EPA-650/2-74-029
April 1974
Environmental Protection Technology Series
j>
55
o
\
LU
a
-------
EPA-650/2-74-029
COMPACT SAMPLING SYSTEM
FOR COLLECTION
OF PARTICIPATES
FROM STATIONARY SOURCES
by
C;irl G. Ringwall
General Electric Company
P.O. Box 43. Bldy. 37
Scheneclady, N. Y. 12.501
Contract No. 68-02-0546
Program Element No. 1AI010
Project Officer: John W. Davis
Chemistry and Physics Laboratory
National Environmental Research Center
Research Triangle Park. N. C. 27711
Prepared for
OFFICE OF RESEARCH AND DEVELOPMENT
ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D.C. 20460
April 1974
-------
This report has been reviewed by the Environmental Protection Agency
and approved for publication. Approval does not signify that the
contents necessarily reflect the views and policies of the Agency,
nor does mention of trade names or commercial products constitute
endorsement or recommendation for use.
11
-------
TABLE OF CONTENTS
Section Page
1 INTRODUCTION 1-1
2 RESULTS, CONCLUSIONS AND RECOMMENDATIONS 2-1
2.1 Results 2-1
2.2 Conclustions 2-2
2.3 Recommendations 2-3
3 TECHNICAL DISCUSSION 3-1
3.1 Function Description of Automatic 3-1
Sampler
3.2 Sensors 3-4
3.3 Hardware Development 3-10
3.3.1 Flow Control Valve 3-10
3.3.2 Fluidic Control Amplifier 3-12
3.3.3 Flow Rate and Flow Totalizing 3-21
3.3.4 Vacuum Pump Selection 3-22
3.4 Hardware Description 3-22
3.5 System Setup and Operation 3-28
4 TEST PROGRAM 4-1
4.1 Dynamic Laboratory Tests 4-1
4.2 Steady-State Tests 4-3
4.3 Field Tests 4-15
4.3.1 Preliminary Field Test 4-15
4.3.2 Field Test on Engineering Prototype 4-23
APPENDIX I A-l
Sensor Test Results A-l
Differential Static Sensor A-l
Co-Flow Sensor A-3
Cross-Flow Sensor A-17
APPENDIX II A-25
Controller Parts Identification A-25
-------
LIST OF FIGURES
Number Title Page
3-1 Block Diagram of Sampler 3-2
3-2 Control Model of Sampling Case 3-3
3-3 Transfer Characteristics of Sampling Case 3-5
3-4 Block Diagram of Controller and Sampling Case 3-6
3-5 Sensor Configurations 3-8
3-6 Flow Control Valve Characteristics 3-11
3-7 Control Valve Schematic 3-13
3-8 Throttle Valve Response 3-14
3-9 Steady-State Characteristics of Control Valve 3-15
3-10 Sensor-Amplifier Interface 3-17
3-11 Fluidic Amplifier Schematic 3-19
3-12 Fluidic Amplifier Characteristics 3-20
3-13 Sample Case Flow Vs. Vacuum 3-23
3-14 Functional Block Location 3-24
3-15 Fluidic Controller and Sampling Case 3-25
3-16 \ Inch Sampling Nozzle and Sensor 3-26
3-17 3/8 Inch Sampling Nozzle and Sensor 3-27
3-18 Fluidic Controller 3-29
3-19 Control Console 3-30
3-20 Pump and Flow Meter 3-31
4-1 Dynamic Test Setup 4-2
4-2 Dynamic Response at Various Gain Settings 4-4
4-3 Dynamic Response - 20 Ft/Sec 4-5
4-4 Dynamic Response - 80 Ft/Sec 4-6
4-5 Steady-State Controller Error-3/8 Inch Nozzle 4-7
4-6 Steady-State Controller Error-J Inch Nozzle 4-8
4-7 Controller Error Vs. Supply Pressure 4-10
4-8 Controller Error Vs. Velocity for +20% 4-11
Supply Variation
4-9 Sampling Error Vs. Filter Pressure Drop- 4-12
3/8 Inch Nozzle
4-10 Steady-State Calibration-3/8 Inch Nozzle 4-13
4-11 Steady-State Calibration-^ Inch Nozzle 4-14
4-12 3/8 Inch Diameter Sampling Nozzle 4-17
11
-------
LIST OF FIGURES (cont.)
Number Title Page
4-13 \ Inch Diameter Sampling Nozzle 4-18
4-14 Calibration Runs-3/8 Inch Diameter Nozzle 4-19
4-15 Calibration Runs-^ Inch Diameter Nozzle 4-20
4-16 3/8 Inch Nozzle 4-21
4-17 \ Inch Nozzle 4-22
4-18 Velocity Ratio Vs. Time on Coal-Fired 4-25
Installation
4-19 Fluidic Controller Output Pressure Vs. Time 4-26
A-l Differential Static Sensor Configurations A-l
A-2 Differential Static Sensor Characteristics- A-4
3/8 Inch Nozzle
A-3 Differential Static Sensor Characteristics- A-5
\ Inch Nozzle
A-4 Effect of Non-Isokinetic Probe Flow on A-6
Static Differential Sensor Reading
A-5 Static Sensor Noise A-7
A-6 Co-Flow Sensor Gain A-8
A-7 Co-Flow Sensor Noise A-9
A-8 Co-Flow Sensor Gain/Null Bias Ratio A-10
A-9 Difference in Two Co-Flow Sensor Indications A-ll
A-10 Co-Flow Sensor Noise Vs. Bandwidth A-12
A-ll Velocity Acceleration at Nozzle Inlet A-13
A-12 Velocity Acceleration at Nozzle Inlet A-14
A-13 Co-Flow Sensor Output Vs. Velocity A-15
A-14 Cross-Flow Signal Upstream Receiver A-18
A-15 Cross-Flow Signal Downstream Receiver A-19
A-16 Velocity Deceleration at Inlet A-20
A-17 Velocity Deceleration at Inlet A-21
A-18 Velocity Deceleration at Inlet A-22
A-19 Cross-Flow Sensor Noise Vs. Bandwidth A-24
A-20 Fluidic Controller-Front Face A-26
A-21 Fluidic Controller-Back Face A-27
A-22 Control Console Connectors A-28
Table 1 Controller Gain Setting 3-33
Table 2 Parts Identification A-29
Table 3 Inlet-Outlet Connectors A-30
iii
-------
ENGLISH TO METRIC CONVERSIONS
English Units
SCFM
PS I
ft
ft3
#/in
in/sec
psi/ft/sec
. 2
in
#/in3
/• 2
sec/in
inches HO
Metric Units
28.317 Liters/min
6895 Newtons/meter2
3.048(10~1) meter
2.831(10~2) meter3
175.125 Newtons/meter
2.54(10~2) meters/sec
22622 Newton sec/meter'
6.45(10~4) meter2
61023.4 Newtons/meter3
o
1550 sec/meter
249.1 Newtons/meter2
IV
-------
ABSTRACT
This report summarizes the work performed on a program
to design, fabricate, and evaluate a controller for
automatically sensing and maintaining isokinetic condi-
tions at the inlet of a particulate sampling nozzle.
The key components developed on the program were 'the
gas velocity sensor and a fluidic control amplifier.
The sensor concept is based on a static pressure
differential between the free air stream and the nozzle
inlet. The fluidic control amplifier which interfaces
directly with the sensor provides the control to auto-
matically maintain isokinetic conditions.
Field tests were performed on the engineering proto-
type system at both oil-fired and coal-fired power
plant installations. Results of these tests showed
that the sensor and controller can function with no
degradation in performance under the adverse environ-
ment of representative power plant stacks. Temperatures
up to 205°C and solid particulate concentrations of 3.50
grams per cubic meter were encountered during the field
testing.
-------
1. INTRODUCTION
This report summarizes all work under Contract 68-02-0546,
Compact Sampling System for Collection of Particulates from
Stationary Sources. An engineering prototype system capable
of automatically maintaining isokinetic sampling conditions
that can be interfaced with presently available air pollution
control sampling trains and Beta gauge monitors was designed,
fabricated and tested.
The specific design goals for the system are:
• Isokinetic sampling of gas streams having velocities in
the range of 20 to 150 feet per second.
• Isokinetic sampling of gas streams having temperatures
of -ic-C to 535°C.
• A sampling rate of 0.5 to 20 standard cubic feet per
minute.
• A response time capable of following flow fluctuations
of +10% with a period of 30 to 120 cycles per minute.
• Automatic control of the sampling rate.
• Totaling automatically the total gas sample flow.
• Provide a visual output reading.
• Electrical outputs to accommodate continuous recording
of sampling rate.
1-1
-------
2. RESULTS, CONCLUSIONS AND RECOMMENDATIONS
2.1 Results
The engineering prototype system developed on this program
meets all of the design objectives, with the exception of the
sampling rate of 0.5 to 20 scfm. Presently available APCO sampling
trains limit the maximum sampling rate to approximately 2.0 scfm.
This limitation is due to the impedance of the particulate
collecting filter and the standard impingers used in the sampling
train. The sensor and automatic controller are compatible with
the specified sampling rate if used with a sampling unit capable
of sustaining the flow rate. High volume air samplers are
commercially available and can satisfy the maximum specified flow.
The key components developed on the program are the gas
velocity sensor and the fluidic controller required to automatically
maintain isokinetic sampling. The developed sensor is based on
measuring the differential static pressure between the free air
stream and sampling nozzle inlet. This sensor:was selected over
two other potential candidates, the cross-flow and co-flow sensor,
on the basis of signal-to-noise ratio.
The automatic controller was designed as an integral
component of a modified commercial sampling case with no change
in outline dimension. The weight of the sampling case was
increased by 4 pounds or approximately a 15% increase.
Operation of the controller requires minimal operator
training. Prior to starting a sampling test, the controller gain
control is set to correspond to a tabulated setting. This setting
is a function of the pressure differential across the S-type pitot
tube on the probe. One additional operation is required to remove
system bias. The bias is removed by adjusting a regulator to
obtain a specified reading on a pressure gauge. No further adjust-
ments are necessary if the velocity of the sampled gas is in the
range of 15 to 40 ft/sec. For the higher velocity ranges one
additional step by the operator is required to compensate for a
sensor bias. After the probe has been inserted in the gas stream,
the flow corresponding to isokinetic conditions is determined
using measured gas temperature and velocity as input parameters.
The bias control on the controller is then adjusted until the
system flow rate meter corresponds to the calculated flow.
The flow totalizing function of the controller utilizes
a mass flow meter and an electromechanical counter. Visual read-
out of sampling rate and totalized flow are provided as well as
an electrical output of sampling rate. The flow meter is a true
mass flow meter and does not require correction factors for gas
temperature or absolute pressure.
Results of the field tests demonstrated that the sensors
and controller can function in an adverse environment with no
degradation of performance. Isokinetic sampling was controlled
to an accuracy of 5% in gas temperatures to 205OC and solid
particulate concentration as high as 3.52 grams per cubic meter.
2-1
-------
2.2 Conclusions
The following specific conclusions can be drawn from the
results of this program:
• Automatic control of sampling rate will signifi-
cantly increase accuracy and repeatability of sampling. Large and
erratic variations in stack velocity were observed during the
field tests. Establishing a time average of isokinetic flow
requires an operator judgment factor. This factor is a variable,
depending on the operator, the environmental stresses present
during the test, and the physiological state of the operator
during the test. The automatic control will, to a large extent,
overcome the effects of these variables.
• Fluidic amplification of the sensor error signals
provides a reliable and economical technique for implementing the
automatic controller. In the lower band of the specified velocity
range fluidics is considered the only practical technique for
reliably sensing and amplifying the small error signals. At 20
feet per second and a gas temperature of 535°C, differential
pressures of less than 0.005 inches of water must be sensed and
accurately controlled. Commercially available electromechanical
pressure transducers have adequate sensitivity and are excellent
laboratory instruments; however, they are subject to excessive
null shift with temperature changes and with abuse, and are not
attractive for a field type operation,
• The limiting source of error in the automatic
controller is in the sensor. The error is introduced by losses
associated with the protective shrouding enclosing the free air
stream sensor. This error can be partially compensated for by a
controller adjustment. The compensation imposes an additional
burden on the operator in that isokinetic sampling flow must be
determined.
The error of the fluidic amplifier and
other control components is less than +2%.
• Contamination of the sensor and associated fluidic
amplifiers can be prevented by backflushing the sensor ports with
clean air. Location of the sensor ports relative to the overall
probe geometry is also a significant factor in achieving insensi-
tivity to contamination.
• Test and evaluation of three sensor configurations,
the cross-flow sensor, the co-flow sensor, and the differential
static pressure sensor, showed that only the differential static
sensor has sufficiently high signal-to-noise ratio to meet the
design goals on system bandwidth.
2-2
-------
2.3 Recommendations
The specific design goal which had the greatest impact on
the overall system design was the specified capability of following
2 Hz perturbations in the gas velocity. This requirement
eliminated both the cross-flow and co-flow sensors because of
inadequate signal-to-noise ratio in the system bandpass.
With the exception of the bandwidth requirement, the co-
flow sensor is attractive for this application and would have been
the logical sensor choice. Both the controller hardware and the
setup and operating procedure could be significantly simplified by
utilizing this sensor. The gain of this sensor is independent of
velocity and an order of magnitude higher than the static probe in
the low velocity range. In addition, shrouding losses can be
negated by a coarse adjustment of sensor flow. These character-
istics eliminate the need for presetting gain as a function of the
S-type pitot reading as well as the determination and adjustment
for isokinetic sampling flow.
In view of these observations, the following recommenda-
tions are made:
• The bandwidth required of a particulate sampling
system should be re-evaluated.
At the extremes of the solid particulate spectrum,
i.e., the large, heavy particles where the sampling nozzle is a
100% efficient impact probe'*' and for the fine particles where
the gas stream lines are followed, the only prerequisite for the
sampling system is ability to maintain an average isokinetic
velocity to obtain totalized flow from which a determination of
particulate concentration can be made. There would be no errors
introduced by the controller's inability to track rapid
perturbations.
In the mid-range of the spectrum there is a
potential error source because of the unsymmetrical relationship
between sampling error and positive or negative errors in sampling
velocities. This relationship has been verified experimentally
and theoretically (2) an(j becomes significant for very large
deviations from isokinetic conditions. Within the design goal
range of +10% deviations from nominal isokinetic velocities, the
dissymmetry is minimal.
The investigation of sampling error vs. bandwidth
could be approached from both theoretical considerations and
empirical testing completely divorced from the hardware developed
on the current program.
• If a reduction in system bandwidth to approximately
0.2 Hz or lower appears feasible, then a retrofit of the engineering
prototype unit to include co-flow sensors is recommended. This
retrofit would require fabrication and replacement of the sampling
nozzles and minor modification to the fluidic controller.
2-3
-------
• If the system bandwidth cannot be reduced, the
only alternative is the sensor delivered with the prototype
system. A modest test program should then be initiated to
further reduce the sensor error.
2-4
-------
3. TECHNICAL DISCUSSION
3.1 Functional Description of Automatic Sampler
A simplified block diagram of the sampler is shown in
Figure 3.1. The free air stream velocity (VR) is converted to
pressure by the free air stream sensor. This pressure is compared
to the output of a similar sensor located in the inlet of the
sampling nozzle. The resultant pressure error signal is amplified
and adjusts the suction on the vacuum pump to give the required
nozzle inlet velocity (VN).
The closed loop transfer function for the simplified block
diagram is given by:
(1)
N
KRG
1+KXTG
N
If KjjO^l and identical sensors are used for free air stream and
nozzle inlet velocity, then KR = KN.
(2)
N
•i"
Providing the free air stream sensor satisfies the
criteria for isokinetic sensing, then isokinetic gas velocities
will be maintained at the inlet of the sampling nozzle. The
accuracy of the controller is directly related to the loop gain
(KNG), as shown in equation (1). To maintain isokinetic
velocities to an error of less than 5% requires a loop gain greater
than 20.
A control model of the sampling case is shown in Figure
3-2. The open loop transfer functions for the model representing
the complete sampling train is given by:
(3) TF
-sT
(R1C2+RC2)S+l[
+ RC
j + RC2)SJ
1 +
where:
RR.| C- CpS
(1+TS)
3-1
-------
Free Air
R >^ Stream
Sensor
(KR)
CO
to
Amplifier
(G)
Control
Valve
& Vacuum
Pump
Sampling
Case
N
Sampling
Nozzle
Sensor
N
Figure 3-1. Block Diagram of Sampler
-------
R
Sampling
Nozzle
Inlet
CO
i
u
Control
Valve
Pump
Figure 3-2. Control Model of Sampling Case
-------
G = gain of forward loop - #/sec/in/sec
A = cross-sectional area of sampling probe - in
1° = density of sampled gas - #/in3
T = time constant of control valve - sec
f = transport lag - sec
C ^ capacitance between sampling nozzle and filter
Cj = capacitance between filter and first impinger - in2
f\
Cg = capacitance of impinger section - in
p
R = filter restriction - sec/in
o
R = impinger restriction - sec/in
The closed loop transfer function relating the sampling
nozzle inlet velocity to free air stream velocity is given by:
p C ~*S V /^ . siTtri\
(4)
s
VR
-sT^
(^A+G '
-CoT+RCoT+RR-C.,
-L ^ ^ _L J
/'ARR
^A
C2] S
X>A + G£ -S T
1C1C2TS3 +
2
+ /"A (T+RIC
/-A
2+RC2) S + 1
+ GS'3^
Equation (4) indicates that to maintain isokinetic
conditions in the sampling nozzle, the gain (G) in units of
mass flow (#/sec) per in/sec of velocity error must be much
greater than the product of gas density and sampling nozzle
area (9 A) .
A measured attenuation and phase plot of a commercially
available sampling case is shown in Figure 3.3. The characteristics
include the flow modulating valve designed for the loop and a
representative 10-foot length of signal transmission line. With
the system gain set for 20 />A, the crossover is a 5 Hz with a
negative phase margin of 20 degrees. Stable operation of this
loop will require frequency shaping of the signal. A lag-lead
circuit is used to lower the open loop crossover to 2.5 Hz. while
still maintaining adequate steady-state gain. In addition to the
lag-lead, a lead circuit is required to get adequate phase margin.
Figure 3-4 is a block diagram of the system as implemented.
3 . 2 Sensors
The gas velocity sensor is the key component in the
automatic sampling control loop. It is exposed to a severe
environment with temperatures as high as 535°C, along with heavily
contaminated gases. In addition to sensing gas velocity, one
3-4
-------
HN--J6S-A (8-30)
GENERAL ELECTRIC COMPANY
SCHENECTAOY. N. Y.. U.S.A.
.1 eye. T 170 Imra Divisions
Figure 3-3. Transfer Characteristic of Sampling Case
-------
R "
u
Fluidic Controller Control Valve Sampling Case
Free
Air
Stream
Sensor
GU + .5S) (1 + .08S)
(1 + 2 STTl + .OlS)
Nozzle
Inlet
Sensor
(1+.01S)
(1+.2S)(1+.IS)
N
Figure 3-4. Block Diagram of Controller and Sampling Case
-------
additional constraint is imposed on the sensor in this application—
isokinetic conditions must be satisfied on the reference sensor
(free air stream sensor).
Three fluidic sensor configurations were tested and
evaluated on the program. They were:
a) Cross-flow sensor
b) Parallel or co-flow sensor
c) Differential static pressure sensor
Figures 3-5a, b and c are schematics illustrating the fundamental
operating principle of each sensor. The cross flow sensor shown
in Figure 3-5a utilizes a jet directed at two receivers. The axis
of the nozzle forming the jet is perpendicular to the free air
stream velocity vector. The angular displacement of the jet axis
from the nozzle axis is given by:
(5) 0 = tan-1 M
s
MT
J
Where Ms is the momentum imparted to the jet from the free air
stream and Mj is the momentum of the power jet. If the receivers
are positioned to give a linear relationship between differential
pressure and jet angle, then the output of the sensor is
proportional to the square of the free air stream velocity and
the sensor gain is directly proportional to the free air stream
velocity. To a first order approximation, the sensor gain
(AP/AVS) is independent of the jet velocity; however, the
maximum range of the sensor is directly dependent on jet velocity.
The co-flow sensor shown in Figure 3-5b utilizes a jet
directed at a single receiver with the jet axis parallel to the
free air stream velocity vector. Interaction between the free
air stream and the jet at the jet boundary imparts a portion of
the free air stream velocity to the jet. The change in recovered
pressure at the receiver is linearily related to free air stream
velocity providing the incremental velocity imparted to the jet is
small compared to the nominal free jet velocity. Sensor gain and
maximum range are directly related to jet velocity.
The differential static pressure probe shown in Figure 3-5c
compares the static pressure in the free air stream to the static
pressure at the inlet to the sampling nozzle. When the static
pressures are equalized the velocity at the inlet to the sampling
nozzle will be equal to the free air stream velocity. In contrast
to the other two sensors, this sensor is not an absolute velocity
sensor — it can sense only equality of gas velocities in a common
duct. The sensor gain is given by:
+ V )
s vf'
3-7
-------
Free
Air
Stream
(V
Sampling v
Velocity r
\ \ \ \
^<\ \\ \ \
.. \ \ \ \ \ \ \
Error
Signal
\ \ \
\ \ \
\\\\n\ \ \ \.\ \ \ \\
a. Cross-flow Sensor
\\\\\\\\ N \ \ V
P Error
S Signal
\\ \\\ \
NT..\\x\\\\\\\ \\
R
b. Co-flow Sensor
^_. <:
,
C q
•
v\\\v\ \\v \v\ \ \\v\\
3
IA\\\\\\\\ \\N\\\\\
st\ \\ \\ \ \ \ \ \ \ \\ \ \ \\
I
I)
\
n
t
Error
Signal
V
r
c. Differential Static Sensor
Figure 3-5. Sensor Configurations
3-8
-------
where:
f = density of gas
Vs = velocity at sampling nozzle inlet
Vf = free air stream velocity
Av = (v -v.)
5 X
AP = output of sensor
At near isokinetic conditions at the inlet to the sampling nozzle,
Vs^Vf and the gain becomes proportional to the free air stream
velocity.
In the initial phases of the program, primary emphasis
was placed on the development of the cross-flow sensor. Because
of the relatively high supply pressures used on the sensor, along
with the capability of operating with back pressured receivers,
it appeared to offer the greatest potential for successful
operation in a contaminated environment. In addition, there was
considerable background experience on application of the sensor as
an anemometer. A sensor with satisfactory gain and range
characteristics was developed on the program. However, the
developed sensor failed to meet the criteria for maintaining
isokinetic characteristics at the free air stream sensor.
In addition, the signal-to-noise ratio of the sensor was
not acceptable. Noise saturation of the signal amplifier stages
occurred when operating at the required steady-state gain and
bandwidth. The sensor was abandoned for these reasons.
The development effort was next directed at the co-flow
sensor. As a preliminary step it was established that this sensor
could meet the criteria for maintaining isokinetic characteristics
with a practical hardware configuration. The developed sensor had
excellent characteristics from the standpoint of scale factor,
range and constancy of scale factor over the complete range. The
signal-to-noise ratio of the sensor is comparable to the cross-
flow sensor and was not adequate to meet the system requirements.
This sensor, though rejected for this particular application,
would be the logical choice for a system with a bandwidth on the
order of 0.2 Hz (an order of magnitude lower than the contract
requirements).
The specific configuration of the differential static
pressure probe was conceived and developed concurrently with the
development of the co-flow sensor. Initial tests on the sensor
concept showed a signal-to-noise ratio of about an order of
magnitude higher than the other sensors and that isokinetic
conditions could be maintained at the free air stream sensor. For
these reasons this sensor was selected as the prime sensor for
further development.
Test results on the three sensor configurations are
reported in Appendix I.
3-9
-------
3.3 Hardware Development
A prime objective in the layout of the system was to
minimize the weight and physical size of the flue mounted sampling
case. Existing units are already heavy and quite difficult to
maneuver while inserting the probe into the three-inch sampling
ports which are typical on existing installations.
The basic control circuit consisting of the sensor, flow
control valve, and amplifiers must be located on the sampling
case. This is predicated by a consideration of signal transport —
the transport lag associated with the vacuum line in the umbilical
cord connecting the sampling case to the ground-based central unit
must be eliminated from the control loop. This cord can range
from 25 to 100 feet in length and may introduce transport lags in
excess of 0.1 sec. Stabilizing a control loop with a 2 Hz band-
width would be very difficult with transport lags of this magnitude.
The physical location of the flow rate and flow totalizing
meters will depend on the configuration of the flow control valve.
Flow in the sample collecting nozzle can be controlled by a
throttling valve placed in series with the nozzle and the vacuum
pump or by a bypass valve around the pump.
With a throttle control valve, the flow in the umbilical
cord is equal to the sampled flow and the flow rate and flow
totalizer meters can be located at the pump end of the umbilical
cord. On the other hand, with a bypass valve, the flow in the
umbilical cord is equal to the sampled flow plus the bypass flow.
With this configuration, the flow meters must be inserted between
the sampling nozzle and the control valve. In the case of the
automatic control loop, this dictates that the flow meters also
be stack mounted.
Both system configurations and a definition of signal
shaping and gain required in the remaining control loop depend on
the control valve; hence, it was the first of the control elements
developed.
3.3.1 Flow Control Valve
Figure 3-6 is a plot of flow through the sampling
nozzle as a function of control valve area for both the throttling
valve and the bypass valve. The flow characteristics were
determined for two values of sampling case impedance. The solid
curves represent a pressure drop of 7.0 psi at a flow of 1.5 scfm;
while the dashed curves are for a 7.0 psi drop at 0.5 scfm. For
a given valve area, the flow control range is somewhat higher with
the throttle valve as compared to the bypass valve. The major
difference between the two approaches is the control range where
the valve becomes non-linear. The throttle valve becomes non-
linear in the maximum flow range, whereas the bypass valve is non-
linear in the minimum flow range.
3-10
-------
3-6. Flow Control Valve Characteristics
3-11
-------
From the standpoint of control characteristics,
either configuration can be adapted to the system. The bypass
valve maintains a relatively constant flow at low suction at the
inlet to the vacuum pump; hence is conducive to long life and
also minimizes pump leakage flow as compared to the throttling
valve. These inherent advantages of the bypass valve over the
throttling valve were not considered of sufficient significance
as compared to the disadvantage of stack-mounted flow meters. On
this basis, the throttling valve configuration was selected.
Figure 3-7 is a schematic of the control valve.
Flow is throttled by a spring-loaded poppet valve. The poppet is
actuated through a bellows seal by a pressure differential across
a flexible diaphragm.
Figure 3-8 shows the frequency response character-
istic of the valve. The valve has the characteristic of a simple
time constant of 0.01 second.
A plot of the steady-state control characteristics
of the valve is shown in Figure 3-9. The valve goes through its
full control range with a control pressure change of 1.5 psi.
Gain of the valve is 1.4 scfm per psi when inserted in the sampling
case.
3.3.2 Fluidic Control Amplifier
The fluidic control amplifier completes the inter-
face between the differential static velocity error sensor and the
control valve. The prime prerequisites of the amplifier are:
• To make a satisfactory interface with the sensor,
the amplifier must be capable of establishing a local reference
slightly positive relative to the stack static pressure. This
gives an outflow of clean air from the amplifier through the sensor
ports and thus prevents contamination of the amplifier and the
sensor.
• The threshold of the amplifier must be well
below 0.01 inches of water. This is the magnitude of error signal
produced at 20 ft/sec gas velocity and a ten percent error in
velocity.
• The steady-state gain must be sufficiently high
to maintain isokinetic conditions as flow impedance of the sampling
case changes with filter loading.
• Amplifier must provide the proper signal
frequency shaping to stabilize the overall control loop.
o To interface with the control valve the output
driver amplifier must have high flow capability to insure adequate
bandwidth in the control valve.
3-12
-------
From Fluidic Controller
Ambient
From
Sampling
Case
Flexible
Diaphragm
Bellows
Seal
To
X Vacuum
Pump
Figure 3-7. Control Valve Schematic
3-13
-------
CO
Figure 3-8. Throttle Valve Response
-------
18(1x250 1mm Divisions
Figure 3-9. Steady-State Characteristics of Control Valve
-------
Figure 3-10 illustrates the method of interfacing
with the sensor. The input stage is a proportional fluidic
amplifier whose supply for the jet nozzle is derived from a
relatively high pressure and a flow restrictor. The spill-over
flow from the jet (i.e., all flow not exiting through the receiver)
is collected and vented to the stack through the stack static
reference tube. The flow impedance of the reference tube and the
magnitude of flow, as determined by the flow restrictor. determines
the pressure in the amplifier vent region relative to the stack
static. In the developed amplifier these parameters were fixed to
establish a pressure level at the amplifier of a positive 0.5 inches
of water relative to the static pressure of the stack. This
pressure level also insures an outflow of clean supply air from
the amplifier control ports and sensor ports.
The steady-state gain requirements of the controller
can be determined by reference to equation (4) in section 3. The
steady-state ratio of the sampled velocity to reference velocity
is given by:
(7) VS G
VR
To maintain the velocity error of less than 5%
requires that VS/VR ;> .95 and the nominal value of G must be
greater than 20 fk. Using a typical sampling nozzle area of
0.1 in^ and a density based on air at 30F, /^A = 4.6/106 #/in.
The nominal gain must then be:
G = 20 f A = ?j_| #/sec/in/sec =
10°
Referring to Figure 3-9, the gain of the valve is
1.4 scfm/psi or 63/106 #/sec/in/H20.
4 At a nominal gas velocity of 20 ft/sec, the sensor
gain is 8.7/10 inches of water per in/sec velocity error. The
product of sensor and valve gain is then approximately 5.5/108 #/in.
To obtain the required gain of 9.2/105 requires a pressure
amplification between the sensor and the flow valve of 1600.
Another argument for determining the steady-state
gain can be based on full utilization of the flow control valve
range to accommodate increased filter loading. Referring to
Figure 3-9 it is apparent that the linear range of the control
valve is traversed with a pressure change of 0.8 psi (22.25 inches
of water). If this range is to be traversed with a 5% velocity
error at 20 ft/sec, the pressure gain must be 2225. This latter
criterion was used in establishing the final amplifier configuration,
3-16
-------
o
CO
I
Free Air
Stream
Sensor
Nozzle
Inlet
Sensor
Stack Static
PStatic
Laminar Flow Restrictor
Restrictor
Vent Flow
To Second
Stage Amplifier
Figure 3-10. Sensor-Amplifier Interface
-------
Figure 3-11 is a schematic of the developed
amplifier, including signal shaping networks. The input amplifier
is followed by four cascaded stages of proportional amplifiers.
These amplifiers are miniature amplifiers (0.01 x 0.01 inch power
nozzle) and provide a maximum pressure gain of 2000. A dual
variable resistor is used between the third and fourth stages to
reduce the gain to accommodate the change in gain characteristics
of the velocity sensor with velocity and also the area
differences of sampling nozzles. A pair of fixed resistors bypass
the input ports to the fourth stage. These resistors can be
switched in of out to accommodate very high velocities and small
sampling nozzles.
The output of the fifth stage goes through a
passive lag-lead circuit. This circuit has a lag break at
0.5 rad/sec and a lead at 2.0 rad/sec and serves the function of
lowering the system crossover frequency to 2.5 Hz (12.3 rad/sec).
Without this lag-lead circuit the system crossover when the steady-
state gain requirements are satisfied is 5.0 Hz. The accumulated
phase lag from the control valve, the signal transport lag and the
sampling case characteristic made it virtually impossible to
stabilize the system at 5 Hz crossover.
The signal is next processed in a lead-lag circuit.
This function is implemented with five cascaded stages of propor-
tional amplifiers with feedback to give the signal shaping. The
lead break of this amplifier occurs at 12 rad/sec with the lag at
90 rad/sec. The function of the lead-lag is to provide adequate
positive phase margin at the system crossover. Two additional
input resistors are provided on the lead-lag circuit. These
resistors function as a summing junction, whereby a bias signal
can be inserted to offset bias introduced by circuit or sensor
dissymetries. This bias control can also be used to establish
control values of Vg/VF greater than unity if so desired.
The last three stages following the lead-lag
circuit function as flow amplifiers. The pressure gain of the
cascaded stages is 2.5 with a flow gain of 50. The output stage
is sized to obtain a 0.01 second time constant on the control
valve.
The resistors associated with the supply are used
to establish the supply pressure levels required for the cascaded
stages. All circuit elements are protected from contamination
entering through the supply by three integral filters, shown on
the schematic. In addition to these filters, a primary filter for
all supply air to the controller is located in the ground-based control console,
The gain vs. frequency characteristic of the
amplifier is plotted in Figure 3-12. In the frequency range of
0 to 10 Hz, the characteristics are dictated by the overall loop
requirements on steady-state accuracy and bandwidth. At the
higher frequencies, the gain can be rolled off to attenuate high
frequency noise components. It is apparent from Figure 3-12 that
even with the maximum high frequency roll off, the amplifier will
3-18
-------
From ^
Sensors \7\
-^
To Control
Valve
Proportional Amp.
Capacitor
To Vent
Figure 3-11. Fluidic Amplifier Schematic
-------
FN-522-B CS-50)
• ENEF-AL ELEC t -UMPANY SCHENECTADv N » i S A.
i Lop >. voles * 90 Divisions
i:i 31; 4 ; 5 : t . 7 . '"air
Figure 3-12, Fluidic Amplifier Characteristics
-------
have a relatively broad bandwidth with a high average gain and
will be vulnerable to saturation by broadband noise. This high
gain-bandwidth product severely limits the maximum tolerable
sensor noise. This product can be roughly approximated by an
average gain equal to the steady-state gain with a bandwidth of
70 Hz. Considering the previously determined minimum gain
requirement of 9.2/105 #/in and a valve with a flow control range
of approximately 2.2/103 #/sec (2 scfm), then the velocity error
required to saturate the system is:
v _ 2.2 (105) .. . ,
V^ - —* = 24 in/sec
& 10* (9.2)
As a rule of thumb,peak-to-peak noise of approxi-
mately one half the saturation level can be tolerated on the output,
If the sensors have a noise source with a Gaussian distribution,
as typical on all sensors investigated in this program, the ratio
of peak-to-peak noise to RMS is 6:1. This requirement translates
into a maximum RMS noise of 4 in/sec measured with a 70 Hz
bandpass filter. Measured sensor noise on both the co-flow and
cross-flow sensors is an order of magnitude greater than this
(see Appendix I). Application of these sensors will require a
severe compromise on gain or bandwidth. For example, a 2:1 gain
reduction in conjunction with a 10:1 bandwidth reduction would be
required.
3.3.3 Flow Rate and Flow Totalizing
The differential static pressure sensor selected
introduces negligible diluent flow into the sample. This permits
the use of standard commercially available flow meters. A
Hasting-Raydist mass flow meter was selected for the flow rate
measuring function. This is a true mass flow meter and can be
located at either the suction or exhaust side of the vacuum pump.
In this application the meter is located in the suction side of
the pump to eliminate errors due to pump leakage.
The meter has a relatively long time constant
(1-2 sec). This is not considered detrimental in a system where
flow rate is controlled automatically in that the prime function
of the flow meter is to get a reasonable time average. Electrical
outputs suitable for remote recordings are provided on the meter.
Flow totalizing is also done with a commercially
available totalizer. The totalizer accepts the analog output of
the flow rate meter, makes an analog-to-frequency conversion, and
accumulates pulses on an electromechanical counter.
Both the flow rate and totalizer functions have
been integrated into the ground-based control console.
3-21
-------
3.3.4 Vacuum Pump Selection
One of the objectives of the program was to
provide for sampling rates of 0.5 to 20 scfm. The upper limit of
20 scfm cannot be achieved with currently available APCO
particulate sampling systems with which the developed system must
interface. The upper flow limit is limited by the filter
restriction and impingers in the sampling train and above a
certain limit is independent of pump capacity and sampling nozzle
area.
The criteria used to size the pump for this program
was to select a pump with adequate capacity so that the maximum
flow is limited by the APCO sampling case and not by the pump.
Figure 3-13 is a plot of sampling flow vs. pump
vacuum. It is apparent that vacuums greater than 8 psi do not
significantly increase the sampling flow. On the basis of this
curve, the pump is selected with sufficient capacity to provide
the flow at a vacuum of 8 psi. A pump with a 7 scfm capacity at
zero vacuum was selected.
3.4 Hardware Description
Figure 3-14 is a schematic identifying the location of
the primary components supplied on the engineering prototype
system. The blocks identified with either the sampling case or
control console are integral parts of these units.
The sampling case is shown in Figure 3-15. This unit
contains the standard impingers, filters, and filter ovens
required to conduct sampling tests as specified in the Federal
Register. Volume 36, No. 247, Pt. II.
The automatic controller is located inside the sampling
case in a rectangular enclosure between the filter oven and the
ice bath container for the impingers. The flow control valve is
located in back of the controller. Heat transfer from the filter
oven will maintain the air temperature around the control unit
and valve at approximately 10°C above the ambient air temperature.
This is an ideal temperature range for both the controller and the
valve.
Three pneumatic lines go from the controller out to the
sampling end of the probe. Flexible Tygon tubing is used inside
the sampling case to connect the stainless steel tubing at the
sampling case end of the probe to AN couplings. The flexible
tubing permits rotation of the sampling probe through an angle of
360°. The three connectors, two for the velocity sensor and one
for stack static reference, are shown on the left side of Figure
3-15. The stainless steel tubes are attached to the outer jacket
of the probe and the two signal lines terminate in an AN connector
at the sample nozzle end of the probe as shown in Figure 3-16.
Figure 3-17 is a closeup view of the sampling nozzle and sensor.
The static pressure ports for the sampling nozzle are drilled in
3-22
-------
FN-156 (8-50)
GENERAL eLZCTRIC COMPANY SCHENECTAOY N. Y. U.S.A.
160x2511 1mm Divisions
Figure 3-13. Sample Case Flow Vs. Vacuum
-------
Sensors &
Sampling Nozzle
Stack v
Mounted^
Ground Based
• Sampling Case
• Control Val\e
• Fluidic
Controller
\
S-Type Pitot
CO
I
to
*..
Vacuum
Pump
and Flow
Meter
Control Console
Flow Totalizing
Flow Rate Readout
Thermocouple Readout
Temperature Controls
S Pitot Readout
Pressure Regulator
Controller Bias
Adjust
Figure 3.14. Functional Block Location
-------
Figure 3.15 Fluidic Controller and
Sampling Case
Page 3-25
-------
Signal Lines
CO
CO
05
Figure 3. 16 1/4 Inch Sampling Nozzle and Sensor
-------
Free Air Stream Sensoi
Nozzle Inlet Sensor
Figure 3.17 3/8 Inch Sampling Nozzle and Sensor
Page 3-27
-------
the wall of the tube protruding axially from the sampling nozzle.
Four ports, 0.02 inch diameter, are used and they are located in
the plane of the nozzle inlet. The free air stream static
pressure ports are in an identical tube positioned outboard of the
sampling nozzle. A protective shroud is used to prevent damage
when inserting or withdrawing the probe from a sampling port.
Figure 3-18 is a photograph of the fluidic controller.
The knob at the upper right hand side of the controller is a
variable resistor. The steady-state gain of the controller is a
function of the resistor setting. The resistor setting required
to achieve a stable control mode is a function of the dynamic
pressure as measured by the S-type pitot tube attached to the
sampling probe and the area of the sampling nozzle. The required
settings are tabulated in Table 1 under System Setup and Operation.
The left hand knob is used as a two position switch. When the
zero is in the vertical position, a bypass resistor is connected
to the downstream side of the variable resistor giving approximateIv
a three-to-one gain reduction. This low gain setting is used w'th
the \ inch diameter sampling nozzle and high gas velocities.
The control console is shown in Figure 3-19. The lower
left hand section contains the mass flow meter indicator and the
flow totalizer. The mass flow meter also has an electrical output
which is brought out to two jacks on a rear panel of the console
for recording.
The inclined manometer is used in conjunction with the S
type pitot tube on the sampling case probe to measure the velocity
head of the sampled gas.
Directly above the manometer are two pressure regulators
and a pressure gauge. The upper regulator sets the supply pressure
to the automatic controller, the pressure gauge indicating the
supply pressure. The lower regulator is used to apply a bias to
the controller to offset any residual bias.
The right hand section of the console contains the
sampling oven and probe jacket temperature controls as well as a
transfer switch and temperature readout for the various thermo-
couples .
The system vacuum pump and flow meter are shown in
Figure 3-20. The pump has a capacity of 7 scfm at 14.7 psia on
the inlet. The flow meter is permanently attached to the pump
inlet. A quick disconnect vacuum line connection is on the right
hand side of the flow meter.
3.5 System Setup and Operation
Setup of the basic sampling case follows the standard
procedures established for manual control. After the sampling
probe has been inserted in the sampling case, the three flexible
lines from the controller are connected to the mating stainless
tubing on the probe by an AN connector. These connections must be
tight as any leakage will cause a system error.
3-28
-------
Gain Changer
li
Gain Adjust
Figure 3.18 Fluidic Controller
3-29
-------
Supply Regulator
Figure 3.19 Control Console
3-30
-------
CO
Vacuum Connection
Figure 3.20 Pump and Flow Meter
-------
The sampling nozzle and sensor assembly is inserted in
the outer end of the probe. Two sensing tubes on the nozzle are
connected to the mating lines on the probe through AN connectors.
These two connections must also be tight to prevent any leakage.
Assuming the approximate velocity head of the gas to be
sampled is known, the next step is to set the knobs on the
automatic controller. Table 1 tabulates the knob setting for the
particular sampling nozzle used and the velocity head. If the
velocity head is a complete unknown, set the knobs to correspond
to the particular sampling nozzle inserted in the probe. For the
4 inch nozzle turn the zero on the left hand knob towards the top
of the sampling case. The left hand knob should be set to a mid-
range position so that the numeral four is on the index mark.
These adjustments are made by turning the screw in the center of
the knob CCW to loosen the knob. The knob is then set to the
required position and locked by tightening the screw.
The umbilical cord, with the exception of the two
pneumatic lines, can now be connected to the sampling case. The
pneumatic lines are left disconnected so they can be purged to
remove any debris that they may have collected in transport or
storage.
The other end of the umbilical cord is connected to the
control console. All connections with the exception of the
vacuum line are made to a rear panel of the console. The vacuum
line connection is made directly to the pump mounted flow meter.
Supply air at a pressure of 35 to 100 psig is brought into
the console through a quick disconnect fitting on the rear panel.
After supplying air to the console the supply regulator on the
front of the console is set to give a 5 psi reading on the gauge.
The second regulator knob (lower knob) is turn CW as far as it
goes. This procedure purges the pneumatic lines to the sampling
case. After a few seconds both regulators are turned all the way
CCV,'. The pneumatic lines are now connected to the sampling case.
The supply line is connected to the supply quick disconnect and
the second line to one of two disconnects by which the bias is
applied to the controller.
The supply regulator on the control console is now set
for a gauge reading of 25 psig which is the system operating
pressure. The second regulator is set so that the needle on the
gauge on the automatic controller is in a range from just leaving
the stop to two divisions above the stop. If this setting cannot
be achieved, the pneumatic line must be switched to the second
connection on the sampling case.
The probe is now ready for insertion into the sampling
port. Supply air must be maintained to the controller continuously
during a sampling run to insure backflushing of the sensor when it
is exposed to a contaminated environment.
3-32
-------
Table 1
Sampling
Nozzle
3/8" Diam.
J" Diam.
Controller Gain Setting
Pitot
Reading
Inches H^O
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
4.5
5.0
Variable
Gain
(Rt.Hd. Knob)
1
1
1
2
2
2
2
3
3
3
3
3
4
4
5
5
6
6
7
7
Fixed
Gain
(Lf.Hd. Knob)
Zero
Right
\
s
Zero
Vertical
V
/
3-33
-------
If there was no prior knowledge of the gas velocity head,
tlie controller may not be set for the proper system gain. Too
high a gain will cause the system to oscillate at a frequency of
about 2 Hz. This can be observed on the pressure gauge on the
automatic controller. If an oscillation is observed, the gain is
lowered by turning the left hand knob on the controller to a
higher number setting until the oscillation stops.
The optimum control range is 0.5 to 1.5 scfm. If the
flow as indicated by the mass flow meter is out of this range, the
sampling nozzle should be changed to a size which will satisfy
this range.
3-34
-------
4. TEST PROGRAM
The test program was directed at an evaluation of the overall
controller. Test results obtained on specific components or
sensors are presented in Appendix I and the sections discussing
the particular component.
Tests were performed in the laboratory, with wind tunnels and
signal generators, to establish steady-state error derivatives and
dynamic response capabilities of the controller. At the completion
of the laboratory tests the controller was tested at a power plant
installation to evaluate the controller's capability of functioning
under representative field conditions. One field test was
performed early in the program, after selection of the final sensor
configuration. This test, performed at an oil-fired power plant,
was primarily designed to test the ability of the sensor to
function in a contaminated environment. The complete engineering
prototype was tested in the second test. This test was made at a
coal-fired power plant.
4.1 Dynamic Laboratory Tests
The dynamic tests are designed to measure the controller's
response to changes in air stream velocity. This response is a
function of the system gain which in turn is dependent on the
nominal dynamic pressure produced by the gas velocity and on the
sampling nozzle area. The controller has a presettable gain to
accommodate these variables. When properly set the overall system
gain and thus response will be constant. The prime objective of
the dynamic tests is to relate the variable resistor setting on
the controller to the dynamic pressure head of the sampled gas.
This is the basis for Table 1 under setup and operating procedure.
These settings are referenced to the dynamic head read on the S-
type pitot tube on the sampling probe. The test setup is
illustrated in Figure 4-1. A means of modulating the free air
stream with a rapid and controllable characteristic is required
for the dynamic tests. The frequency range of interest (0-2 Hz)
is beyond the capabilities of available wind tunnels hence special
fixturing was required. The modulated air source is generated by
a relatively small plenum fed from a constant flow source (choked
upstream orifice). The plenum feeds two nozzles — one nozzle
approximately two diameters larger than the sampling probe produces
»a free jet which is directed at the probe. The second nozzle
functions as a variable bleed from the plenum, where the discharge
coefficient is varied in a sinusoidal fashion by a rotating cam at
the nozzle exit. Step changes in air flow are produced with the
same fixture by replacing the rotating cam with a spring loaded
shutter.
Dynamic closed loop performance is determined by measuring
the free jet velocity and comparing to probe inlet velocity as
measured by a pitot tube at the probe inlet. These two velocity
measurements yield the closed loop output/input ratio of the
controller.
4-1
-------
I
to
Ps
CHOKED ,
ORIFICE
PLENUM
JET NOZZLE
PITOT TUBES
P/E
I
FREE VELOCITY INLET VELOCITY
SPEED
CONTROL
CAM
MOTOR
'SAMPLE CASE
— TO SERVO ANALYZER
ERROR SIGNAL
VALVE AND
FLUIDICS
TO PUMP
Figure 4-1. Dynamic Test Setup
-------
The average velocity for these tests was measured by both
flow measurements and by the pitot static tube at the inlet of the
sampling nozzle. These were then correlated to the S-type pitot
tube by steady-state tests in a wind tunnel.
Figures 4-2 through 4-4 show typical performance curves
of the controller. In Figure 4-2 the velocity was held constant
and the system gain varied by changing the variable resistors. A
gain reduction of 2:1 from nominal results in a decrease in
bandwidth to approximately 70% of the nominal with degraded
damping. An increase of gain by a factor of 2:1 results in
approximately a 2:1 increase in bandwidth and the control loop is
on the verge of instability. The measured relationship between
bandwidth, damping and gain tend to follow the predicted
characteristics. The signal shaping networks were selected to
give maximum phase margin at a bandwidth of 2.5 Hz and system
damping will degrade as the bandwidth deviates from the nominal
in either a high or low direction.
4.2 Steady-State Tests
The prime objectives of the steady-state tests were to
obtain a quantitative measure of steady-state error derivatives
in sampling rate as a function of nominal sampling velocity,
pneumatic power supply, filter loading and component drift. The
controller gain settings established in the dynamic tests were
used in all steady-state testing.
The first series of tests were designed to identify those
errors associated with the fluidic controller. These tests were
conducted with the sampling nozzle in the wind tunnel and the
controller gain set to correspond to the nominal air velocity.
The output of the velocity sensor was monitored independently with
an inclined manometer. The automatic controller was turned on
with the bias adjust set to zero and the sampling rate recorded.
The wind tunnel was then shut down and the bias control adjusted
so that the needle on the controller pressure gauge moved off the
stop. The wind tunnel was then turned on and the sampling rate
recorded. This procedure was repeated with the bias adjusted so
that the pressure gauge indicated one mark and two marks from the
stop.
A base reference flow rate was established by removing
the automatic controller and manually controlling the flow valve
to null the velocity sensor. The fluidic controller error is the
deviation from the flow rate obtained with the sensor nulled.
Results of these tests are summarized in Figures 4-5 and
4-6. The curves are normalized to the flow rate obtained with
the sensor nulled. Referring to Figure 4-5, it is apparent that
the controller can contribute errors as large as 15% at a gas
velocity of 20 ft/sec if not compensated with the bias adjust.
The uncompensated error tends to decrease with increasing gas
velocities, indicating that the major source of the inherent bias
in the controller is in the preamplifier stages or the gain
4-3
-------
GENERAL EUECTPIC COMPANY
SCHENECTADY. N » U.S A
.» i'Vi l 170 Imm Mi »>.>..iv
:...., j •
Figure 4-2. Dynamic Response at Various Gain Settings
-------
(.8-30)
GENERAL ELECTRIC COMPANY.
SCHENEC >OY. N. Y. U.S.A.
.» eye. * 170 I mm Divisions
Figure 4-3. Dynamic Response - 20 Ft/Sec
-------
KN-265-A 18-50)
GENERAL ELECTRIC COMPANY.
£CHENECTAD». N. »., U.S.A.
3 eye. i 170 1mm Divisions ,
Figure 4-4. Dynamic Response-^0 Ft/Sec
-------
I N-I5h (S Sill
180x250 1mm Divisions
Figure 4-5. Steady-State Controller Error-3/8 Inch Nozzle
-------
180x250 lin.n Division
ii:r, ]
.U: 1±=
j' M' I !
-t-i dt—1- I -tr" -+ -
:fe;ftt5!,;,
tz m
•
Figure 4-6. Steady-State Controller Error-J Inch Nozzle
-------
changing resistors. If the inherent bias were introduced by a
component following the gain changing resistors, the error would
be independent of the nominal gas velocity. The error contributed
by the controller will be less than 5% if the bias control is
properly adjusted.
The influence of pneumatic supply pressure on the
controller accuracy is summarized in Figures 4-7 and 4-8. The
controller bias was adjusted with a nominal supply pressure of
25 psig. Deviations in flow were recorded as the supply was
varied from the nominal. Flows are normalized to the flow at the
nominal supply. Figure 4-7 is a plot of two specific tests at 20
and 40 ft/sec while Figure 4-8 is a summary of all the tests. The
controller has more than adequate insensitivity to supply pressure.
The primary concern would be undetected supply pressure changes
during a sampling run. These changes would be minimal in that the
control console supplies the automatic controller through a
regulated supply.
Sampling rate error as a function of filter loading was
measured by placing calibrated orifices in the outlet of the
filter holder. Results of several representative tests are shown
in Figure 4-9. The effects of filter loading are minimal
providing the system is not near the mass flow limit at the start
of a sampling run. This is illustrated by the nominal 1.7 scfm
test. This is almost at the 2.0 scfm limit imposed by a clean
filter and the sampling case impingers, and additional pressure
drops cannot be readily accommodated by the controller. A range
of pressure drops representative of field conditions is 4 inches
of H20 to 20 inches of H20. This range encompasses effects of
filter loading and sampling flow rates of 0.5 to 1.5 scfm.
Results of calibration tests of the automatic controller
prior to the final field test are shown in Figures 4-10 and 4-11.
In each instance the automatic controller was set up with the pre-
determined gain setting and the bias adjust varied to bring the
pressure gauge needle to the second mark above the stop.
As indicated in Figure 4-10, the controller is maintaining
isokinetic conditions to better than .10% over a velocity range of
15 to 40 ft/sec and that the major source of error is in the sensor.
The sensor characteristic which was obtained independently by nulling
with an inclined manometer is shown as the dashed curve. At the
maximum velocity of 54 ft/sec the controller is contributing significant
error because of system flow limiting (1.8 scfm flow rate). The upper
curve defines the maximum flow rate control range that can be estab-
lished with the bias control. Within this range the sensor error can
be compensated for by calculating isokinetic flow from measured gas
temperature and the S-type pitot pressure differential. The bias
control can then be adjusted until the system flow meter indicates
the correct flow.
The \ inch diameter sampling nozzle has a constant error with
velocity as shown in Figure 4-11. The error is approximately 15% and
of a polarity to cause the sampled velocity to be lower than the free
air stream velocity.
4-9
-------
I'N-lSh <.s - •<
• nin • mn>
1
1
1
.' -1;:
.-.J-...
fcpitJ .:}:
-•f !••{•
"r :• -- f
~J I
-, -I - r
Figure 4-7. Controller Error VsVSupply Pressure
-------
I \-156 (3 =(i
180x250 Inii- Division-
' -hi- •-t-i-4r
Figure lf-8. Controller Error Vs. Velocity for +20% Supply Variation
-------
18i %.'.->(> Iniiu divisions
4-^ , (
!--.- 4-14 J
-M -H— I F f- t
f-
t
-f—- < j-:
4"L
_;j_.
,44^
tzt:, xxLl
J^:Lr:!-^
—t-•— -t- • t •-I '•
-; Inches of H00
! --I--:
. I
r r-
U „
-- 4-
Figure 4-9. Sampling Error Vs. Filter Pressure Drop-3/8 Inch Nozzle
-------
ELEC7H ' COM PAN Y. S7 I, .= >
180x250 1mm Divisions
—f— ---}-- +-4--.-J. -T-:
Q;
--.- —-f-^rf—f~H^
• 1 • I
~T~ ' T— tr.T-1 -'/!. —T—:—
el f"if* "1 TXT 1 « ' *f ^" y fi'AO
4.V^jAvjr:j^ | J-'|b/ f?
-------
GEN KAL E.i
I8nx$>0 Imm Divisions
I -jL ...
ffip"
:'ti.L
- ., .
::"r
^.:..
•i; -
ii i
<
r
---•— -4 L»-- -
i '
m
-
•\
M
1
5 j^a
: i !
i — ». . .
i
L ;:
i
-:1
..!.-
rrrrr .
I-44--M
... . .1 '. ..
. . , ! . — --
i
:~
:; , -
•
*ii
: i__
- i - !
i '• i
Tr~
.. i...
•l-|-1-
I
i ' i
fin
-_-I(!g,s jv.ftl(
.
:::.
• • :
Figure 4-11. St(
~r-
•-•.I
sady
'.I—
....
j
-State
*
,
Lt:
•
*
' .
r
_ -p +• /Gfig~*
± _\/_ / OV W
;
._r
~r . r
•;."•
W
i —
t
Calibration-:
—
_
• • •
r
*
, ,
/i.:
. \ ::
Inch
30-
h T
.-._,. —
-
>-- --- .-
:' 'l;: '
I-: (: L i
• • '
; [
, .»
•t
1
"t
Nozzle
-------
4.3 Field Tests
Two field tests were conducted on the equipment. A
preliminary field test was directed at evaluating the basic sensor
concept. Isokinetic flow was maintained manually for this test.
Several required design changes in the sensor were identified in
the preliminary test.
These changes were subsequently incorporated in the
prototype sensor which was evaluated in the second field test.
Sampling rates were controlled by the automatic controller during
the second test.
4.3.1 Preliminary Field Test
Test Location - Number three stack of the General
Electric Power Station at Schenectady, New York. This plant is an
oil-fired installation. Tests were conducted on January 19, 1973.
Test Procedure - The test procedure and sampling
train configuration of Method 5 (Determination of Particulate
Emissions from Stationary Sources) as specified in EPA Standards
of Performance for New Stationary Sources in the Federal Register
dated December 23, 1971, was used in conducting the tests. The
tests were conducted by Mr. David Wilson of the General Electric
Technical Services Laboratory.
The test nozzles were standard nozzles modified to
incorporate free air stream and nozzle inlet static pressure ports.
After insertion of the sampling nozzle into the stack, the sensor
was maintained at null by manually controlling the vacuum pump
suction. The null was monitored with an inclined water manometer
with a reading accuracy of better than 0.01 inches of water.
With the sensor nulled the pressure differential
across the calibrated orifice was recorded and compared to the
calculated pressure differential corresponding to isokinetic nozzle
flow. The calculation was performed with a standard nomograph
and dry and wet molecular weights as determined by a prior test
on the stack.
Two sample nozzle sizes were tested. A 3/8 inch
diameter nozzle was considered the primary test nozzle—isokinetic
flow on this nozzle corresponds to about 0.85 scfm, which is an
optimum range for the control unit instrumentation. The second
sensor had a \ inch diameter nozzle with an isokinetic flow of
approximately 0.33 scfm. This is a low range for accurate
quantitative measurements. This sensor nozzle was operated with
the nozzle axis misaligned 30 degrees relative to the free air
stream velocity. This orientation yields a component of particulate
velocity normal to the surface of the sensing tube and the inlet
of the sensing ports and represents a "worse case" operating mode.
The output of a fluid amplifier used to establish
a backflushing flow into the sensor and to amplify the sensor
4-15
-------
differential pressure was monitored throughout the test. A
positive pressure of 0.5 inches of water relative to stack static
(averaged approximately 0.5 inches of water) was maintained by
this amplifier.
Test Results - Figure 4-12 is a plot of the ratio
of nozzle inlet velocity with the sensor nulled to the computed
inlet velocity on the 3/8 inch diameter sampling nozzle. The
nominal value of 0.9 agrees quite well with the results of a pre-
test calibration run made in a laboratory environment. On the
basis of dynamic head the gas velocities encountered in the stack
correspond to a velocity range of 26 to 34 ft/sec on the calibration
curve. Total variation of the nozzle inlet velocity was compared
to the computed value was 5% over the two-hour run. After 80
minutes of test time the four-inch diameter filter used in the
sample case became so heavily loaded that the pump could no longer
maintain the sensor at null. During the subsequent removal and
replacement of the filter, the sensor was left in the stack and
was exposed to a severe anisokinetic operating condition for 15
minutes. During the remainder of the test the sensor was operated
at null. The volume of dry gas collected during the run was 102
cubic feet, referenced to standard conditions.
Figure 4-13 shows the results obtained on the J
inch nozzle. The change in ratio as a function of test time
remains within 5%; however, because of nozzle misalignment, the
nominal ratio departs considerably from the laboratory calibration.
The volume of dry gas collected during this run was 27 cubic feet
referenced to standard conditions.
The output of the fluid amplifier was monitored
throughout the run. It had an initial offset before the sensor
was inserted in the stack equivalent to 0.02 inches of water on
the input. This offset remained virtually unchanged as the probe
was inserted into the stack and nulled.
After the field test was completed the sensors
were rechecked in an "as is" condition in the calibration setup.
Figures 4-14 and 4-1.5 show the comparison of before and after the
field test. The change in characteristics is considered accept-
able. The maximum change of 6% occurred on the ^ inch nozzle in
the range of 33 ft/sec.
Figures 4-16 and 4-17 are photographs of the two
sensors taken after the stack tests. The 3/8 inch diameter nozzle
is shown in Figure 4-16. The sensors are made of 1/8 inch OD
stainless steel tubing. The free air stream sensor is located
above and slightly to the right of the nozzle inlet. The straight,
upright tube is used to establish a positive backflushing pressure
relative to the stack static pressure. The contamination deposition
pattern is clearly evidenced in the photographs with the major
deposition occurring on the leading surfaces of the tubes. Build-
ups of as much as 0.03 inches occurred on the stack static reference
tube. The surface of the sensing tubes containing the static
pressure ports are parallel to the gas flow and exhibit virtually
4-16
-------
FN-521-A (8-50)
GENERAL ELECTRIC COMPANY. SCHENECTADY N. Y.. U.S.A
70X100 1/10 Inch Divi.ioni
Figure 4-12. 3/8 Inch Diameter Sampling Nozzle
-------
FN-521-A (8-50)
GENERAL ELECTRIC COMPANY. StHENECTADf, N. V.. U.S.A
70Xl(Xf 1/10 Inch Division.
1}
n
m+H-ff
tti
-h
m. i • u_
tt
±r
±i
tad
+
±r±
MrlKK
1
fiT
ffft-H-'H-l ! I I Mil
js;
8^*4^
{3n4\
ffi
-4-4--.-I-_-
I I
-r
---
-Irtr
t-H
._!.
t
t I-
r -)-
-. J_i-._
1
^(^rt
i^K
-1
1:1'
5«
::n±
w
s
*
.±f
00
^
7, r?r
qa
±
-h
» i
--,»/-!
2;
xij
I i I
I
ic
h
20
^
off
-i n-1-H^
-i—\-
.'_!_.;..
®
i-H
N
O
^
H4444=
o3; 5s::
irf+WH---
-dij
4C
-ffl
inu4^
P
r
^
.A
[r
4 H^t
i.C
i in..
55j-
t
nrnrf
J L.
-fr] +;
-i±£
4---
ff
--4- H-
==s
f ._
~
-t f -+
H-p--t
ah±
ITT
«JN
41
;-f^
t±
j • n
P i .. -
rm:
Figure 4-13 f Inch Diameter Sampling Nozzle
-------
FN-521-A (8-50)
GENERAL ELECTRIC COMPANY. SCHENECTADV. N. Y.. U.S.A
70X100 1/10 Inch Division.
, f — I I | I | I
- 1 . . _]_'. J. !.
Figure 4-14. Calibration Runs-3/8 Inch Diameter Nozzle
-------
FN-S21-A (8-5
GENERAL ELECTRIC COMPANY. SCHENECTADY. N. Y.. U.S.A.
70X100 1/10 Inch DivUioiu
Figure 4-15.Calibration Kuns-^ inch Diameter Nozzle
-------
Contamination
V
Contamination
Figure 4. 16 3/8 Inch Nozzle
4-21
-------
Contamination
X
.
Figure 4. 17 1/4 Inch Nozzle
4-22
-------
no buildup of contaminates. Although not evident in the photograph,
some buildup did occur on the blunt nose of the nozzle inlet sensor.
Examination of the sensor under a microscope showed no contamination
in or near the free air stream sensor ports. A small amount of
contamination occurred on that portion of the nozzle inlet sensor
tube inside the nozzle. The contamination was deposited in small
clumps, rather than the homogeneous layer which occurred on external
tube surfaces and is believed to be caused by the 15-minute period
required to change the sample case filter.
Figure 4-17 shows the \ inch diameter nozzle. Its
configuration is the same as the 3/8 inch nozzle with the exception
that the sensors are fabricated out of 0.085 inch OD stainless
tubing, as compared to the 1/8 inch tubing used on the larger
nozzle. The band of contamination on the 1/8 inch stack static
pressure tube is indicative of the degree of misalignment between
the free air stream and the nozzle axis; the free air stream
velocity being almost normal to the plane of the photograph. The
free air stream se,nsors remained clean even though there is a
component of velocity normal to the tube surface and the sensing
port. The surface of the inlet sensor tube upstream of the nozzle
inlet and on the side exposed to a normal component of velocity
did have a small amount of contamination and contamination buildup
was starting on the edges of the sensing port.
4.3.2 Field Test on Engineering Prototype
Test Location - This test was conducted on the
number two boiler of the General Electric Power Station at Erie,
Pennsylvania. This is a coal-fired installation. The probe was
inserted in the horizontal ducting between the boiler and the
precipitator. The tests were conducted on January 9, 1974.
/
Test Procedure - A sampling train configuration, as
specified under Method 5 (Determination of Particulate Emissions
from Station Sources) as specified in EPA Standards of Performance
for New Stationary Sources in Federal Register #247, Pt. II, was
used in these tests. The full complement of cyclone and porous
filters was used in the sampling case.
The initial gain settings and sampling nozzle size
were based on logged data from previous tests made at this location.
The controller gain was set to correspond to an S-type pitot
reading of 0.85 inches of water. Available data on gas temperature
and velocity indicated that the optimum flow control range of
between 0.5 and 1.5 scfm would be obtained with the \ inch sampling
nozzle.
The controller bias adjustment was made and with
the controller on, the sampling probe was inserted in the duct and
the test started. Isokinetic sampling flow rate was calculated
from the observed pitot readings and gas temperature and compared
to the flow rate established by the controller. The sampling
nozzle flow rate was 15% lower than the calculated isokinetic flow
but corresponded almost exactly to the room temperature calibration
tests presented in Figure 4-11. The controller was operating stably
and no further adjustment of the gain setting resistors was required.
4-23
-------
These initial settings were retained for the first
test. On subsequent tests the controller bias was adjusted to
establish isokinetic sampling flow.
On all the tests the following data was recorded:
• Stack Temperature
• Stack Velocity
• Sampling Flow Rate
• Control Valve Pressure Differential
Test Results - The test probe was operated in the
duct for a total elapsed time of 3 hours and 15 minutes. Operation
was continuous with the exception of three short interruptions to
change filter membranes.
Figure 4-18 is a plot of the velocity ratio main-
tained during the test. The first portion of the test was run
with no adjustment of the controller bias other than the initial
adjustment prior to inserting the probe in the duct. The
controller maintained sampling flow rate to within a band of better
than +5% of the value determined by laboratory calibration.
At the end of 1 hour and 8 minutes of test time,
the controller was shut down and the filter replaced. After
replacing the filter the controller bias was adjusted to make the
sampling rate equal to the calculated flow for isokinetic sampling.
The bias control was left at this setting for the duration of the
test.
The sampling case filter was replaced once during
the test. As apparent from the plot of velocity ratio, there is
a trend towards a decrease of sampling nozzle gas velocity with
time. This is indicative of increased pressure drop in the
sampling case due to the filter becoming loaded. In the latter
portion of the test the controller was operated continuously for
l£ hours without replacing the filter. Isokinetic sampling was
maintained to an accuracy of better than 6%.
Figure 4-19 shows the flow control valve input
pressure as recorded during the test. This parameter is
significant in that it is a direct indication of filter loading
and can be used as a criteria for judging when the sampling case
filter should be changed. The maximum change, between filter
replacements, is 0.30 psi. Referring to Figure 3-9, it is apparent
that this is less than 20% of the linear control range of the valve
and indicates that the control valve has adequate range capability.
The total gas sample collected during the test was
118 standard cubic feet. Of this total, 80 cubic feet were
collected while maintaining isokinetic velocities to an accuracy
of 5% or better. The remainder was collected with sampling
velocity approximately 15% lower than the stock velocity. The
weights of solid particulates collected during the test were
9.45 grams by the cyclone filter and 1.78 grams by the porous
4-24
-------
IS1-\ 250 1 mm Di\ isimis
ife-l: 4 i . j.Ciabg BL )£i{iiH- -t-t -4- T- -
4 -I --•:-.-
•- ; . t-
Figure 4-18. Velocity Ratio Vs. Time on Coal-Fired Installation
-------
mini Division*
-i't: * r- • I-
^— a~z^c~* •—-* j- —-——
u4-.J -i->4 4'--
_U_U
! '
.u+a:^ ] j
.*.. i 7 .
J1.L i. 1 . -4..
Figure 4-19. Fluidic Controller Output Pressure Vs. Time
-------
filter. Considering only the participates collected while main-
taining accurate isokinetic sampling, the particulate concentration
is 0.094 grams per standard cubic foot.
After completion of the test the controller was
returned to the laboratory and disassembled. All of the fluidic
amplifier modules were removed and each lamination given a
microscopic examination for evidence of contamination. No
evidence of contamination was found, indicating that backflushing
and referencing the input amplifier to the duct static pressure
effectively prevents particulates from entering the sensor ports
and signal lines.
Visual inspection of the sampling nozzle and the
sensor showed virtually no adherence of particulates on any part
of the sensor or nozzle. This is in contrast to experience on the
oil-fired installation where particulates adhered and caused a
significant buildup of contaminates on all leading edges of the
nozzle and sensor tubes.
4-27
-------
APPENDIX I
ISOKINETIC SENSOR TESTS
-------
APPENDIX I
Sensor Test Results
The major development effort on the program was devoted to the
test and evaluation of the gas velocity sensors. The sensor is a
critical component in that it must function in a heavily contami-
nated environment, at elevated gas temperatures, and over a broad
range of stack velocities. Three basic sensor configurations were
fabricated, tested and evaluated. These configurations were the
cross-flow sensor, the co-flow sensor, and the differential static
probe.
The sensor characteristics of primary concern in this application
are:
• Ability to function with air flow exiting from all signal
ports. This is considered a prime prerequisite in avoiding
malfunction from contamination.
• Isokinetic velocities must be retained at the free air
sensor. Flow disturbances introduced by the sensor result
in an error in measuring free air stream velocity and in
corresponding error at the inlet to the sampling nozzle.
• The signal-to-noise ratio of the sensor must be compatible
with the design goals on system bandwidth.
• A sensor gain independent of velocity is highly desirable
though not absolutely necessary.
• A high sensor scale factor is highly desirable.
e The diluent flow introduced into the sampling nozzle should
be a minimum. This flow must be accounted for iri the
totalized flow on a sampling run.
• Sensor configuration must be compatible with insertion and
withdrawal through existing' sampling ports of three inches
in diameter.
Differential Static Sensor
This sensor went through three design evolutions as illustrated
in Figure A-l. The original concept placed the free air stream
sensor several inches in front of the sampling nozzle inlet. This
configuration is ideal from the standpoint of flow disturbance at
the free air stream sensor. This configuration was abandoned
because of the physical constraints imposed by existing sampling
ports. By displacing the free air stream sensor axially as shown
in A-lb the overall length is reduced to under three inches and
the sensor can be inserted into existing sampling ports. This
configuration was field tested, and operating experience gained on
that test indicated that protective shrouding was necessary,
leading to the final configuration shown in A-lc.
A-l
-------
Nozzle Inlet
////// /////////
F *--•••* —
Free Air Stream
a
///7~7T7
Shroud
\ \ \ \ \ \ \ \\V\\ \ ^y \ \
: C Q"
/"~7~7~7~7~7
I
// '
r\ \.\ \
\ \ \ \ \ \ \ \ \ \ \
Figure A-l. Differential Static Sensor Configurations
A-2
-------
Figure A-2 shows the performance characteristics of the three
configurations. These tests were performed in a wind tunnel. The
sensor was nulled and the sampling nozzle flow measured. The
sampling nozzle inlet velocity was computed from the measured flow
area and the effective area of the sampling nozzle. As apparent
from Figure A-2 the protective shroud introduces losses which
cause the sampling nozzle inlet velocity to be smaller than the
free air stream velocity.
This effect becomes more pronounced at the higher velocities as
evidenced by Figures A-2 and A-3.
The sensor gain was determined by varying the nozzle inlet velocity
relative to the free air stream velocity and noting the differential
pressure across the sensor.
Figure A-4 is a plot of the sensor gradient. The x axis represents
probe velocity normalized to isokinetic velocity, while the y axis
is the measured probe static differential pressure referenced to
kinetic head. The measured sensitivity is approximately 707f of
the theoretical sensitivity shown by the dashed line on Figure A-4.
This discrepancy, attributed to the inadequate spacing between
static probes, does not affect the probe's ability to sense
isokinetic conditions.
The sensor noise characteristics are shown in Figure A-5. The
peak-to-peak noise output is approximately six times the RMS values
shown on Figure A-5. In the lower velocity range a quantitative
noise measurement was not obtained because instrumentation back-
ground noise exceeded the sensor noise.
Co-Flow Sensor
The pertinent performance characteristics of the co-flow sensor
are summarized in Figures A-6 through A-13. Sensor gain vs. supply
pressure is shown in Figure A-6. The lower gain curve was obtained
when the receiver was backpressured to a value which insured out-
flow at the maximum air stream velocity of 150 ft/sec. Back-
pressuring reduces the sensor gain to approximately 60% of a sensor
operating with no back pressure. For both cases, sensor gain
varies approxiamtely as the 2/3 power of supply pressure.
Figure A-7 gives the equivalent sensor noise in ft/sec. Observed
peak-to-peak values of noise are five times the RMS values. Sensor
noise is essentially independent of supply pressure.
The ratio of sensor gain to the DC pressure level at nulled
condition is shown in Figure A-8. This curve is significant when
considering the sensor's susceptibility to drift from variations
in back pressure and amplifier input impedances. A high ratio is
desired; hence, the co-flow sensor should be operated with the
minimum acceptable nozzle supply pressure. A marked degradation
in sensor gain and linearity occurs at supplies lower than 3 psig.
A lower limit of 5 psig was selected to give a reasonable operating
margin.
A-3
-------
GENERAL l-..f.i. .: J COXF* -.CHEK£CTAD-| . V U.S.A
BF-^-r —T~
I -i-
t!ti-,
!
r**rirt^
_ |--f4~. !
-J-•• " it'
•^.:H•..,!...4-.. -f
Cfr
-r—t-t-•
P ^- t~f :i^—^ H
.-
- - — -f'-r4—,---^^--^M^'--^^-^- -— - • ^-rh-r-T ^
^4i-k-- J^.rfe'jy :r^^- i^fe^p-^ ' ii^Slydii
r T 7™ ' I' ' .~T^ """. I"1 , ^^n ^~* -•"•>' ' ! >'•*'- *T* H'^-i-f •?"*•<—
Figure A-2. Differential Static Sensor Characteristics-3/8 inch Nozzle
-------
KN-155 i«
GENERAL ELEC'KIC COMP,V CHENEC-^DY. N. •{.. U.S..".
i)'\!3i'j|i.-
. „ L:TPT: , .t -.-.
--—T i r i i Hi;
11 .. ';. 1.
^ ! i-^'f
_..'. 4- --:t_—i~ -iT-
, -;-r . j,.
.:.) FH-nvt,.^^:
rrn rn ^T ... .^»
Figure A-3. Differential Static Sensor Characteristics^ Inch Nozzle
-------
K.V-S21-A (S-SO)
CiNERAL ELECTRIC COMPANY. SCHENECTADY. N. Y.. U.S.A.
70X1OO I/SO lr.0 L'ivi»ic.<\»
t
1
OS
1 1
1
L.
-
-
-
:
-
P
p
:
—
-
"
-
r
-
•—
-
.
:
--
T
"(
<
H
l-l
5
1
N
^.
•i*
-4-
1
J
£
C
4»
e
5
1
l
:
-
7
J
-<
H
J
5
-H
3
1
2
•H
,}
1
!
1
3
4
_:. .
H-
1
i -
T
f
-
-
:
--
, — ,
-
.j
-
j
.-
-
;
•
...
:
c
•"-
--
:
-
•
,-
_;
/
f
1
-
[
-
r
L
<
L
-
J_
n
-;5
L
_
..L
F
r
>
1
_
—
:
.
\
x
:
_
_
I
N
i n
-
II
-
-
v
V.
••
•*-.
-
-
•s
i
-
-
-
I
IT
t
-
r
t
$
-
-
~
-
. J_
__
•
—
-
:
...
--
:
-
:
-
--
-
•-
-
-
;
_
-
—
r
-
-
•1
-
#
-
~
— '
..
I
-
-
-
-
(
-
-
.
rf"'
'
'
-
-
-
-
-
-
i
i'
ILL
I
-
i
f
, ,
-
_,
-
-
--
jj
"
-j-i
-
=
i i
|
4^
~
:
j
-j..
-
•-
-
-
--
-
^_
i
-j-
^
-
:
/*
-\
^
4-
I
-
—
"
TT
3
-
-
-•
_i_
7
_
-
-
L
V
__
-
•
•
\
1
i
-
\-
-
_.
_
_
\
\
-
, .
}
d
-
^
-
-
-
-
/
r
h-U
-
5z
/
-
-
-
•
-
-
-
i
i
i
i-
-
-
_
3.
:
t
-
f
I
-
-
/
i
/ i-
/
--
-
r 1— j—
-
-
:
z
/
>
^.
.
-
-
j
a
/
j.
.
/
i
-(-
!
•
/
p
->
\
11
<
Sf
JC
1
_
i
-
£
.-
r
.
so
ic
3
-
c
\
-
Lc
Lc
I
-
T
^
c
\
-
-
1
»1
3
K
-/
^5
i.'.'.
-
i
o
...
r
_
._
c
-
'
;t
£
-
-
--
I"
_
i
e
-
ft,
-
:
I
•~t
c
-
FJ
-r
CM
-1-
—
-
-
-
-'
-
-
•
-
-
-
Figure A-4. Effect of Non-Isokinetic Probe Flow on Static Differential Sensor Reading
-------
.ChENECTAUY. i 1 U.'j.f..
(. -,
ra-^rr
4
: L __,._ • '!;
""f;r T TT
i
^4- fTr%V-Miife:J-L VJnJttfln'rii.infifi
m^\
m&^
t!$*H*tf
rii-tiliii
•M J!ll(
tit***?
.L_ . j !. ...
OWL
.T.ljresfa
-------
FN-521-B (8-50)
GENERAL ELECTRIC? COMPANY. SCHENECTADY. N. Y.. U.S.A.
100X140 7/100 Inch Division!
Figure A-6. Co-Flow Sensor Gain
-------
FN-521-B (8-50)
GENERAL ELECTRIC COMPANY, SCHENECTAD*. N. Y.. U.S.A.
100X140 77100 Inch Dn>*ioni
Figure A-TT Co-*
-------
GENERAL ELECTRIC COMPANY. SCHENECTAD". N. Y.. U.S.A.
100X140 7/100 Inch Dm.ion.
W
•i--i
w
H
tf
~
TF
;i
TTTT
TTTT
s
Si
til :
Ml -
hi
2^&
tt
I!
i
I M i
I
m
HUJ
I J-4-T
r
Li ! ;
1-
ff
4 i '
4-U
» i
ii
,414
..; .,
i ftf'
: 1-U-
4 • .
i
n
ttfr
I
P]1
^
it I
ri ;
lit
t I
,f
o
o
p
f
l!\
) M
LLLL
ti
- *
i("!
•H44
IH
4 ! J-l
I
M
O
D
-il-
i±*
H-
4-f--
ia
t i »
H
Ui
n
II
11
L-
iJi
•-t-+
st
cm
lii
•i-n
CQ
"3
i-tt;
H
.-
-•f
ii1^
in
Bix
::Hl:
TTT
I .
HE
tri
S
4-rO
I 1. I .
i-l 4
1.11
Art;
•
fi
m
-i-l -i
iTTT
Hi
_
i
f'i
\
lilJ
i * r
,8
i.Lf!
mi
till
r f
141
uHT
U .1 I
Hi
fi
:i 1
ir
Sbpplj
-
If
I L :
sjiMi-i;
Hri
1
-
'
a
la
iti
M
H:
iHT
d±E
Figure A-8. Co-Flow Sensor Gain/Null Bias Ratio
-------
FN-S21-B (8-50)
GENERAL ELECTRIC COMPANV. SCHENECTADY. N. Y.. U.S.A.
100X140 7/109 Inrh U.-.isinrn
Figure A-9. Difference in Two Co-Flow Sensor Indications
-------
FN-S21-B (8-50)
GENERAL ELECTRIC COMPANY. SCHENECTAOY. H. Y.. U.S.A.
100X140 7/100 Inch Divinioiu
isor-$2-L -Sei&or
Figure A-iO. Co-Flow'Sensor Noise Vs. Bandwidth
-------
CENF.R*! ELECTRIC C3'-«» C.rilKf.C : :'.\. N. 1 -.;
r i -ftf
± L L
:fetet^
^3fcT:-;£^r ;
T^^^^Tii ^T
« -U-J.J ..~ ..-i .- '.
- J—,-'i-4-j—(•*+- — .._-f-j-»,|_i_ -_.,.i ;.
mffi* I-: :^£ ±£U.
Figure A-ll. Velocity Acceleration at Nozzle Inlet
-------
.EiERAL CLCC'IIIC IOMPA' -L EL '
..;.-.[:r.±_, .
~ '
_
?q -4._^_.4.-__^
a r^"
U— J~ . -4- .
tfrtrrfr-3±t!±|
, ; : -fc.
! •
^-M—-f-.-~Ly- -r--
-i-—-i-__j.__L. - T 1' 1.
- : '
Figure A-12. Velocity Acceleration at Nozzle Inlet
-------
GENERAL ELECTRIC COMPA' 'CHENEC'AOY. N. V.
. ! J. \ ]Ml 'II-
•irrTjfx-^ZL-H-— |--.lt •-•-!::
— .-.-, .-..----r-.r .-
rr r • !»i|i
-U--4- -44—a,—P—t—i'4—i+i.
, +,.:
..j tiUJ Lf, ; U^*;
. r T L;>-:- •- p- :
.... i - .1. 4 -... i.. ;4_t, . i
,__j j '; i ^
^:':!iJi.
:•^"f.:'.','ir:".- t-J-JJ. -"L"''..;Xur;iL;,_
feMf^^Sg
---•^-r- f - '-r^rT-t-f
^s^p
•;T*r!ii~ c±tti-t;-
^r-feHSor-
Figure A-13. Co-Flow Sensor Output Vs. Velocity
-------
Figure A-9 shows the differential pressure between two sensors
exposed to the same air velocity. The sensors were nulled at
20 ft/sec air velocity by adjusting the receiver back pressures
with variable series resistors. The resistors were then left
fixed and air velocity varied. At the maximum velocity of
160 ft/sec, the sensor error is 6%. The error vs. velocity is
well behaved, indicating one sensor has slightly higher gain.
This type of characteristic can be compensated by adding two more
variable resistors, shunting the sensor receiver.
Figure A-10 shows the noise vs. bandwidth characteristics of two
sensors. The differential noise between the reference sensor and
the sampling probe sensor will be the effective noise applied to
the amplifier following the sensor. The noise has a Gaussian
distribution and absolute magnitude is proportional to the square
root of bandwidth. Considering a 2 Hz closed loop bandwidth on
the automatic sampling control, the RMS noise is approximately
0.7 ft/sec which on the basis of an overall assessment is very
acceptable. The full impact of sensor noise does not become
apparent until a specific mechanization of the control is
considered. For example, to meet a steady-state accuracy of
better than 5% requires a forward loop gain of at least 20 times
the product of gas density and probe area. This requirement also
translates into a gain of 20 from an air velocity error signal to
the controlled output velocity. Noise in the frequency range
between the closed loop bandwidth (2 Hz) and the lowest tolerable
bandwidth on a noise filter (approximately 10 Hz) will then be
amplified by approximately 20. The RMS noise in this band can be
obtained from Figure A-10 and is 1.4 ft/sec with peak-to-peak
values of 7 ft/sec. If no saturation or filtering occurred, peak-
to-peak noise values of 140 ft/sec would occur in the output. The
sampling case is an effective filter in this frequency range;
hence the noise would not show up on the output. However, the
flow control valve would respond and would be commanded to a level
corresponding to total excursions of 140 ft/sec. The velocity
range that can be achieved in the sampling probe is inherently
fixed by the sampling case pressure drops and typically will be a
range from 0 ft/sec (with throttling valve) to a maximum approaching
two times the nominal control velocity. Assuming a nominal of
70 ft/sec, the control valve will be overdriven, by noise, by a
factor of 7:1 — going to a fully closed to fully open position.
An acceptable value would be on the order of 50% of full range,
indicating a required noise reduction of approximately 15:1. A
filter to yield this reduction requires a bandpass of .04 Hz and
would restrict the closed loop bandwidth to approximately 0.2 Hz
as compared to the design objective of at least 2 Hz.
The sensor noise has the characteristic of noise from jet turbulence
and has about the expected magnitude; hence, it is unlikely that
any significant reduction in sensor noise can be achieved.
Figures A-ll and A-12 show the effect of sensor flow on the inlet
velocity to the shroud. As shown by Figure A-ll, the sensor
aspirates and increases the inlet velocity relative to the free
air stream velocity. For any given sensor configuration it is
A-16
-------
apparent that the sensor flow can be selected to compensate for
inlet losses. The inlet velocity is strongly influenced by shroud
length, as shown by Figure A-12. With a length to diameter ratio
of 2:1 the inlet velocity is relatively insensitive to sensor flow,
staying well within a 5% error range for sensor supply pressures
up to 40 psig. The sensor output as a function of free air velocity
is shown in Figure A-13. A linear output with a scale factor of
.004 psi/ft/sec (10 psig sensor supply) was obtained over the test
range of 0-160 ft/sec.
Cross-Flow Sensor
Characteristics of the cross-flow sensor are summarized in Figures
A-14 through A-19. A sketch of the sensor configuration is shown
in the upper right hand corner of Figure A-14. Both upstream and
downstream receivers were tested. The data plotted in Figure A-14
was obtained on the upstream receiver. The upstream receiver has
a usable range of 20 to 140 ft/sec with a gain variation of
approximately 5:1 over the range. The gain is relatively constant
over the range of 80 to 140 ft/sec.
The output from the downstream receiver is shown in Figure A-15.
The optimum range for the sensor is 20 ft/sec to an upper limit
determined by the supply pressure.
The specified velocity range of 20 to 150 ft/sec can be accommodated
by using the upstream receiver for the high velocity range and the
downstream for the low velocities.
The sensor scale factor and range are strongly influenced by the
nozzle supply pressure with the optimum supply pressure being the
maximum that can be utilized in the sensor. In view of the
optimum trend a series of tests were performed in a wind tunnel to
empirically determine the maximum sensor flow vs. diameter and
length of the shroud. The probe inlet velocity was measured and
compared to the free air stream velocity. Three representative
test results are shown in Figures A-16, A-17 and A-18. Figure A-16
shows the results obtained with the sensor enclosed by a 0.375
inch diameter shroud. The inlet losses associated with the shroud
are excessive as indicated by the inlet free stream velocity ratio
with no flow injected into the shroud. Introduction of sensor
flow significantly decreases the sampled velocity, as shown by the
series of curves obtained at different supply pressures. Figures
A-17 and A-18 were obtained with a 0.625 inch diameter shroud
enclosing the sensor. This diameter is about the minimum diameter
that can be used to maintain inlet losses to an acceptable value.
The percent error resulting from sensor flow has been significantly
reduced as compared to the 0.375 inch diameter shroud. The
influence of shroud length is not particularly strong, as shown by
comparison of Figures A-17 and A-18. The empirical relationship
derived from this test series is given by the following:
A-17
-------
1- \-155 i:-1
CE'lEPAL ELEl 'RIC COM°i -CHENECMDY N v .E
:. . -'.•'-4- -j- -;••; 4- - -- *• -r-j—
i 1 :t '.t-
--
:~l
flQM" sepliratlcm
: A
• f •
r.: ; rf;..:
tn±±t:tj**T • • i: -- tttiiirl
-i;H-: ; lE l::'l.:
K*
,. - -
_,. , >r. __ .. _ ___ L , •
.
Figure A-14. Cross-Flow Signal Upstream Receiver
-------
. E'iERAL • :
C.HI -it':
_. --L
I
- •!
i
t_j -••• 4— 4- .. -
1 i i
:
! ! .(. ••; i
; f • ' ;
. ,. ' :
P ':'•'•: ITm
_
:J,1j
i : i .
• -' r-fr+^---H-rH--1- r
, ,-•- K;--rq'
^^ y^i^-'Hiia
-*^-1-"
.Stream V^loclrty - Ft/Siec
Figure A-15. Cross-Flow Signal Downstream Receiver
-------
K.V-155 (8 '
GENERAL ELECTRIC COMPA' 'JCHENECTADY. N. Y.. U.S.A.
Dlvisioiu
to
o
Figure A-16. Velocity Deceleration at Inlet
-------
'•V-15S ',.*
ELECTRIC COMF»' CKF :(. - )Y N ( . U.b.A.
- 1- f. r -J^
''11 ~
-i'*r~: ' 2tr-ps^
...Jli.}.;. i—'.!—ua4^---ri~>1
i-u?.... 4 - -!—i/\nr,':~ '. i in
j. :i.j • tfTp- !., Hr-rr'-J
JTCf-j4-:;h;:-^^i|j
-,-r' f '-rr rrH^'~ r^~
f ..^i-,
-^--1; .-it ':!-:-;.
iJH
j. .
,,.-i j • .{. . -, ;::i :
Figure A-17. Velocity Deceleration at Inlet
-------
i-X-155 (S
GENERAL ELECTRIC COMPA1 "iCHENECTADY. N. Y.. II V '.
"| t':' • i
T-j .-) ^ ... .
r
]" —j 7- -r --?•• t - j — . -j ••-•- h-H
-Li .-i- ; >']!
r-rf:—tf- -f—4r—*
• -1 41 •(..
t . i
...4- i ----- '-.- , -L. — -i—
31 i,_. |
, r -f T '. -
.. -;,r;::=-ehi
-n-j* 'H-rH--' -:(-*-^- •;
Sl!±S
*Mrr.fet*fc*ri4r-
r •
Ei.
Figure A-18. Velocity Deceleration at Inlet
-------
% of error oC w = (1-V/V ) 100
~^7^~
o
where w = sensor mass flow
A = shroud area
V = free stream velocity
Extrapolation of these results to 20 ft/sec and 1000F yields a
minimum shroud diameter on the order of 3.5 inches and a length
of 7 inches to maintain the error to less than 5% with a 40 psig
supply. In addition to the prohibitive size, the advantage of
being able to use identical geometry for the free air stream and
sample probe sensor is lost.
The sensor noise characteristics are plotted in Figure A-19.
A-23
-------
. \-155
. > '.EHAL f ECrr COYPV ICHtWtCTAJ
!(-u\Ji" ;»»iC:.|
^.. -r ""• f ' .•
J.. -£.--.4 • '---Tj- ;,il
-r 7-H
-•--• 4- --
"f~ •<-.- "f h~ :!—~
-:•!.! : U-a
,-1^...-- . -1- ^- — .
M rt"', '.-f-•-.•]-}• —^
tS
:f •-ttT"-
r '- i" -'r/^- :
width Hz
-.,._.. 4 } ~. -4 -j . ...
Figure A-19. Cross-Flow Sensor Noise Vs. Bandwidth
-------
APPENDIX II
FLUIDIC CONTROLLER PARTS IDENTIFICATION
-------
APPENDIX II
Figures A-20 and A-21 show the location of the component parts
making up the fluidic controller and the input-output connections,
Figure A-22 shows the location on inlet and outlet connectors to
the console.
The component parts, with the function and General Electric model
number, are listed in Table #2. The input-output connectors are
identified in Table #3.
A-25
-------
Figure A-20 Fluidic Control^: - Front Face
Page A-26
-------
Figure A-21 Fluidic Controlla: - Back Face
Page A-27
-------
Figure A-22 Control Console Connectors
Page A-28
-------
Table 2
COMPONENT PARTS
Front Face of Controller
Location Description
a Input stage
b Gain adjust
c Preamplifiers
d Preamplifier supply and filter
e Lead-Lag feedback resistors
f Lead-Lag gain block
g Output amplifier and filter
h 1st stage supply resistor
i Gain changer
Part Number
CR280QA1045
CR280RV32
CR280QA1046
CR280QA1047
CR280RF32
CR280AM12B
CR280QA1048
CR280QA1049
CR280QA1050
Back Face of Controller
c By-pass resistor
e Lead-Lag capacitors
f Lead-Lag input resistors and bias resistors
h Lead-Lag supply and filter
CR280RF32
CR280CF32
CR280QA1051
CR280QA1052
A-29
-------
Table 3
INLET-OUTLET CONNECTIONS
Back Face of Controller
Location Description
a Input from sensor
b Stack static reference input
d Bias adjust inputs
f To stabilizing volumes
g Output to valve
i Air supply inlet
Control Console
a Flow meter
b Flow meter output jack
c Pump power switch
d Main power switch
e Pump outlet
f Umbilical cord connectors
g Inclined manometer
h Pneumatic supply pressure
A-30
-------
REFERENCES
1. Whiteley & Reed, "The effect of Probe Shape on the Accuracy
of Sampling Flue Gases for Dust Content".
2. V. Vitols, "Theoretical Limits of Errors Due to Anisokinetic
Sampling of Particulate Matter", Journal of Air Pollution
Control Association, Feb. 1966, Vol. 16.
3. Carbonar, Colin & Olivari, "The Deflection of a Jet By a
Crossflowing Stream and Its Application to Aneomometry",
Paper R2, Fourth Cranfield Fluidics Conference, Mar. 1970.
4. Hayes, Tanney & Templin, "The Co-flowing Jet Velocity
Sensor", Paper Jl, Fifth Cranfield Fluidics Conference,
June 1972.
5. Smith, Martin, Durst, Hyland, Logan & Hager, "Gas Sampling
Improved and Simplified with New Equipment", APC Paper
#67-119.
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
REPORT NO.
EPA-650/2-74-029
3. RECIPIENT'S ACCESSION-NO.
4. TITLE AND SUBTITLE
Compact Sampling System for Collection of
Participates from Stationary Sources
B. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
Carl G. Ringwall
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
General Electric Company
P.O. Box 431, Bldg. 37
Schenectady, New York 12301
10. PROGRAM ELEMENT NO.
1AA010
11. CONTRACT/GRANT NO.
68-02-0546
12. SPONSORING AGENCY NAME AND ADDRESS
EPA, Chemistry & Physics Laboratory
National Environmental Research Center
Research Triangle Pk., N.C. 27711
13. TYPE OF REPORT AND PERIOD COVERED
Final Report
14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES
161 ABSTRtn^s report summarizes the work performed on a program to design, fabricate,
and evaluate a controller for automatically sensing and maintaining isokinetic
conditions at the inlet of a particulate sampling nozzle.
The key components developed on the program were the gas velocity sensor and
a fluidic control amplifier. The sensor concept is based on a static pressure
differential between the free air stream and the nozzle inlet. The fluidic
control amplifier which interfaces directly with the sensor provides the control to
automatically maintain isokinetic conditions.
Field tests were performed on the engineering prototype system at both oil-
fired and coal-fired power plant installations. Results of these tests showed
that the sensor and controller can function with no degradation in performance
under the adverse environment of representative power plant stacks. Temperatures
up to 205°C and solid particulate concentrations of 3.50 grams per cubic meter
were encountered during the field testing.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS C. COSATI Field/Group
Isokinetic Sampling
Automatic Isokinetic Sampling.
EPA Train
Sampling
Stationary Source SAmpling
Pollution Monitoring
Particulars
Stack Monitoring
Mass Concentration
18. DISTRIBUTION STATEMENT
Release Unlimited
19. SECURITY CLASS (ThisReport)
Unclassified
21. NO. OF PAGES
106
20. SECURITY CLASS (Thlspage)
Unclassified
22. PRICE
EPA Form 2220-1 (9-73)
-------
INSTRUCTIONS
1. REPORT. NUMBER
Insert the EPA report number as it appears on the cover of the publication.
2. LEAVE BLANK
3. RECIPIENTS ACCESSION NUMBER
Reserved for use by each report recipient.
4. TITLE AND SUBTITLE
Title should indicate clearly and briefly the subject coverage of the report, and be displayed prominently. Set subtitle, if used, in smaller
type or otherwise subordinate it to main title. When a report is prepared in more than one volume, repeat the primary title, add volume
number and include subtitle for the specific title.
5. REPORT DATE
F.ach report shall carry a date indicating at least month and year. Indicate the basis on which it was selected (e.g.. dare of issue, date of
approval, date of preparation, etc.).
6. PERFORMING ORGANIZATION CODE
Leave blank.
7. AUTHOR(S)
Give name(s) in conventional order (John R. Doe, J. Robert Doe, etc.). List author's affiliation if it differs from the performing organi-
zation.
8. PERFORMING ORGANIZATION REPORT NUMBER
Insert if performing organization wishes to assign this number.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Give name, street, city, state, and ZIP code. List no more than two levels of an organizational hlrearchy.
10. PROGRAM ELEMENT NUMBER
Use the program clement number under which the report was prepared. Subordinate numbers may be included in parentheses. *
11. CONTRACT/GRANT NUMBER
Insert contract or grant number under which report was prepared.
12. SPONSORING AGENCY NAME AND ADDRESS
Include ZIP code.
13. TYPE OF REPORT AND PERIOD COVERED
Indicate interim final, etc., and if applicable, dates covered.
14. SPONSORING AGENCY CODE
Leave blank.
IS. SUPPLEMENTARY NOTES
Enter information not included elsewhere but useful, such as: Prepared in cooperation with, Translation of. Presented at conference of,
To be published in, Supersedes, Supplements, etc.
16. ABSTRACT
Include a brief (200 words or less) factual summary of the most significant information contained in the report. If the report contains a
significant bibliography or literature survey, mention it here.
17. KEY WORDS AND DOCUMENT ANALYSIS
(a) DESCRIPTORS • Select from the Thesaurus of Engineering and Scientific Terms the proper authorized terms that identify the major
concept of the research and are sufficiently specific and precise to be used as index entries for cataloging.
(b) IDENTIFIERS AND OPEN-ENDED TERMS - Use identifiers for project names, code names, equipment designators, etc. Use open-
ended terms written in descriptor form for those subjects for which no descriptor exists.
(c) COSATI FIELD GROUP - Field and group assignments are to be taken from the 1965 COSATI Subject Category List. Since the ma-
jority of documents are muttidisciplinaiy in nature, the Primary Field/Group assignment(s) will be specific discipline, area of human
endeavor, or type of physical object. The application(s) will be cross-referenced with secondary Field/Group assignments that will follow
the primary posting(s).
18. DISTRIBUTION STATEMENT
Denote reusability to the public or limitation for reasons other than security for example "Release Unlimited." Cite any availability to
the public, with address and price.
19. & 20. SECURITY CLASSIFICATION
DO NOT submit classified reports to the National Technical Information service.
21. NUMBER OF PAGES
Insert the total number of pages, including this one and unnumbered pages, but exclude distribution list, if any.
22. PRICE
Insert the price set by the National Technical Information Service or the Government Printing Office, if known.
EPA Form 2220-1 (9-73) (Rcvtrse)
------- |