 55
15 Heat Exchanger Performance Estimate 56
16 Exhaust Flowmeter Calibration Test Setup 58
17 Exhaust Flowmeter Calibration Data 59
VI
-------
FIGURES (Continued)
No.
18
19
20
21
22
Flowmeter Read-Only Memory Function for the Signal
Comparison of Vortex Flowmeter Signals (Varying Plenum
Volume)
Pag
62
64
65
67
70
23 Effect of Acoustic Beam Obstruction on Flowmeter
Signals 72
24 Effect of Acoustic Beam Collimators Length on Flowmeter
Signals 74
25 Comparison of Flowmeter Signals for Heated Air Flows . . 76
26 Comparison of Flowmeter Signals for Vehicle Exhaust Flow
(Cruise Condition) 77
27 Comparison of Flowmeter Signals for Vehicle Exhaust Flow
(Acceleration Condition) 78
28 Typical Exhaust Flow Trace Over a Complete Cycle of the
EPA Urban Driving Schedule 80
VII
-------
TABLES
No. Page
1 Sample Diversion and Purge Valve Lcfgic 18
2 Comparison of Signal Processor Design Approaches .... 37
3 Multiplier Sequence of Operations 43
4 Signal Processor Parameter Binary Scaling Factors .... 45
5 Exhaust Gas Transport Properties 61
viii
-------
LIST OF ABBREVIATIONS AND SYMBOLS
ABBREVIATIONS
A/D analog-to-digital
CL chemiluminescent
CVS constant volume sampler
CVS-CH cold/hot start CVS (procedure)
°C degree Celsius
cm centimeter
D/A digital-to-analog
DIP dual in-line package
FID flame ionization detector
Hz Hertz
hr hour
ISB intermediate significant bits
°K degree Kelvin
kgm kilogram
kJ kilojoule
kPa kilopascal
LFE laminar flow element
LSB least significant bits
1pm liter per minute
IX
-------
ABBREVIATIONS (Continued)
MSB
m
mm
ms
NDIR
ROM
SAR
sec
VAC
VDC
w
Mm
Us
SYMBOLS
AQ, A^ A,
Ar
CHY
CO
co2
D
d
f
f
c
f
m
most significant bits
meter
millimeter
millisecond
non-dispersive infrared
read-only memory
successive approximation register
second
volts, alternating current
volts, direct current
watt
micro-meter
micro-second
coefficients used in Equation 16
argon
fuel formula
carbon monoxide
carbon dioxide
duct diameter
strut diameter
vortex frequency
cutoff frequency
maximum frequency
-------
SYMBOLS (Continued)
H absolute humidity
HC hydrocarbon
HO water
k proportionality constant
£ inductance tube length
M. mass of ith specie emitted
m. mass of ith specie in sample
N nitrogen
NO oxides of nitrogen
X.
0 oxygen
PR barometric pressure
? standard barometric pressure
P, pressure at flowmeter
P.! actual pressure at valve inlet
P, vacuum at condensate trap
Q volume flow rate
Q actual volume flow rate
A
Q,., standardized volume flow rate
q maximum sample flow rate
q average sample flow rate
s
RT resistance at temperature T
XI
-------
SYMBOLS (Continued)
R ice-point resistance
Reynolds number
S Strouhal number
(SVP)1 saturated water vapor pressure at
temperature T
(SVP). saturated water vapor pressure at
temperature T,
T temperature
T temperature at flowmeter
T temperature at valve inlet
T standard temperature
t time
t valve cycle interval
t valve-open interval
o
t test interval
U gas velocity
U average gas velocity
V cumulative exhaust volume
V' cumulative volume of dried exhaust
V plenum tank volume
v cumulative sample volume
X. volume fraction of ith specie
X1.1 volume fraction of ith specie in dried
sample
xn
-------
SYMBOLS (Continued)
X" CO fraction in dried sample
eo v
X" C00 fraction in dried sample
V^Wrt i
X
HO HO fraction in exhaust
X* HO fraction in dried exhaust
n~U /
Xu ~ HO fraction in dried sample
Y fuel hydrogen/carbon ratio
ot,6 coefficients in Equation 20
AQ. actual exhaust volume per flowmeter pulse
f\
AQ standard exhaust volume per flowmeter
pulse
Aq sample volume per flowmeter pulse
At time interval between flowmeter pulses
m
X effective/full-open sample flow ratio
A, leakage/full-open sample flow ratio
y exhaust gas viscosity
V viscosity temperature exponent
p. standard density of ith specie
xiii
-------
ACKNOWLEDGMENTS
The consistent support throughout the program of Dr. Ronald Bradow and
Mr. Peter Gabele of EPA is gratefully acknowledged.
The personnel of Northrop Services, Inc. were instrumental in the instal-
lation and evaluation testing of the proportional sampler at the Triangle
Park facility.
Mr. Robert Campbell of J-Tec Associates provided valuable assistance in
the development of the exhaust flowmeter.
The ultimate success of the program was due primarily to the concerted
efforts of the personnel within the Industrial and Environmental Products
Operation of the Aeronutronic Division. In particular, Mr. R. L. Kopp
was responsible for the detailed design and implementation of the signal
processor electronics.
xiv
-------
SECTION 1
INTRODUCTION
The purpose of this program was to develop and evaluate a proportional
sampler for determining automobile exhaust mass emissions. The principal
requirement for this device is to sample the exhaust at a mass flow rate
which is a constant, known fraction of the instantaneous total exhaust
mass flow rate. Unlike a constant volume sampler (CVS) the proportional
sampler does not dilute the exhaust with ambient air. This can result
in an order of magnitude increase in the sample concentrations, which
in turn could permit quantitative identification of trace pollutants in
the exhaust.
Specific design requirements were as follows:
(1) Maintain sampling proportionality throughout the standard
EPA Urban Driving Schedule (LA-4-S-3, as specified in the
Federal Register, Vol. 37, No. 221, dated Wednesday,
November 15, 1972), including idle, acceleration, cruise,
and deceleration modes.
(2) Provide a condenser and trap for removal of water and
heavy hydrocarbons from the sample.
(3) Maintain the sample lines preceding the trap at a minimum
of 93°C.
(4) Hold the entire volume of sample obtained from an
LA-4-S-3 test in a single bag.
(5) Withstand exposure to high temperature, corrosive exhaust
gases. Only 316 stainless steel or Teflon or equivalent
materials shall contact the sample.
(6) Sample the exhaust at a minimum rate of one part sample
per thousand parts exhaust.
(7) Provide a sample flow rate of at least six liters per
minute under maximum conditions.
-------
(8) Insure that the sample is not diluted in any manner.
(9) Display cumulative exhaust volume and instantaneous
system pressures and temperatures.
(10) Provide 0.5 second response in the measurement of exhaust
temperature.
Performance objectives of the device were as follows:
(1) Agreement within +3 percent by comparison with a constant
volume sampler for masses of hydrocarbon, CO, oxides of
nitrogen, and C02 in an LA-4-S-3 test. Test-to-test
repeatability of +5 percent for these specie mass data.
(2) Exhaust flow metering accuracy (cumulative volume) of
+1 percent.
(3) Exhaust sampling rate of at least 40 Hertz under maximum
conditions.
(4) Exhaust flow range capability of 10 to 850 cubic meters
per hour (m^/hr).
(5) Tailpipe pressure disturbance of 1.24 kiloPascals (KPa)
or less under maximum conditions.
(6) Achievement of performance goals for exhaust temperatures
up to 450°C.
(7) Satisfactory completion of at least 8000 LA-4-S-4 tests
without major component failure.
The program activities encompassed the design, manufacture, checkout,
calibration, and test of the device. Engineering liaison was provided
during the installation and initial evaluation testing of the device at
the EPA Triangle Park facility. Operating and maintenance instructions
were also provided for the device.
This Final Report documents all technical activities performed during
the program. A summary of the program is provided in Section 2, includ-
ing the equipment design, test results, conclusions and recommendations.
Detailed discussions of the equipment design and test results are then
presented in Sections 3 and 4, respectively. A detailed Requirements
Specification for the proportional sampler is provided in Appendix A.
Appendix B is an Evaluation Test Plan for use at the EPA Triangle Park
facility.
-------
SECTION 2
SUMMARY
EQUIPMENT DESIGN
The proportional sampler design is based upon the following basic concept.
A gas flowmeter is used to measure the total dynamic exhaust mass flow
from the tailpipe. A sample of the exhaust is drawn from the total flow
through a sample flow control valve. Operation of this valve is regulated
by signal processing electronics so as to maintain the sample flow at a
constant mass fraction of the total exhaust flow. The sample flow is kept
above its dewpoint using a heated line up to the inlet of a refrigerated
condenser and trap. The dried sample is filtered and collected in a
sample bag. As with a constant volume sampler (CVS), the sample bag then
contains a mass-weighted average of the vehicle pollutant concentrations
over a dynamic driving schedule.
The exhaust flowmeter selected for the proportional sampler uses the
principle of vortex shedding to measure the gas volume flow rate. This
type of flowmeter features good repeatability, good dynamic range, rapid
response, low pressure loss, and good durability. The flowmeter consists
of a duct with a small diameter rod positioned crosswise to the flow.
As the flow passes the rod a stable pattern of vortices is established
behind the rod. The rate (i.e., frequency) at which vortices are shed
from the rod is approximately proportional to the volume flow rate of gas
through the duct. The vortex shedding frequency is detected using an
ultrasonic beam transmitted laterally across the duct behind and perpen-
dicular to the rod. Each vortex scatters the ultrasonic beam, and the
result is an amplitude-modulated signal. This signal is converted by
flowmeter electronics into a pulse train with a frequency equal to the
vortex shedding frequency.
The sample flow control valve is a solenoid driven flapper valve operated
in an on/off pulsed mode. A fixed valve-open interval is used, so that
the average sample flow is then (approximately) proportional to the valve
pulse frequency. This approach to the valve design was selected to pro-
vide good stability and insensitivity to contamination in the sample flow
metering. The valve is a modified missile control hot gas valve with
exceptional durability for the proportional sampler application. The
valve response is proportional to its instantaneous pulse frequency, which
-------
may be a maximum of 40 Hertz. Since the valve frequency is approximately
proportional to the exhaust flow rate, the valve response to dynamic
pollutant concentration variations tends to minimize distortion in the
final mass-weighted-average sample.
The principal function of the signal processing electronics is to control
the sample valve pulse frequency, and thereby maintain the sample mass
flow at a constant fraction of the exhaust flow. This requires the
signal processor to compensate for (1) flowmeter nonlinearity, (2) vari-
able exhaust pressure and temperature, (3) variable gas temperature at
the sample valve inlet, and (4) leakage through the sample valve in the
valve-closed state. Secondary functions of the signal processor are to
totalize the exhaust mass flow and provide system parameter output
signals. The signal processor consists of a special purpose digital
computer which evaluates the system equations in real-time. An absolute
pressure transducer and two platinum-wire resistance temperature sensors
comprise the essential instrumentation.
The complete sample flow system includes a probe, pre-filter, flow con-
trol valve, heated transfer line, refrigerated condensor, condensate
trap, filter, pump, diversion valves, collection bags, purge air pump,
and supplemental system pressure and temperature indicators. The propor-
tional sampler operates on 115 VAC power. The electrical system includes
DC power supplies, signal processor and sequential control electronics,
pumps, valves, heated line controller, refrigeration unit, and cooling
fans. The equipment is housed in a three bay console, with an externally-
mounted exhaust heat exchanger.
Operation of the device is consistent with either the 1974 CVS or 1975
CVS-CH federal test procedures. An operator's control panel provides
for (1) selection of exhaust flow range, (2) automatic test sequencing
with selectable test intervals and optional remote start, (3) independent
transfer of any sample to an analysis system, (4) selection of probe
flush, condensate trap drain, or bag purge functions, (5) readout of
cumulative test time, cumulative exhaust volumes, and key operating
parameters, and (6) output of signal processor parameters to external
recorders.
Development tests of the proportional sampler established the need for
acoustic muffling and cooling of the exhaust upstream of the flowmeter.
Both needs were met by using an open cycle (i.e., no water return) water-
to-exhaust heat exchanger. Ambient temperature facility water is passed
in counterflow through a cooling coil assembly mounted inside an exhaust
plenum tank. The plenum tank, besides housing the heat exchanger, acts
as a low-pass acoustic filter against tailpipe pressure fluctuations.
To maintain the proper time-phase relation between the exhaust and
sample flows, the sample probe is located upstream of the plenum tank.
In this configuration the transport delay of exhaust from the tailpipe
to the sample probe is minimized, while the delay in flowmeter response
to exhaust flow rate changes remains negligible.
-------
TEST RESULTS
Testing activities during the program included (1) development and cali-
bration of the exhaust flowmeter, (2) development and calibration of the
sample flow control valve, (3) development of the proportional sampler
configuration, and (4) evaluation of the final device configuration.
The results of these tests are summarized below.
The exhaust flowmeter was developed and initially calibrated by the
supplier, J-Tec Associates. The flowmeter configuration was similar to
that tested in an earlier program (Reference 1) involving direct measure-
ment of automobile exhaust flow. The supplier's calibration data, using
a laminar flow element (LFE) as the reference, indicated satisfactory
operation over the flow range from 10 cubic meters per hour (m^/hr) up
to over 850 m-Vhr. The calibration was independently verified at
Aeronutronic using three LFE's over the range of 10 to 586 m^/hr. As
expected, the flowmeter signal is significantly nonlinear at flow rates
below 50 m-Vhr. This nonlinearity is compensated for by the signal
processor.
Development of the sample flow control valve principally involved min-
imizing the leakage flow in the valve-closed position. A maximum leakage
limit of three percent of full-open flow could not be achieved with the
original flapper/orifice design. A small stainless steel ball was used
in a modified orifice design to compensate for flapper/orifice misalign-
ment. This approach was successful in reducing the leakage flow to one-
half percent of full-open flow. The valve was then flow calibrated using
a differential weight technique. Valve flow was found to be linear up
to 40 Hertz pulse rate, with a flow rating of 8.5 liters per minute (1pm)
at 40 Hz.
Development testing of the proportional sampler was conducted in two
phases. All testing was done in the Aeronutronic Emissions Test Cell.
Vehicles with engines from 2.3 liter (4 cylinder) up to 6.6 liter (V-8)
were used in both steady-speed and dynamic cycles on a chassis dynamom-
eter. A CVS was used in series (downstream) with the proportional sampler
so that the two measurements of vehicle C02, CO, NOX, and HC mass emis-
sions could be compared for each test.
The first phase of development testing involved principally the exhaust
flowmeter. Initial tests showed large (30-50 percent low) errors in the
flowmeter output. Monitoring of the flowmeter internal electronic signals
revealed a severe interference in the vortex detection process, which in
turn led to loss of output signal pulses. Additional diagnostic testing
indicated the presence of acoustic noise in the flowmeter duct at frequen-
cies within the pass band of the vortex-shedding process. This suggested
the use of a plenum volume upstream of the flowmeter to filter the acoustic
disturbance. An experiment with two 208 liter drums connected in series
was successful in drastically improving the flowmeter signal quality.
This, along with minor adjustments in the flowmeter electronics, resulted
in good agreement between the flowmeter and the CVS in tests using the
-------
EPA urban driving schedule. Further testing and analysis led to the
design of a 237 liter tank to be installed between the sample probe and
the flowmeter.
The second phase of development testing was conducted after completion
of the proportional sampler assembly, including the external plenum tank,
but prior to development and installation of the exhaust heat exchanger.
Initial tests of this configuration showed poor flowmeter signal quality
and consequent errors in the measured mass emissions. These appeared to
be caused by a flowmeter malfunction and by resonant acoustic waves in
the ducting. Repair of the flowmeter improved the signal, but did not
wholly account for the excessive output signal error. Attempts to
attenuate the acoustic resonance conditions in the plenum tank led to
the discovery that the flowmeter error varied with the exhaust gas tem-
perature at the flowmeter. Tests with heated ambient air showed that
the flowmeter signal quality deteriorated rapidly for gas temperatures
above 65°C. An experimental heat exchanger was then tested with vehicle
exhaust. By reducing the gas temperature from over 200°C to below 50°C
the flowmeter signal error was eliminated. A permanent exhaust heat
exchanger was then designed, fabricated, and installed inside the plenum
tank.
Final testing of the proportional sampler at Aeronutronic demonstrated
the adequacy of the exchanger performance, flowmeter signal quality,
and C02 mass emissions comparisons with the CVS. The equipment was then
delivered to the EPA Triangle Park facility for evaluation testing and
acceptance. Initial testing at this facility indicated generally satis-
factory performance of the device, even for vehicles equipped with
oxidation catalysts and producing exhaust temperatures up to 450°C.
CONCLUSIONS
This program has resulted in the successful development of a proportional
sampler for characterization of automobile exhaust mass emissions. The
equipment design requirements and performance objectives were satisfied,
subject to the following limitations:
(1) Additional evaluation testing is required to more firmly
establish the accuracy and repeatability of the device
for the primary species of C02> CO, NOX, and total
hydrocarbons.
(2) Achievement of an accurate exhaust flowmeter signal
requires substantial cooling of the exhaust gas and
attenuation of acoustic noise from the tailpipe. This
typically results in condensation of water vapor from
the exhaust flow, which could complicate interpretation
of mass emissions measurements.
-------
(3) For larger engines, the additional pressure loss through
the exhaust heat exchanger can induce a tailpipe back-
pressure disturbance in excess of 1.24 kPa. However,
excursions above this limit occur only momentarily
during peak vehicle accelerations (typically once dur-
ing the EPA urban driving schedule).
(4) The exhaust flowmeter, as finally configured, has an
upper limit of 500 m^/hr for accurate operation. This
limitation is unimportant for even the largest engines
used in light duty vehicles.
(5) The response of the exhaust temperature sensor generally
exceeds 0.5 second, but is adequate due. to the large
thermal inertia of the exhaust heat exchanger.
RECOMMENDATIONS
The following general recommendations are offered for the use and possible
further development of the proportional sampler:
(1) The accuracy and repeatability of the device should be
firmly established using comparisons of C02 and CO mass
emissions between the proportional sampler and a CVS
in series.
(2) Comparisons of C02 and CO mass emissions between the
proportional sampler and a CVS should be used as
experimental controls for all testing with the device.
(3) It may be desireable to modify the sample flow system
to improve the sample integrity for some pollutants of
interest. Modifications could include using a higher
temperature heated line, bypass of the condenser and
trap, or the use of special filter media. These would
be subject to the limitation that the sample flow
remain choked across the sample flow control valve
orifice.
(4) Future versions of the proportional sampler could
employ several design improvements, including a
simplified flowmeter configuration, a signal processor
with reduced self-test capability, a membrane sample
dryer in place of the refrigerated condenser and trap,
a simplified heat exchanger assembly with reduced
exhaust pressure loss, a closed cycle exhaust temper-
ature control system, and more compact packaging of
the equipment.
-------
SECTION 3
EQUIPMENT DESIGN
The design rationale for the proportional sampler and its principal com-
ponents are presented in this section. A general discussion of the
device's theory of operation, fluid system, electrical system, console
layout, and operating characteristics is provided first. This is fol-
lowed by more detailed discussions of the exhaust flowmeter, sample flow
control valve, signal processor, and exhaust heat exchanger.
GENERAL
Theory of Operation
The following equations describe the essential characteristics of the
exhaust flowmeter, sample flow control valve, the valve control problem,
and exhaust emissions calculations.
The exhaust flowmeter provides a direct measurement of gas volume flow
rate. The gas absolute pressure and temperature must be measured at the
flowmeter to convert the volume flow rate into a mass flow rate (i.e.,
volume flow rate at standard pressure and temperature). The equation
is:
Qs - QA pi V(ps V (1)
3
where Q = exhaust standardized volume flow rate (m /hr at 20°C
S and 101.325 kPa).
3
Q = exhaust actual volume flow rate (m /hr).
A
P.. = exhaust gas pressure at the flowmeter (kPa, absolute) .
P = standard pressure (= 101.325 kPa).
S
TI = exhaust gas temperature at the flowmeter (°K).
T = standard temperature (= 20°C = 293.15°K).
-------
The sample flow control valve operates in an on/off pulsed mode. A
sample pump downstream of the sample valve provides sufficient vacuum at
the valve outlet to maintain choked (i.e., critical) flow across the
valve orifice. During the valve-open portion of a complete valve cycle
the sample flows at its maximum rate. During the valve-closed part of
the cycle the nominal flow is zero. Two non-ideal effects must also be
considered. First, the valve leaks slightly in the valve-closed posi-
tion. Second, the transient flows during the opening and closing transi-
tions may not exactly balance out. The result is the following general
equation for the average sample flow over a complete valve cycle:
A t + A, (t - t )
o o 1 c o
where q = average sample standardized volume flow rate (liters
S per minute at 20°C and 101.325 kPa).
q = sample standardized volume flow rate for the full-
m open valve (1pm at 20°C and 101.325 kPa valve inlet
conditions).
P^ = actual valve inlet pressure (kPa, absolute).
T = actual valve inlet gas temperature (°K).
A = ratio of effective/full-open sample flow during
valve-open interval.
A, = ratio of leakage/full-open sample flow during valve-
closed interval.
t = valve-open interval (sec).
t = valve cycle interval (sec).
= valve-open interval plus valve-closed interval.
The signal processor must then turn the sample valve on and off (i.e.,
open and closed) to maintain the relationship:
qo = k Q /60 (3)
S S
where k = sample proportionality constant (parts per thousand).
-------
c
t
o
QA k
q ' 60 '
(A - A,)
O J-
T /T,
s / 2 ,
Tl V Ts 1,
Combining these equations yields the following expression for the valve
cycle interval (assuming equivalent flowmeter and valve inlet pressures)
(4)
By using a fixed valve-open interval, the dynamic variables in Equation
4 are limited to QA, T^ and ^2-
The use of this sample valve technique implies discrete rather than con-
tinuous sampling of the exhaust. The possible distortion this might
create in the sample can be evaluated using the general sampling theorem
of modern communication theory (Reference 2). This states that any
2 x fm independent, discrete samples per second are sufficient to charac-
terize a function varying at a maximum frequency of fm cycles per second.
In the proportional sampler application, if the exhaust specie concentra-
tions vary at a frequency of one cycle per second, the required valve
pulse rate for a distortion-free sample would be two cycles per second.
The upper limit for the valve pulse rate is determined by the need to
provide a constant value of Ao independent of the value of tc. Valve
calibration data will show that for tc X).025 sec this constraint is
met. Thus the valve can be cycled at frequencies up to 40 Hz, for an
equivalent specie concentration bandwidth of 20 Hz. This maximum valve
pulse rate can be employed for any given test by selection of the appro-
priate peak vehicle exhaust flow rate. This "range change" capability
is provided in the signal processor and is equivalent to selecting the
proportionality constant, k.
The above equations do not account for removal of water vapor from either
the sample or the main exhaust flow. In the final equipment configura-
tion both flows will be substantially dried. Assuming that no species to
be measured are lost in the condensation processes, correction for water
removal requires only adjustments for the decrease in gas volume in the
sample collection bag and in the exhaust flow at the flowmeter. These
corrections are applied in the final calculations of pollutant mass
emissions, as described below.
The proportional sampler may be used to determine mass emissions in two
ways. First, the pollutant concentrations in the collection bag may be
measured using a separate analysis system. The pollutant mass for a
given test is then:
Mi = pi Xi V (5)
where M. = mass of ith specie emitted (grams).
p. = density of ith specie at standard temperature and
pressure
10
-------
X. = volume fraction of ith specie in the exhaust,
averaged over the test.
V = cumulative exhaust standardized volume over the test
Typically the exhaust heat exchanger will remove most of the water vapor
in the original exhaust. The signal processor provides for accumulation
of the temperature- and pressure-corrected flowmeter signal as follows:
V
where V = cumulative dried exhaust standardized volume over the
test (m3).
t = test interval (sec).
Since only water vapor is (partially) removed, the metered volume will
differ from the original exhaust volume by the factor:
V = V (1 - X/d - X) (7)
where X' = volume fraction of water vapor in the exhaust
2 gas at the flowmeter.
X^ _ = volume fraction of water vapor in the original
rl_ U
2 exhaust.
The partially dried sample will also experience a volume decrease, with
the following effect on measured specie concentrations:
where X1.1 = volume fraction of ith specie in the dried sample,
averaged over the test.
X|! _ = volume fraction of water vapor in the dried sample
Equations 7 and 8 may be used with Equation 5 to determine the final
pollutant mass emissions:
M. = p. XJ V (1 - X0)/(l - X0) (9)
11
-------
The two water vapor fractions required in the evaluation of Equation 9
may be estimated using the following expressions:
where (SVP)
(SVP)
XH20 = (SVP)4/(PB ' V
saturated water vapor pressure at temperature
T! (kPa).
saturated water vapor pressure at the refrigerated
condensate trap temperature, TA (kPa) .
P-n = local barometric pressure (kPa, absolute) .
o
P, = vacuum reading at the condensate trap (kPa,
vacuum gage) .
In evaluating Equations 10 and 11, the assumption of average parameter
values is sufficiently accurate for normal testing. Note that estimation
of the water vapor fraction in the original exhaust is not required for
this type of mass emissions measurement.
The second method of determining mass emissions is to directly measure
the mass of pollutant contained in the sample flow extracted from the
total exhaust over the test interval. This could be done by filtering
the sample flow with specie-selective traps, and measuring the mass of
each collected specie, m^. This is equivalent to measurement of the
specie volume fractions and the cumulative sample standardized volume:
mi
= Pi Xi V
where
6oToo
(dt)
and where v = cumulative (undried) sample standardized volume
over the test (m-^)
Then Equations 3, 6, 7, 12 and 13 may be combined to yield the relation:
, (1 - w
m
i = pi xi v
ooo
12
-------
Finally, Equation 5 is employed in rearranging Equation 14 to yield the
following result for the total mass of the ith specie:
(1 ~ XH n}
1000 2U ,.,,.
M. = m. - : - . ~r- - - - r- (15)
i k (1 - X
The proportionality constant, k is known from the sample valve calibra-
tion data. The water vapor correction factor in Equation 15 may be esti-
mated using the following expression:
(1 - XJ Q)
Tl - x'n) ' Ao + Al XC09 + A2 XCO
h_U 2.
with the coefficients defined as:
Ao = (1 - X^ Q) (1 + 0.23 H)
(1 - X^ Q)
A = 0.5y (1 + 0.115 H) T - ~-r
1 (1 - X)
(i - x^ Q)
A = [0.5y - (0.5 - 0.25y) 0.23 H] T - ^
2 (1 " X}
and where H = ambient air absolute humidity (gm water per kgm
of dry air) .
y = hydrogen/carbon atomic ratio in test fuel.
X" = volume fraction of CO 2 in dried sample,
2 averaged over the test.
X" = volume fraction of CO in dried sample, averaged
over the test.
Equation 16 is derived assuming combustion of fuel to C02, CO, and t^O
at air/fuel ratios greater than stoichiometric. Note that measurement
of ambient humidity is a standard requirement in most emissions test
facilities.
13
-------
Fluid System
The proportional sampler fluid system is illustrated schematically in
Figure 1. The system consists of three principal flow networks main
exhaust, sample, and purge air. These are described below.
Main Exhaust Flow - The vehicle or engine exhaust first passes through
an inlet duct where the sample is extracted. The exhaust then enters
the external heat exchanger, passing across a column of cooling coils
and out through a return duct to the flowmeter. The design of these
major components of the fluid system is presented in subsequent
paragraphs.
One temperature and three pressure measurements are provided in the
exhaust ducting. First, at the sampler inlet, is a low-range pressure
gage. This gage is mounted on the lower front panel, and provides a
dynamic measurement of exhaust backpressure for visual monitoring only.
An absolute pressure transducer is mounted on the flowmeter tube to mea-
sure the gas pressure P]_. This transducer is the strain-gage diaphragm
type and its signal is input to the signal processor. The gas tempera-
ture at the flowmeter, T]_, is measured using a platimum-wire resistance
temperature sensor. This probe is mounted in the exit duct just down-
stream of the flowmeter. As with the Pj_ transducer, the T]_ probe inter-
faces electrically with the signal processor. Either signal may be
visually monitored using a meter on the operator's control panel.
Sample Flow - The exhaust sample train includes the probe, prefilter,
sample flow control valve, heated line, refrigerated water bath and con-
densate trap, sample filter, vacuum adjustment valve, sample pump, bag
input solenoid valves, sample collection bags, and bag output valves.
The principle features of these components are noted below.
The sample probe is located in the inlet duct. The probe rises vertically
to the prefilter. The prefilter has a stainless housing and element
with Teflon seals. The probe orientation and prefilter are intended to
minimize the entrainment of particulates in the sample flow. A variety
of elements may be used in the prefilter. The minimal filtration level
is 40 to 50 )_im, with an approximate pressure drop of 0.25 kPa at the
maximum sample flow rate.
The design of the sample flow control valve is reviewed separately in a
subsequent paragraph.
The sample line from the sample valve outlet to the condenser coil in
the refrigerated water bath is temperature-controlled up to 93°C. The
line is 4.7mm inside diameter Teflon traced with electrical heating wire,
thermal insulation and a protective casing. The heating wire extends
around the base of the sample valve, around the valve inlet tubing, and
around the sample probe down to the fitting on the main inlet duct.
Separate pieces of thermal insulation and protective casing are clamped
to these segments. The heated line junction box is located at the inlet
14
-------
0)
JS
o
co
S
OJ
H
3
n)
CO
a)
C
o
H
4-1
n
O
(X
O
)-l
Pn
60
rH
15
-------
to the condenser coil. The junction box houses the line temperature
sensor (a thermistor capsule) and lead wire/heating wire connections.
The refrigerated water bath assembly consists of the water tank, refrig-
eration condensing unit, refrigeration evaporator coil, temperature con-
trol switch, sample gas condenser coil, water trap assembly, submersible
pump assembly, and water tank cover.
The water bath assembly condenses water vapor from the sample gas stream,
collects the water droplets, and provides a means of removing the con-
densed water from the console system. The sample gas is passed through
the condenser coil and water trap. The coil and trap are immersed in
the refrigerated water bath. Since the water temperature in the bath
is maintained at 2 +1°C, water vapor in the sample gas stream condenses,
forming droplets on the inner walls of the condenser coil. The droplets
are moved by gravity and the influence of the moving gas stream to the
water trap assembly.
The water trap assembly separates the water droplets from the gas stream.
A 50 cm3 storage reservoir stores the condensed water. Level sensing;
probes indicate when the trap is 3/4 full and completely full of water.
Separation of the water droplets is achieved by means of a spiral dif-
fuser in the upper part of the trap assembly. The diffuser imparts an
outward vortex motion to the gas stream. The diffuser is designed so
that the droplets are blown onto the inner vertical cylindrical surface
of the trap reservoir. The water droplets run down the wall, into the
bottom of the reservoir. The collected pool of liquid in the reservoir
rotates under the influence of the swirling gas stream. This rotation
keeps the top surface of the liquid pool flat and thus prevents splash-
ing. Splashing of the surface would cause droplets to be picked up by
the gas which exits at the center, top of the trap assembly. Further
protection against droplet pickup in the exit gas stream is provided by
the outlet baffle.
The condensate water level probes are fabricated of single strand pure
nickel wire. They are arranged so that the liquid level will contact
one pair at the 3/4 full point and the other pair when the trap is full.
The probes are electrically insulated and are unaffected by water droplet
formation or the bath water.
The final sample filter is a clamshell style which uses 7 cm diameter-
filter papers. The filter is plumbed so that the contaminated side of
the element is visible when the filter is open. For general testing
Whatman grade GFC or GFA glass fiber filter papers should be used.
A needle valve is used between the filter and sample pump to control the
vacuum at the condensate trap during test. The valve is adjusted to
achieve a vacuum of 40 kPa at the maximum sample flow of 8.5 1pm. A
higher vacuum than this could result in an above ambient dewpoint for
the sample leaving the condensate trap.
16
-------
The sample pump is a two-headed diaphragm type with a. 250 w split-phase
motor. The internal surfaces of the pump are teflon coated. The two
pumping chambers are plumbed in series. This pump is identical to the
purge pump. The sample pump relief valve is set to crack at 35 kPa.
Valves VI through V6 control the flow into and out of the sample collec-
tion bags in accordance with the three test phases. These valves, as
well as the condensate trap drain valve V7, are identical normally-closed
solenoid valve with 4.7mm orifices and viton seals. The valves are
rated for continuous duty on 24-28 VDC and will open against a maximum
inlet pressure differential of 689 kPa. The diversion valve logic is
summarized in Table 1.
The three sample bags are fabricated from clear Tedlar material. The
edges are double heat-sealed. A 9.5mm diameter stainless distribution
tube is inserted diagonally into the bag to avoid stratification of the
sample gas. The bags are hung from support tubes mounted on the top of
the console. The distribution tubes are cantilevered from bulkhead
fittings on the internal structure.
The sample line instrumentation includes the following:
(1) Sample valve inlet temperature, T2, measured using
a platinum-wire resistance temperature sensor
(identical to Tl probe) mounted in the sample valve
inlet tube (horizontal orientation). The probe
excitation, signal conditioning, and readout are
included in the signal processing electronics.
(2) Heated line temperature, T3, measured using a
thermistor probe mounted inside the heated line
junction box. Readout of T3 is accomplished
indirectly through the temperature controller
mounted on the console front panel.
(3) Refrigerated water bath temperature, T4, measured
using a liquid-expansion thermometer. The probe
is immersed in the water bath with a remote dial
mounted on the control panel for visual monitoring.
(4) Sample line vacuum/pressure, P2, measured using a
compound bourdon-tube gage mounted on the control
panel.
(5) Sample manifold vacuum/pressure, P3, measured using
a compound bourdon-tube gage mounted on the control
panel.
Purge Air Flow - Four purge functions are provided to (1) backflush the
pre-filter and sample probe, (2) drain the condensate trap, (3) fill the
sample bags, and (4) evacuate the sample bags to complete the bag rinse
17
-------
Table 1. SAMPLE DIVERSION AND PURGE VALVE LOGIC
Function
Test Phase I
Test Phase II
Test Phase III
Analyze Bag 1
Analyze Bag 11
Analyze Bag III
Probe Flush
Trap Drain
Bag Fill
Bag Evacuate
Valve
VI
X
X
X
V2
X
X
X
V3
X
X
X
VA
X
V5
X
V6
X
V7
X
X
V8
X
X
X
V9
X
V10
X
Vll
X
X
X
V12
X
V13
X
18
-------
cycle. Ambient air is used as the purge gas. The air is drawn from the
console interior through a filter and moisture separator. The purge air
is pumped by a two-headed diaphragm pump (identical to sample pump)
plumbed with the chambers in parallel. The purge flow control valves, V8
through V13, are identical normally-closed solenoid valves with 9.5mm
orifices. The valves are rated for continuous duty on 24 to 28 VDC and
will open against a maximum inlet pressure differential of 13.8 kPa.
Since the purge pump develops pressures greater than this the purge func-
tions must be implemented with the sample lines vented to ambient prior
to initiating the desired function. The purge pump relief valve is set
to crack at 35 kPa. The trap drain valve, identical to the sample diver-
sion valves, is oriented in a reverse flow direction to prevent leakage
from ambient into the sample line during test when the trap is under
vacuum. The purge valve logic is summarized in Table 1.
Electrical System
The proportional sampler electrical system is diagrammed in Figure 2.
The power distribution and general electrical functions of the console
are reviewed below. A detailed discussion of the signal processor design
is reserved for a subsequent paragraph.
The console operates on two 115 VAC, 60 Hz, single phase power inputs,
each with a maximum current of 15 amps. Facility power is routed into
the relay and fuse box (lower right, rear of console) through circuit
breakers. The two primary distribution lines and a power control line
are each separately fused. The console power switch energizes a two-
pole power control relay (Kl). One primary power line serves the sample
pump, purge pump, and refrigerated water bath. The other primary power
line serves the electronics and valves' power supplies, exhaust flowmeter,
heated sample line and cooling fans.
Four DC power supplies (lower, right, front of console) are used. The
+5 VDC power is distributed within the electronics assembly for the
transistor-transistor logic. The +15 VDC supply is used for analog
input and output signal conditioning and excitation. The +28 VDC supply
is used for the solenoid valves and relays (K2, K3, K4). The +10 VDC
supply is used in series with the +28 VDC supply to provide +38 VDC to
the coils of the sample flow control valves. The supplies are all
adjustable and are short-circuit protected.
The sample flow control valve operates on +38 VDC to two opposing
electromagnets. Dropping resistors are used in series with the coils
to limit the steady-state current. Solid-state coil drivers are used,
under control of the signal processor logic, to energize one or the
other coil (never both). The coils are energized only when the "Sample
Valve Power" switch is closed. This switch also energizes relay K2 to
turn on the sample valve cooling fan.
19
-------
D S
I1 2
SI 2
o d
CC
t-
1
I 5 y
1 1' i
8 & 2
W 1 LU
ni i
LU LUI Jii
_l 1 ^
111 1
1 . '
§ §
in t"
+ *~
>- J CC
_J Q. UJ
ft! D. (-
S D CO UJ
D CO Z 5
) BAG VALUES, V4-V6
|
|
1 ) ANALYZE BAG 1, 1
|
H SAMPLE PUMP |
"i
L
{ PURGE PUMP |
HEATED SAMPLE LINE
HTEMPERATURE CONTROLLER! ,
J i!
LO .
s LJ
TRAP FULL SENSOR
REFRIGERATED
WATER BATH
SUBMERSIBLE PUMP
3
J
-L i LU U
liisi
_r
i
i
i
i
i
j
tO
t~>
6C
0)
0)
4J
03
>^
0)
cfl
O
t-l
4J
U
1-1
0)
to
CO
O
r!
O
ft
O
H
P-i
01
Q)
bO
H
20
-------
The sample pump and purge pump use a common 115 VAC line controlled
through relays K3 and K4. The 115 VAC input to K3 is through the
normally-closed contact of K4. Thus when K4 is energized, the purge
pump is turned on and the sample pump is turned off independent of
whether K3 is energized or not. The purge pump is turned on through K4
by the purge pump logic whenever one of the four purge functions is
selected at the operators control panel. The sample pump is turned on
through K3 by the sample pump logic when the "Sample Pump" switch is
closed. The sample pump logic will de-energize K3 if a trap-full indica-
tion is received from the condensate trap full sensor. Thus, the sample
pump is protected from ingesting water from a full trap. Water in the
pumping chambers could damage the pump due to overpressurization.
The heated sample line receives 115 VAC power through a temperature con-
troller mounted on the console front panel. The controller uses a solid-
state Triac device to regulate the load voltage. A 5 ohm load resistance
(relay and fuse box) is used in series with the heated line to obtain
the minimum load for stable control. An adjustable current limit is pro-
vided with the controller. A thermistor is used to sense line tempera-
ture. The load voltage is then controlled in proportion to the set point
versus sensor temperature difference.
The water bath refrigeration unit is thermostatically controlled with an
adjustable switch. The switch is set to the desired temperature (about
2°C). When the cooling water rises above the set temperature, the switch
closes and turns on the compressor to circulate the refrigerant. The
submersible pump in the water bath runs continuously. The trap-full
sensor within the condensate trap is electrically isolated from the
assembly.
Five cooling fans are used in the console to circulate ambient air. Two
fans are located above the electronics assembly (top, right, front of
console) and run whenever console power is on. Two fans are located on
the console floor by the exhaust flowmeter transmitter and receiver
crystals. These are turned on by an independent control panel switch.
The fifth fan is used to cool the sample valve electromagnets (lower,
left, front of console) and is on whenever the sample valve power switch
is closed.
Console Layout
The physical construction of the proportional sampler is illustrated in
Figures 3 through 7. The overall dimensions of the console are 234 cm
wide by 91.5 cm deep by 206 cm high, including the heat exchanger tank.
The exhaust enters the front of the console at the lower left. After
passing the sample probe, the exhaust is ducted into the heat exchanger
tank. It leaves the tank near the top and is returned to the console
through a vertical duct. The main flowmeter duct passes across the lower
rear of the console and exits at the right side.
21
-------
a
o
0)
rH
O
CO
g
o
cfl
co
n)
§
H
4-1
p
O
P.
O
H
PH
-------
OPERATOR'S CONTROL PANEL
FLOWMETER ELECTRONICS
ELECTRONICS ASSEMBLY
HEATED LINE
CONTROLLER
Figure 4. Control panel and electronics assembly
23
-------
REAR OF
ELECTRONICS
ASSEMBLY
POWER
SUPPLIES
Figure 5. Console interior (right side view)
24
-------
CU
O
B
0)
60
I
O
-------
REFRIGERATED
WATER BATH
ASSEMBLY
DIVERSION
VALVES
(V1 - V6}
SAMPLE PiROBE,
PREFILTER,AND
VALVE
Figure 7. Console interior (left side view, with heat exchanger removed)
26
-------
At the center front bay of the console is the operator's control panel,
flowmeter electronics, sample filter, heated line controller, and tail-
pipe backpressure gage. Console operations are reviewed in a subsequent
paragraph.
The electronics assembly, including sequencing logic, signal processor,
purge logic, pump logic, and valve, relay, and coil drivers, is housed
in the right front bay of the console. The assembly consists of a verti-
cal swing-out file with a hinged door (shown open in Figure 4). The
individual components are mounted on 36 cards (6 rows by 6 columns)
attached to the file frame. Most cards are connected to a +5 VDC and
ground busses with integral noise suppression. The cards used for digi-
tal circuits feature integrated-circuit dual-in line-package (DIP)
sockets. Individual digital components are simply plugged into these
sockets. The sockets have pins extending through the card for wire-
wrapping from the back side. This method of construction is uniquely
suited for easy troubleshooting and repair of the electronics. The cards
used for analog circuitry are specially fabricated from vector-board and
feature primarily soldered connections within the card. Two cards are
dedicated to input/output cables, one to the control panel and one to
the sample valve, solenoid valves, relays, etc.
Visible in the interior views are the rear of the electronics assembly,
power supplies, flowmeter, sample bags, pumps, water bath assembly,
solenoid valves, heated line, sample valve coils, and transducers. In
operation the console can be completely enclosed by the cabinet doors.
Operating Features
Operation of the proportional sampler involves the following typical
pretest, test, and post-test steps.
Pretest - The operator first completes a purge cycle using the Purge
Control selector switch. This sequence includes evacuation of residual
sample from the bags, a flush of the sample probe, draining the sample
condensate trap, filling the bags with purge air, and a final evacuation
of the bags. The sample filter is changed and the heat exchanger tank
is drained of exhaust condensate from the previous test.
The time intervals for the four phases of testing (in accordance with
the 1975 CVS-CH procedure) are dialed in using the thumbwheel switches.
Analog recorders may be set up to record any of five dynamic test
parameters exhaust actual volume flow rate, exhaust standardized
volume flow rate, gas temperature and pressure at the flowmeter, sample
temperature at the sample valve inlet, and the sample open/close command
signal.
The operator may select one of eight maximum expected exhaust flow rates,
such that the sample valve pulse rate is maximized up to the 40 Hz limit.
The operator checks the heated line and water bath temperatures and
insures that the bag valve and flowmeter self-test switches are off.
27
-------
Just before the test starts, the operator opens the exhaust cooling
water valve to obtain the desired water flow rate, turns on power to
the sample valve, and turns on the sample pump.
Tjsst. - The test sequence is initiated by momentarily closing the Start
Test switch. Panel jacks are provided in parallel with this switch to
enable setup of a remote start switch. The console then automatically
sequences through the four test phases, including switchover of sample
flow from bag to bag. The current test phase and elapsed time in that
phase are continuously displayed on the control panel. The cumulative
volume of exhaust (cubic feet at standard conditions) is continuously
displayed for the three sample collection phases (e.g., cold transient,
stabilized, and hot transient).
The operator monitors the various pressures and temperatures and esti-
mates average values of sample line vacuum, water bath temperature, and
exhaust temperature at the flowmeter. Values of ambient barometric pres-
sure, temperature, and humidity are usually required.
Post-Test - After completion of the test sequence the sample pump sample
valve power and cooling water are turned off. The essential test data
are recorded for each phase, and the bag samples are transferred to an
analysis system using the individually controlled bag drain valves (V4,
V5, V6). Note that, if desired, each sample may be analyzed immediately
after completion of its corresponding test phase without affecting subse-
quent samples.
EXHAUST FLOWMETER
The purpose of the exhaust flowmeter is to measure dynamically the
vehicle exhaust volume flow rate. The flowmeter output signal is input
to the signal processor as the basic parameter for controlling the
sample flow rate.
The flowmeter selected for the proportional sampler is the vortex-
shedding type. A summary of the theory of operation of this flowmeter
is presented in the following paragraph, and is based upon the review
provided in Reference 3.
The flow of fluid across a bluff (e.g., cylindrical) body will form a
wake downstream of the body due to separation. For sufficiently high
flow velocity, the wake will consist of a stable pattern of vortices,
which have detached from alternate sides of the body and are moving down-
stream with the flow.
Numerous investigations have shown that over a wide range of flows, the
following non-dimensional parameter, termed the Strouhal number, is
practically constant:
S = fd/U (17)
28
-------
where S = Strouhal number
f = vortex shedding frequency
d = body dimension normal to flow
U = fluid velocity
The principal deviation from constancy occurs at flow velocities just
above initiation of the stable vortex pattern. However, the deviation
is correlatable with the flow Reynolds number:
Re = pUd/y (18)
where Re = Reynolds number
p = fluid density
y = fluid dynamic viscosity
In the flowmeter application the following relationship is employed:
Q % D2 U ^ D2 df/S {Re} (19)
where Q = fluid volume flow rate
D = duct diameter
U = average fluid velocity in the duct
The repeatability of the phenomenon -permits calibration of such a flow-
meter over a wide range of operation. Development efforts have focused
on optimum body shapes and reliable methods of vortex frequency
detection.
The vortex frequency is detected, in the J-Tec unit, by passing an ultra-
sonic sound beam through the vortex trail. As the vortices intersect
the sonic beam a modulation is imparted to the sonic signal, produced
by a refraction-like effect when the sonic rays have their apparent
velocities modified by the rotational velocity components of the vor-
tices. Therefore, the received sonic signal contains an amplitude modu-
lated component whose modulation frequency is the vortex-shedding fre-
quency. The modulation amplitude imparted by the vortices varies as a
function of the fluid velocity and the fluid density. However, since
only the modulation frequency is of interest, these factors have no
effect upon the measurement. The modulation frequency is detected and
processed into a square wave whose frequency is again the vortex-shedding
frequency and this is the principal output of the flowmeter. This can
be utilized in a frequency to voltage converter to produce an analog
output or can be directly counted in a digital counter for display pur-
poses or for external usages.
29
-------
Of importance in this type of flowmeter is the fact that the vortex
frequency/volume flow scale factor is only a function of the vortex rod
diameter. This implies that the system does not require periodic recali-
bration except in those cases where the vortex rod is eroded. Also,
since the output is a frequency, amplifier drifts or offsets do not
affect accuracy. The flowmeter electronics include bandpass filters to
reject DC signal shifts and noise outside the vortex frequency range of
interest.
An alternate flowmeter supplier, Eastech, Inc., was considered at the
outset of the program. The Eastech design employed a thermistor as the
vortex frequency detection method. The thermistor was mounted external
to the main duct with connecting tubes to either side of the bluff body.
The vortex trail would induce small flow perturbations in the tubing at
the shedding fequency, and the thermistor signal could thus be converted
into a pulse train. The two flowmeter designs were compared on the
basis of metering accuracy, dynamic range, response, pressure loss,
durability, cost, and development experience in the direct metering of
automobile exhaust. The J-Tec design showed slight advantages in
dynamic range, response, and pressure loss. However, the most signifi-
cant factor was judged to be the development experience acquired with
the J-Tec design (Reference 1). Although both designs appeared feasible,
the Eastech model had not been tested with actual exhaust flow. Since
the J-Tec model had been tested with apparent success in such an appli-
cation, it was selected for use in the proportional sampler.
Detailed design of the flowmeter was performed by J-Tec, based upon the
previous exhaust flowmetering application. The design parameters of
significance include duct diameter, bluff body shape and cross-flow
dimension, acoustic beam frequency, beam transmitter and receiver con-
figuration, and duct length. The design criteria are:
(1) The strut Reynolds number (pUd/y) must be greater
than 50 for a stable vortex pattern to form.
(2) The duct Reynolds number (pUD/y) must be greater than
900 to prevent relaminarization.
(3) The maximum flow velocity measurable is limited by
downstream drift of the acoustic beam, away from the
receiver. For a given maximum volume flow this may
require a larger duct diameter than that compatible
with the above criteria. Alternatively, the meter's
dynamic range may be compromised.
(A) The receiver electronics were limited to 10 kHz maximum
shedding frequency.
(5) Efficient acoustic coupling is required between the
transmitter/receiver transducers and the gaseous fluid
medium.
30
-------
(6) The beam transducers must be protected from the high
temperature exhaust.
(7) The vortex scattering of the acoustic beam must have
sufficient signal/noise ratio to avoid interference
from fluid turbulence in the detection of the shedding
frequency.
(8) The flow velocity profile across the duct should be
essentially fully-developed.
Finally, the performance requirements for the flowmeter were originally
specified as follows:
(1) Meter ambient air and internal combustion engine exhaust.
(2) Volume flow range from 17 to 510
(3) Static repeatability of 1/2 percent of reading.
(4) Resolution of 15,000 pulses per cubic meter (minimum).
(5) Time response equivalent to five pulse interval.
(6) Pressure loss less than 0.21 kPa at 510 m3/hr.
(7) Gas temperature from 15°C to 260°C, with a goal of 425°C.
The resulting flowmeter configuration consists of a duct assembly and
separate electronics chassis. The main duct is 137 cm long by 7.62 cm
outside diameter. Fabricated from 0.165 cm thick stainless steel tubing,
the internal flow diameter is 7.29 cm. The vortex-shedding strut is
114 cm downstream of the inlet flange, providing a flow development
length of 15 diameters. A flow straightener consisting of 0.3175 cm
cell size by 2.54 cm length stainless steel honeycomb is installed in
the duct inlet to reduce flow turbulence.
The strut is a 0.239 cm diameter stainless steel rod mounted vertically
across the duct. The ultrasonic beam is transmitted laterally across
the duct approximately 0.5 cm downstream of the rod. The rod diameter,
being slightly less than originally planned, provides a resolution of
approximately 24,000 pulses per cubic meter of gas.
The beam transmitter and receiver transducers are piezoelectric crystals.
The crystals are mounted in thermal-standoffs for protection from high
temperature gas. Aluminum honeycomb was used inside each thermal stand-
off's cavity to collimate the beam. It was found necessary during
development testing of the proportional sampler to replace the aluminum
honeycomb with stainless steel honeycomb for improved durability.
31
-------
The electronic chassis includes a power supply, transmitter signal
driver, receiver signal conditioning, pulse and analog output signal
drivers, and a self-test function which electronically simulates a flow
of approximately 440 m^/hr. The acoustic beam frequency was optimized
at about 151 kHz. Relatively minor changes in the receiver signal con-
ditioning were implemented during the development testing.
In light of the development test results, the use of thermal standoffs
for the beam transducers was not required. This would permit future
designs to be somewhat simpler and more reliable.
SAMPLE FLOW CONTROL VALVE
The sample flow control valve is an electromagnetically driven flapper
valve operated in an open/close pulse mode. A schematic diagram of this
type of valve is presented in Figure 8. The working end of the flapper
opens and closes the inlet restrictor periodically drawing an exhaust
gas sample. The valve critical flow orifice inlet meters the sample
flow rate. The only moving part in the valve is the flapper-armature
assembly, a cylindrical beam which is flexurally pivoted at its center.
One end of the assembly serves as a common armature between two opposed
electromagnets and the other end serves as a flapper between the restric-
tor and a fixed stop. The armature end is isolated from the valving end
with a bellows mounted at the flexural pivot point. When an electrical
command is given to one of the electromagnets, the resulting magnetic
field draws the armature to the energized magnet. The flapper on the
other end of the assembly is then forced to seat on the inlet restrictor
blocking the inlet flow. Conversely, an electrical command to the
opposite electromagnet reverses the armature motion. The flapper is
forced away from the seat onto the fixed stop opening the valve. The
valve may be operated at rates up to 40 samples per second with response
times less than 4 milliseconds.
The internal construction of a valve similar to that used in the propor-
tional sampler is illustrated in Figure 9. The actual sample valve
design was derived from an existing hot gas (1200°C) valve used in a
ballistic missile attitude control system. The valve body, flapper,
flexure, and bellows are constructed of corrosion resistant stainless
steel and nickel alloys. The armature and pole pieces are a Permendur
2V magnetic alloy. Each pair of pole pieces has a solenoid coil attached
with a top bar to complete the magnetic circuit. Each coil is wound with
1000 turns and has 22 ohms resistance. A 38 VDC power supply is used to
drive the coils, but in series with 5 ohm dropping resistors to limit
the steady-state current to 1.4 amp per coil.
32
-------
X
01
Q
LU
N
5
Is
111 V)
q
H
CO
O
Q.
Q
111
CO
O
H
CO
O
0.
LU
a.
O
cfl
t-i
H
O
H
01
O
tn
0)
.H
n)
o
o
y-i
01
tH
ex
6
to
00
0)
S-i
3
00
H
33
-------
POLE PIECES
FOR
ELECTROMAGNET
Figure 9. Flapper valve cross-section
34
-------
The principal modifications of the existing valve design were in the
restrictor/orifice configuration. These included:
(1) Reducing the flow area of the restrictor to obtain the
desired full-open sample flow of about 16 1pm. An
orifice diameter of 0.142 cm was used. To insure
pulse-to-pulse repeatability the orifice throat was
located upstream of the flapper seat.
(2) A 0.397 cm stainless steel ball was incorporated into
the restrictor at the flapper end. This insures proper
alignment of the sealing surface (spherical) against
the restrictor outlet. The ball is trapped in a cavity
at the restrictor outlet by the flapper.
(3) A threaded, rather than welded, jam nut is used to
retain the restrictor in the valve body. A silicone
0-ring is used to seal the nut against the body.
This design permits disassembly of the restrictor and
sealing ball for cleaning and/or replacement.
(4) The total flapper stroke was reduced from 0.102 cm
down to 0.063 cm. This was permitted since only one
restrictor is used, and the orifice flow area is
reduced. The stroke reduction improves the valve
response and flexure fatigue life.
(5) The second restrictor and an unrestricted part in the
valve body were capped off since they are not required.
(6) The end cap on the valve body was replaced with a pipe
thread adaptor (welded on) for use as the valve outlet.
This change was made primarily to facilitate cleaning
and reassembly of the valve.
SIGNAL PROCESSOR
The principal functions of the signal processing electronics are to
(1) receive the flowmeter, pressure, and temperature signals, (2) pro-
vide open and close command signals to the sample flow control valve,
(3) compute and display the cumulative standardized exhaust volume for
each test phase, and (4) provide analog output signals of exhaust actual
and standardized volume flow rates, flowmeter pressure and temperature,
sample inlet temperature, and valve open/close commands. The selection
of the design approach, processor equations, architecture, scale factors,
and hardware implementation are reviewed in the following paragraphs.
Three approaches to the signal processor design were initially considered.
These were: (1) analog computation, (2) general purpose microcomputer,
and (3) special-purpose digital computation. An informal comparison of
35
-------
the three approaches was made on the issues of system accuracy, develop-
ment effort, development risk, and equipment serviceability. Although
the basic requirements were defined, the extent of nonlinearity compen-
sation, computation speed capabilities, and reprogramming flexibility
were uncertain factors in the evaluation. The results of the comparison
are summarized in Table 2. The special purpose digital computer was
the selected approach. The key factors were its inherent flexibility
in tailoring of accuracy and speed characteristics, its minimum and
predictable development effort, minimum risk, and its ease of service
in the field.
Processor Equations
The processor functions may be described in terms of the computational
tasks as follows:
(1) Linearize the temperature sensors' signals.
(2) Linearize the exhaust flowmeter signal.
(3) Determine the point in time at which the sample valve
open command should be given.
(4) Determine the point in time at which the sample valve
close command should be given.
(5) Accumulate and display the exhaust standardized volume.
(6) Determine the actual and standardized exhaust volume
flow rates.
Temperature Sensor Linearization - The temperature sensors are identical
platinum-wire probes whose resistance function follows the Callendar
equation (Reference 4):
100
100
(20)
where T = temperature (°C)
R = resistance at temperature T (ohms)
R = resistance at 0°C (ohms)
o
a, 6 = constants for platinum
36
-------
Table 2. COMPARISON OF SIGNAL PROCESSOR DESIGN APPROACHES
Points of
comparison
System
accuracy
Development
effort
Development
risk
Equipment
serviceability
Analog
computation
Marginal, due
to ampli-
fier drifts
Equivalent to
S/P digi-
tal, but
less
predictable
System com-
plexity
could
increase to
achieve
required
accuracy.
Marginal, due
to exten-
sive cali-
bration
require-
ments.
General purpose
microcomputer
Good, but could be
speed-limited
Highest, due to
software develop-
ment not compen-
sated by hardware
reductions
Best flexibility, but
could require
increase in pro-
gramming effort to
achieve satis-
factory speed.
Good, but could
require specialized
programming support
Special purpose
digital computer
Can be designed
to meet
resolution and
speed
requirements
Equivalent to
analog, but
more pre-
dictable
Minimum, due to
use of read-
only memories
for non-
linearity
compensation.
Best, due to
minimal
calibrations.
37
-------
The sens:-r£ are excited using fixed current sources and the resultant
voltage amplified for input to the digital processor. Rather than use a
complex computational algorithm to convert the nonlinear voltage signal
into a temperature, the signal is linearized using what is essentially
a table look-up method. The look-up table consists of a matrix of read-
only memories (ROM's) programmed to provide a point-by-point description
of the function. The voltage is input to an analog/digital (A/D) con-
verter with a 10-bit unipolar binary output. This binary number is used
as the address for the ROM matrix. The contents of each memory location
is a 12-bit binary number representing the temperature at the sensor.
The result is in the form of a normalized exhaust gas temperature and
(using the same ROM matrix) a normalized sample valve inlet temperature:
VTs
" /T
-I1 s
= Function of temperature sensor voltage(s) (21)
Flowmeter Linearization - At the low end of the flowmeter's dynamic
range, a Ja/nolds number effect becomes significant and the actual volume
flow is not quite proportional to the vortex shedding frequency. The
Reynolds number will vary primarily with vortex frequency, secondarily
with gas temperature (through both density and viscosity), and only
slightly with gas pressure. It is convenient to measure the vortex fre-
quency by measuring the elapsed time between flowmeter pulses:
At = = flowmeter pulse interval (sec) (22)
m r
Without the Reynolds number effect the exhaust volume for flowmeter
pulse would be constant:
QA At
AQA = o^n"1 ^ f At = constant (23)
A JbUU m
where AQ. = exhaust volume per pulse (nH)
A.
This suggests the use of the flowmeter pulse to trigger a computational
sequence involving the exhaust volume increment, the corresponding sample
volume increment, and a standardized volume increment. This leads to a
relatively simple, but general, method for linearizing the flowmeter
output. The Reynolds number becomes:
<24)
38
-------
where the approximate gas viscosity temperature relation has been used:
U ^ (T)V (25)
and where V = gas viscosity temperature exponent
Since the gas pressure does not vary widely its dynamic effect on
Reynold's number is neglected. From Equation 24 it can be noted that a
single variable will describe the variation in Reynolds number with vor-
tex frequency and gas temperature. This suggests the use of a ROM matrix
in the same manner as the temperature sensor linearization. Using a
normalized temperature the form of the equation is:
AQ. = Function of At (T./T )1+V> (26)
A mis
The independent variable in this function must first be computed from
the input parameters. This can be done in two steps. First a ROM matrix
is used to evaluate the temperature power function:
[(TT/T )1+V] = Function of (T,/T ) (27)
-L s Is
Then this ROM output is multiplied by the measured value of the flowmeter
pulse interval:
Atm * (%/k) (3o)
i J
39
-------
(3) Square-root of sample inlet temperature (using succes-
sive approximation divisions)
(T /T )1/2 = (T /T ) f (T /T )1/2 (31)
j fa j£ S j£ S
(4) Standardized sample volume per flowmeter pulse
AQ (P /P )
[tQ (XQ - \^} (35)
cycle
(8) Reset cumulative standardized sample volume (after valve-
open command)
E [(Aq ) - (A At )]
INITIAL S l m
(36)
where AQ = standardized exhaust volume per
pulse (m3)
Aq = standardized sample volume per flow-
meter pulse (seconds, due to normaliza-
tion by qm)
and where the quantity t0 (As - AI) is a lumped input to
the processor.
40
-------
Cumulative exhaust standardized volume - The cumulative volume is
obtained in two steps:
(1) Pressure compensation of exhaust volume
AQs = [AQs (Pg/P^J x (P1/Pg) (37)
(2) Summation of standardized exhaust volume increments over
the test phase
V = £ (AQ ) (38)
TEST
Exhaust volume flow rates - Since these are not used directly in the
processor, they must be calculated for digital to analog (D/A) conver-
sion and output to the operator's panel. The equations are simply:
(1) Actual volume flow rate
} (39)
(2) Standardized volume flow rate
Pr_O£e s sor Architecture
A simplified block diagram of the signal processor is provided in Fig-
ure 10. The heart of the processor is a 12-bit by 12-bit multiplier
which is sequenced through the equations step by step. Division is per-
formed using the multiplier in a successive approximation mode. The
other principal components are the three ROM matrices, input switches
for digital constants, registers for intermediate parameters, and
accumulators for addition and subtraction.
The multiplier sequence of operations is summarized in Table 3. The full
sequence of computations is performed (within approximately 100 |js) for
each flowmeter pulse, and is initiated by the pulse. Evaluation of
Equations 21 and 27 is essentially immediate. Then Equation 28 is eval-
uated in the first step of the multiplier sequence, and also results in
immediate evaluation of Equation 26. Equations 29 through 33 are then
evaluated in multiplier steps 2 through 6 respectively. Equations 34,
35 and 36 are evaluated independent of the multiplier in the valve on/off
control. Meanwhile Equation 37 is evaluated in multiplier step 7 with
Equation 38 evaluated subsequently in the exhaust volume accumulator.
The volume display is incremented from the accumulator each second of
the test period. Meanwhile the multiplier sequence is completed with
the valuation of Equations 39 and 40 in steps 8 and 9 respectively.
After each sequence the analog signal inputs are updated for the next
sequence.
41
-------
t
r
00
Ki
>
0.
0.
00
'
r
o-
'
i
a
a
w
C
<
'
r
^
/i
n
<
a
t
*
c
*
£
A
0
L
a
I-
C
OUTPUT
k
5
>
]
k
C
J
J
L
J
D
»
/
c/
a
4
c
<
4
-i
0
5
£
a
2
/*
<
1-
3.
O
r-
a.
t
T-
<
E
**
3
|
h
a
2
a
j
i
r"
LL
a
5
<
(/:
^
LL
O
2
LU
<
i
<
i
i
<
:
t
P
c
^
7
C
C
k
on
B1
^
^
U
>
J
f
>
J
D
C
?
3
J
-i
<
4
M
C]
h
C
2
LL
LL
C
O
I
1
1
',
r-
<
«_r
O
H
3
1
OUTPUT 1
1
r
D
3.
z
k?
a.
.
^
-^
t?
t
h
c
C
(-
i
4 j
^ <
J
h
c
^
i-
3 :
: &
3 1
3 C
i
h
k J
N
H -.<
' J
h
2 E
* :
«
^
i
>
i
N
k
^
3
\
IN
cfl
^
«l
n)
H
O
O
01
H
4-1
a.
H
CO
i~l
o
w
M
0)
o
o
)-l
a,
c
M
H
t/3
M
H
42
-------
O
M
H
w
P-.
O
W
U
25
W
C/
H
i-l
^
,0
tfl
H
s~*
o
4-1 4-i
U 0)
3 !-
T) 01
0 S
cu J2
C
f-< f\)
3 "°
fj
O
rt
4J
(0
M
(11
O
0)
C 0
O £
H 4)
AJ ^-j
30)
t j ^
tr > C C
rH 0 O
P. -H >rl
H TO CO
4J -H -H
J *^ ^
r~1 ^ ^
3 -H -H
S Q O
^^ x*^ JX-N
OO CT> O
cs) cs n
/ N
^~ N C"
U} 4-i
H <]
~~^
EH 0- ,< CT O' O'
^-^ (X
M ^ 6 "-^
H 0^
-» tn M
CNJ C/ C'
H /* « x"^
CO M M
H H PJ
. >> >->
O flj rH rH rH
H 3 ft ft ft CU (U
CO O1 -H -H -H t3 T3
H M 4-1 4J 4-1 -H fl
H _3 ^ 3 -H -H
« « a
^^ ^^>i X^« /^ XS ^ V
rH CM ro r- cr> O
C C C C- C CS
. « . CO 0,
43
-------
The binary arithmetic scaling factors (into engineering units) for the
processor parameters are provided in Table 4. These factors were derived
in conjunction with an overall system accuracy analysis. The analysis
considered potential errors arising from the following sources:
(1) Flowmeter and sensor installations (false reading of
true parameter).
(2) Calibrations (reference reading not equal to true
parameter).
(3) Uncompensated nonlinearity.
(4) Transducer and analog signal conditioning
nonrepeatability.
(5) Digital signal processing errors (resolution or
truncation effect in the operations of A/D conversion,
ROM function generation, multiplication and division,
and D/A conversion).
The last of these error sources is primarily controlled by proper selec-
tion of the digital word length (e.g., 12 bits or 1 part in 2^-2
resolution).
Truncation errors can be assumed to accumulate in a root-sum-square
fashion for independent operations (Reference 5). The individual trunca-
tion error will decrease as the inverse of 2n where n is the number of
bits per word. Thus a 12-bit word has one-quarter of the truncation
error of a 10-bit word, and one-sixteenth that of an 8-bit data word.
It was found desirable to use a 12-bit word length for the processor,
except for the ROM address words wherein the size of the ROM matrix
doubles for each additional address bit. Also the A/D and D/A conver-
sions could be performed with sufficient accuracy using 10-bit word
lengths.
Electronic Implementations
The following paragraphs provide summary descriptions of the electronics
involved in the signal processor.
Flowmeter Input - The flowmeter pulse train is timed using a high-speed
(3/8 microsecond interval) clock input to a 16-bit counter. Receipt of
a flowmeter pulse gates the counter total into a storage register, resets
the counter to zero, and enables the counter to accumulate clock pulses
for the next flowmeter pulse.
Analog Inputs - The pressure and temperature signals are derived from
passive transducers. Excitation is provided from a +6.35 VDC reference
voltage obtained from the +15 supply using a stable zener diode. The
44
-------
o
H
U
o
X>
CO
H
1 01
rH rH
rH CO
3 0
Pn CO
4-1
ti
Cfl
O
H
n_t
T-l
C
00
T-l
CO
4 '
CO
trj
01
4-1
ti
cd
o
H
i[ [
H
c
60
iH
CO
C* 4-1
o co
H o
rj
rH 1
O
CO CO
a
H
C
0
H
4-1
Cfl
U
o
r4
M
01
4-1
cu
J^
cd
P-J
cu
3
rH
CO
3
rH
tfl
y*-^
pq
t ]
^
4-1
H
XI
4-1
tfl
O
H
14-4
H
Pi
60
T-l
CO
4-1
Cfl
tO
0)
1
CO
^4
o
CO
CU
0
o
ft
tfl
4-1
H
a
3
60
H
M
0)
CU
£3
1-1
60
C
01
CO
1
1
m
«^-
CM
\O
1
W
m
r^
CO
O
CO
4J
I-l
X*
vd
rH
S-l
CU
4-1
CO
1-1
60
CU
M
4-1
3
C
M
^^
CO
*o
p3
o
o
td 4-
N
H
V4
td
0)
C
i-l
-1
CO
rH
CM
H
' f
*J$
^^
CO
H
-^
H
to
H
rH
H_
r-^
m
^~
CM
CO
1
W
vD
O
CO
4J
H
XI
CM
rH
CM
^
O
4-1
a
H
e
o
a
cu
p
^
0
m
i-i
CO
CM
II
r^* in
m m
*^-
rH ^
O CM
CO
4-1
H
XI
co o
4J rH
H
XI
«a- co
CM S
*^j-
co
t I ^ I
0 3
U D.
cu
e <
3 -«-
^^
*fcO
o
rH
CO
1
W
-a-
o
rH
Cfl
4-1
rH
XI
o
rH
to
4J
3
a,
4-1
3
O
0
<^
^^
CO
P-,
*^^
m
^o
o
rH
CO
1
W
^o
CM
0
CO
4-1
H
XI
CM
rH
1
I-l
CU
H
rH
a.
T-l
4-1
rH
3
Cfl
PM
m
CM
co
rH
O
H
II
00 r-.
sr o o
rH CM CN1
O\ rH rH
-* O O
CO CT» CO
1 rl 1,
CO
rH
CM CM rH
rH f- O
to
4J
rH
_O
Cfl CO O
4-1 4-1 rH
rH -H
XJ XI
CM -3" CO
'^^"^^"^
4-1
rH 3
ft
4-1 r-l C
30) rH (H
O. TH
4J rH 4J S
3 ft U O
O TH 3 Crf
4J TJ
S rH 0 <
O 3 t-i 0-
PS S P-, <
r -t
'^>
-^.
rH ^^
^-s CO
W T3
? H a
+ ^ o
rH rH O
^-v H CU
CO ^ CO
r i v_v
""^ £
CO CO
1 1
w w
CO CO
OO OO
o o
*^" -a*
0 0
VO CT>
1 1
U-4 M^
r- co
o\ CM
o o
o o
CO CO
4-1 4-1
H -H
XI XI
CM ^r
rH CM
OO
c£J
CM
4-1 CO
3 r4
ft O
4-1 4-1
3 td
0 V4
01
0 3
<£ &
s-^.
CO
£
6 rH rH rH 4-1
-------
a)
3
o
u
a)
rH
s
H
1
rH
i-H
3
fj
O
iH
4J
3
rH
O
CO
0)
«
0)
rH
cd
O
CO
C
cd
o
H
4-1
r-4
a
00
rH
CO
1 j
CO
cd
0)
4-1
a
cd
CJ
rH
iH
C^
00
f-t
CO
4-1
ca
o
(
CO
a
rH
d
o
4-1
cd
o
o
LI
CU
4J
CU
8
cd
rl
cd
SO so
SO -3- SO CO SC
SO SO O O C
rH so m m P
O O O O r-
sO O* CO CO SC
II II 1
MM MM [i
SO SO CM
o cfi r^ ^* o
* CO OO CM i-
O O CTi i-H f
CD 0 rH CD C
^%
4J
*W
a)
4J
u-i co
H 4J
X! iH
CO XI
CO CO CO 0
4-1 4-1 4J CO 4J 4.
iH iH iH -rl T
to , n f\ ,n j
CM ^ CM CO CN C*
rH CM s_> g ,-H ,_
4-1
J3
P.
r*^ »ri
CO H
M X! O f
CU M O 4J
HO 4J cd 4.
rH 4J rH Jj 1
o. cd a -H «
iH M to B T
4-1 CU O +
rH 0 (X, C <
.33 H cu :
S 53 Q Q C
jr"N
U
0) r
CO /-"^
"** rH
4 co eu
v-> CO
0 PLI
r i V^/
_^
-. co
a Je o*
D^ '^
"i U_>»
) CO CO
1 I
] U W
5 l-» OO OO
} SO -3- *
i m o o
H rH CM CM
) CO CT> so
1 1 1
3 MUM
r- i-H
'^ CM CM
H 00 CM
i co rH m
5 O O O
ca
4J
H
XI
5 ca co CM
i 4J 4J rH
4 iH -H
j XI XI
1 CM < CO
^ rH CM S
cu
4-1
ca
«^^«>k ^^**~*~* iH
00
J_(
^ ^4 rH 0 rH 0 1
y& c5-g tx w
^""'^ ^*-\
CO
co CM a
-o *-» o
Jrf a rH U
^ O ^-N CU
go co co
or at H ^
CO -^
s-> CM CO
H Cf
O
CO
co
o
O
CO
1
M
m
in
CM
CO
o
co
4-1
H
XI
CO
CO
^rj
4-1
3
CX
C
iH
XI
O
4-1
&
CO
pL|
r-H
°
r
r<
CO
co
oo
o
o
so
1
W '
in
co
o
CM
CO
4-1
iH
x»
CM
rH
SO
M
01
«H
fX
iH
4-1
rH
^
S
H
CO
1
OO
O
CM
Q^
1
M
rH
CM
CM
rH
O
CO
4_f
IH
XI
^>
CM
so
CJ
3
0
H
s~+
CO
0
c
o
o
o>
co
SM/
4-t
<3
r-
f<
CO
l
oo
0
CM
SO
1
in
o
ca
4-1
**H
XI
CM
rH
CO
5^
IH
a)
4J
CO
i-4
00
CU
t-l
H
o
-^J
rt
4-1
H
Q
3
to
3
H
^3"
SO
CM
CO
O
O
CO
1
a)
CM
-i
o
01
4-1
t^j
XI
OO
CO
*g*j
4-1
3
p
$
xi
u
4-1
1
p-l
M
C
*
1
,-<
v_.
^
SO
CM
CO
O
O
so
1
M
m
o
CO
4J
H
XI
so
rH
cu
4J
co
«r4
00
cu
^4
c
o
*H
4J
6
Q
3
en
H
CO
pj
o o
o
ca
O «-'
46
-------
o
II
CO
0)
3
o
o
cd
b-.
1
i-H
rH
3
C
0
l-{
3
rH
O
(0
01
0)
rH
cd
0
CO
4-1
C
cd
0
ri
H
G
00
H
CO
4-1
CO
td
01
4-1
pj
cd
o
-_, j
ri
H
a
60
H
CO
4J
10
o
I
CO
s
C
rl
a
o
H
cd
o
o
J^
01
4-1
0)
t-l
cd
P-I
01
3
i-H
cd
0)
3
rH
cd
^-^
W
to
~]
Xrf^
4-1
r-l
^
4J
C
cd
o
rl
rl
C
H
CO
4_>
CO
cd
CD
1
C/i
j_(
o
CO
a>
o
o
M
(X
CO
4J
C
3
£>Q
c*
rl
^4
O)
0)
iH
00
a
a)
oo r-»
^j" ^~
CM CM
0 0
* *
O O
rO vO
1 1
p-J M
\0 iO
o> r^
o o
0 O
CO
r-l
^Q
co
OO 4-)
H
^£)
C/>
O
4J pj
rl O
t* 1
CO 0)
pk
P-l rH
rH Cd
O >
^ ^
co
o
C3
O
O
0)
CO
v'
O
4J
CO CO
1 1
td td
O 0
r*« i"***
t-H r-1
*
o o
cr» so
i i
td UJ
m rH
m CM
O CO
rH ~3-
0 0
*
o o
CO
4-1
H
f>
en CM
rl
^o
^1
O
4-1
td
0)
a
3
00
V0
-d-
\O
1
fd
rH
CM
CO
^-
0
O
ca
H
rQ
O
CO
01
4-1
CO
H
00
01
}^
pj
o
T-l
rt
rt
R
3
CO
^-"x
CO
3
v ^
>
oo
CO
v£>
^j-
CO
1
w
CM
CO
CO
CM
CO
4J
H
J3
-4-
to
^
rH
cd
0
H
0
0)
TJ
"^x^
£
nJ
c
rl
«
o ^^
ON iH 4J
rH & CJ <4H
OO Oi
CM *-*
CO
1 U
td -rl
rQ
CJ 4-1
CM Ol
CO rH 01
CO >4-J
O
CM "-'
CO
rH
cd
&
CJ
cU
*rj
-*
CO
r*~<
cd
i-H
\4 pu
o tn
4_t Bf-t
cd ^
J_l
Cd rH
£g
0 -rl
CJ CJ
0>
o
rH r^ co oo
r^ ^O CT^ OO
in in CM CM
O\ Ol OO OO
CO rH m rH
co m CM o
CO CO O rH
CM Cfr CM CO
O O CO
CO CR
4-1 4-1
H -rl
tr> to
03 O WO
r-l T-l
^n ^fi
CM C/3 CM tO
rH *i''^ rH XM
OO O>
4J 4J
4J 3 4-13
C tX C! ft
- 3 --.
0-Q 0- O
*- H ^-~N
1-J 1-1
tf~{ jf^
"^^ ^^»
CO CO
vS 3
<; w
O' Cr1
47
-------
signals are amplified up to 0-10 VDC range and input to the A/D con-
verters. Each signal has a dedicated 10-bit A/D converter.
Read-only Memory Functions - The three ROM functions employ identical
matrices of memory chips. Each chip is organized into 256 words, by
4-bit word length. The 12-bit output is obtained by combining three
chips least-significant, intermediate-significant, most-significant
bits (LSB, ISB, MSB) into a column. Since the input resolution is
10 bits, each ROM matrix has four columns, with each column using the
most-significant 8 bits of the input. The least significant 2 bits
determine which column is selected. Since the memory chips are tri-
state output devices, the column outputs are wired in common. Only the
selected column is active.
Multiplier - The multiplier forms a 24-bit product from two 12-bit
inputs. The product is formed from eighteen 4-bit sub-products obtained
in parallel multiplications using 4-bit by 4-bit multiplier chips. The
parallel multiplication requires less than 1 microsecond for completion.
Typically only the most significant 12 bits of the 24-bit product are
carried forward in the computation. Each input to the multiplier is
gated through an 8-into-l digital multiplexer. A single multiplexer
chip is used for each of the 12 input bits. The specific input to be
used in each step of the multiplier sequence is controlled by the pro-
cessor timing through selection of the multiplexers' input channel.
Division - Division, including a square-root operation, is performed as
a series of multiplications. A successive approximation register (SAR)
is used as a quotient generator. The SAR begins a division with the
MSB = 1 and all other bits = 0. Then a multiplication is performed
between the SAR output and the denominator input to yield an estimate
of the numerator as the product. This estimate is compared with the
actual numerator input and the SAR MSB is reset to zero if the estimate
is greater than the actual numerator. This routine is repeated for each
bit of the SAR in descending significance. The five numerators in the
computational sequence are gated into the product/numerator comparator
through a multiplexer in accordance with processor timing. Note that
for the square root operation the SAR is input to both inputs of the
multiplier. The final SAR quotient is gated to the appropriate storage
register by the processor timing.
Sample Valve On/Off Control - For each flowmeter pulse the incremental
sample mass is added to a storage register, the valve leakage increment
is subtracted from this register, and the cumulative result is compared
to the selected cycle scaling input switch to (
-------
reset this sum for the next valve cycle. The valve-open and -close com-
mands are output to jacks on the operator's control panel as well as to
the sample valve coil drivers.
Exhaust Mass Accumulator - For each flowmeter pulse the incremental
exhaust standardized volume is added to a storage register. The register,
reset only at the beginning of a test phase interval, maintains a running
total. However, this total is in the binary form in this register. This
binary number must be converted to a decimal number for display on the
control panel. This conversion is done by gating a clock pulse train
simultaneously into a binary counter and into a decimal counter. The
value in the binary counter is compared with the value of the cumulative
exhaust mass register. When the two binary numbers agree, the clock
pulses are stopped, and the value in the decimal counter is sent to the
display drivers.
Exhaust Volume Displays - The three exhaust standard volume displays on
the control panel consist of four decimal digit display modules. Each
module is a seven-segment incandescent lamp which operates off of the
+5 VDC power supply. Each digit is driven by a standard seven-segment
driver chip which decodes the binary-coded decimal (BCD) input from the
decimal counter. Each 4 digit display includes an input multiplexer and
storage register as part of the signal processor self-test capability.
The volume displays are updated each second of the test interval.
Analog Outputs - With the exception of exhaust gas pressure, the analog
output signals are obtained from the processor through 10-bit D/A con-
verters. The output voltage range is 0 to 10 VDC. Since the pressure
transducer features excellent linearity this signal is not corrected
within the digital signal processor. Thus the analog output is derived
directly from the analog input.
Self-Test Aids - A substantial self-test capability has been built into
the signal processor to aid users in troubleshooting the electronics.
The self-test features include: (1) switchover of input parameters to
dual in-line package (DIP) switches which are under manual control,
(2) static display of nearly all internal parameters via the control
panel volume displays, and (3) dynamic display of key parameters during
a test on the volume displays. Individual bits may thus be checked in
any of the processor modules. These features are in addition to a static
system test capability implemented through the exhaust flowmeter.
EXHAUST HEAT EXCHANGER
When it was determined that the flowmeter would not perform accurately
for high temperature (>65°C) flows, the incorporation of an exhaust heat
exchanger into the proportional sampler was selected as the most expedi-
tious resolution of the problem. A flowmeter gas temperature of 40°C or
less was specified as the principal heat exchanger performance require-
ment. The exhaust cooling load was estimated as a function of exhaust
inlet temperature, outlet temperature, and flow rate. The results are
49
-------
presented in Figure 11. For outlet temperatures below about 50°C the
heat rejection associated with the condensation of water vapor becomes
significant. The assumption is made that the exhaust is saturated at
the outlet temperature.
An open-cycle (i.e., no return of coolant) water-to-exhaust heat
exchanger concept was selected as the basic design approach. The
theoretical cooling capacity of such an exchanger is presented in Fig-
ure 12 as a function of water flow rate and temperature rise. Figures 11
and 12 indicate that a water flow of 20 to 40 1pm would provide adequate
cooling power for the exhaust flows and temperatures expected. The
design constraints for a practical exchanger configuration were as
follows:
(1) The nominal facility water supply pressure is 345 kPa.
No auxiliary water pump should be used.
(2) The total exhaust pressure loss through the proportional
sampler, including the heat exchanger, should not exceed
1.25 kPa at 400 m3/hr. The allowable loss due to the
heat exchanger was estimated at 0.5 kPa.
(3) The exchanger should be designed for packaging within
the (pre-existing) exhaust plenum tank (40.6 cm diam-
eter by 182.8 cm height). This constraint was con-
sidered most practical in view of the allowable exhaust
pressure loss and the desire to avoid water vapor con-
densation prior to sampling of the exhaust. A secondary
benefit was the possibility of further attenuating
acoustic noise within the plenum tank.
Within these constraints the most promising configuration appeared to be
a vertical column of coiled tubing through which the cooling water would
pass. The exhaust would be forced to pass back and forth four times
across the tubing by the use of baffles within the tank, as diagrammed
in Figure 13. The water inlet and outlet connections would pass through
bulkhead fittings at the front of the tank. Although a counter-flow
arrangement appeared most efficient, a parallel-flow design could be
obtained by simply switching the external water connections.
Stainless steel was selected for the tubing on the basis of its durabil-
ity in the exhaust and cooling water environments. A standard tubing
size of 1.905 cm outside diameter by 0.124 cm wall thickness was selected
on the basis of an acceptable water pressure drop at a flow of 20 1pm.
The tubing was available in lengths of about 6 m, necessitating the use
of twelve coil segments to achieve a total tubing length of approximately
70 m. The segments were joined using 90 degree elbow, flareless tube
fittings. The coil diameter was selected as the maximum compatible with
assembly inside the plenum tank. The inner and outer baffles were cut
from stainless steel sheet stock. The spacing between each coil of tub-
ing was specified to provide for a maximum exhaust-side heat transfer
50
-------
CO
01
w
01
60
C
H
rH
O
O
o
en
3
a)
3
w
a)
1-1
a3AAOd 9NHOOO
51
-------
0)
00
c
tfl
u
CO
p
tfl
s
0)
I
OJ
rH
a
o
i
0)
a,
o
M
o
4-1
O
tfl
O.
n)
a
M)
o
o
o
-------
OUTER
BAFFLES (2)
INNER
BAFFLES (3)
PRESSURE
GAGE
CONTROL
VALVE
WATER
INLET
WATER
OUTLET
EXHAUST TO
FLOWMETER
CLAMPED JOINT
AT TANK MIDDLE
(REMOVABLE TOP)
EXHAUST FROM
INLET DUCT
TANK DRAIN
FITTING FOR EXHAUST
CONDENSATE
Figure 13. Exhaust heat exchanger schematic diagram
53
-------
coefficient consistent with the allowable pressure loss of 0.125 kPa per
stage. This spacing, nominally 0.0635 cm, was controlled by weaving
stainless steel wire around each coil.
The final cooling coil assembly is illustrated in Figure 14 just prior
to installation of the top half of the tank over the assembly. Note the
0.635 cm tubing frame around the coil at the top and middle of the assem-
bly. Together with a third such frame at the tank bottom, these and
the two outer baffles provide five-point lateral support for the coil
within the tank. The heat exchanger and plenum tank together weigh
approximately 85 kgm, including 15 kgm of water in the coil assembly.
The methods of Reference 6 were used to estimate the performance of the
final heat exchanger design. The results are presented in Figure 15,
which illustrate the generally satisfactory level of exhaust temperature
reduction. The resistance to heat transfer from the exhaust-to-tubing
outer surface constitutes 98 percent of the overall resistance to heat
transfer from the exhaust to the cooling water. This, along with the
cooling load, accounts for the variation in exhaust outlet temperature
with flow rate. The actual variation in exhaust outlet temperature is
relatively small due to the large thermal inertia of the exchanger
assembly. Thus the actual outlet temperature corresponds principally to
the average exhaust flow rate over about one minute.
54
-------
(A) OVERALL VIEW
LATERAL SUPPORT FRAME (1 OF 3)
TOP OUTER BAFFLE
WATER INLET FEED TUBE
INLET VALVES GAGE
BOTTOM HALF OF PLENUM TANK
(B) DETAIL VIEW
TOP OUTER 8AFFLE
COIL SEGMENT JOINT (1 OF 11)
LATERAL SUPPORT FRAME
Figure 14. Heat exchanger coil assembly
55
-------
CO
tu
0)
o
e
-------
SECTION 4
TEST RESULTS
The results of the exhaust flowmeter calibration, sample flow control
valve calibration, and proportional sampler development testing are
presented in this section.
FLOWMETER CALIBRATION
The flowmeter supplier, J-Tec Associates, provided an initial calibration
obtained using a 0-680 m^/hr laminar flow element (LFE). The objective
of the subsequent calibration testing at Aeronutronic was to verify the
J-Tec data, particularly in nonlinear flow ranges from 17 to 70 m-Vhr.
Three LFE's, with flow ranges of 0-34, 0-170, and 0-680 m3/hr, were used
in the calibration test setup shown schematically in Figure 16. Each
LFE was placed in series with the vortex flowmeter to provide equivalent
air mass flows through the two meters. The steady-state metered flow
was adjusted by opening or closing the bypass butterfly valve. The
critical flow venturi upstream of the vacuum-producing turbo-compressor
provided an approximately constant sum of metered and bypass flows. The
instrumentation was used to determine the vortex flowmeter's actual
volume flow rate as a function of the vortex pulse frequency. The fre-
quency was measured using several ten-second pulse counts for each flow
setting. The instrumentation and ducting were thoroughly leak-checked
prior to each series of tests.
During the series of tests with the 0-34 and 0-170 m-Vhr LFE's it was
found necessary to modify the test setup slightly. Acoustic noise was
being generated at the mixing point of the metered and bypass flows.
This noise was of sufficient strength to interfere with the vortex
shedding signal at the lowest metered flows. A plenum volume was inserted
between the vortex flowmeter and the mixing point to attenuate the noise.
This was successful to the point of eliminating all false pulses, even
with the metered flow completely blocked off.
The calibration data finally obtained are presented in Figure 17, along
with the data provided by J-Tec. The 0-34 and 0-680 agree reasonably
well with the original calibration up to about 100 m^/hr. Above that
point the data diverge slightly to a maximum difference of 5 percent at
57
-------
a.
3
0
58
-------
Q
Z
LU
a
111 LU
LU LL li.
-oo
S-S
666
£yyy
-1 zz z
o OOO
StC.CC.fC.
\- H H
' D D D
O Z Z Z
o oo o
LU OC flC QC
H- LU LU LU
-i «
o
rI
M-J
4J
CO
60
s
CM
LO
O
LO
(O
CM
(O
CO
CM
CO
O
CO
(90l X
U313IAIMOld U3d
-------
600 m^/hr. The 0-170 nrVhr LFE did not produce consistent data in com-
parison with any other measurements. This LFE was judged to be out of
calibration, and the data were not used. The final calibration curve
employed for the vortex flowmeter is indicated in Figure 17. Also shown
are the upper and lower theoretical calibration factor limits for a con-
stant Strouhal number. These limits correspond to vortex shedding at
the average and peak duct velocities respectively.
As discussed in the design of the signal processor, the flowmeter signal
nonlinearity is appropriately characterized by determining the volume
per pulse as a unique function of the parameter Atm (T]_/Tg)l+v, termed
the temperature-compensated flowmeter pulse interval. However, a compli-
cation arises in converting air flow calibration data for use in metering
exhaust gas. This is due to differences in the dynamic viscosities of
the two gases. The Reynolds numbers for air and exhaust flows will not
be the same for equivalent values of the temperature-compensated pulse
interval. Reynold's number equivalence can be obtained by an adjustment
of the exhaust parameter:
- FAt (T /T )1+V
[ (
where: y<, = gas dynamic viscosity at 20°C
air ^ values for air flow
exh ^ values for exhaust flow
The transport properties of air and exhaust were estimated using the
methods of Reference 7 and the data of Reference 8. The results are
presented in Table 5. These were used to obtain the value of the
exhaust viscosity temperature exponent, V, and the ratio of air /exhaust
standard (20° C) viscosities. The flowmeter calibration data were then
adjusted and converted to the functional format shown in Figure 18. The
data of Figure 18 were then used directly in the programing of the read-
only memory matrix for the signal processor.
It should be noted that the flowmeter nonlinearity and the viscosity
correction factor are of generally reduced significance in the final
proportional sampler configuration. This is because the heat exchanger
precludes large values of T^/Tg.
SAMPLE VALVE CALIBRATION
The purpose of the sample value calibration testing was to accurately
determine the sample flow as a function of cycle frequency for a fixed
60
-------
Table 5. EXHAUST GAS TRANSPORT PROPERTIES
Temperature (°K)
Specific Heat
( kJ )
Dynamic Viscosity
m . S
Thermal Conductivity
(m "~S~ 5K)
Prandtl Number
300
1.060
1.653xlO~5
2.528xlO~5
0.692
400
1.085
2.055x!0"5
3.200xlO"5
0.695
500
1.110
2.421xlO~5
3.860xlO~5
0.697
600
1.144
2.763xlO~5
4.527xlO~5
0.697
700
1.177
3.083xlO~5
5.180xlO~5
0.698
Notes: (1) Assumed combustion reaction is
0.1346 CH, oc + 0.938 (0.78 N0 + 0.21 00 + 0.01 Ar)
i . o j 2. 2.
-> 0.1346 C02 + 0.1245
+ 0.7315 N + 0.0094 Ar
(2) Air dynamic viscosity = 1.841 x 10 5 kgm/(m . S) at 300°K,
= 1.374 x exhaust viscosity at 300°K
(3) Viscosity temperature exponent ^ 0,747
61
-------
CO
ai
O
s
. §
o
LJJ
V)
<
cc
z
UJ
CO
Q.
ID uj
K-
O
_i
* "-
O
UJ
CO
(N
UJ
DC
QC
UJ
Q.
S
111
o
CO
CO
0)
o
o
cfl
c
M
-------
valve-open interval. The data were checked for linearity and consistency
with leakage and full-open flow measurements. The final calibration data
permit evaluation of the sample flow proportionality constant, k.
The test setup is shown schematically in Figure 19. The method involved
flowing a measured mass of dry air at a steady rate for a measured period
of time. The parameters measured are indicated in Figure 19. The
principal limit to the calibration accuracy is the scale resolution of
0.005 kgm. A minimum mass of air equal to 0.7 kgm was used for each
test run to achieve an acceptable accuracy. The setup was leak checked
with liquid leak detector at the beginning of each run.
The pump provided sufficient vacuum to insure choked flow across the
valve orifice with a nominally atmospheric inlet pressure. The pressure/
vacuum gage and water U-tube manometer provided indication of the valve
inlet pressure. The needle valve in parallel with the differential pres-
sure regulator provided the final stage of regulation of the air flow.
The needle valve was adjusted to maintain zero gage pressure at the valve
inlet. Thus the valve inlet absolute pressure was simply the measured
barometric pressure. The air-bottle pressure regulator provided an approx-
imately constant input pressure to the flow regulator even though the air
bottle pressure was continually decreasing. The heat exchanger coil
allowed the air entering the flow regulator to be at a constant temperature
during the run. The air temperature at the valve inlet was monitored using
a thermocouple.
The sample valve coils were driven by the electronics used in the final
proportional sampler assembly. Signal inputs were derived using available
test equipment. Since coil temperature affects valve response, a cooling
fan was used to stabilize the coil temperature. Approximately 5 minutes
of 50 percent duty cycle warm-up were provided for the coils.
The above described setup was used to obtain data for 20, 30, and 40 Hz
cycling rates, as well as for the full-open condition. For leakage
measurements, however, this setup required impractically long run times.
Instead, a low range rotameter (Brooks type 1110-01F1G1A, 0-0.85 1pm
full-scale) was used to estimate the leakage flow rate.
The results of the calibration testing are summarized in Figure 20.
First, two tests at the full-open condition were conducted to verify the
repeatability of the setup. Then the three tests at the indicated cycl-
ing rates were performed. These data exhibited very good linearity, but
did not agree with the results of numerous leakage tests. It appeared
that the basic setup had a systematic error of about 0.34 1pm, which was
independent of the flow rate. The source of this error could not be
determined, but was probably associated with the bottle weight measure-
ments. The final data were adjusted to agree with the leakage flow
measurement.
63
-------
c
o
1-1
o
-------
(ti
T3
C
O
cfl
M
tfl
O
0)
rH
OS
-------
Using Equation 2 of Section III, the following valve parameters were
then established as input constants to the signal processor:
(1) t = 0.010 second
o
(2) AX = 0.005
(3) A = 1.297
o
(4) t (A -A.,) = 0.01292 second
o o l
DEVELOPMENT TESTING
Testing of the proportional sampler was conducted in two phases. The
initial phase had the primary objectives of (1) verifying the flowmeter
accuracy for actual vehicle exhaust, (2) verifying the stability of the
sample valve calibration after exposure to exhaust, and (3) evaluation
of the flowmeter and sample valve durability in the exhaust environment.
This phase was completed prior to the final configuration and integration
of the signal processor, control panel, and sample flow system into the
console. The second phase had the general objective of verifying the
complete system's functional and performance characteristics. The results
of these tests are presented in the following paragraphs, in essentially
a chronological discussion.
Initial Test Phase
The development testing was conducted at the Aeronutronic Emissions Test
Facility. The initial test setup is shown schematically in Figure 21.
Two vehicles were used, one with a 6.55 liter V-8 engine and one with a
2.3 liter 4-cylinder in-line engine. The chassis dynamometer was a
45 kw electric type with inertial, grade, and road load simulation
capabilities. The constant volume sampler (CVS) was an Aeronutronic
Model CVS-4 with dilute exhaust temperature control and a critical flow
venturi as the flow metering element. The gas analysis equipment
included Beckman 315B C02 and CO analyzers and a Beckman Model 400
flame-ionization-detector (FID) for measurement of hydrocarbons (HC).
These were housed in Aeronutronic Series 300 emissions analysis consoles
which provided sample conditioning and electronic controls. Standard
test equipment was used for all other measurements indicated in Figure 21.
During each test, whether steady-state or dynamic, a continuous sample
of raw exhaust was drawn in parallel through the sample valve and a
probe upstream of the CVS mixing point. This sample was partially dried
in a refrigerated water bath prior to analysis for C02» CO, and HC volume
fractions. Simultaneously, ambient and dilute exhaust samples were col-
lected by the CVS as in a normal emissions test. These samples were
then analyzed following completion of the test. Readings of exhaust
duct pressure and flowmeter vortex frequency were manually recorded
66
-------
ex
3
W
Q)
G
0)
Cu
O
r-H
0)
>
0)
a.
CO
a
o
o
a,
o
CM
0)
3
H
67
-------
throughout each test. Exhaust duct temperature was measured continuously
using a thermocouple and multi-point recorder. An analog signal of
exhaust volume flow from the flowmeter was also recorded on a two-pen
strip chart along with the vehicle speed signal.
For each test the standardized exhaust volume was calculated from the
flowmeter data and compared with that calculated from the CVS data using
a carbon balance method. A correction for the effects of sample drying
was included in these calculations. The overall method is similar to
that described in Appendix B of this report. The first tests conducted
were with the V-8 engine at the steady-state conditions of loaded idle,
45 km/hr, 70 km/hr, and 85 km/hr. The standardized exhaust volumes from
these tests, as indicated by the vortex flowmeter, were from 32 to 51 per-
cent below those calculated from the CVS data. The 4-cylinder vehicle
produced an obviously anomalous flowmeter signal prior to conduct of
formal tests.
A four-trace oscillscope was used to monitor the flowmeter's internal
electronic signals. The principal signals of interest were the amplitude-
modulated ultra-sonic carrier signal, the demodulated audio frequency
vortex shedding signal, and the digital pulse output signal. The vortex
shedding signal appeared to have a strong interference pattern, indicat-
ing a second source of modulation for the ultrasonic beam in addition to
the vortex-shedding process. The consequent irregular nature of the
signal led to so-called pulse-dropping in the output signal. This was
caused by the inability of the pulse-triggering electronics to dis-
criminate between the two modulation sources. Consultation with the
flowmeter supplier led to modification of the electronic signal filter
parameters. In effect this resulted in a narrower bandpass for the
input to the pulse-triggering electronics. However, the spectrum of the
spurious modulation source overlapped the vortex shedding bandpass such
that significant interference and pulse-dropping remained.
It was suspected that the secondary modulation source was the presence
of pressure waves in the flowmeter duct. These were thought to be caused
by noise and exhaust manifold pressure fluctuations which were not suffic-
iently attenuated by the vehicle's muffler. This hypothesis was confirmed
in a series of tests to develop a supplementary acoustic muffler configura-
tion. A variety of experimental configurations were evaluated including:
(1) A second automobile muffler in series with either a
208 liter or a 416 liter plenum tank.
(2) A butterfly valve in series with either a 208 or
416 liter plenum tank.
(3) A small diameter (4.5 cm) inductance tube in series
with either a 208 or 416 liter plenum tank.
In all cases the flowmeter signal quality was significantly improved. A
comparison of the flowmeter signals between umnuffled and muffled exhaust
68
-------
flow is provided in Figure 22. The four traces, beginning with the top-
most, are the modulated carrier (shown as a half-wave in parts b and c) ,
the demodulated vortex shedding signal, the corresponding output pulse
train, and the pressure in the flowmeter duct. This last trace was
obtained using strain-gage diaphragm pressure tranducer.
Part (a) of Figure 22 illustrates the effect of severe engine pulsations
(rough idle) on the flowmeter signals. The strong pressure waves scatter
the ultra-sonic beam with nearly 100 percent modulation of the carrier
signal, thus swamping the vortex shedding modulation. The result is
extensive intermittent pulse dropping (gaps) in the output pulse train.
Part (b) illustrates a more typical level of pressure pulsation, but
still with a significant percentage of pulse-dropping in the output
signal. Note that although the basic vortex shedding frequency can be
visually determined from the photograph, the pulse-triggering electronics
can only detect a zero-crossing of the demodulated signal.
Part (c) of Figure 22 shows the effect of an order-of-magnitude reduction
in pressure pulsation magnitude at the flowmeter. (The pressure trace
voltage scale is 10 percent of that for parts (a) and (b).) The demodu-
lated signal is relatively regular and very few pulses are lost in the
output signal. This particular muffler arrangement included a second
automobile muffler as well.
Tests using the first 505 seconds of the EPA Urban Driving Schedule
finally yielded very good agreement (less than 0.5 percent error)
between the flowmeter and CVS-derived standardized exhaust volumes.
Further tests were then conducted to determine an optimum configuration
for a permanent supplemental muffler.
The muffler tests showed the plenum tank volume and downstream inductance
to be the most important parameters. This arrangement is analogous to a
parallel L-C electrical low-pass filter circuit in that the input noise
spectrum is attenuated above a cutoff frequency:
fc ^
where: D = inductance tube diameter (m)
T = gas temperature (°K)
3
VT = plenum tank volume (m )
& = inductance tube length (m)
The desire to minimize this frequency was balanced by the need to maintain
a low exhaust pressure loss through the ducting, and practical limits on
the plenum tank volume. There was also concern over maintaining a time-
correlated sample with the dynamically varying exhaust flow. This led
to incorporation of the plenum tank downstream of the sample probe, such
69
-------
(A) NO PLENUM VOLUME.
SEVERE ENGINE PULSA-
TION. PEAK-TO-PEAK
PRESSURE * 4.0 KPA
(B) NO PLENUM VOLUME.
NORMAL ENGINE PULSA-
TION. PEAK-TO-PEAK
PRESSURE *2.7 KPA
(C) PLENUM VOLUME
NORMAL ENGINE PULSA-
TION. PEAK-TO-PEAK
PRESSURE *0.2 KPA
Figure 22. Comparison of vortex flowmeter signals (varying plenum volume)
70
-------
that the transport delay of exhaust from the tailpipe to the sample probe
is minimized. The final configuration, described in the preceding sec-
tion, included a 237 liter plenum volume with any one of six 1.2 m long
inductance tubes. The tube inside diameters ranged from 3.8 cm up to the
full diameter of the flowmeter duct, 7.29 cm. The corresponding cutoff
frequencies were from 9 to 17 Hz. These are substantially less than the
lowest vortex frequency of interest, 100 Hz. Interchangeable tubes were
provided to permit trade-off of the cutoff frequency for reduced exhaust
pressure loss.
In addition to this primary result of the initial development testing,
the following durability results were obtained:
(1) The sample valve leakage and full-open flows were
remeasured and found to be identical to their original
values.
(2) The stainless steel prefilter element had become
clogged, indicating a need to occasionally clean or
replace the element. A 50 ym pore size was substituted
for the original 15 ym element.
(3) The aluminum honeycomb material used in the flowmeter
to collimate the ultrasonic beam within the transmitter
and receiver cavities had become significantly corroded
in the high temperature exhaust environment. They were
subsequently replaced with collimators made from stain-
less steel honeycomb. This apparently straightforward
substitution led to a further development problem dis-
covered in the second phase of testing.
(4) The above-mentioned collimators tended to accumulate
condensed water vapor from the exhaust. In one case
this obstructed the ultrasonic beam sufficiently to
affect the flowmeter signal quality. This is illu-
strated in Figure 23, where the reduced amplitude
of the carrier is apparent. The flowmeter electronics
include compensation for variations in carrier signal
strength, but in this case the compensation circuitry
had become saturated. This would have been a minor
problem in the original configuration, and the use
of an exhaust heat exchanger virtually eliminated it
altogether.
Final Test Phase
The setup for the final phase of testing was similar to that for the
initial phase, but with the completely assembled proportional sampler.
The basic test plan for the final tests was the EPA Evaluation Test
Plan (Appendix B). The principal performance objective was agreement
71
-------
(A) BEFORE CLEANING AND
DRYING OF ALUMINUM
HONEYCOMB COLLIMA-
TORS. (WEAK CARRIER
SIGNAL)
(B) AFTER CLEANING AND
DRYING OF COLLIMA-
TORS. (NORMALCAR-
RIER SIGNAL)
Figure 23. Effect of acoustic beam obstruction on tlowmeter signals
72
-------
between the proportional sampler and the CVS for CQ2 and CO mass emis-
sions. The C02 mass would primarily reflect accurate exhaust flow
metering, while the CO mass in conjunction would reflect accurate sample
proportioning by the signal processor and sample valve. The test setup
included the Aeronutronic Model CVS20 and advanced prototype analysis
equipment, as well as the previously used analyzers.
The first tests of this phase revealed flow metering errors in the range
of 10 to 30 percent compared to the CVS values. Monitoring of the flow-
meter signals led to the discovery of an abnormal carrier signal wave-
form. This is illustrated in Figure 24, and was the effect of the changed
beam collimators. The flowmeter was then returned to J-Tec for repair
and readjustment, with the following results:
(1) One of the ultrasonic transducers was found to be faulty,
and was replaced.
(2) The stainless steel collimators were replaced by slightly
longer (2.54 cm versus 1.78 cm) stainless steel honeycomb
pieces. Although not as long as the original aluminum
pieces (3.8 cm), the carrier signal quality was acceptable
(Figure 24).
(3) The transducer orientations and ultrasonic beam (i.e.,
carrier) frequency were readjusted to obtain good signal
quality for air flow in the range from 17 to 500 m-Vhr.
After reinstallation of the flowmeter, steady-state tests with vehicle
exhaust continued to result in flowmeter pulse dropping and consequent
mass emissions errors of up to 28 percent. A series of tests were then
conducted to isolate the cause of the acoustic beam secondary modulation.
This included use of the special duct pressure transducer, retest of
prior experimental muffler configurations, and evaluation of alternate
muffler approaches. The results were as follows:
(1) The flowmeter tended to perform best at idle flows,
indicating satisfactory attenuation of the lowest
engine pulsation frequencies.
(2) The experimental muffler arrangements provided better
flowmeter accuracy than the permanent muffler, indi-
cating the importance of muffler shape as well as
plenum volume.
(3) An experiment with a sound absorbent liner in the
plenum tank reduced the flowmeter errors by more
than half.
From these tests it appeared necessary to modify the muffler to further
attenuate pressure fluctuations in the flowmeter duct. The source of
these fluctuations was thought to be standing pressure waves in the
73
-------
(A) SHORT (1.78 CM) STAIN-
LESS STEEL HONEYCOMB
COLLIMATORS. (NOTE
ASYMMETRIC MODULA-
TION OF CARRIER
SIGNAL.)
(B) MEDIUM (2.54 CM)
LENGTH STAINLESS-
STEEL HONEYCOMB
COLLIMATORS. (NOTE
NORMAL SYMMETRIC
MODULATION OF CAR-
RIER SIGNAL.)
Figure 24. Effect of acoustic beam collimators length on
flowmeter signals
74
-------
plenum tank at frequencies corresponding to the second and higher
harmonics of the longitudinal wavelength. Analyses indicated the
desireability of using an absorbent liner in the tank. A fiberglass
material suitable for the high temperature exhaust environment was
obtained and installed.
Two series of three 505 second (transient phase of the EPA Urban Driving
Schedule) dynamic tests were conducted, one with and one without the
plenum tank liner. Both series of tests involved the following sequence
cold start 505, 10-minute vehicle soak with the ignition off, warm
start 505, and a final hot start 505 cycle with no vehicle soak. The
results were clear trends of increasing error from cold to warm to hot
cycles, and from the unlined to lined plenum tank. This led to the
hypothesis that the flowmeter signal was at least partially affected
by the exhaust temperature.
A test of this hypothesis was conducted using a resistance heater in
conjunction with an ambient air flow. Gas temperatures of 65°C and
122°C were obtained using 115 and 208 volt AC power, respectively. The
results were as illustrated in Figure 25. It can be seen that the
remaining secondary modulation source is due to the elevated gas tempera-
ture. The modulation mechanism is not clear, but probably involves
thermal turbulence (waves) in the hot flow, or possibly beam reflections
due to temperature gradients along the beam axis. Attempts to resolve
the physical phenomena involved were not pursued since a more direct
solution of the problem appeared feasible.
The most obvious approach to the problem was the incorporation of an
exhaust heat exchanger in the plenum tank. A test of this approach was
conducted using an available commercial water/air heat exchanger. The
exchanger was located between the tailpipe and the plenum tank inlet.
An ambient-temperature water flow of 20 1/m provided an exhaust tempera-
ture reduction of up to 160°C as the gas temperature at the flowmeter
was kept below 50°C. The resulting flowmeter signal quality was excellent,
implying that the signal interference problem experienced in the second
phase of testing was due solely to gas temperature effects. Comparisons
of the cooled and uncooled exhaust flowmeter signals are provided in
Figures 26 and 27.
Following this test the detailed design, fabrication, and installation
of the final exhaust heat exchanger was completed. Final tests of the
proportional sampler resulted in good agreement for the CC>2 (average
error of 0.7 percent for six tests) and CO (average error of 3.8 percent
for three tests) mass emissions in comparison with the CVS. The heat
exchanger performance was as predicted, with the exhaust temperature
maintained below 35°C. However, the exhaust pressure loss was higher
than desired. Tests with ambient air indicated an overall loss of
1.25 kPa at a flow of 265 m3/hr. This will be acceptable for testing of
light duty vehicles in the current emissions and fuel economy driving
schedules. The peak exhaust flow occurs in the second cycle of the
Urban Driving Schedule. A typical dynamic exhaust flow history (as
75
-------
(A) T, = 20°C
1
75 M3/HR
DROPPED PULSES <0.5%
(B) TT = 65°C
QA * 75 M3/HR
DROPPED PULSES
(C) TT = 122°C
QA * 75M3/HR
DROPPED PULSES « 12.5%
Figure 25. Comparison of flowmeter signals for heated air flows
76
-------
(A) WITH HEAT EXCHANGER
INSTALLED.
T, <
35°C
95 M3/HR
(B) PRIOR TO INSTALLATION
OF HEAT EXCHANGER
TT * 130°C
QA « 90M3/HR
Figure 26. Comparison of flowmeter signals for vehicle exhaust flow
(cruise condition)
77
-------
(A) WITH HEAT EXCHANGER
INSTALLED.
TT < 35°C
QA * 125M3/HR
(B) PRIOR TO INSTALLATION
OF HEAT EXCHANGER.
TT * 150°C
QA * 140 M3/HR
Figure 27. Comparison of flowmeter signals for vehicle exhaust flow
(acceleration condition)
78
-------
traced from a strip chart with reversed time scale), is shown in Fig-
ure 28 for this cycle. Note that peak flows always occur during acceler-
ation modes, which by their nature are of limited duration. As a cruise
mode is entered the flow will be reduced, and will decline to nearly the
idle flow during deceleration modes. The largest engines may exceed the
tailpipe pressure disturbance at about 195 seconds into the urban driving
schedule, but the duration of this excursion will certainly be less than
2 seconds.
After completion of the second phase of development testing at Aeronutronic,
the device was shipped to the EPA Triangle Park facility, along with com-
plete operating and maintenance instructions. The proportional sampler
was installed and performed satisfactorily in preliminary evaluation
testing. Included in these tests were vehicles equipped with oxidation
catalysts which produce exhaust temperatures up to 450°C.
79
-------
(UH/EW) 3J.VH
Mend awmoA
w
0)
,c
4-1
14-1
O
0)
rH
O
>-,
a
a)
4-1
cu
,-t
t
o
o
M
01
O
0)
o
rt
n
4-1 Q)
O T3
t-H CU
4-1 J2
a
4-1 CO
CO
3 M
a) C
01 -H
M
rH TJ
cd
o c
-H cd
00
CN
0)
M
1 1
§ S
1 1
S S
1
8 £
1
g S
,
. i
(UH/WX) Q33dS 3TOIH3A
80
-------
REFERENCES
1. Olson Labs, Phase 1 report on CRC contract to develop an exhaust
flowmeter.
2. Black, H. S., Modulation Theory. Princeton, N. J., Van Nostrand
Company, Inc., 1953. Chapter 4.
3. Vortex Volumetric Flowmeter. J-Tec Associates, Cedar Rapids, Iowa.
December 1973.
4. Handbook of Physics. E. V. Condon and H. Odishaw, Ed. New York,
McGraw-Hill, 1958. Page 5-37.
5. Gold, B. , and C. M. Rador. Digital Processing of Signals. New York,
McGraw-Hill, 1969. Chapter 4.
6. Chapman, A. J., Heat Transfer. New York, Mac Millian Co., 1960.
7. Brokaw, R. S., Approximate Formulas for Viscosity and Thermal
Conductivity of Gas Mixtures. NASA Tech. Note D-2502. November 1964.
8. Svehla, R. A., Estimated Viscosities and Thermal Conductivities of
Gases at High Temperatures. NASA Tech. Report R-132, 1962.
81
-------
APPENDICES
Page
A. Proportional Sampler Requirements Specification 83
B. Proportional Sampler Evaluation Test Plan 94
82
-------
APPENDIX A
PROPORTIONAL SAMPLER
FOR AUTOMOBILE EXHAUST EMISSIONS TESTING
SPECIFICATION OF REQUIREMENTS
1.0 SCOPE
Specified herein are the functional, performance, and critical
component requirements for a Proportional Sampler (hereafter called
"device"). The purpose of the device is to permit determination of
specie masses emitted in the exhaust of an internal combustion engine
during a simulated road test on a dynamometer. In this context the
principle function of the device is to collect and hold for analysis an
exhaust sample drawn at a mass flow rate in constant, known proportion
to the instantaneous total exhaust mass flow rate.
2.0 FUNCTIONAL REQUIREMENTS
2.1 General
The device shall sample the engine exhaust at a mass flow rate
in constant proportion to the total exhaust mass flow rate at all normal
engine operating modes. Sampling proportionally shall be maintained
throughout the EPA Urban Driving Schedule (LA-4-S-3, as specified in the
Federal Register, Vol. 37, No. 221, dated Wednesday, November 15, 1972),
including idle, acceleration, cruise, and deceleration modes. The device
shall not dilute the exhaust sample in any manner.
2.2 Exhaust Flow Metering
The device shall directly meter the flow of total engine exhaust.
The instantaneous flowmeter signal shall be used to control the sample
flow regulating component (i.e., sample flow control valve) and to deter-
mine the cumulative exhaust mass emitted during the test.
2.3 Exhaust Flow Conditioning
The device shall include all components to condition the exhaust
flow as required to assure acceptable flow metering accuracy. Such com-
ponents may include straightening vanes, acoustic mufflers, heat exchang-
ers, or other items as necessary.
2.4 Sample Conditioning
The device shall provide for the following sample conditioning
functions:
(a) Extraction of the sample gas from the exhaust
stream without entrainment of particulates.
83
-------
(b) Removal of water vapor and heavy hydrocarbons
from the sample.
(c) Maintenance of the sample temperature at or
above 90°C between the point of extraction
and the condensate trap.
(d) Filtering of fine particle contaminates from
the sample.
(e) Pumping of sample through the conditioning
system into the collection bag.
(f) Collection of the sample flow so that a mass-
weighted average sample is obtained for the
test. The collection bag shall be sufficient
to hold the entire volume of sample from an
LA-4-S-3 test. A total of three collection
bags shall be provided for use in tests involv-
ing three-phase (cold transient, stabilized,
hot transient) driving schedules. Transfer
of a sample to an analysis system shall be
permitted simultaneous to the filling of any
other bag with sample.
2.5 Sampling Components Purge
The device shall provide for ambient air purge of the sample
extraction probe, condensate trap, and sample collection bags. Evacua-
tion of purge air from the collection bags shall also be provided.
2.6 Durability
The device shall be capable of handling undiluted internal com-
bustion engine exhaust at gas temperatures up to 450°C. Only stainless
steel or Teflon or materials of equivalent durability shall be exposed
to the exhaust and sample gases.
2.7 Control and Data Display
The device shall provide for manual operation and data acquisi-
tion. In addition, the three-phase sample collection procedure shall be
capable of being implemented with a single start command, at either the
control panel or at a remote station. The device shall provide the follow-
ing data at the control panel:
(a) Visual readout of cumulative exhaust mass for
each of three test phases.
84
-------
(b) Visual monitoring of operating parameters
including exhaust pressure and temperature,
sample pressures and temperatures, exhaust
volume and mass flow rates, and sample flow
control valve duty cycle.
(c) Analog signals for external recorders
including exhaust volume and mass flows,
exhaust pressure and temperature, and
sample flow control valve inlet tempera-
ture and duty cycle.
3.0 PERFORMANCE REQUIREMENTS
3.1 Exhaust Flow
The device shall be capable of operation over the range of
exhaust volume flow from 17 to 500 m^/hr and the range of exhaust temper-
ature from 5°C to 450°C.
3.2 Sample Flow
The maximum obtainable sample flow shall be a minimum of 8.5 1pm.
The sample flow proportionality constant shall be selectable from the
following nominal values:
(a) 3.00 parts/thousand (ppk) for 170 m^/hr peak
exhaust flow.
(b) 2.00 ppk for 255 m3/hr peak exhaust flow.
(c) 1.70 ppk for 300 m^/hr peak exhaust flow.
(d) 1.50 ppk for 340 m^/hr peak exhaust flow.
(e) 1.33 ppk for 380 m-^/hr peak exhaust flow.
(f) 1.20 ppk for 425 m3/hr peak exhaust flow.
(g) 1.00 ppk for 510 m-Vhr peak exhaust flow.
(h) 0.75 ppk for 680 m^/hr peak exhaust flow.
3.3 Accuracy
The device shall provide the following accuracies over the
operating ranges specified above.
(a) Sample proportionality constant within +2 per-
cent of specified value.
85
-------
(b) Cumulative exhaust readout within +1 percent
of value. (Above 0.3 m^.)
(c) Analog recorder signals within +1 percent of
full scale values.
3 . 4 Response
The device shall meet the following dynamic response criteria.
(a) Ducting hold-up volume less than 0.25 m^ between
the inlet and the exhaust flowmeter, including
acoustic muffler, but not including ducting
from the engine to the muffler inlet.
(b) Response to step change in exhaust flow rate:
T,- < 0.01 + 0.025 Q . /Q,, .
flow xpeak final
where
Tf1 = equivalent low-pass (dissipative)
filter time constant (seconds)
Q , = selected peak exhaust flow (170,
P 255, 300, 340, 380, 425, 510,
680 m3/hr)
0- . , = exhaust flow rate following the
xfinal _ , , o,, . e
step change (m-Vhr)
(c) Response to step change in exhaust temperature:
where :
T = equivalent low-pass (dissipative)
filter time constant (seconds)
3
Q = exhaust flow rate (m /hr)
3.5 Exhaust Back-Pressure
The exhaust pressure drop through the device shall not exceed
1.240 KPa including losses due to flow conditioning.
86
-------
4.0 COMPONENT REQUIREMENTS
4.1 Exhaust Flowmeter
The exhaust flowmeter employed in the device shall be of the
vortex shedding type, and shall use an ultrasonic beam scattering tech-
nique for quantitative detection of the vortex shedding frequency. The
flowmeter shall satisfy the following performance criteria:
(a) Flow range from 17 to 500 m3/hr.
(b) Calibration as shown in Figure 1.
(c) Nonlinearity of calibration less than 10 per-
cent of value at any point in the flow range.
(d) Nonrepeatability less than +1/2 percent of
value at any point in the flow range (based
on 100 pulse count interval).
(e) Negligible dynamic distortion for flow
variations at frequencies less than 1 Hertz
per m^/hr over the flow range.
(f) Pressure drop less than 0.375 KPa at
680 m3/hr.
4.2 Sample Flow Control Valve
The sample flow control valve used in the device shall be a
frequency modulated on/off type. The valve shall incorporate a critical
flow orifice to meter the sample flow, and a solenoid driven flapper to
open and close the orifice. The valve shall satisfy the following per-
formance criteria:
(a) Full-open flow of at least 16.2 1pm for an
inlet temperature and pressure of 20°C and
101.325 KPa absolute, respectively.
(b) Full-closed (leakage) flow less than
1/2 percent of the full-open flow value.
(c) Calibration as shown in Figure 2.
(d) Negligible nonlinearity and nonrepeat-
ability of calibration for input
frequencies up to 40 Hertz, with an
input valve-open command of 10.0 msec
duration.
87
-------
CO
LU
i
CM CO
<£)
in
co
CM
CO
Q
o
LU
CO
cc
LU
H
Z
01
CO
3
Q.
tr
LU
LU
I
Q
LU
CO
o
V
LU
OC
D
Q.
5
LU
C
O
O
C
C
O
-r-l
CO
O
0)
4J
<1)
O
,4
<4-(
03
3
n)
J2
X
w
60
-r-l
d o o
3Sind U313IAIAA01d/3IAimOA
88
-------
8
DC
LU
LU
I
LL
u
LU
O >
-LO
00
-------
(e) Withstand repeated exposure to unfiltered
exhaust gas at inlet temperatures up to
450°C without degradation in performance.
4.3 Instrumentation
The device shall incorporate the following instrumentation as
required to perform the exhaust and sample flow metering functions:
(a) Exhaust pressure - 0 to 103.421 KPa absolute
pressure transducer of the strain-gage
diaphragm type with +1/2 percent overall
accuracy.
(b) Exhaust and sample valve inlet temperatures -
platinum wire resistance temperature sensors
with +1/4 percent accuracy (not including
error due to nonlinearity) over the range
from 5°C to 450°C.
4.4 Signal Processing Electronics
The device shall employ analog and digital signal processing
electronics to perform the following functions:
(a) Receive the exhaust flowmeter, pressure
transducer, and temperature sensor signals
and determine the cumulative exhaust volume
at the standard conditions of 20°C and
101.325 KPa for display on the control
panel.
(b) Provide valve-open and valve-close command
signals to the sample flow control valve
such that the mass flow rate of sample is
a constant fraction of the exhaust mass
flow rate.
(c) Provide output signals of exhaust volume
and mass flow rates, exhaust pressure and
temperature, sample valve inlet temperature,
and the valve open/close commands.
The following equations shall be implemented in the signal
processing electronics (nomenclature per Table A):
(a) Exhaust flowmeter volume/pulse -
AACF = Function of AT (Ij1
IU.G L G L .i.
90
-------
TABLE A
NOMENCLATURE FOR SIGNAL PROCESSING EQUATIONS
AACF = Exhaust volume per flowmeter pulse (m3)
AT = Exhaust flowmeter pulse duration (seconds)
meter
T, = Exhaust gas temperature at flowmeter (°K)
T = Standard temperature = 273.16°K
O
m = Exhaust gas viscosity temperature exponent
ASCF = Exhaust mass per flowmeter pulse (m3 at 20°C and 101.325 KPa)
P, = Exhaust gas pressure at flowmeter (KPa, absolute)
PS = Standard pressure - 101.325 KP£
SCF = Cumulative exhaust mass over the test interval
£ ^ Summation of flowmeter pulses over the test interval
TEST
T .. = Sample flow control valve cycle time, i.e., between valve-
open commands (seconds)
Z ^ Summation of flowmeter pulses over the valve cycle time
CYCLE
K = Sample flow proportionality constant (ppk)
(SCFH) = Sample valve full-open flow rate at standard conditions
(m3/hr)
91
-------
TABLE A (Continued)
NOMENCLATURE FOR SIGNAL PROCESSING EQUATIONS
T = Gas temperature at sample valve inlet (°K)
q = Ratio of leakage/full-open sample flow during valve-closed
J.G3K.
interval
q = Ratio of average/full-open sample flow during valve-open
interval
AT = Sample valve open time, i.e., between valve-open and valve-
close commands (seconds).
ACFM = Exhaust volume flow rate (nr/hr)
SCFM = Exhaust mass flow rate (nrVhr at standard conditions)
92
-------
(b) Exhaust flowmeter mass/pulse -
T,. ,
AACF
(c) Cumulative exhaust mass -
SCF = E ASCF
TEST
(d) Sample flow control valve cycle time -
T 1 = E AT _
Cycle CYCLE meter
With the summation limit determined by the
criterion:
E [3.6K
CYCLE (SCFH)
1/2
MCF 'S»'21 -1 AT
leak meter
> AT (q - q )
open open leak
(e) Exhaust volume and mass flow rates -
ACFM = 3600 AACF/AT
meter
SCFM = 3600 ASCF/Ai
meter
The signal processsing electronics shall include compensation
for input signal nonlinearities and shall be implemented such that the
overall accuracy requirements of the device are satisfied, with due
consideration for the accuracy limitations of the exhaust flowmeter,
sample flow control valve, and instrumentation.
4.5 Exhaust Flow Conditioning
The device shall include an exhaust heat exchanger and plenum
tank located between the point of sample extraction and the exhaust
flowmeter. The heat exchanger shall be an open cycle water/exhaust type
requiring less than 40 1pm ambient temperature cooling water to maintain
the exhaust temperature below 65°C at the flowmeter. The plenum tank
shall have sufficient volume to eliminate acoustic interference at the
flowmeter.
93
-------
APPENDIX B
PROPORTIONAL SAMPLER EVALUATION TEST PLAN
I. OBJECTIVE AND SCOPE
The objective of the Proportional Sampler evaluation testing
is to verify the accuracy and repeatability of the device. The testing
is to be conducted at the EPA Research Triangle Park facility. The
device will be installed in series with a CVS in an emissions test cell.
A referee engine will be driven through the LA-4-S-3 driving schedule.
A minimum of 10 such tests will be performed. The mass emissions of C02,
CO, HC, and NOX will be determined from both the Proportional Sampler and
the CVS for each test. Statistical tests will be performed with these
data to demonstrate the accuracy and repeatability of the Proportional
Sampler data in comparison to the EPA CVS.
The purpose of this test plan is to define the procedural
requirements for the installation, calibration, test and data analysis
phases of the evaluation.
II. INSTALLATION
The evaluation test setup is shown schematically in Figure 1.
Installation of the Proportional Sampler in the test cell should proceed
in accordance with detailed Operating and Maintenance Instructions. The
steps involved include:
(1) Tie-down of console floor jacks.
(2) Attachment of facility ducting from engine exhaust
and to CVS inlet.
(3) Attachment of sample transfer and condensate trap
drain lines.
(4) Filling of the refrigerated water bath.
(5) Hookup to facility power.
(6) Checkout of console adjustments.
(7) Checkout of console functions.
(8) Attachment of the heat exchanger and plenum tank
to the proportional sampler ducts.
(9) Completion of system leak checks.
94
-------
ROMETEfl
3
UJ
oc
YBULB
MPERATU
K UJ
0 1=
UJ
K
TBULB
MPERATU
g?
0)
CO
c
o
H
4_>
CO
3
, I
UJ
>
0)
a.
B
CO
CO
CO
c
o
rl
o
t-l
PM
3
00
rl
fa
95
-------
In addition to these basic steps, the entire CVS and sample
analysis plumbing should be thoroughly leak checked. All sample lines
which are under vacuum must be made leak free so that no uncontrolled
dilution of sample gas can occur.
If desired, the vehicle driver can be provided with a remote
start switch from the Proportional Sampler control panel. This may
simplify the overall synchronization of the test between the vehicle,
Proportional Sampler, and CVS.
III. CALIBRATIONS
All test equipment calibrations should be checked prior to the
series of evaluation tests. The Proportional Sampler will be fully
calibrated prior to shipment to the EPA. It will only be necessary to
verify the digital switch settings in the signal processing electronics,
and check the pressure and temperature sensor calibrations. The remain-
ing calibrations of principal importance are the CVS blower, the sample
gas analyzers, and the test fuel. The requirements for these are summar-
ized in the following paragraphs.
A. CVS Blower
It is understood that the CVS to be used in the evaluation tests
will be a conventional Rootes blower type. The volume per blower revolu-
tion should be accurately determined using a laminar flow element (LFE)
in accordance with existing EPA procedures. (Reference Federal Register,
Vol. 38, No. 124, Appendix III, Thursday, June 28, 1973, pp. 17167-8.)
Since even the use of an LFE does not guarantee calibration to a uniform
standard between the Proportional Sampler and the CVS, it is recommended
that the CVS blower calibration be performed using both the EPA and the
Aeronutronic LFE's. The Aeronutronic LFE will be shipped along with the
Proportional Sampler for this purpose. This dual calibration of the EPA
CVS will identify any differences in the two flow standards prior to the
evaluation tests. Resolution of such differences can then be made inde-
pendent of the evaluation test comparisons between the Proportional
Sampler and CVS.
Calibration of the other CVS readouts (revolution counters,
pressure and temperature recorders) should also be completed. By assur-
ing the CVS accuracy, any differences between the Proportional Sampler
and CVS specie masses can be isolated to the Proportional Sampler.
B. Sample Gas Analyzers
The C02, CO, NOX and FID analyzers should be completely recali-
brated prior to the evaluation tests. The Proportional Sampler raw sample
and CVS dilute sample will probably require different analyzer ranges.
This can lead to spurious differences in the specie masses unless each
range utilized is calibrated separately. The calibration gases should
be characterized to within +1 percent of measured concentrations. The
existing EPA calibration procedures should be employed.
96
-------
IV. TEST OPERATIONS
Although the evaluation test setup is designed to minimize the
impact of vehicle, driver, and test cycle variabilities, it is recommended
that the series of 10 tests employ a single vehicle, driver, and driving
cycle. The vehicle should have a V-8 engine with approximately 5.75 liter
displacement. The driving cycle should be the LA-4-S-3 implemented with
a hot start (i.e., temperature stabilized engine).
The following paragraphs contain recommendations for the scope
of the various test operations. Further detailed test procedures are left
to the discretion of EPA personnel.
A. Cell Preparation
The pre-test preparations include the following general tasks
for the Proportional Sampler, CVS, and vehicle:
(1) Turn on Proportional Sampler console power.
(2) Turn on CVS power and blower.
(3) Purge and evacuate all sample collection bags.
(4) Replace sample filter elements.
(5) Set heat exchanger water flow at desired value.
(6) Fuel vehicle and run the engine to stabilize the
temperature, and warm up the dynamometer.
(7) Perform Proportional Sampler function check,
including sample valve power.
(8) Set test interval and exhaust flow selector switches
on Proportional Sampler control panel to desired
values.
(9) Turn on Proportional Sampler and CVS sample pumps.
(10) Check instrumentation hookups and readings for CVS
and Proportional Sampler.
B. Pre-Test Data Acquisition
The pre-test data to be recorded include the following:
(1) Barometric pressure (PB).
(2) Ambient dry-bulb temperature (Tj)).
97
-------
C. Test Sequence
The test sequence should proceed in accordance with normal EPA
procedure. The "start test" signal should be input to the Proportional
Sampler coincident with the CVS. The Proportional Sampler will automati-
cally enter the end-of-test mode in accordance with the previously
selected test intervals.
D. Test Data Acquisition
During the actual test the following data should be recorded:
(1) Proportional Sampler water bath temperature (T4).
(2) Proportional Sampler sample line vacuum (P2).
(3) Exhaust Temperature (T^) at the flowmeter.
(4) CVS blower inlet vacuum (PI).
(5) CVS blower differential pressure (APCVs).
(6) CVS blower inlet temperature (Tp).
E. Sample Analyses
After completion of the test, the bag samples should be analyzed
for C02, CO, HC (FID), and NOX (CL) in the following order:
(1) Proportional Sampler raw exhaust sample.
(2) CVS dilute exhaust sample.
(3) CVS ambient air sample.
The analyzers should be zeroed, spanned, and zero-checked before
and after each sample. Span gases should have +1 percent relative
accuracies.
F. Post-Test Data Acquisition
After the test the following data should be recorded:
(1) Proportional Sampler cumulative exhaust volume
readout (Vpg).
(2) CVS blower cumulative revolutions (N).
98
-------
i = CO , CO,
(3) Raw exhaust sample concentrations (X.)'
1 IT O
(4) Dilute exhaust sample concentrations (X_.)riTT j_ ~~9,
, . . . . . n\J , NO
(5) Ambient air sample concentrations (X.) x
i ArliS
V. DATA ANALYSIS
A. Proportional Sampler Specie Masses
The masses of C02, CO, HC, and NOX emitted by the vehicle over
the LA-4-S-3 test, as measured by the Proportional Sampler, should be
calculated using the following equations:
(1) Specie masses
(CO ) =o V CX 1
k 2'PS PCO PS ^ CO ;PS
(CO)PS - pco VPS &CQ>VS
(HC)PS = PHC VPS ^HC^S
-------
(X ) = Volume fraction of the ith specie
(i = C02, CO, HC, NOX) in the raw
exhaust sample, corrected for water
vapor removal
(2) Specie volume fractions
(Vps (Vps t1
where: Wpc = Volume fraction of the ith specie in
the raw exhaust sample, as measured
by the gas analyzers
(X 0)pS = Volume fraction of water vapor in the
2 cooled exhaust, as calculated below
(X^ 0)pS = v°lume fraction of water vapor in the
2 partially dried raw exhaust sample,
as calculated below
(3) Cooled exhaust water vapor volume fraction
(SVP)TI
D
where: (SVP)_,, = Saturated vapor pressure at the aver-
age gas temperature at the exhaust
flowmeter kPa
P = Barometric pressure kPa
B
and where the temperature Tl may be read out at the
proportional sampler control panel.
(4) Sampler water vapor volume fraction
(SVP)
PS Pn - P2
i. a
where: (SVP)T, = Saturated vapor pressure at the
refrigerated water bath temperature
T4 kPa
P2 = Sample line vacuum kPa, as indicated
at the Proportional Sampler control
panel
100
-------
and where the temperature T4 is indicated at the Proportional
Sampler control panel.
B. CVS Specie Masses
The reference masses of C02, CO, HC, and NOX emitted by the
vehicle over the LA-4-S-3 test, as measured by the CVS, should be calcu-
lated using the following equations:
(1) Specie masses
= pco vcvs
(co)cvs = pco vcvs (xco}cvs
(HC)cvs = PHC vcvs (XHC}CVS
^VCVS = PN00 VCVS (XNO >CVS
2 x
where: ^pwc = Cumulative dilute exhaust volume
(cubic meters), corrected to
standard conditions of 101.325 kPa
and 20°C
(X.) = Volume fraction of the ith specie in
the dilute exhaust sample, corrected
for ambient concentrations
(2) Cumulative dilute exhaust volume
(P P ) 293.16
? = V . N .
CVS o 101.325 (T + 273.16)
where: V = Volume of gas pumped per revolution of the
CVS blower (cubic meters), as determined
from the blower calibration curve for the
measured average differential pressure
across the blower
N = Number of CVS blower revolutions during
the test, as measured by the revolution
counter
P = Vacuum at the inlet to the CVS blower
kPa, averaged over the test
T = Temperature of dilute exhaust at the CVS
blower inlet °C, averaged over the test
101
-------
(3) Specie volume fractions
(Vcvs = DIL - (XI>AMB (1 -
where: (X.) = Volume fraction of the ith specie in
the dilute exhaust sample , as measured
by the gas analyzers
(X.). = Volume fraction of the ith specie in
the ambient air sample, as measured by
the gas analyzers
DF = CVS dilution factor, as calculated
below
(4) CVS dilution factor
(xco )PS+(XCO)PS+(XHC)PS~(XCO ^AMB'^CC^AMB^
DF = - £ - £ - . - . --
(XC02)DIL+(XCO)DIL+(XHC)DIL~(XC02)AMB~(XCO)AMB~(XHC)AMB
C. Statistical Analysis
The Proportional Sampler accuracy and repeatability should be
determined through statistical analysis of at least 10 comparison tests
(i.e., trials), with the EPA CVS. The random variables should be defined
as:
. _ (C°2)PS
C°2 "
A _ (CO)PS - (co)cvs
Aco - (co)cvs
_ (HC)pg
HC -
(N°x)PS - CVS
CVS
102
-------
Then for each specie the mean and standard deviation should be
calculated using the following:
A = £ (A ) /n
j=l,n J
S = { £ [(A ) - A ]2/(n-l)}1/2
j=l,n
where: A. = Mean difference for ith specie
A. = Random variable for the ith specie
n = Number of trials
j = Summation index
S. = Standard deviation for ith specie
The desired value of the mean differences is zero. The limits
of systematic (i.e., non-random) inaccuracy for the specie masses should
be determined using the T distribution for small sample tests. With a
probability of 0.9 the absolute value of the systematic error will be
less than:
Vn
where: |A.| = Absolute value of the mean difference for the
ith specie
T = Value of the T statistic up to which the area
under the T distribution is 0.9.
= 1.383 for 9 degrees of freedom (10 trials)
The limits of random inaccuracy (i.e., nonrepeatability) for
the specie masses should be determined using the chi-squared distribution.
With a probability of 0.9 the absolute value of the nonrepeatability will
be less than:
3(J. = 3 . S. (
0.9
103
-------
2
where: Xn q = Value of the chi-squared statistic beyond which
the area under the chi-squared distribution is
0.9.
= 4.168 for 9 degrees of freedom (10 trials)
The performance goals for the Proportional Sampler are (1)
accuracy within +3 percent and (2) repeatability within +5 percent.
These goals correspond to the following criteria, respectively:
e. < 0.03
i
3a. <_ 0.05
which for 10 trails imply the following approximate limits for the means
and standard deviations:
|A | £ 0.0250
S. <_ 0.0113
Should these goals not be met for the group of 10 trails, addi-
tional diagnostic testing could be performed. For example, if the mean
differences are uniformly larger than desired for all species, this would
indicate an erroneous flowmeter signal. This discrepancy could be due
to the flowmeter calibration, water vapor removal correction, or inade-
quate suppression of pressure fluctuations at the flowmeter.
A second example would be acceptable accuracy for the C02
specie, but poor accuracy for the CO, HC, and NOX species. This would
indicate an adequate flowmeter signal, but an erroneous sample proportion-
ality. Additional testing under steady state conditions would then be
desired to determine if dynamic distortion was the cause of the inaccuracy.
A third example would be acceptable accuracy for C02 and CO,
but not for HC or NOX. This would indicate acceptable flow-metering and
sample proportioning accuracies, but would reveal limitations in the
handling of raw exhaust samples. Inaccurate or nonrepeatable HC masses
would indicate hang-up in the sample plumbing. It might be possible to
further isolate this problem using a methane analyzer. Comparisons of
methane mass (if of significant magnitude relative to the total HC mass)
could verify the Proportional Sampler basic accuracy since methane is
not subject to hang-up.
Inaccurate or nonrepeatable NOX masses would indicate either
loss of NOX in the condensate trap, or an increased instability of NOX
species at concentrations typical of raw exhaust. Tests to isolate these
104
-------
possibilities might include a pH measurement of the water from the con-
densate trap and measurement of NOX decay rate in the raw exhaust sample.
Once satisfactory accuracy and repeatability have been estab-
listed for the Proportional Sampler, the testing may be expanded to
include cold starts and alternate vehicles.
105
-------
TECHNICAL REPORT DATA
(/'tease read /uitmctioiis i»t the rci'cru1 tic/ore completing)
1 REPORT NO
EPA-600,
276169f
4 TITLE AND SUBTITLE
DEVELOPMENT OF A PROPORTIONAL SAMPLER FOR
AUTOMOBILE EXHAUST EMISSIONS TESTING
S. REPORT DATE
June 1976
6. PERFORMING ORGANIZATION CODE
3 RECIPIENT'S ACCESSION-NO
7 AUTHOR(S)
8. PERFORMING ORGANIZATION REPORT NO
Harold J. Haskins
9 PERFORMING ORGANIZATION NAME AND ADDRESS
Industrial and Environmental Products Operation
Aeronutronic Ford Corporation
Newport Beach, California 92663
10. PROGRAM ELEMENT NO.
1AA010
11. CONTRACT/GRANT NO.
68-02-1755
12. SPONSORING AGENCY NAME AND ADDRESS
Environmental Sciences Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina
13. TYPE OF REPORT AND PERIOD COVE.RED
Final , 7/74 - 3/76
14. SPONSORING AGENCY CODE
EPA -ORD
15. SUPPLEMENTARY NOTES
16. ABSTRACT
This report describes the development of a device that is capable of
sampling gaseous emissions from automobiles. The device samples exhaust
gases at a mass rate that is proportional to the total exhaust gas
mass flow rate, which is measured using an ultrasonic vortex flowrneter.
The flowmeter delivers signals, which are conditioned by process control
electronics, to a sample valve. Non-standard temperature and pressure
conditions at both the vortex flowmeter and the sample valve are compensated
for in the process control electronics. The report focuses primarily
on development of the vortex flowmeter, the sample valve, and the process
control electronics. These three components comprise the heart of the
proportional sampling system.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
* Air Pollution
Design
^Development
* Sampling
* Test Equipment
Controller characteristics
^Proportioning
*Mass flow
*Exhaust emissions
Automobiles
c. COSATI I lold/Croup
13B
14B
20D
13F
21B
3. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
19 SECURITY CLASS (This Report!
UNCLASSIFIED
20 SECURITY CLASS (This
UNCLASSIFIED
21. NO OP PAGES
120
22 PRICe
EPA Form 2220-1 (9-73)
106
------- |