EPA-650/2-74-089
OCTOBER 1974
Environmental Protection Technology Series
::]^j^:^
-------
f
EPA-650/2-74-089
INVESTIGATION OF EXTRACTIVE
SAMPLING INTERFACE PARAMETERS
by
K. J. McNulty, J. F. McCoy,
J. H. Becker. J. R. Ehrenfeld, and R. L. Goldsmith
Walden Research Division of Abcor, Inc.
201 Vassar Street
Cambridge, Massachusetts 02139
Contract No. 68-02-0742
ROAPNo. 26AAP
Task 27
Program Element No. 1AA010
EPA Project Officer: James B . Homolya S
Chemistry and Physics Laboratory
National Environmental Research Center
Research Triangle Park, North Carolina 27711
Prepared for
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D.C. 20460
October 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
-------
FOREWORD
This document is the result of a twelve month, three-phase investigative
program with the intent of providing EPA with sufficient information to permit
the establishment of minimum specifications for the design of continuous
extractive sampling interface systems. An extractive sampling interface system
is the equipment associated with an instrumental source measurement system which
extracts, transports and conditions a sample of the source effluent. The work
in this program was directed toward an investigation of interface systems for
use on Category 1 sources for the following instrumental techniques:
a. non-dispersive infrared analyzers (NDIR)
b. ultra-violet analyzers (NDUV)
c. electrochemical cell analyzers
d. chemiluminescent analyzers
Specifically, this program was charged with performing a survey of source
monitor manufacturers to obtain information on their respective analyzers and
the commercial products required in their sampling systems. This information
would then be used to provide a basis for the specification of all the sampling
system requirements for each analytical technique. This activity was Phase I
of the program. Phase II involved a laboratory study of gas sample losses in
various components of the sample interface system, particularly:
a. possible gas-solid reactions that could occur on particulates
collected on the probe tip filter
b. sample losses or gas-phase reactions within the sample lines
c. gas interactions within various moisture removal systems
Additionally, the feasibility for calibrating the total source measurement
system through injection of calibration gases at the probe inlet was investi-
gated. On completion of Phase II, an Interim Report was submitted which
included the results of laboratory studies and contained sampling system
design information to be used by EPA personnel.
iii
-------
Phase III of the program was charged with demonstrating by field mea-
surement an adequate sampling system for each analytical technique. The
objective of these field tests was to locate potential operational problems,
correct these problems and demonstrate one month of continuous operation.
During continuous operation only normal maintenance was to be permitted,
and calibration checks were to be performed automatically and through the
filter probe, if feasible.
This final report is presented in two parts. This structure best re-
flects the dual (design criteria and field demonstration) aspects of the program.
Part One contains the information presented in the above-mentioned
Interim Report, modified to include results of the Phase III field program.
It is intended to stand alone as a source of design information for inter-
face systems. Presently there is no other single reference source available
on the design of the sampling system with this amount of quantitative engineering
data.
In order to conform with present EPA report requirements for the use of
SI units (Systeme International d'Unite), the data presented in the Interim
Report in the usual engineering units have been converted to SI units for
presentation in Part One. These units will undoubtedly be inconvenient to
— p
the reader; since units such as kilo-Newtons per square meter (kN/m ) will
not impart the same physical sense as the equilivent inches (or even centi-
meters) of water. However with experience (and conversion tables) the SI
units should not be any serious impediment to the designer of interface
systems.
Part Two of this report relates the experience of demonstrating
"minimum" type sampling systems in the field. This part is not independent
of Part One since the demonstration systems implemented were designed
based on the information presented in the Interim Report, (See Part One).
IV
-------
In assessing the contribution of this program to the state-of-the-art,
it is felt that several contributions have been made, particularly in the use
of low-cost sampling lines, a lower-cost probe-tip filter approach, less
severe moisture removal requirements and calibration of the measurement system
through the probe tip filter. It is somewhat disappointing that all elements
of a "minimum" type sampling system have hot been successfully demonstrated.
Further work is required to specify "minimum", yet reliable techniques for
moisture removal and for sample pumping. In this context the wouldbe designer
of a new sampling interface system has through this report a considerable
data base for sampling system design. However, any new system which does not
reproduce an existing successful system will contain experimental elements.
There remains no "algorithum" for designing "minimum" sampling systems.
-------
ACKNOWLEDGEMENTS
In performing a program of this type one is grateful for the assistance
and wisdom of others. Principal acknowledgement in this regard is to
Mr. James B. Homolya of EPA who was the prime mover for initiating this in-
vestigation and served as the Technical Project Officer for the program. In
the collection of information on existing practice we gratefully acknowledge
the helpful discussions, with the representatives of many manufacturers of
sampling instruments and industrial users of source monitoring equipment.
We are particularly grateful to Mr. Robert Saltzman of DuPont who was respon-
sible for loaning to the program a DuPont Model 400 NDUV analyzer plus freely
contributing his knowledge of sample interface systems in response to our many
questions. The field program was helped by the cooperation of the Boston Edison
Company and The Public Service Company of New Hampshire who provided test sites.
inle express our gratitude to these companies and their employees for their
assistance to this program. At Maiden we would like to acknowledge the efforts
Df Mr. Roger Lisk who in the course of his technican duties was exposed to the
hardships and discomforts of the worst kind of New England Winter weather.
VI
-------
CONTENTS
Title Page
LIST OF TABLES ix
LIST OF FIGURES ix
PART I
SUMMARY 1
INTRODUCTION 7
SOURCE CHARACTERISTICS 11
Power Plants 11
Nitric Acid Plants 11
Sulfuric Acid Plants 12
ANALYZER REQUIREMENTS 17
DESIGN OF SAMPLING INTERFACE 21
Design Strategy 21
Control of Sample Flow Rate 24
Control of Sample Temperature 27
Control of Sample Pressure and Venting of Sample 28
Positioning of Sample Pump 32
Materials of Construction 35
Sample Extraction 41
Calibration 47
Sizing of Sample Lines 54
Heat Transfer from Sample Lines 57
System Response Time 60
Pump Requirements 64
Moisture Removal 67
Fine Filtration 77
Instrumentation 79
Specific Source/Analyzer Combinations 80
Sources of Sampling System Components 87
Interface for Field Demonstration 88
SAMPLE INTERACTION WITH THE INTERFACE SYSTEM 89
Reactions 89
Absorption 94
Adsorption 99
vii
-------
Title Page
REFERENCES 103
APPENDIX A. CALCULATION OF STACK GAS COMPOSITIONS 107
APPENDIX B. CALCULATION OF FLOW RATE VS PRESSURE DROP 113
APPENDIX C. HEAT TRANSFER FROM SAMPLE LINES 117
APPENDIX D. EFFECT OF DEAD-END VOLUMES ON RESPONSE TIME 127
APPENDIX E. RESPONSE CHARACTERISTICS OF BACK-MIXING VOLUMES 131
APPENDIX F. CONDENSER COOLING REQUIREMENT 137
APPENDIX G. SPECIFICATIONS FOR SAMPLING INTERFACE 139
APPENDIX H. SOURCES OF SAMPLING SYSTEM COMPONENTS 143
APPENDIX I. CALCULATION OF S02 LOSS AND pH OF CONDENSATE 149
APPENDIX J. CALCULATION OF N0£ ABSORPTION RATE 157
PART II
INTRODUCTION 161
DESCRIPTION OF DEMONSTRATION SYSTEM 163
Background 163
Sampling and Instrumental Locations 163
Fabrication of Sampling System 165
Calibration and Sampling Lines 168
Instrument Racks 170
OPERATION 177
General 177
Operational Data 177
Operational Problems 177
DATA 183
DISCUSSION 191
TECHNICAL REPORT DATA SHEET 194
vm
-------
LIST OF TABLES
•/
Table Title Page
1-1 Federal Regulations for Stationary Source Emissions 7
1-2 Stack Gas Characteristics for Fossil-Fuel Fired Power Plants .... 13
1-3 Stack Gas Characteristics for Nitric Acid Plants 14
1-4 Stack Gas Characteristics for Sulfuric Acid Plants 15
1-5 Sample Conditioning Requirements for Various Analyzers 18
1-6 Barometric Pressures and Standard Deviations 29
1-7 Chemical Resistance of Various Materials 36
1-8 Maximum Continuous Operating Temperatures for Plastics 38
1-9 Costs of Various Sample Line Materials Based on 30.48 Meters
of 6.35 mm o.d. Tubing 39
1-10 Filtering Properties of Sintered Stainless Steel 44
1-11 Sampling Line Lag Time 56
1-12 Response Times for Various Sampling Line Materials 101
II-l Approximate Length of Sampling Line from Probe to Instrument Rack. . 170
11-2 Operational Data 178
II-3 Plant Operation 180
11-4 Background Data 184
II-5 Response Time to Span Gas for Interface System Plus Analyzers. ... 185
II-6 Oil-Fired Plant S02 Data 186
II-7 Coal-Fired Plant S02 Data 187
II-8 Coal-Fired Plant NOX Data 188
s
LIST OF FIGURES
Figure Ti tle_ Page
1-1 Flow Schematic of Monitoring System for Combustion Sources 22
1-2 Sampling Interface using BPR to By-Pass Excess Flow 25
1-3 Alternatives for Sample Venting to Atmosphere 31
1-4 Alternatives for Positioning of Sample Pumps 33
1-5 Probe-Tip Filter Arrangements 42
1-6 Flow vs Pressure Drop for 3.175 mm Thick Sintered Stainless Steel. . 44
1-7 Pressure Drop Across Alundum Thimble Filled with Fly Ash 45
ix
-------
Figure Ti tle Page
1-8 Through the Probe Tip Filter Calibrate System 49
1-9 Technocheck Check Valve (Techno Corporation, Erie, Pa.) 50
1-10 Schematic of Timers for Automatic Calibration 52
1-11 Flow Rate vs Pressure Drop for Various Line Sizes 55
1-12 Pump Output vs Pressure (Thomas Industries Inc.) 66
1-13 Ball Float Trap for Removal of Liquid Condensate (Armstrong
Machine Works) 72
1-14 Condensate Trap Using Barometric Leg 74
1-15 AT vs AP for Trap and Analyzer 76
1-16 Gravity and Glass Wool Demister 83
1-17 Reaction Rate Data for S02 Oxidation 91
1-18 Experimental Apparatus for Solubility Loss Tests 95
1-19 Measured S02 Level and Temperature Conditions 96
1-20 Adsorptive Interactions with Various Sample Line Materials 100
1-21 Partial Pressure SOg vs pH 152
1-22 Liquid Concentration of S02 vs pH 153
II-l General Arrangement of Pump in Sampling Systems 164
II-2 Oil-Fired Site 166
II-3 Coal-Fired Site 167
11-4 Probe Assembly 169
II-5 Schematic of Sample and Calibration Gas Connections for Analyzers . 172
II-6 Instrument Rack 173
II-7 Schematic of Sample and Calibration Gas Connections for •
Instrument Racks 174
11-8 Front and Control Panels of Instrument Rack 175
-------
INVESTIGATION
OF
EXTRACTIVE SAMPLING INTERFACE PARAMETERS
PART I
DESIGN DATA AND INFORMATION
XI
-------
SUMMARY
Design of an adequate extractive Interface for the continuous monitoring
of NOX and S02 from Category I sources involves a number of engineering
choices and trade-offs. The objective of this program is to examine the
choices available and to provide information pertinent to the design of an
adequate sampling interface. An "adequate" or a "minimum" sampling inter-
face is the simplest and least expensive system that will permit continuous
monitoring within certain specified tolerance limits.
The design of a sampling interface must be predicated on the specific
source/analyzer combination for which it is intended. Typical source char-
acteristics are given for power plants, nitric acid plants, and sulfuric
acid plants. The sample conditioning requirements are given for eight
commercially available analyzers including NDIR, UV, electrochemical, and
chemiluminescent instruments.
Although it is difficult, in the absence of field test data, to recom-
mend a specific system which will achieve an overall cost minimum, a number
of design options are presented and discussed in relation to the "minimum"
design criteria.
Most analyzers are sensitive to pressure changes, and some means must
be provided for controlling the pressure at a fixed value. The simplest
method of controlling pressure is to vent the analyzer to atmosphere.
Barometric pressure variations are well within the tolerance limits for
accuracy.
Sample flow rate is not a critical parameter and the need for flow con-
trol devices or back-pressure regulators is not anticipated. A flow rate
on the order of 1 slpm is typical of the analyzers surveyed.
Sample temperature is not a critical parameter as long as it remains
within fairly wide limits. All analyzers which are sensitive to sample
temperature provide internal control.
-------
In order to control pressure by venting to atmosphere, the sample pump
must be positioned upstream of the analyzer. However, several alternatives
are still available for the positioning of the pump in relation to other inter-
face components. Advantages and disadvantages for these alternatives are
discussed.
The proper choice of materials of construction is an important part of
the interface design. Information is presented on the chemical resistance,
heat resistance, and cost of various materials. Of the less expensive mate-
rials, polypropylene appears to be quite suitable for extensive use, except
where concentrated sulfuric or nitric acid is encountered.
For combustion sources, coarse filtration is required before the sample is
withdrawn from the stack. Two types of filters can be used: 1) the "external"
or exposed filter in which the filter element is supported within the stack
and sample passes through the filter to the sample line, and 2) the "internal"
or in-line filter which is also supported within the stack but is exposed only
to the withdrawn sample. (These are illustrated in Figure 1-5, p. 42). The in-
ternal arrangement is recommended when calibration gases are to be injected
at the probe-tip. Internal filters are available in sizes sufficient to
collect particulates over a period of several months. Therefore, automatic or
semi-automatic back-flushing of the filter is unnecessary. The pressure drop
across a loaded alundum thimble was found to be quite acceptable.
Calibration of the analyzer by injection of zero and span gases at the
probp tip was successfully demonstrated in pitot plant tests. The requirement
of once-daily automatic calibration is met by using a simple timing device to
actuate solenoid valves.
Information is provided for proper sizing of sample lines. For a typical
analyzer flow rate on the order of 1 SLPM, 6.35 mm OD tubing is recommended.
The sc. -pie line response lag is on the order of half a minute.
T'ie rate of heat transfer from sample lines is determined and is shown
to be fairly rapid. Less heat resistant plastic materials can be used for
Cample lines, except for the first few meters in the vicinity of the stack
•/hen sampling hot combustion gases. Unheated, self-draining sample lines
-------
are recommended where condensate freeze-up is not a potential problem.
Heated, self-draining lines are recommended for sub-freezing ambients,
with heating sufficient to hold the sample line temperature above 0°C.
It is impractical to use high sample flow rates in unheated or uninsulated
lines to prevent condensate freeze-up.
The tolerance limit on system response can be easily met without
extracting excess sample and by-passing the excess to vent. Equations
and procedures are given for calculating the system response time.
It is shown that under practical conditions, dead-end volumes have little
effect on response time. Large mixing volumes, however, can have a
significant effect.
When pressure is controlled by venting to atmosphere, the two most
suitable types of sampling pumps are the bellows pump and the diaphragm
pump. Both are available in sizes which meet specifications for sampling
system use.
Moisture removal is an important part of the interface design. Only
NDIR instruments require the maintenance of a constant moisture level.
A refrigerated condenser is recommended for most applications. Where
absorption of N02 by the condensate is a potential problem, a permeation
dryer can be used or, in some cases, the sample can be maintained above
its dew point. The use of dessicants is not recommended. Various con-
densate traps and removal schemes are discussed. The simplest appears
to be a ball-float trap when operated under positive pressure and a
barometric trap when operated under negative pressure. Information
is given on the temperature-pressure requirements to prevent condensa-
tion in the analyzer.
The considerations involved in fine filtration are discussed. A
number of filter media will give good parti oil ate removal, including a
tube packed with glass wool.
Sophisticated instrumentation is not required. Useful measurements
include the sample flow rate, the pump suction and discharge pressures,
-------
and the condenser temperature. If heated sample lines are used, sample
line temperatures are also useful.
Some modifications of the above general design considerations are
required for different source and analyzer combinations. These modi-
fications are discussed. Methods of demisting sulfuric acid stack
gases are also considered.
Information is provided on commercial sources of various components
suitable for use in sampling systems, and the specifications are given
for the interface to be demonstrated in field tests.
An important requirement of any sampling interface is that inter-
actions between the interface and sample must not exceed certain
tolerance limits. Three types of interaction, viz., reaction, absorp-
tion, and adsorption, are considered in detail.
Loss of S02 by catalytic oxidation to $03 on particulates collected
in the probe-tip filter has been shown to be negligible both by calcu-
lation and by experiment. Loss of NO by gas phase oxidation to N0£ can
be appreciable if the residence time within the sampling system is very
long. However, under normal sampling conditions, the loss of NO will
be within the tolerance limits on accuracy. Catalytic oxidation of NO
on probe-tip particulates was shown, by experiment, to be negligible.
Loss of N02 by reduction is not anticipated.
Loss of S02 by absorption(dissolution) in condensate is shown to be
negligible both by calculation and experiment. The solubility of NO
is much less than S0£ and losses are again negligible. On the other
hand, N02 is very soluble and, if equilibrium is reached, complete N02
lo^s can occur. However, this is of little concern in most monitoring
applications. Combustion stack gases typically contain only a few percent
of t..c total NOX in the NOg form, and complete loss will not exceed
the tolerance limit which is set on NO . In the few applications where
significant N0« concentrations are expected, means are recommended for
eliminating absorptive losses.
-------
Adsorption on interface components can effect the system response
time. Experimental tests on various sample lines indicated negligible
adsorption effects for Teflon, 316 SS, and polypropylene; small effects
for nylon; and large effects for polyethylene and Tygon. The latter two
are not recommended for extensive use in sampling systems.
-------
INTRODUCTION
Under authority of the Clean A1r Act, the U.S. Environmental Protection
Agency has set forth standards of performance for new and modified station-
ary sources (1). Table 1-1 lists the sources covered, the limitations placed
on various pollutants, and the requirements for monitoring. Continuous
monitoring of NOX and/or SO- is required for power plants, nitric add
plants, and sulfuric add plants.
TABLE 1-1
»
FEDERAL REGULATIONS FOR STATIONARY SOURCE EMISSIONS
Source
Pollutant
Limit
Continuous Monitoring
Required?
Steam Generators
Gas Fired
Oil Fired
Coal Fired
Nitric Acid Plant
Sulfuric Acid
Plant
Incinerators
Portland Cement
Parti culates
NOX
Parti culates
so2
NO,
X
Parti culates
so2
NOX
N0x
so2
H2S04 mist
Parti culates
Parti culates
0.043 kg/G joule
0.086 kg/G joule9
0.043 kg/G joule
0.344 kg/G joule
0.129 kg/G joule
0.043 kg/G joule
0.516 kg/G joule
0.301 kg/G joule
1.5 gm/kg acid3
2.0 gm/kg add
75 rug/ kg acid
.1668 mg/sl
150 mg/kg kiln feedd
Yesb
Yes
Yesb
Yesc
Yes
Yesb
Yes
Yes
Yes
Yes
No
No
No
Plant
50 mg/kg kiln feed*
(a) Expressed as N02
(b) Continuous monitoring by smoke
detector
(c) Except where low sulfur fuels are
burned and representative daily
sulfur analyses are performed
(d) Emission limit on kiln
(e) Emission limit on clInker
cooler
-------
In order to continuously monitor stack gases, an analyzer must be used
which Is capable of providing an electronic output that is proportional to
the concentration of pollutant in the sample stream. However, even the best
of analyzers cannot provide reliable monitoring without a properly designed
sampling interface. The sampling interface must be capable of performing
the following functions:
- Removing a representative sample from the stack;
- Maintaining sample integrity (within specified limits)
during transport to the analyzer;
- Conditioning of the sample for compatibility with the analyzer;
- Providing gas switching to calibrate the analyzer.
Although there are a number of commercially available extractive sampling
interfaces, the rationale behind the particular designs used has not been
thoroughly reported or examined. The objective of this program and of this
report is to provide information pertinent to the design of S09 and NO ex-
£ A
tractive sampling interfaces with particular emphasis on design criteria
for a "minimum" system. In this respect, a minimum sampling system may be
defined as the simplest and least expensive system that will permit analysis
within certain specified tolerance limits over an "acceptable" period of
time.
There are several tolerance limits which are particularly applicable
to the design of stack sampling interfaces:
- Accuracy must be within +. 10% based on the mean output and +. 20%
at the 95% confidence interval. Accuracy may be determined by
injection (preferably at the probe tip) of zero and span gases
or by verification with wet chemical analysis.
- System response must be 95% complete within eight minutes.
System response may be determined by Injection at the probe of
a step change in concentration and measuring the time required
for 95% of full response.
- The operational period must be at least seven days during which
no replacement, repairs, or corrective maintenance may be per-
formed other than routine analyzer adjustments.
- The analyzer must be automatically recalibrated once every 24
hours by injection of zero and span gases at the probe tip, if
feasible.
8
-------
Other limits such as zero drift, calibration draft, and repeatability
are more directly applicable to the analyzer operation and will not be
considered as limitations on the design of the sampling Interface.
In addition to the above limitations, the design of the sampling Inter-
face will be strongly Influenced by the source/analyzer combination. The
obvious design procedure 1s:
1. determination of source characteristics at the most feasible
sampling sites,
2. select the best sampling site,
3. determine the requirements of the analyzer, and,
4. design an appropriate sampling Interface that will provide the
analyzer with a compatible sample.
-------
SOURCE CHARACTERISTICS
The three sources of particular Interest in relation to federal re-
requirements for continuous monitoring are power plants, nitrfc acid plants,
and sulfuric acid plants. It is, of course, Impossible to provide pre-
cise information on the temperature and composition of stack gases from each
of these sources. The stack gas characteristics depend on a wide range of
operating variables and are strongly influenced by emission control processes.
The design of a sampling interface should be predicated on information from the
specific source for which it is intended. Nevertheless some "typical"
values can be presented and some general comments made.
POWER PLANTS
The major stack gas constituents resulting from combustion of fossil
fuels can be easily approximated from an assumed fuel composition and an
assumed excess of air. The assumptions and calculations are given in Appendix
A. The results for gaseous, liquid, and solid fossil fuels are given in
Table 1-2. The typical maximum emissions are based on the Federal standards
given in Table 1-1.
More detailed information is available on emissions from coal fired
(2) and oil fired (3) power plants. This data gives actual emissions for
various facilities, fuels, and operating procedures.
NITRIC ACID PLANTS
Most nitric acid is produced by the "pressure" process 1n which
an ammonia-air mixture under pressure is catalytically oxidized to NO and N0«.
The N02 is subsequently absorbed in water to form nitric acid. The tail
gas containing NO and unabsorbed N0~ from the absorber is the primary source
of NO emissions from nitric acid plants. Table 1-3 gives typical stack gas
A
characteristics from two plants: one with no further treatment of absorber
tail gases and one using catalytic reduction to convert NO and NO- to nitrogen.
Catalytic treatment involves mixing the tail gases with a fuel such as hydrogen
or natural gas and passing the mixture through a reactor. The nitrogen oxides
are reduced to nitrogen and the fuel is oxidized to combustion products
(C02 and H20).
11
-------
Catalytic treatment of waste gases is by far the most commonly used
method of reducing NO emissions. However, it is also possible to remove NO
n
and NO by caustic absorption to form nitrite and nitrate salts. In this
A
case, the stack gases will have approximately the same composition as the
gases from a plant with no waste gas treatment except that the concentration
of NO will be reduced by about an order of magnitude.
SULFURIC ACID PLANTS
Most sulfuric acid is produced by the contact process in which
any one of several sulfur containing feeds (elemental sulfur, hydrogen sul-
fide, sulfide ore, spent sulfuric acid, etc.) is oxidized to S02 in a sulfur-
burning furnace. The S02 is mixed with air and catalytically oxidized to
S03. The SO. is then absorbed in 98 to 99 percent sulfuric acid. The tail
gas from the absorber is the primary source of emissions from sulfuric acid
plants. Typical stack gas compositions are shown in Table 1-4. Note that there
is essentially no free water in the stack gases leaving the absorber. The
typical emission limits are again based on the federal standards. It is
obvious that some means of emission control must be used to reduce S0? and
H2S04 mist.
Many types of S02 recovery systems have been proposed but until
recently only a few such systems have been installed. Alkaline scrubbing
has been used to reduce S02 to about 300 ppm in two stages. The use of any
type of wet scrubbing process will saturate the stack gases with water at
the prevailing conditions within the scrubber.
12
-------
TABLE 1-2
STACK GAS CHARACTERISTICS FOR FOSSIL-FUEL FIRED POWER PLANTS
Stack Temp (°C)(a)
Typical Stack Gas Composition^ '
co2
N2
°2
Typical Maximum Emissions^0'
Particulates
so2
N0x
Gaseous Fuel
135-205
8%
16*
73%
3%
0.119 mg/sl
127 ppm
Fuel Oil
150-205
12%
9%
76%
3%
0.127 mg/sl
394 ppm
205 ppm
Coal
150-205
14%
* 6%
77%
3%
0.125 mg/sl
572 ppm
464 ppm
(a) Ref. 15
(b) Calculated on the basis of 20% excess air. Assumptions and calculations
given in Appendix A.
(c) Calculated from Federal emission standards (1). See Appendix A.
13
-------
TABLE 1-3
STACK GAS CHARACTERISTICS FOR NITRIC ACID PLANTS
Before Waste Gas Catalytic Waste Gas
Treatment Treatment
Gas Temperature3 21.1-37.8 °C 204.4-260 °C
Typical Stack Gas
Composition3
NOX 0.3% 0.01%c
H20 0.7% 3.8%
N2 96.0% 94.2%
02 3.0%
C02 -- 2.0%
Typical Maximum
Emissions'1
NO 202 ppm
A
(a) From Ref. 16, p. 10.
(b) Calculated from Federal emission standards (1) assuming an
average stack gas rate of 3621 si/kg acid produced
(Ref. 16, p. 22).
(c) For specific installations, NOX content may vary considerably:
from .21% to <.0002% (Ref. 16, p. 22).
14
-------
TABLE 1-4
STACK GAS CHARACTERISTICS FOR SULFURIC ACID PLANTS
Contact Plant with Mist Eliminator
Stack Gas Temperatures*3' 23.9 - 100 °C
Typical Stack Gas Composition^
S02 0.26%
S03 and/or HgSO. mist
09 and N2 99.74%
Typical Emissions L1m1t^
S02 255 ppm
H2S04 mist .0272 mg/sl
(a) Ref. 17
(b) Calculated from Federal standard assuming an average stack gas rate
of 2747 si/kg acid produced (ref. 17).
15
-------
ANALYZER REQUIREMENTS
The requirements of the analyzer play an Important part 1n the design
of a sampling Interface. Requirements that must be considered Include sample
temperature, pressure, and flow rate; part1culate level, moisture tolerance,
and Interfering species. A survey of analyzer manufacturers was conducted
In order to ascertain limits on various analyzer requirements. Two manufac-
turers were contacted for each of the following types of monitors:
- NDIR (Beckman, Intertech)
- UV (Dupont, Teledyne)
- Electrochemical (Dynasciences, Envlrometrics), and
- Chemlluminescent (Aerochem, Thermo Electron)
The results are summarized in Table 1-5.
17
-------
SAMPLE CONDITIONING REQUIREMENTS FOR VARIOUS ANALYZERS
00
Principle
Manufacturer
Model
Species
Sample Flow
Rate
Sample
Temperature
Sample
Pressure
Moisture
Content
Particulate
Level
Analyzer
Sensitivity
to Pressure
Interfering
Species
Beckman
864, 865
S02, NO
0.5-1.0 slpm
-1. 11-37. 8°C
"Approx. atmospheric
Dewpoint = 1.67°C
Maintain same dewpt.
in sample & cal .
gases
"clean"
(probably < 1 vim)
Output directly
proportional to P
Sensitivity to H90 =
100 to 1 e-
NDIR
Intertech
Uras 2
S02, NO
0.5-1.0 slpm
10-45°C
-2.02 x 103N/m2 to
0.103 M N/m2
Dewpoint = 1.67°C
Maintain same dewpt.
in sample & cal .
gases
< 1 Vim
Output directly
proportional to P
H20
duPont
400
S02, N0x
"not critical"
(0.5-10 slpm)
Ambient to 204. 4°C
(typical 104.4°C)
"Vacuum" to
0.446 M N/m2
Keep sample temp.
above dewpoint
< 20 vim
Output directly
proportional to P
None
UV
Teledyne
611, 612
S02, N02
"not critical"
(0.5-10 slpm)
Ambient to 176.7°C
(typical 79.4°C)
'.'Vacuum" to
1.48 M N/m2
Keep sample temp.
above dewpoint
< 10-20 ym
Output directly
proportional to P
None
-------
TABLE 1-5 (continued)
Principle
Manufacturer
Model
Species
Sample Flow
Rate
Sample
Temperature
Sample
Pressure
Mol sture
Content
Parti cul ate
Level
Analyzer
Sensitivity
to Pressure
Interfering
Species
Electrochemical
Dynasciences
NX! 30, NR230, SS330
S02, N02, NOX
0.25-1 slpm
4.44-37.8°C
-1.49 k N/m2 to
"slightly above
atmospheric"
40-80% rel. hum. or
sample temp. > dew-
point
"Free of suspended
material" (< 1 urn)
Output directly
proportional to P
N02
Envirometrics
NS300
SO,, NOX
1-2.5 slpm
4.44-48.9°C
"Approx. atmospheric"
Sample temp. > dew
point
Remove 98% > 0.7 ym
100% > 1.7 ym
Sensitive to pressure
N02
Chemi luminescent
Aerochem
AA
NO, NOX
0.5 slpm
automatically fixed
3.89-60°C '
.101 x 106± 2.66 k N/m
Recalibrate for
other pressures
Dewpoint < 21.1°C
< 1 ym
Flow rate internally
regulated
Minor sens, to C02
Thermoelectron
10A
NO, NOX
0.5-2.0 slpm
4.44-48.9°C
2 -16.4 k N/m2to
68.94 k N/nr Re-
calibrate for other
pressures
Dewpoint below
analyzer temperature
Remove 91%
> 0.6 ym
Pressure internally
regulated
-------
.-' y' r- ' r /
DESIGN OF SAMPLING INTERFACE
DESIGN STRATEGY
The design of a "minimum" sampling interface is not a completely
straightforward problem. To be acceptable, the sampling interface must de-
liver a continuous sample flow that is conditioned to meet the analyzer re-
quirements. However, there are a number of different interface designs that
can perform this task. The minimum-system design is that one which repre-
sents the greatest economy in terms of capital Investment, operating costs
(labor) and maintenance costs. This may involve a number of trade-offs.
For example, the use of cheaper materials of construction may result in a
higher rate of corrosion and higher maintenance costs. Or, a highly auto-
mated system will require a larger capital investment but may save on oper-
ating costs. These trade-offs are difficult to assess in the absence of
actual in-field operating data. Furthermore, a number of options are avail-
able in the placement of sample conditioning components. There may be no
obvious overriding advantage for one design over another. As a result, it
is unrealistic to attempt to propose a rigid sampling interface design for
a particular source/analyzer combination and claim that it achieves an ab-
solute cost minimum. Nevertheless, it is worthwhile indicating the general
directions in which such a minimum lies.
Figure 1-1 shows a sampling interface that simultaneously measures
(NDIR) and NO (electrochemical) emissions from a coal-fired or oil-fired
n
power plant. This system is the most demanding in terms of sampling condi-
tioning requirements and will, therefore, be used to illustrate the con-
siderations involved in designing a sampling system. Specific interface re-
quirements for other source/analyzer combinations will be given in the sub-
section entitled Specific Source/Analyzer Combinations . The strategy
used in arriving at the design of Figure 1-1 was to keep the system as
simple as possible and to use automation only where absolutely necessary,
i.e., for automatic daily calibration checks. It is reasonable to presume
that the simpler the system the fewer the potential failures.
Sample is withdrawn from the stack through a probe tip filter, which
removes coarse particulates, and is drawn through a sample line which, for
21
-------
-ixi-rH *
IO
3
(J
IO
4->
Alund
xParti
7\ Filte
1
J
um
culate
A
!
M
10
3
01
3
O
+j
in
^~
Ball
~ FioaL *^
Trap
r^ T
(PI-2j
NV viy
X i^vi
j PI-I)
4?
i/*^
Re frig.
/" X. "*" Condenser
V J ' 1
1.
V^/ T
C i
Ball
n n F1oat
SsvXQS T
T
(PR) (pT) (PR")
^1
(TV
1 1
Zero
S02 NOX
Span Span
Glass
wool
:ilter
Gl
,„_ un.
Fil
PR =
SV =
PI =
FI =
r
I
ass
Dl
ter
pre
sol
pre
flo\
Analyzer
Elect.
Chem.
Vent
Atmos.
Vent
Figure 1-1. Flow Schematic of Monitoring System for Combustion Sources.
-------
outdoor service, is heated only enough to prevent freezing of condensate in
the lines. The sample passes through a pump fitted with a by-pass valve to
regulate the discharge pressure. For the NDIR instrument the sample then
passes through a refrigerated condenser to obtain a constant low moisture
level, and condensate is continuously removed by a ball-float trap. For the
electrochemical instrument, only the condensate trap is necessary. A fine
filter is used to remove particulates above the submicron range. The flow
rate of each sample is adjusted with the needle valves, and is measured at
the flow indicators before passing through the analyzers. Calibration gases
are automatically injected at the probe tip once every 24 hours. The pres-
sure at the pump suction is measured to check for excessive pressure drop
across the probe tip filter and sample line; the pressure at the pump dis-
charge is measured to check for excessive pressure drop across the fine fil-
ter and to set a reasonable pump discharge pressure.
The design requirements for each specific component in the interface
will be considered in subsequent sections after a more detailed treatment of
the design rationale for the overall system.
23
-------
CONTROL OF SAMPLE FLOW RATE
Most analyzers are not particularly sensitive to minor variations in
the flow rate; the analyzers shown 1n Table 1-5 can typically tolerate a two-
to four-fold variation in flow rate. It is interesting that the recommended
sample flow rate for all of the analyzers surveyed is on the order of one
standard liter per minute (SLPM). This flow rate will be used as "typical"
in several subsequent design calculations.
Since sample flow rate is not a critical factor, the use of a flow
controller is unnecessary. For the design of Figure 1-1, sample flow rate is
determined by the pressure at the fine filter outlet and the setting of the '
needle valve upstream of the flow meter. With this arrangement, there
should be only a minor long-term variation in flow rate. Potential sources
of variation are: changes in pump discharge pressure due to loading of the
probe tip filter, plugging of sample lines, plugging of the pump by-pass
line, temperature variations, etc.; and changes in the fine filter discharge
pressure due to particulate loading. It is anticipated that the flow rate
variation will not exceed the allowable range (Table 1-5) over a seven day
period.
Most commercially available sampling interfaces withdraw more sample
from the stack than is actually needed by the analyzer. The excess sample
is then vented through a back pressure regulator as shown in Figure 1-2. The
by-pass flow to vent is generally several times as great as the flow rate to
the analyzer. The primary advantages that may be cited for this arrangement
are: (1) the response time for the sampling Interface 1s less since a higher
«
flow rate is used, and (2) the regulated pressure maintains a constant flow
rate to the analyzer. However, it will be shown in a later section that if
the ampling lines are properly sized, the 95% response time of the sampling
•>ystem will fall well within the eight minute limit. Thus, the use of-high
f'o.: r; .«s for improved response is really unnecessary. Moreover, the prac-
t'ce o' extracting excess sample has a distinct disadvantage in that filters
are loaded more rapidly, sample lines plug more rapidly, corrosion and par-
iculate destruction of pumps and other components is Increased, and, in
jeneral, the maintenance costs are greater.
24
-------
rHXh
Sample
Line
Refrigerated
Condenser
ro
en
Fine
Filter
Trap
Figure 1-2. Sampling Interface Using BPR to By-Pass Excess Flow.
-------
As pointed out previously, it 1s probably not necessary to use a back
pressure regulator to maintain an acceptably constant flow rate to the analyzer.
However, if a constant flow were required, one possible arrangement would be
to use a back pressure regulator in a redrculation loop to the pump suction
as shown by the dashed line in Figure 1-2. In this way the pressure could be
regulated without having to vent a portion of the sample. Except where main-
tenance of a constant flow rate is essential, the use of a needle valve is
preferred to a BPR since the needle valve is simpler, cheaper, and less
susceptible to malfunction.
26
-------
CONTROL OF SAMPLE TEMPERATURE 'URE
Without exception, all of the analyzers surveyed had provisions for
internal temperature control. Therefore, 1t is only necessary for the
sampling interface to provide a sample within the rather wide temperature
limits of Table 1-5. In general, the sample temperature equilibrates quite
rapidly with the surroundings (quantitative information is provided in the
subsection entitled Heat Transfer from Sample Lines).so that the sample
enters the analyzer at essentially room temperature. It is therefore
necessary that the enclosure housing, the analyzer, and the conditioning
components of the sampling interface be kept above freezing and wfthin the
temperature limits of Table 1-5. There is no need to accurately control
the ambient temperature at a fixed value.
27
-------
CONTROL OF SAMPLE PRESSURE AND VENTING OF SAMPLE
In general, analyzers are quite sensitive to changes in sample pres-
sure. For most analyzers output varies in direct proportion to pressure. It
is therefore necessary in most cases to provide some means of pressure con-
trol at the analyzer. The simplest and most convenient means of controlling
the pressure is by venting the analyzer to atmosphere. In this case, the
analyzer will operate at atmospheric pressure provided the vent line is prop-
erly designed to give a negligible pressure drop.
Atmospheric pressure is, of course, subject to variation. Data on
the variation of barometric pressure (4) is given in Table 1-6. This data
covers an 18-year period from 1946-1963 with barometric pressure readings
taken once every three hours. The largest monthly variation is for the
northeast region of the United States Where a standard deviation of 8.75 mm
of mercury was measured for the month of^February. The probability that
a pressure reading will fall beyond two st^dard deviations from the mean
is less than 5% (4.55%). Therefore, at a confidence interval of 95%, the
maximum pressure variation may be taken as 2330\|/m2 or about 2.3%. This
is a very small vacation when compared to the al\^wable 20% variation at
the 95% confidejs^e interval.
The veA line will generally be fairly short C6 m or less) and even
with 6.35 ip/o.a. tubing, the pressure drop w^ll be small. For example, at
a flow r/ce of 1 SLPM, the pressure drop across'-6 meters of 6.35 mm o.d.
x l.£2'mm wall tubing is only 241 N/m . The onlyStime this pressure drop will
have any effect at all on the accuracy of the analyW output is when differ-
ent flow rates are used during calibration and samplAnalysis. If, for ex-
ample, calibration is done at 2 SLPM and sample analysi^at 1 SLPM, the
pressure drop will be 482 N/m during calibration giving a^pressure differ-
ence of 241 N/m2. This translates into an error of only 0.23% which is
entirely negligible.
At the other end of the jspectrum there may b^ some concern in making
the vent line too large in diameter and too short. In this case, it is pos-
sible that air may diffuse back |:hrough the vent line and cause an inaccurate
28
-------
TABLE 1-6
BAROMETRIC PRESSURES AND STANDARD DEVIATIONS
Values in inches of mercury? eight observations per day 1946-1963
Jan
Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Annual
Alameda, Mean 30.101 30.076 30.042 30.006 29.971 29.920 29.927 29.923 29.905 29.982 30.058 30.097 30.000
Calif. S.D. 0.172 0.169 0.165 0.126 0.103 0.089 0.073 0.078 0.088 0.105 0.145 0.168 0.146
Total Obs. 6184 5646 6191 6000 6200 6000 6432 6448 6239 6447 6239 6447 74473
Boston, Mean 29.985 29.955 29.910 29.912 29.934 29.910 29.933 29.959 30.033 30.029 29.980 29.993 29.961
Mass. S.D. 0.317 0.344 0.286 0.265 0.198 0.182 0.157 0.152 0.205 0.241 0.274 0.295 0.253
Total Obs. 4462 4063 4460 4317 4464 4313 4464 4452 4319 4462 4317 4462 52555
ro
Chicago, Mean 29.366 29.339 29.275 29.247 29.269 29.250 29.296 29.313 29.341 29.345 29.299 29.343 29.307
Illinois S.D. 0.259 0.260 0.256 0.237 0.176 0.147 0.119 0.119 0.164 0.199 0.252 0.256 0.215
Total Obs. 4215 3834 4214 4077 4211 4053 4074 4010 3942 4395 4315 4462 49802
Miami, Mean 30.105 30.070 30.041 30.026 29.992 29.989 30.040 29.991 29".937 29.943 30.028 30.085 30.021
Florida S.D. 0.123 0.124 0.111 0.107 0.086 0.078 0.059 0.071 0.086 0.099 0.101 0.102 0.110
Total Obs. 5698 5196 5701 5517 5704 5515 5702 5454 5279 5455 5279 5455 65955
To convert inches of mercury to Newtons/square meter, multiply values by 33874.
-------
analyzer reading. However, calculation of the rate of back-diffusion against
flow indicates that this is an extremely unlikely possibility. It is fair to
assume that at the point where the vent line attaches to the analyzer suffi-
cient restriction has been provided to prevent atmospheric back-diffusion.
It is important in installing the vent line to consider the possibil-
ity of moisture condensation. If the sample is vented outdoors where the
temperature is below the dewpoint of the sample, i.e., the temperature of the
condensate removal trap, condensation will occur. Condensate may run back
through the vent line into the analyzer causing severe problems or, if outside
temperatures are low enough, the condensate may freeze and plug the vent line.
Outdoor vents should be as short as possible and pitched so that condensate
runs away from the analyzer rather than toward it, as shown in Figure 1-3.
Alternatively, liquid condensate can be handled by a barometric trap as
shown in Figure 1-3. The trap should be filled with water before startup so
that samples are not vented within the instrument enclosure. A third al-
ternative, shown in Figure 1-3, is to heat trace the portion of the vent line
which extends beyond the instrument housing. The short pitched vent is the
simplest of the three alternatives.
30
-------
Analyzer
Analyzer
Heat
Traced
(a)
(b)
Analyzer
1
,
Drain
(c)
Figure 1-3. Alternatives for Sample Venting to Atmosphere; (a) Pitched line;
(b) Barometric trap; (c) Heat traced vent.
-------
POSITIONING OF SAMPLE PUMP
There are a number of options available in the placement of
sample conditioning components for a given source/analyzer combination.
The arrangement shown in Figure 1-1 is only one of several possibilities.
It is difficult,™ the absence of in-field operating experience to
properly weight the advantages and disadvantages of each arrangement.
Nevertheless, it is worthwhile pointing out these advantages and dis-
advantages and making some qualitative comments.
One major alternative lies in the placement of the sampling
pump. Two different arrangements are shown in Figure 1-4. Other pump
positions are also possible. For example, the pump may be placed down-
stream of tne analyzer in which case the sample would be "pulled" through
the entire system rather than "pushed". This alternative would require a
more elaborate means of pressure regulation at the analyzer and also maximizes
any air in-leakage that may occur. However, for the chemiluminescent analyzers,
which have built-in flow control provisions, this is the preferred arrangement
(see the subsection entitled Specific Source/Analyzer Combinations).
For the two alternatives shown in Figure 1-4, the following advant-
ages may be listed for alternative A (pump upstream of condenser and fine
filter).
(1) Condensate removal occurs at the point of highest pressure
in the monitoring system and is therefore more complete than in alternative
B. The mole fraction of water vapor in the sample is equal to the vapor
pressure of water at the condenser temperature (or trap temperature if
no condenser is required) divided by the total pressure. Thus, the amount
jf water vapor in the sample is inversely proportional to total pressure.
2
if, for example, the pressure at the condenser is 13790 N/m for alternative
A and -6895 N/m for alternative B there will be 18% less moisture in the A
sample. This difference in moisture levels will probably be of little
consequence even for analyzers which are sensitive to water vapor. However,
-------
Condenser
Fine
Filter
Analyze
Trap
T
Alternative A
Condenser
Fine
Filter
Trap
T
Alternative B
Figure 1-4. Alternatives for Positioning of Sample Pumps
33
-------
for alternative B the analyzer operates at a higher pressure than the
condenser (or trap) and care must be taken to operate the analyzer at
a higner temperature also in order to prevent condensation (see the sub-
section entitled Moisture Removal).
(2) The possibility of air in-leakage is minimized since tne
condenser, trap, and fine filter are under pressure rather than vacuum
as in alternative B.
(3) Since in alternative A the condensate removal trap is under
positive pressure a simple ball float trap can be used to continually
remove condensate. For alternative B, a barometric leg must be used, or
the trap must be intermittently pressurized to remove condensate (see the
i
subsection entitled Moisture Removal).
The advantage of alternative B is that the pump is protected from
both liquid condensate and particulate matter. It may turn out that this
is an over-riding advantage. Field experience gained by the Thermo-
electron Corporation (5) indicates that the pump is the one component most
prone to failure. For this reason they strongly prefer alternative B.
On tne other hand, Beckman Instruments (6), Leeds and Northrup (7), and
others use alternative A for sampling combustion sources. In tne field
tests to be conducted in Phase III of this project alternative A in Figure
1-4 will be used initially. If rapid pump failure occurs, alternative B
will then be implemented.
i
*The results of the work in Phase III (see Part Two) favor Alternative
B over A. With experience it seems that the pump needs more protection
Cat least from coal-fired sample gas) than provided by alternative A.
34
-------
MATERIALS OF CONSTRUCTION
The choice of proper materials of construction is a very important
part of designing a sampling interface. Acceptable materials of construc-
tion must meet three criteria: (1) the materials must have sufficient
chemical resistance to withstand the corrosive constituents of the sample;
(2) the materials must not exhibit excessive interaction (reaction, absorp-
tion, adsorption) with the sample gases; and (3) materials used in or
near the stack must be heat resistant. For a sampling system of minimum
design the above criteria must be weighed in the light of material costs.
Chemical Resistance
The corrosive constituents encountered in monitoring Category
I sources are: nitrogen oxides, sulfur dioxide, dilute nitric acid, dilute
sulfurous acid, dilute to concentrated sulfuric acid (wet S03 or acid
mist). The chemical resistances of various materials to these constituents
have been collected from a number of sources (8-12) and are summarized in
Table 1-7. All materials are evaluated at room temperature and may be
considerably less resistant at higher temperatures.
Of the metals, Carpenter 20 stainless is the most resistant
(and also the most expensive) followed by 316 SS, 304 SS, and finally
aluminum. Glass and Teflon are quite resistant to all components of the
sample. PVC and Tygon are somewhat less resistant but still good for all
constituents except concentrated nitric acid which is not anticipated to
be present in the sampling interface. Polyethylene and polypropylene have
very nearly the same chemical resistance and are acceptable for use except
where concentrated nitric acid is encountered (unlikely) or where concen-
trated sulfuric acid is encountered (acid mist from sulfuric acid plant).
Nylon is a material of apparent limited usefulness in sampling systems;
while sources differ on the performance of Viton during resistance tests.
35
-------
CO
en
TABLE 1-7
CHEMICAL RESISTANCE OF VARIOUS MATERIALS
Material
Dry
so2
Dry
Oil.
HN03
Oil.
H2S03
Oil.
H2S04
Cone. *
HN03
Cone.
304 SS
316 SS
Carpenter 20 SS
Aluminum
Glass
Teflon
PVC
Tygon
Polyethylene
Polypropylene
Nylon
Viton
S
(some pitting
observed)
S
S
S
S
S
S
S
S
S
Sc - Ud
*High concentrations of HN03 are not
**Quantities in
(a) Ref. 8
parentheses indicate
(b) Ref. 9
S
S
S
_
S
S
s
s
s
s
s
s
S
S
(<.051)**
S
S
(.127-. 508)
S
(<.127)
S
S
Se-Qb
Sc -Qb
S
S
s
anticipated in the
corrosion rates in
(c)
Ref. 10
Q
S
S
S
(.127-. 508)
_
S
S
s
s
s
u
s
U
S-Q
(<.508)
S-Q
Q
(.508-1.27)
S
(<-127)
S
Sa -Qb
S
S
s
u
s
S
S
(<.508)
S
U
(>1 .27)
S
(<.127)
S
U
Qe - Ub
U
U
u
sd - ue
U
U
(>1.27)
S
U
(>1.27)
S
(027)
S
Sa -Qb
Sb - Qe
QC - Ub
QC - Ub
U
Sd - QC
sampling interface.
nm per year.
(d) Ref. 11
(e)
Ref. 12
S = Satisfactory
Q = Questionable
U = Unsatisfactory
-------
Heat Resistance
The heat resistance of various plastic materials Is given in
Table 1-8. Typical stack temperatures (Table 1-2) are below the temperature
limit for Teflon and some may be below the temperature limit for Viton.
Depending on the exact temperature at the point of extraction, these two
materials are suitable, in terms of heat resistance, for use at any point
in the sampling interface. Less heat resistant plastics cannot be used in the
vicinity of the probe when sampling combustion sources. The rate at which
stack gases cool to a temperature compatible with other plastics is generally
rapid and will be considered in the section entitled Heat Transfer from
Sample Lines.
Surface Interactions
Materials of construction can interact with the sample by catalytic
reaction, bulk absorption or surface adsorption. These factors are considered
in detail in the section entitled Sample Interaction with the Interface System.
The results, in relation to materials of construction, can be
briefly summarized as follows. No permanent losses, e.g., by reaction, are
anticipated for any of the materials listed in Table 1-7. Absorption and
adsorption by the walls of the system must necessarily be transient phenomena
since saturation will eventually be reached and, at steady state, the
correct concentrations will be measured. Sorption (both absorption and
adsorption) by the interface walls will be manifest in a slower system re-
sponse to changes in concentration at the probe tip. Experimental measure-
ments of system response using sample lines of various materials indicate
that adsorption and absorption are negligible for 316 SS, Teflon, polypropyl-
ene, and nylon; are moderate for polyethylene; and are large for Tygon.
Costs
Costs are an important consideration in designing a minimum
system. Local distributors of various materials were contacted for list
prices on 6.35 mm diameter tubes. The materials surveyed are listed in
Table 1-9 in order of decreasing cost.
37
-------
TABLE 1-8
MAXIMUM CONTINUOUS OPERATING
TEMPERATURES FOR PLASTICS
Material
Teflon
Viton
*
Polyethylene
Polypropylene
Nylon
CPVC
Tygon
Maximum Temp. (°c)
250
150
80-125.6
110
121.1
110
60-82.2
^Depends on type used.
38
-------
TABLE 1-9
COSTS OF VARIOUS SAMPLE LINE MATERIALS
BASED ON '100 FT OF 6.35 MM OD TUBING
Material
Wall Thickness
List Price per 30.48 m
Heat Traced Teflon
Heat Traced 316 SS
Carpenter 20 SS
316 SS
304 SS
Viton
Tef1 on
Tygon
Aluminum
Glass
Nylon
Polypropylene
Polyethylene
.889
1.016
.889 welded
.889 seamless
.889 welded
.889 seamless
.889 welded
1.575
.787 stiff wall
1.575
1.575
.889
8mm OD x 6mm ID
.762
.767
1.016
350.00
325.00
196.00
184.45
80.95
113.07
71.66
145.00
107.00
63.00
15.60
11.00
7.33
4.40
4.31
3.70
39
-------
Practical Experience
A considerable amount of practical experience on materials
of construction for sampling systems has been accumulated by manufacturers
of stack gas monitors, although usage may not always be based on minimum
design considerations. In general, Teflon sampling lines are preferred
since they are more corrosion resistant and less expensive than stainless
steel. Corrosion resistant plastics are preferred (5) at least through
the point where liquid condensate is removed from the system. Both 304
and 316 SS have been found (13) to adequately resist corrosion when the
sample is kept above its dew point, but 304 is significantly less resist-
ant to warm acid condensate. Acceptable materials for internal pump
construction include Teflon, Viton, and 316 SS (5, 6). Ball valves have
been found (5) to operate more reliably than others since the ball is
wiped free of condensate and particulates with each operation.
40
-------
SAMPLE EXTRACTION
To obtain a representative participate sample or acid mist sample,
sampling must be done isokinetically. That is, the velocity and direction
of flow into the sample-line opening must be the same as the velocity and
direction of flow in the stack. However, for gas sampling it is desirable
to eliminate as much particulate matter as possible at the probe inlet.
This can be done by nonisokinetic sampling in which the probe opening
faces downstream (toward the top of the stack). Any particulates
entering the probe must then go through a directional change of 180°.
This eliminates many particulates particularly in the larger size range.
Even with the probe tip protected from direct impingement of
particulates, enough fine particulate matter is carried into the sampling
system to cause particulate fouling of various interface components. It
is essential, for combustion source sampling, to filter the sample at the
probe tip. In general there are two types of filtering arrangements that
may be used: an "external" or an "internal" arrangement. These are illu-
strated in Figure 1-5.
In the external arrangement a porous cylinder is used as the
filter media. The cylinder is typically constructed of sintered 316 SS
although other corrosion resistant metals may be used in addition to
sintered glass, quartz, ceramic, and porous silicon carbide. It is advan-
tageous to fabricate a semi-cylindrical baffle for the under side of the
probe to protect it from direct impingement of particulates.
As an example of the availability of this type of probe tip filter,
the Pall Trinity Micro Corporation manufactures three different shapes of
sintered metal probe tip filters. Standard fabrication is 316 stainless
steel but other materials are available including inconel, monel, nickel,
347 SS, 410 SS, other 400 SS, silver, copper, and high nickel molybdenum
alloys-. In the cylindrical configuration,filter areas from 0.016 to .214
sq. m. are available. Six different porosities are manufactured as listed
41
-------
Porous
Cylinder
Stack
'Wall
Stack
Gas
External Filter
Porous
Filter
Element
Stack
Stack
Gas
Internal Filter
Figure 1-5. Probe-Tip Filter Arrangements
42
-------
in Table 1-10, and the pressure drop/flow characteristics for each grade
/. ' are shown in Figure 1-6. For a flow rate of 2 SLPM (.07 CFM) and the
smallest available probe filter area (0.016 sq. m.), the flow through
the filter is 118.9 1/sq m. The corresponding pressure drop for the
2
finest available porosity is on the order of 68.95 N/nr for the clean
filter. The filters can be regenerated to a certain extent by back
flushing, and for badly clogged filters a chemical cleaning procedure
is recommended. No information is available on pressure drop as a function
of particulate loading in combustion stack gas monitoring. These filters
are relatively inexpensive: a price of $24.45 was quoted for the cylindri-
cal filter with 0.016 sq m. area in grade D porosity.
For the internal arrangement shown in Figure 1-5, the sample passes
through a probe tube before entering the filter. Any one of a number of
in-line filtering devices may be used. A Western Precipation alundum
thimble holder with a medium porosity thimble (pore size = Sum) has per-
formed very well in laboratory tests. This device has an advantage over
some other filter configurations in that it permits high particulate load-
ing with only a minimal increase in pressure drop.
The pressure drop across a clean alundum thimble is negligibly
small for all normal sampling flow rates. The pressure drop for a thimble
loaded to capacity with 71 grams of fly ash was measured experimentally
and is shown in Figure 1-7. At a flow rate of two standard liters per
minute (enough to simultaneously feed two analyzers) the pressure drop is
only about three inches of water.
The 71 gram capacity of the standard thimble corresponds to an
operational period of 180 days assuming, as a worst case, iso-kinetic
sampling at a rate of 2 SLPM of stack gas containing the typical maximum
particulate concentration of .125 mg/sl (Table 1-2). Thus, the alundum
thimble should be able to operate for long periods without back flushing
and without changing the filter element. Other filters and filter holders
43
-------
TABLE 1-10
FILTERING PROPERTIES OF SINTERED STAINLESS STEEL
Removal Ratings, Microns
Grade
C
D
E
F
G
H
Mean
Pore Size
(microns)
165
65
35
20
10
5
When Filtering Liquids
Nominal
(98%)
55
22
12
7
3
2
Absolute
(100*)
160
55
35
25
15
12
When Filtering Gases
Nominal
(98%)
45
8
4
1.3
0.7
0.4
Absolute
(100%)
110
20
,
11
3
1.8
1.0
68.95
(kN/m2)
6.895
.6895
.3048
3.048 ~ 30.48
(kl/min/nT)
304.8
Figure 1-6. Flow vs Pressure Drop for 3.18 mm Thick Sintered Stainless Steel
44
-------
ALUNDUM FILTER
24
20
16 -
FLOW
12
8
LOADED WITH 71 GRAMS
OF DUST
Figure 1-7. Pressure Drop Across Alundum Thimble
Filled with Fly Ash.
2.49
4.98 7.47 9.97
A PRESSURE ACROSS FILTER
(kN/m2)
12.46
14.95
45
-------
may have a somewhat lower capacity and require more frequent maintenance.
However, the above results indicate that, in general, back-flushing the
filter is unnecessary. Therefore, the use of automatic blow-back, as
in many commercially available systems, is not recommended for a minimum
design since it adds cost, complexity, and maintenance problems while
providing only a minimal increase in operating convenience.
Since combustion gases contain significant water vapor, it is
very important that the filter be kept above the water dew point; other-
wise, it will quickly become clogged with condensate. If the filter is
placed outside the stack it may be necessary to electrically heat the
filter or at least to insulate it in order to prevent condensation. This
can be circumvented by placing the filter inside the stack where it is
kept at the temperature of the gas sample.
An interesting alternative to the alundum thimble holder is a
plug of pyrex glass wool contained in a holder made from ordinary stainless
steel pipe fittings. While such a filter may be inexpensive and quite
efficient in removing particulates, packing the glass wool to the correct
density is a matter of trial and error. Laboratory experience has shown
that a straight-through glass wool filter has a significantly higher
pressure drop under loaded conditions than an alundum thimble.
Of the two arrangements shown in Figure 1-5, the internal arrange-
ment is preferred since it is much easier with this configuration to inject
calibration gases upstream of the filter. For the external arrangement an
elaborate injection system would have to be built in order to pass cali-
bration gases through the filter. The internal arrangement using a alundum
thimble was demonstrated in field tests.
*
At the time of publishing this report, Balston Inc., Lexington,
Massachusetts is bringing out a cylindrical glass fiber instack filter.
This filter has about the capacity of an alundum thimble plus has a
higher collection efficiency (90% at .6 urn) and reputedly a lower pressure
drop. The price of the filter holder 1s also lower than for the alundum
thimble holder.
46
-------
CALIBRATION
The sampling Interface requirements regarding calibration are that the
system shall be automatically recalibrated once every 24 hours, and that the call
bration gases (zero and span) shall be introduced upstream of the probe tip
filter. The rationale behind this second requirement is that if any losses
are occurring in the sampling system, e.g., catalytic oxidation of SIL at the
probe tip filter, the same or similar losses will be experienced by the cali-
bration gases and will be corrected for in the calibration.
Three methods were considered for calibration by injection at the
probe tip:
(1) Gas flooding the filter, i.e., injecting span and
zero gas ahead of the filter to a pressure greater
than the stack gases.
(2) Valving off the stack gas and then injecting span
and zero gas.
(3) Spiking the sample gas with a known high concentra-
tion of the test gas - known addition method.
The method suggested by (2) is the one selected for development.
This approach represents some economy of span and zero gases over (1), gas
flooding.
The method suggested by (3) has the advantages of calibrating the sys-
tem with the minimum of span gas and using the stack gases as background.
However, there are implementation problems with this method in that the con-
centration increase in the sample stream is given by the following:
where: AC is the concentration increase
C1 is the span concentration
C is the sample concentration
q is the span gas flow rate
Q is the gas flow through the filter
47
-------
Therefore, both q and Q must be accurately known to assure calibration accu-
racy. This would require additional flow control equipment and instrumenta-
tion adding to the complexity and cost of the system.
Method (2) has been implemented for laboratory tests as shown in Fig-
ure 1-8. A check valve is used to interrupt the sample flow when calibration
gases are injected. This arrangement has worked well during the short dura-
tion of the tests. One problem that may be anticipated is the failure of the
check valve to seat because of particulate buildup. This may be circumvented
by using a butterfly type check valve as shown in Figure 1-9. This valve is
routinely used in the pneumatic conveying of particulate solids (flour,
starch, cement, etc.) and is available in PVC with a tygon seal ($22) for
use below 82.2°C and in 316 SS with a viton seal for use below 204°C ($78).
Teflon seals are also available for use above 121°C.
Even if the check valve fails to close completely, the calibration
system of Figure 1-8 can be used by operating in the gas flooding mode. It is
only necessary to make sure that the calibration gases are delivered at a
pressure higher than the stack pressure.
The arrangement of solenoid valves for the automatic introduction of
calibration gases is shown in Figure 1-1. One two-way solenoid valve is used on
each calibration gas and a three-way valve is used on the line to the stack. The
three-way valve is optional and may be eliminated. As will be pointed out in the
subsection entitled System Response Time, it is unnecessary from the viewpoint of
response time considerations to reduce the dead volume of the calibration line.
A number of options are available in the timer circuit used to ac-
tuate the solenoid valves. One simple possibility, which will be used in
the field tests, is shown in Figure I-10. Two timers are used; timer #1 has
f 24-hour cycle with a 15 minute on-time, timer #2 has a cycle time of 33
Problems with check valves sticking (closed as well as open) was found
in the field tests. Experience indicates that gas flooding is more reliable
and probably justifies the high consumption of calibration gases.
48
-------
SHUT-OFF
VALVE
NOZZLE
\
CALIBRATC GAS
CERAMIC
Fl LTER
•> SAMPLE TC INSTRUMENT
10
DIRECTION
OF GAS
STEAM
CHECK
VALVE
Figure 1-8. Through the Probe Tip Filter Calibrate System.
-------
en
o
Figure 1-9. Technocheck Check Valve (Techno Corporation, Erie, Pa.).
-------
minutes and Independently times four events. The operating sequence is given
in Figure 1-10. Calibration can also be done manually using the toggle switches
shown, and pilot lights indicate when calibration gases are flowing.
51
-------
S4
S3
—cr
CAM '
PPCGRM-1ER
•—f
---H
o-1
CLOCK
TIMER
PILOT
COMMON
O
LINE
Figure 1-10. Schematic of Timers for Automatic Calibration.
52
-------
Figure I-10 (continued). Sequence of Operations.
(1) Clock timer - 24 hour cycle. On for 15 min, off for 23 hours
45 min. Actuates cam timer.
(2) Cam programmer - 33 minute cycle
Contact #1: On 30 min, off 3 min. Allows motor of cam
timer to operate after clock timer opens. Actuates three-
way solenoid valve (V-l).
Contact #2: On for 10 min, off for 23 minutes. Oper-
ates solenoid valve V-2.
Contact #3: Off for 10 min, on for 10 min, off for 13
min. Operates solenoid valve V-3.
Contact #4: Off for 20 min, on for 10 min, off for 3
min. Operates solenoid valve V-4.
53
-------
SIZING OF SAMPLE LINES
Two considerations must be kept In mind when specifying the size of
sample lines. For a given required flow through the line the diameter must
be large enough so that the pump can handle the pressure drop and yet small
enough so that the response time is not excessive.
Flow rate as a function of pressure drop is shown for various line
sizes in Figure 1-11. The assumptions and equations used in calculating these
results are given in Appendix B. For a flow rate of two standard liters per
minute, sufficient to simultaneously feed two analyzers, 6.35 mm o.d. tubing
gives a pressure drop between 1379 and 2758 N/m2 per 30.48 m of tubing (de-
pending on the tube wall thickness). This pressure drop is quite acceptable
for most sampling pumps. 3.175 mm tubing could conceivably be used
but it is less convenient to work with and may not be readily available in
some materials of construction (e.g., heat traced Teflon). It is therefore
recommended that 6.35 mm o.d. sampling lines be used.
The response time of the sample line may be simply calculated as-
suming no wall effects and no axial dispersion. The lag time, t, for a
sample line of volume V (liters) and flow rate F (liters/min) is:
t = £ (1)
The flow rate, F, is related to the standard flow rate, FS, as follows:
F = F 1-8T+ 492 3774.3 (2)
S 552P + 37743
where T is the gas temperature in °C and P is the pressure in N/m2 gauge. Table
11 gives lag times for 30.48 m of sampling line with various inside diameters
"or flow rates of 1 and 2 SLPM at 25°C and 202707 N/m2. Lag times for other
rlow rates and lengths may be obtained by ratio. For the 6.35 mm o.d.
cubing a lag time of about 30 seconds or less may be expected for a 1 SLPM
flow. This 1s considerably less than the allowable eight minutes for 95%
response. It would therefore appear that 6.35 mm o.d. tubing is acceptable
both from the point of view of pressure drop and response time for flow rates
on the order of 1-2 liters per minute.
54
-------
.2261
2.261 22.61
Pressure Drop (kN/m2/100 m)
90.44
Figure 1-11. Flow Rate vs Pressure Drop for Various Line Sizes.
55
-------
TABLE 1-11
SAMPLING LINE LAG TIME
Lag Time per (30.48 m) Length
1 st. liter per min . 2 st. liter per
Tubing Size 1 st. liter per min . 2 st. liter per min
1.6:1 (3.175 o.d. x .762 wall) 3.58 sec 1.79 sec
3.175 (3.175 i.d.) 13.26 sec 6.63 sec
4.318 (6.35 o.d. x 1.016 wall) 24.54 sec 12.27 sec
4.572 (6.35 o.d. x 0.889 wall) 27.51 sec 13.75 sec
4.826 (6.35 o.d. x .762 wall) 30.64 sec 15.32 sec
6.35 (6.35 i.d.) 53.06 sec 26.53 sec
7.747 (9.525 o.d. x .889 wall) 1.32 m1n 0.658 min
9.525 (9.525 i.d.) 2.00 min 0.995 min
10.92 (12.70 o.d. x .889 wall) 2.62 m1n 1.31 min
12.70 (12.70 o.d.) 3.54 min 1.77 min
56
-------
HEAT TRANSFER FROM SAMPLE LINES
An exact solution to the rate of heat transfer from sample lines
would be quite complicated and would probably not be justified for present
purposes. Reasonable approximate results can be obtained by simplified pro-
cedures. The assumptions, equations, derivations, and calculational proce-
dures are given in Appendix C.
Three specific calculations are made in Appendix C:
(1) The rate of cooling of combustion stack gases is determined so
that the feasibility of using less heat resistant materials such as poly-
ethylene and polypropylene can be assessed.
(2) A calculation is made to assess the feasibility of using un-
heated sample lines in subfreezing ambient.
(3) The relative resistance to heat transfer of the tube material
is determined.
Plastic materials such as polyethylene and polypropylene cannot be
used in the near vicinity of the sampling probe on combustion sources be-
cause of their heat limitations (approximately 93.3°C). However, the calcula-
tions of Appendix C indicate that the sample gas cools quite rapidly. For a
flow rate of 2 SLPM through 6.35 mm o.d. x 1.016 mm wall tubing with a stagnant
ambient at 37.8°C, the sample cools from 260°C to 93.3°C in less than .61 meters of
tubing. Therefore, the lower heat resistance of plastic materials is not a
serious disadvantage and they can be used for all but the first few meters of
the sample line. It should be noted that the length required for a given
temperature change is proportional to the flow rate.
The use of unheated sample lines in subfreezing ambients is, in gen-
eral, not feasible when the gases contain condensible moisture. The calcula-
tion of Appendix C shows that the gas sample cools from 260°C to 0°C in less
than .61 meters for flow at 2 SLPM through 6.35 mm o.d. x 1.016 mm tubing in
a 16.09 Km/h wind at -6.67°C. In order to prevent freeze-up and eventual plug-
ging, the sample line must be maintained above 0°C by heating.
57
-------
An Interesting possibility for shortening sample lines is to locate
the analyzer in a heated cabinet in the near vicinity of the sampling loca-
tion and to run the output signal to the instrument house. This would be a
less convenient location, however, from the standpoint of operation and main-
tenance. On the other hand, it would eliminate the need for long sample
lines and in such cases, it may be possible to avoid the requirement of
heated sample lines. Short sample lines could be insulated to a sufficient
thickness, or sample flow rate could be increased to avoid freeze-up. This
alternative may be useful in certain installations but is not recommended in
general because of the less convenient location.
The third calculation given in Appendix C compares the resistance to
heat transfer of the tube wall with the internal film resistance. For a stain-
less steel tube (1.016 mm wall thickness), the resistance to heat transfer
is only about 0.15% of the internal film resistance. For a teflon tube of
the same dimensions, the tube resistance is about 10% of the internal film
resistance. Thus, the tube material has little effect on the rate of cool-
ing of sample gases. This is particularly significant in relation to the
refrigerated condenser, i.e., it is unnecessary to use a stainless steel cooling
coil to obtain high heat transfer rates. Teflon or other plastics will pro-
vide nearly equal rates.
In general, it is recommended that low cost plastics such as
polypropylene be used for the minimum sampling system wherever con-
densate freeze-up is not a potential problem, e.g., for ambient tempera-
tures > 0°C; for dry stack gases; and for indoor lines. For locations
where condensate freeze-up may occur, heat traced sampling lines are rec-
ommended. Heat traced teflon is commercially available in packaged form
and probably represents an overall economy over fabrication of heat traced
lines from cheaper materials.
To prevent freeze-up, it is only necessary to maintain the sample
line above 0°C. This can be done using a simple variac and a bimetallic
thermometer in the sample line. A given variac setting will correspond to
approximately a constant temperature differential between ambient and the
sample line. Precise temperature control is unnecessary.
58
-------
The advisability of keeping the entire sample line above the sample
dewpoint will be considered in the section entitled Sample Interaction with
the Interface System. In general, this practice is not necessary and is
recommended for a minimum system.
59
-------
SYSTEM RESPONSE TIME
An important requirement of the sampling system is that its 95%
response time must be less than eight minutes. In general, the sampling
system can be divided into plug-flow volumes, dead-end volumes, and
back-mixing volumes. The overall system response depends on the type
and number of these volumes.
Plug-Flow Volumes
A plug-flow volume is one in which the length-to-diameter
ratio is large (~ 10 or greater) and through which the gases pass at
a relatively high velocity (on the order of feet per second). Examples
of plug-flow volumes are sample lines, condenser coils, and tubing con-
necting the interface components. For plug-flow volumes, axial dis-
persion can be neglected, and the response time is simply the volume
divided by the volumetric flow rate in consistent units [see Equation
(1)]. The 95% and 100% response times are the same since it is as-
sumed that axial dispersion is negligible.
When several plug-flow volumes occur in series, the over-
all system response time is the sum of the response times of the indi-
vidual volumes. An actual sampling system will contain a combination
of both plug-flow and back-mixing volumes. The effect of each plug-
flow volume will be to add a constant time increment to the system
response equal to its volume divided by its flow rate. The overall
response lag of the system due to plug-flow volumes is therefore de-
termined by summing the individual plug-flow volumes and dividing by
the volumetric sample flow rate.
Dead-End Volumes
A dead-end volume is a volume which is connected to the main-
stream but through which no bulk flow passes. Dead-end volumes can
potentially affect the response time of the system. An example of a
dead-end volume is the calibration line in Figure 1-1. If span gas were
flowed through the system before switching to the sample at time zero,
calibration gas will diffuse into the sample line creating a potential
response lag.
60
-------
It is possible to calculate the rate of diffusion from cali-
bration line (or from any dead-end volume) to the sample line. This
calculation is given in Appendix D for the following situation. At
times less than zero, SOg span gas is flowed through the calibration
line and through the sample line to the analyzer. At time zero sample
containing no 502 ""s drawn through the sample line and the calibration
line becomes a dead-end volume which is assumed to be infinitely long.
The calculation determines the time required for the concentration of
S02 (leaving the junction of the calibration and sample lines) to drop
to 5% of its initial value. For a sample flow rate of 1 SLPM and a
6.35 mm x 1.016 mm calibration line, the 95% response time of the dead-
_3
end volume is about 10 seconds. This value is reasonable since the
sample flow rate is so much greater than the diffusive input from the
dead-end volume. This result indicates that under all practical con-
ditions, dead-end volumes have a negligible effect on the system
response.
Back-Mixing Volumes
A back-mixing volume is one in which the length-to-diameter
ratio is small and through which the linear velocity is small. In this
case, axial dispersion tends to mix the contents and the simple plug-
flow equations are invalid. Examples of mixing volumes are the con-
den sate removal trap, the cell of some analyzers, and the sample pump
when piped as shown in Figure 1-1. This sample pump arrangement is a
mixing volume, since the discharge to suction recirculation is large
compared to the input and output from the recirculation loop.
Since all plug-flow volumes can be summed to give an over-
all system plug-flow volume which adds a constant lag to the instrument
response, and since dead-end volumes can be neglected, the remaining
mixing volumes can be considered to be connected in series, end-to-end
with neglibible transport time between volumes. Then, the overall sys-
tem response is the response calculated for the required number of
sequential mixing volumes plus the time lag due to the plug-flow volumes.
61
-------
The response characteristics of mixing volumes are treated
in detail in Appendix E. The response characteristics of a given vol-
ume depend on its time constant, T:
i=f (3)
where T = time constant (min)
V = volume (liters)
F = actual volumetric flow rate (LPM)
At times less than zero, the concentration is constant at a mole frac-
tion of Y. in all volumes. At time zero sample containing zero con-
centration is flowed into the first volume of the sequence. For the
special case in which the time constant is the same for each consecutive
volume, the response characteristics follow the general equation:
YN / N i-n \
Y7 - (l + £ HT^) «P <-'/*> <«>
where Y^ = mole fraction leaving Nth volume
Y. = initial mole fraction at t = 0
N = number of consecutive mixing volumes
t = time
T = time constant
For the more general case when the time constants are not
equal, the results are considerably more complex and do not lend them-
selves to presentation in generalized form. Equations for one, two, and
three consecutive mixing volumes are given in Appendix E.
A sample calculation of system response is shown in Appendix
E for the following situation:
Plug-Flow Volume: 60.96 m of sample line and interconnecting
tubing of 6.35 OD x 1.016 mm wall.
Back-Mixing ,
Volumes: Pump and recirculatiqn loop = 200 cm
Trap volume = 500 cnr
Analyzer cell = 400 cm3
62
-------
Sample Flow Rate: 1 1PM
The 95% response time calculated for this hypothetical system
is 3.28 minutes, 27% of which is due to plug-flow volumes and the re-
mainder to mixing volumes. Response time may be reduced either by in-
creasing the flow rate or reducing the volume. In this respect, it is
important not to use greatly oversized components in the sampling inter-
face. In addition, the pump recirculation loop should be kept as small
as is practically possible.
If response time problems are encountered in the field
tests, the sample flow rate will be increased and the system will be
modified to by-pass the excess to vent as shown in Figure 1-2. If this
contingency proves to be necessary, only enough sample will be by-
passed by decrease the system response to the eight minute limit.
63
-------
PUMP REQUIREMENTS
Diaphragm and bellows pumps are generally conceeded to be
superior to other types of pumps for gas sampling, at least when
positioned upstream of the analyzer. Their primary advantages are:
- they do not require a shaft seal and are therefore not
subject to potential failure of the seal and contamination
of the sample by air in-leakage;
- they do not require internal lubrication which could con-
taminate the sample;
- they develop quite adequate suction and discharge heads at
flow rates well above those required for sampling systems;
- they are relatively inexpensive.
When the pump is placed downstream of the analyzer, and the sample is
"pulled" through the entire system, a simple water on air aspirator can be
used. These have a distinct advantage over the diaphragm or bellows pumps
in the fact that they have no internal working parts and are, therefore,
less prone to failure. However, they require a continuous supply of
regulated air, steam, or water which may not always be conveniently available.
As pointed out in the subsection entitled Positioning of Sample Pump, systems
operating with the pump downstream of the analyzer must make some provision
for regulating the pressure (e.g., a pressure regulator or vacuum breaker)
at the analyzer. For analyzers with a built-in pressure regulator (e.g.,
TECO), the air aspirator may be preferrable to mechanical pumps.
When positioned upstream of the analyzer, the sampling pump should
be capable of pulling a reasonable vacuum at the suction and providing a
reasonable positive pressure at the discharge when operated at the intended
sample flow rate. The higher the suction and discharge heads, the longer
t •• system can operate without changing filter elements. In general, high
vacuums should be avoided at the suction side, since this increases the
po3sibility of air in-leakage. Minimum specifications for the pump are
flexible, but it is advisable to require at least 2 SLPM into 20.68 k N/m2
at a suction of -13.79 k N/m .
64
-------
Pump performance curves are shown for small diaphragm pumps In
Figure 1-12. It is apparent that these pumps considerably exceed the
minimum specifications (the manufacturer supplies smaller pumps, but
at the same cost). For a suction head of -13.79 k N/m (4 in Hg vac.) and
2
a discharge head of 20.68 k N/m , the pump delivers about 8.5 SLPM. This
is plenty of flow to cover contingency plans for sampling at higher flow
rates.
The control of pump discharge pressure and flow rate should be
implemented in accordance with the manufacturer's recommendations. The
two most common methods of control are a throttle valve on the pump
discharge and a by-pass valve from pump discharge to suction (see Figure 1-1).
The latter method is recommended for diaphragm pumps when operated
significantly below the pump curve (i.e., at flow rates considerably
below what the pump will deliver at a given suction and discharge head).
Severely throttling the discharge causes the pump to work against a
much higher discharge pressure and considerably reduces pump life.
When placed upstream of the condensate trap the sampling pump
must be capable of passing liquid condensate. The pump should be Icoated
above the condensate trap so that condensate can drain away by gravity flow.
Field experience reportedly shows (5) that sampling pumps have
a high rate of failure because of corrosion and particulate abrasion.
Other factors being equal, pump selection should be based on reliability
and maintenance consideration. This requires field-test information for
various types of pumps. For the field demonstration portions of this
program a diaphragm pump (Thomas Industries) was selected since it is
available with completely Teflon-coated internal construction and is more
resistant to corrosion than the 316 SS internals of the bellows pump.
Problems were encounted during the field demonstration with particulate
plugging of the diaphragm pump as described in Part II, page 21 of
this report. These problems could probably have been eliminated by position-
ing the pump as shown in alternative B of Figure 1-4 rather than alternative
A. Others (37) have also reported problems with diaphragm pumps in
sampling applications.
65
-------
i:».82r
16.99}-
14.16
19.82.
-PRESSURE '-5 VOLUME
T ' -i. I07CA H
101.4 .135.8 170.3 204.8
(k N/m2)
239.2
19.88
16.99
14.16
11.33
8.50
5.66
2.83
(1/m)
f
PRESSURE vs VOLUME
MODEL I07CAI8
101.4 . 135.8
(k N/nT)
170.3
204.8 239.2
273.7
273.7
Figure 1-12. Pump Output vs Pressure (Thomas
Industries Inc.)
PRESSURE VS VOLUME
MODEL IG7CAI6 ,
(1/m)
101.4 135.8 170.3 204.8 239.2 273.7
(k N/m2)
PR ESS U RE v* VOLUME
MODEL I07CA20
239.2 273.7
(k N/m2)
-------
MOISTURE REMOVAL
Combustion stack gases contain significant quantities of water
vapor which condense out as the sample cools from the stack temperature
to room temperature. For some analyzers which are insensitive to water
vapor and can operate at high temperatures (e.g., the UV instruments)
the problems of moisture removal can be avoided by keeping the tempera-
ture of the sample above its dew point. This will, of course, require
heated sample lines and heated filters and possibly, depending on place-
ment, a heated pump. Use of a totally heated system appears to be in-
consistent with the criteria for a minimum design, but there may be situa-
tions in which it achieves an overall cost minimum.
For analyzers which cannot operate at very high temperatures,
moisture removal is a necessity. This is most readily accomplished by
allowing the sample to cool as it passes through the sample line. It
is very important to pitch the sample lines in a downward direction so
that condensate drains away from the probe tip toward the condensate trap.
It is also advisable to avoid low spots in the line where condensate can
collect.
Some commercially available systems maintain the sample above
its dew point until it reaches the condenser,at which point it is rapidly
cooled. This is done primarily to avoid solubility losses of the species
being measured. The advisability of this practice will be considered in the
section entitled Sample Interaction with the Interface System.
Some instruments, the NDIR's in particular, are sensitive to water
vapor. This sensitivity can be circumvented by one of three methods: (1)
the use of optical filtration within the analyzer, (2) removal of all
water vapor, or (3) the use of a constant water level in both the sample
and calibration gases. The use of optical filtration is a function of
analyzer design and will not be considered further.
67
-------
Substantially all water vapor can be removed by the use of
dessicants or permeation dryers. It is of interest to determine the
amount of water vapor that must be removed in order to operate within
the accuracy limits of +10% of span. The effect of water vapor will
increase as lower concentrations of S09 or NO are measured. The mea-
£ X
surement of NOX from a gas fired power plant represents a "worst case"
since the typical maximum emission limit is only 127 ppm (Table 1-2). A
typical water sensitivity for an NDIR instrument is on the order of 100 to
1. That is, 100 ppm water registers as 1 ppm on the analyzer. For this
sensitivity the water level must be reduced to 1270 ppm which corresponds
to a dew point of -17.78°C.
The use of a constant moisture level in both the sample and cali-
bration gases is the most common method of eliminating the effects of
water vapor. This can be done by passing both through a refrigerated
condenser controlled at 1.67°C.
Care should be taken not be dry out the condenser by operating
for long periods with calibration gases. Under most circumstances this
would not be a problem since a flow of dry gas at 2 SLPM would pick up
only about 0.6 ml of water per hour at 1.67°C.
Refrigerated Condenser
Refrigerated condensers are available in packaged form (e.g.,
Hankison Corporation) for cooling sample gas streams. A number of other
temperature-regulated cooling devices could be easily modified for use.
One particularly interesting possibility, in relation to a minimum design,
is the use of a small domestic-type refrigerator containing a cooling
roil for the sample and the condensate trap. Lines could be fed into
id out of the refrigerator by removing a small piece of molding around
the refrigerator door. As pointed out in the subsection entitled Heat
Transfer from Sample Lines, a Teflon coil (or other plastic) could be
used with only a slight loss in heat transfer rate.
68
-------
The heat requirements for cooling and drying the sample are
very small. As shown in Table 1-2 the maximum typical moisture content
expected for Category I sources is 16% for a gas-fired power plant.
For a sample flow rate of 2 SLPM the cooling requirement is 23 k joule/hr
(see Appendix F). This requirement increases in direct proportion to
flow rate.
The other quantity of importance in the design of the con-
denser is the surface area required to cool the gas. This depends to
some extent on the internal construction of the condenser. If it is
assumed that the condenser is capable of maintaining the outside wall
of the cooling coil at 1.7°C, then the coil length is determined by the
internal film resistance to heat transfer. In general, the heat transfer
rates can be calculated from the information in Appendix C. Enough
surface area must be provided to remove the latent heat of vaporization
in addition to cooling the gas stream.
Dessicants
Dessicants are inconvenient and expensive for removal of
water vapor. Even if used after a refrigerated condenser to remove the
remaining 0.68% water vapor, frequent maintenance is necessary. For
example, a commercially available Drierite column containing .567 kgm
of Drierite and placed at the condenser outlet would last only three days
at most. In addition, there is the possibility of interactions with
constituents of the sample stream other than water. Some loss of S(L on
dessicant materials has been observed (6). Therefore, dessicants are not
recommended for use in sampling systems.
Permeation Dryers
Permeation dryers (e.g., Perma Pure Products, Inc.) have been
used with success (18) in drying stack gases. Stack gases are passed
through the tube side of a tube-and-shell-type dryer in which the tubes,
made of ion-exchange membranes, selectively pass water but retain other
69
-------
stack gas constituents. The driving force for water permeation is provided
by either evacuating the shell side of the dryer or purging the shell side
with dry air. The sample must be kept above its dew point since permeation
occurs from the gas phase. Thus the sample line must be heat traced up to
and including the first 15 cm of the dryer. At flows of 2 to 3 SLPM, water
reductions of 2000 to 1 are claimed. Dew points of -18°C and below are
routinely obtained.
Several persuasive advantages can be listed for the permeation
dryer: (1) materials of construction do not come into contact with liquid
condensate and are therefore less prone to corrosion; (2) a condensate trap
is not required for removal of liquid; (3) there is no possibility of sample
loss by solution in liquid condensate; and (4) it is competitively priced with
refrigerated condenser units. The disadvantages are that some problems have
been encountered with particulate plugging in field tests on combustion sources,
and that the operation of a permeation dryer is somewhat less convenient than
the use of a refrigerated condenser. A vacuum pump and controls must be main-
tained on the shell side of the dryer, or a regulated supply of dry purge air
must be used. If the dryer is located in the instrument house the entire
sample line must be kept above the sample dew point. If located in the vicinity
of the stack, the heat tracing problem is reduced but the accessories and
controls (vacuum pump, pressure gauge, flow meter and valve) would be in a
less accessible location. Long lines could be run to and from the dryer but
"•hese are again inconvenient. Nevertheless, the permeation dryer offers a
very interesting alternative, and in many cases, particularly for dust-free
stacks, the advantages may be over-riding.
A commercially available refrigerated condenser was selected
ior the field demonstration according to the following rationale. Some
dc, bt still remains as to whether the permeation dryer is suitable for
use on combustion stacks because of particulate plugging. Also, assuming
the corrosion rate for the condenser is acceptably slow and the solubility
70
-------
losses are small (see section entitled Sample Interaction with the Inter-
face System) the over-riding advantages for the condenser become its
simplicity of operation and the lack of heat traced sample lines held above
the sample dew point. The domestic type refrigerator certainly provides
the most economical approach in terms of initial investment, but both the
time required for modifications and the reliability of its service are
questionable.
Removal of Condensate
Condensate can be removed from the sample line by allowing
it to collect in a volume or trap while the gas passes on to the analyzer.
This trap can be drained manually, as part of a weekly maintenance routine,
or automatically. For a flow rate of 2 SLPM the rate of condensate accumula-
tion is about 18 ml per hour (see Appendix F), assuming a total moisture
content in the feed of 16%. This amounts to slightly over three liters
collected per week. Thus the trap could consist of a simple gallon con-
tainer that is manually drained each week. The advantages of this alter-
native are that the trap would be very cheap and simple and would not be
prone to mechanical failures. The primary disadvantage is that such a
large volume would add considerably to the response time of the system.
In addition the trap would be prone to human neglect and possible overflow.
However, this alternative may achieve a cost minimum particularly where
the stack gases contain less moisture and a smaller trap volume can be
used.
A simple device for automatically removing condensate is the
ball-float trap, an example of which 1s shown in Figure 1-13. As condensate
collects,the float rises, opening the drain valve at the bottom of the
trap. Traps of this type are available from steam trap manufacturers but
the materials of construction are typically not compatible with the corro-
sive condensate that must be handled. A number are available with stainless
71
-------
18.73 cm
Figure 1-13. Ball-Float Trap for Removal of Liquid
Condensate (Armstrong Machine Works).
72
-------
steel working parts but with cast Iron or mild steel enclosures. For
most of these traps the capacity Is much greater than required and some
are fairly expensive. Ball-float traps are also manufactured for
removal of condensate from compressed air lines, but again the materials
of construction are not particularly corrosion resistant. Although an
extensive search was not conducted, the ball-float trap of Figure 1-13
is the only one found at a reasonable cost that appeared to be suitable
for the present application. The potential disadvantage of the ball-float
trap is that corrosion and/or particulates may prevent the drain^valve
from seating properly and thus cause sample loss into the instrument house.
A second alternative for automatically removing condensate
is to have an automatically controlled drain valve which is actuated either
by a level switch within the trap or by a timer. An external timer is
more reliable than inexpensive level switches and less expensive than
sophisticated level-detecting devices. A reasonable time cycle can be
easily calculated from the sample flow rate, moisture content, and trap
volume.
When the trap operates under positive pressure as in alterna-
tive A of Figure 1-4, condensate is removed simply by opening a valve to
atmosphere. When operated under negative pressure as in alternative B
of Figure 1-4, removal is not so simple. For manual removal, on a weekly
basis,'the pump could be shut down while the trap is drained, or the trap
could be pressurized with compressed air. For automatic removal, a trap
with a barometric leg could be used as shown in Figure 1-14. For a pump
2
suction of 13.79 k N/m vacuum, a length of 1.4 m would be required for the
barometric leg. In fact, if constructed of glass or transparent plastic,
the trap could be used to measure the sample line pressure and thereby
the pressure drop across the probe-tip filter. This trap has an additional
advantage of containing no internal working parts to malfunction. The
disadvantage of the barometric trap is that it is not very compact and
not very flexible in terms of the operating pressure range. However,
contingent plans for the field demonstration include the use of such a
trap if alternative B of Figure 1-4 is found necessary.
73
-------
a,
L
Figure 1-14. Condensate Trap Using Barometric Leg.
-------
For traps In which a drain valve is actuated by a level
switch or a timer, a solenoid valve must be simultaneously actuated to
admit compressed air to the trap forcing the condensate out against
atmospheric pressure.
5. AP-AT Relation
As pointed out in the subsection entitled Positioning
of Sample Pump, moisture removal is a function of the trap temperature
and pressure. If the trap is located upstream of the pump and if it
operates at the same temperature as other components in the interface,
then further condensation can occur when the sample goes through the
pump and its pressure is increased. It is a simple matter to calculate
the temperature increase required to prevent condensation for a given
pressure increase. Figure 1-15 shows the required AT for a given AP
below atmospheric. For example, if the trap operates at 25°C and
-34.47 k N/m^ while the analyzer is operating at atmospheric pressure,
the analyzer must be operated at 7.2°C above trap temperature or 32.2°C
in order to be sure that no condensation will occur in the analyzer.
75
-------
0.5: • j
AP (k N/m ) (Pressure of Analyzer-Pre^surs of Trap)
Figure 1-15. AT vs AP for Trap and Analyzer.
76
-------
Fine Filtration
The participate removal requirements for the analyzers listed in
Table 1-5 generally call for substantially complete removal of particles above
one micron in size. This is best accomplished by a fine filter in the
vicinity of the analyzer. Filters can be divided into two broad categories:
surface filters and depth filters.
Surface filters remove particulates by presenting a fine porous struc-
ture to the gas streams. These pores prevent the passage of particulates
which collect at the surface of the filter element. Examples of surface fil-
ter elements are paper, polymer membranes, metal membranes, porous metals and
ceramics, etc. Depth filters, on the other hand, collect most of the par-
f.culates within the bulk of the filter element. Examples of depth filters
are loosly packed fiberous materials and relatively large diameter granular
materials.
Surface type filters are best suited for use on dry solid particulates.
When a surface layer of these particulates builds up on the filter the increase
in pressure drop will not be excessive provided the surface layer remains dry
and porous. However, if the surface layer of particulates becomes moist or
if the particles are gummy and can coalesce, the filter quickly becomes fouled
and the pressure drop becomes excessive. Surface filters when used in gas
filtration can generally remove particles smaller than the pore size of the
filter element. This is due to electrostatic capture of smaller particles
and may result in retention of particles an order of magnitude below the pore
size.
Depth filters can be used for gummy solids and for moist gas streams
as well as for dry solid particulates. Therefore, they offer the advantage
of flexibility in case of a system malfunction. A fine grade depth filter
can retain essentially all particulates. Particles larger than about O.Jj^m
can be captured by direct inertial impact with the filter medium while particles
less than about 0.5/m exhibit Brownian motion which greatly increases their
chances of collision with and retention by the filter medium. Large particles
are held by mechanical forces while small particles are retained by van der
Waals forces.
77
-------
There are a number of commercially available surface filters and depth
filters that can be supplied with housings compatible with the corrosive nature
of the sample stream (see subsection entitled Sources of Sampling System Com-
ponents). These can be obtained in porosities that will efficiently remove
particles well below one micron in size.
Efficient filtration can be obtained by using a tube packed with fine
glass wool. Glass wool packing will function as a depth filter although, de-
pending on the packing density, larger particles may accumulate primarily
at the surface. A glass tube holder is preferable since the conditions of
the filter media can be checked visually. Construction details for gas
sampling filters made of glass wool and other materials have been given (14),
but for flow rates considerably greater than required for continous monitoring.
3
A packing density of 0.08g/cm was recommended with a bed depth of two inches
and a linear flow rate, based on the filter cross sectional area, of less
than .61 m/sec. It is important to make sure the glass fibers lie normal to
the direction of gas flow, as far as possible, to prevent channelling. It
may be necessary to use some trial and error in packing the filter to the
proper density. Commercially available filters have an advantage in reliability,
reproducibility, and convenience. Nevertheless, a glass tube filled with
glass wool will be used for the field demonstration because of its obvious
cost advantage.
78
-------
INSTRUMENTATION
As a general rule, instrumentation should be kept as simple as pos-
sible on all equipment intended for operation by plant personnel. This is
particularly true for a minimum system.
It is necessary to measure the flow rate of sample to each analyzer, but
as pointed out in the subsection entitled Control of Sample Flow Rate, it is
usually unnecessary to accurately control the flow rate. Small rotameters of
10 to 20% accuracy are quite sufficient to give an indication of flow. Some
analyzers will have a built-in sample flow meter obviating the need for any
flow instrumentation in the interface.
There are two points at which temperature indication may be helpful:
(1) at the cold end of a heated sample line to make sure the temperature
remains above freezing, and (2) at the refrigerated condenser to set the de-
sired control temperature. Temperature can be simply and effectively
measured by using a bimetallic dial thermometer Inserted into the sample
line through one branch in a tee.
Pressure measurement at the pump suction is useful for determining
the pressure drop across the probe tip filter and sample line. This indi-
cates when the filter element needs to be changed or when the sample line
is becoming plugged. This pressure measurement is not really essential If
the filter is a high-loading type which requires infrequent maintenance.
Pressure measurement at the pump discharge is useful for two purposes:
4
measuring the pressure drop across the fine filter and adjusting the recircu-
lation through the pump bypass to keep the discharge pressure low. Without
this pressure check, the pump could be run at high pressure/low flow con-
ditions which shortens pump life.
79
-------
SPECIFIC SOURCE/ANALYZER COMBINATIONS
Previous sections have presented general design considerations based
:,n the source/analyzer combinations.of Figure 1-1. The conditioning require-
ments for other combinations are generally less demanding. This section con-
siders separately the requirements for monitoring power plants, nitric acid
plants, and sulfuric acid plants in relation to the four types of analyzers
listed in Table 1-5.
Power Plants
The basic sampling system components required for monitoring
combustion sources are:
- Probe tip filter (for particles ^ 10 urn and greater)
- Sample pump
- Refrigerated condenser (NDIR instruments only)
- Condensate trap
- Fine filter (may not be required for UV instruments
depending on pore size of coarse filter at stack)
These components have been discussed in detail in preceding
sections.
Sulfuric Acid Plants
The exact nature of the sampling interface for sulfuric acid
plants depends to some extent on the type of emission control process that
is used. As shown in Table 1-4, typical S02 concentrations in uncontrolled
plants are an order of magnitude above the emission limit. If a dual-ab-
sorption process is used to reduce S(L emissions, the stack gases are dry
and moisture removal components are unnecessary; however, acid mist must
br removed. If a wet scrubber is used to reduce S0«, acid mist removal is
unnecessary, but water vapor must be removed or controlled at a fixed level
for NDIR instruments.
The basic components required for sampling dry sulfuric acid
stack gases are:
80
-------
- Probe tip (no filter required)
- Demister
- Sample pump
- Fine filter
When off-gases from a wet scrubber are monitored, the basic in-
terface components are:
- Probe tip (no filter required)
- Sample pump
- Refrigerated condenser (NDIR instruments only)
- Condensate trap
- Fine filter
It may not be necessary to provide automatic condensate withdrawal when
sampling off-gases from a wet scrubber. If the gases are assumed to be
saturated at 37.8°C, cooling to 25°C,for analysis of a sample flowing at
1 slpm (only one analyzer must be fed),will produce only 270 ml of conden-
sate per week. This could be handled in a relatively small volume on a
weekly maintenance basis.
Of the interface components required for S0~ stack sampling,
the demister is the only one that remains to be considered in detail. The
simplest demister is a large settling volume in which the gas velocity is
low and liquid particles can settle by gravity. This is effective for re-
moving large particles but is ineffective, under practical conditions, for
removing smaller particles. For example, the terminal velocity (29) of a
100 um particle of specific gravity 1.5 is about .3048 m/sec. For a sample
flow rate of 2 slpm, 100 ym particles will settle out in a 1.27 cm di-
ameter tube. However, 10 um particles have a terminal velocity of .3 cm/sec
and a 12.7 cm diameter tube is required for removal.
Cyclone separators can be used to demist the sample by centrif-
ugal force. However, these may be expensive and prove to be too large to
obtain efficient operation with the small flow rates used in sampling.
81
-------
The third type of demister is an impingement separator. When an
obstruction is placed in the path of the gas flow, the gas is easily diverted,
but liquid droplets are carried forward by their momentum, impinge on the
surface, and are collected. The most commonly used industrial separator of
this type is the wire mesh demister. The same effect can be achieved by
using glass fibers or Teflon fibers which are more corrosion resistant.
Taking the typical emission limit of Table 1-4 for sulfuric acid
mist and assuming isokinetic sampling at a rate of 1 slpm, the rate of sul-
furic acid mist collection is only 0.27 grams per week. Even at the much
higher "uncontrolled" emission rate of .131 mg/sl, only 1.3 grams of mist
are collected per week. Therefore, it is unnecessary to continuously re-
move acid mist condensate from the system. This can be done manually on a
monthly or even semi-annual basis.
The demister to be demonstrated in field tests combines both
gravity separation and impingement on glass wool as shown in Figure 16. It
is constructed of ordinary PVC pipe fittings and should be simple and inex-
pensive to fabricate.
Nitric Acid Plants
The sampling of stack gases from nitric acid plants may require special
handling to prevent NOg solubility losses. Sample interactions with the interface
system will be considered in detail in the section entitled Sample Interaction with
the Interface System. It is sufficient for present purposes to indicate that sig-
nificant N02 losses can occur by dissolution in the stack gas condensate.
It is obvious from Table 1-3 that nitric acid plants must use some
type of waste gas treatment to reduce NO output. The two most common types
A
of waste gas treatment are catalytic reduction and wet scrubbing with water or
ce stic. For catalytic treatment, a typical water vapor concentration of
abou . 4% can be expected, and for scrubbers, the stack gases will be saturated
at the scrubber temperature. Thus condensation can be expected for both types
82
-------
Glass Wool Packing
Gravity Separator/Collector
Figure 1-16. Gravity and Glass Wool Demister.
83
-------
of waste gas treatment when the sample is cooled to room temperature.
Furthermore, it appears from the few data (16, pp. 24 and 25) located on
N0« emissions, that NO- can be a significant fraction of the total MO
for both types of waste gas treatment.
Solubility losses of NOp can be prevented by one of two means:
1) the entire interface can be heat traced to keep the sample above its
dew point, and 2) a permeation dryer can be used to selectively remove water
vapor while retaining other gases. It is again difficult, in the absence
of operating experience with these two alternatives, to choose the one which
represents an overall cost minimum. Complete heat tracing is unacceptable
for the NDIR instruments, which are sensitive to water vapor, and for the
Aerochem instrument which cannot tolerate dewpoints above 21°C. Heat tracing
may also be unacceptable for the other chemiluminescent and electrochemical
analyzers if the dewpoint is above their maximum allowable sample temperature
(Table 1-5). Heating tracing is, of course, a viable alternative for the UV
instruments.
Permeation drying, on the other hand, can be used with all
instruments and may represent a cost minimum even for the UV analyzers.
The sampling interface components required are:
- Probe tip (without filter)
- Permeation dryer (with accessories)
- Sample pump
- Fine filter
NDIR Analyzers
The distinguishing feature of NDIR analyzers is their sensitivity to
w cer vapor. As pointed out in the subsection entitled Moisture Removal, this
retires either removal to a dewpoint of below -17.8°C or maintenance of a con-
slant moisture level in both the sample and calibration gases.
84
-------
UV Analyzers
The distinguishing feature of UV analyzers is their high oper-
ating temperature and the manufacturers' recommendation that the sample be
kept above its dewpoint. This mode of operation is really only required
when N02 solubility losses are to be prevented, and significant N02 concen-
trations are encountered only in nitric acid stack gases. For other
applications, the higher cost of a completely heat traced interface, as com-
pared to a room temperature condensate trap, does not appear to be justified,
at least in the short run.
Another noteworthy feature of the UV analyzers is their higher
tolerance of particulates. If sufficient removal is obtained in coarse
filtration, fine filtration is not required.
Electrochemical Analyzers
There are no particularly noteworthy requirements of the electro-
chemical analyzers.
Chemiluminescent Analyzers
The chemiluminescent analyzers are unique in the fact that they
must have some means, internal to the instrument itself, to control the flow
rate of sample to the reaction chamber. The TECO instrument provides in-
ternal pressure regulation and the Aerochem instrument provides internal
flow regulation. For either instrument, it is advisable to follow the in-
terface design recommended by the manufacturer. For example, the TECO in-
strument draws sample through the interface and analyzer using a pump on
the downstream side of the analyzer.
85
-------
The internal pressure regulator maintains the pressure at
2
about -17 k N/m vacuum. Therefore, there Is no option In the placement
of the pump. A pump is typically supplied with both the TECO and Aerochem
instruments. Other design variables such as particulate removal, moisture
removal, sample line materials, etc., can be specified as previously dis-
cussed. For both instruments, moisture can be removed by a room temperature
trap provided the analyzer is kept above room temperature.
86
-------
SOURCES OF SAMPLING SYSTEM COMPONENTS
One of the problems faced 1n Implementing an Interface design is
locating manufacturers of suitable components. There are a number of com-
ponents that can be fabricated without problem, but in many cases the cost
of labor for fabrication makes it less expensive to buy commercially avail-
able components. It should be emphasized that our survey of sampling sys-
tem components was by no means exhaustive. There may be many other manu-
facturers whose products are acceptable or even preferable over the ones
listed in this section. Furthermore, specific field experience has not
been obtained for most of the components listed. Nevertheless, manufac-
turers of products which appear to have potential for sampling systems
are listed in Appendix H. A number of these listings are taken from
Reference 30.
87
-------
INTERFACE FOR FIELD DEMONSTRATION
The major interface design considerations have been presented in
previous sections. Some of the more mundane details of construction can
be best obtained by an examination of the engineering specifications for a
specific system. Appendix G gives the specifications for an interface
using an NDIR and an electrochemical analyzer on power plant stacks.
-------
SAMPLE INTERACTION WITH THE INTERFACE SYSTEM
Regardless of the exact design of the sampling interface, it is
essential that the sample be transported from stack to analyzer with
tolerable losses and interactions. There are several mechanisms by
which interaction can occur, including:
- reaction;
- absorption;
- adsorption; and
- dilution.
Dilution can occur by air leakage into portions of the system which are
under vacuum. Its importance is primarily a function of the quality of
fabrication rather than design, and will not be considered further.
REACTIONS
Gas phase species can be lost both by homogeneous gas phase
reaction and by heterogeneous catalytic reaction on system components
or collected particulates. Reaction losses will be examined for each of
the three species of interest, viz., SOg, NO, and NOg.
SO? Losses
Catalytic oxidation of SC-2 to $03 is the most likely means
of S02 loss by reaction. Reaction will occur only when the gases are
at a high temperature, i.e., at the probe-tip filter of combustion
sources. Materials of construction, such as stainless steel, Teflon,
glass, and ceramics, are generally very poor catalysts and would not be
expected to catalyze the reaction. However, particulates collected at
the probe-tip filter could conceivably be catalytically active, although
this is unlikely. To check the magnitude of potential S02 losses, it is
assumed that the alundum thimble is loaded to capacity with 71 grams of
fly ash, and that the fly ash has the same catalytic activity as a
good commercial S02 oxidation catalyst (platinum or alumina). Assuming
a sampling rate of 2 SLPM and a stack concentration of 1000 ppm S02,
89
-------
the space velocity through the fly ash 1s 7.54 x 10~5 g-moles S02/(hr)
(gram fly ash). Kinetic data are shown in Figure 17 (31) for a com-
mercial S02 oxidation catalyst and a reaction mixture of 1000 ppm S02
and 0 ppm $03. Since the reaction rate is proportional to PQ1» and
p-0.5
$03 , this rate curve represents a severe case. When the data is ex-
trapolated to. a reasonable sampling temperature (204. 4°C), the reaction
rate is 1 x 10'6 g-moles S02/(hr) (g-catalyst), this rate would give
only a 1.3% loss of the SOg in the sample. Since the reaction rate
is directly proportional to the partial pressure of S02 the same
percentage loss would be incurred by more dilute samples. It is there-
fore concluded that catalytic loss of S0£ will be negligible under all
practical sampling conditions.
This conclusion was verified experimentally by loading a
stainless steel filter element with 7 grams of fly ash and sampling 1200
ppm S02 in air at a flow rate of 1 SLPM. This corresponds to a space
velocity of 4.59 x 10"4 g-moles S02/(hr) (g-fly ash). No measurable
loss of S02 was observed for operating temperatures of 21.1°C, 204. 4°C,
and 371. 1°C, as determined by shunting the sample through a "clean" filter.
NO Losses
The oxidation of NO to N02 is a very interesting reaction:
2 NO + 02 t 2 N02
It is one of the few examples of a third-order homogeneous reaction.
That is, two molecules of NO and one molecule of 02 must simultaneously
collide before reaction can occur. The rate of oxidation is:
- ^T- - "1 PNO p02 -
The rate of oxidation is proportional to the square of the NO partial
pressure so that the rate decreases greatly at low NO concentrations.
Lie other interesting fact concerning the oxidation is that the rate
constant k-j decreases as the temperature increases, which is opposite
tc the typical temperature dependence of rate constants. From
90
-------
Figure 1-17. Reaction Rate Data for S02 Oxidation.
1x10
-2
1x10
-3
1000 ppm S02 - 0 ppm SO.
0.2% Pt-ON-A*0
«0
S IxlO-4
o
£
0
1
CM
0
C/J
I/I
01
E
i
cn
IxlO"5
IxlO"6
^^^^p^^— \
— \
^—^— y
E \
- \
\
\
E \
\
\
- \
400CF
I \l
91
io3/T
-------
Equation (5), the rate of oxidation is a maximum when the partial pres-
sure of NOg is zero. Assuming P02 » PNO, then PQ is approximately
constant. Equation (5) integrates to:
(PNO)
- i + k, (PNO)0 P02 t
where
(PNO)0 - Initial partial pressure of NO
PNQ = Partial pressure of NO at time t
P02 = Partial pressure of 02
kj = Forward rate constant
For a "worst case" analysis, an average sample line temp-
erature of 40°F is assumed. At this temperature, k-| is about 32.5 atnr2
sec"1 (32). From Table 1-2, the typical maximum NOX limit is about 500 ppm
and the oxygen partial pressure is about 0.03 atm. The time required
for a 10% conversion of NO to N02 at these conditions is 3.80 min. Thus,
the average residence time for the sampling system must be less than
3.8 min. Average residence time is determined by dividing the system
volume by the flow rate. For a flow rate of 1 SLPM, a system volume
of 3.8 liters is required for a 3.8 min. residence time. For most
sampling systems, this requirement is easily met. For example, 60.96 m
of 6.35 OD x 1.016 mm tubing contains a volume of 0.893 liters. This
leaves approximately 3 liters for other system components.
The conversion of NO to N02 does not really represent a
loss, since the emission limit is set on NOX, and one mole of N0£ is
produced for each mole of NO reacted. However, it will be shown later
tliat conversion to N02 will increase solubility losses. Nevertheless,
t e above calculations show that, under practical sampling conditions,
the loss of NO by reaction (and subsequent absorption) will not exceed
the 10% tolerance limit.
92
-------
The possibility of catalytic oxidation of NO to N02 on
fly ash collected in the probe-tip filter was checked experimentally.
No loss was observed when flue gas doped with NO was passed through a
filter containing fly ash. The control equipment was run with a
clean filter. The sampling temperature was 232.2°C; NO concentration,
1300 ppm; and Q/M, 42 SLPM/gram fly ash.
NO? Losses
Nitrogen dioxide can be reduced to NO and N£ as in the
catalytic process for treatment of waste gas from nitric acid manufac-
ture. However, this requires a catalyst, high temperature, and a re-
ducing atmosphere. While the first two requirements may be met for
combustion stack gases, the use of excess air in combustion rules out
the possibility of a reducing atmosphere in the stack. Therefore,
loss of N02 is not expected. Furthermore, combustion stack gases
typically contain only less than 5% of total nitrogen oxides (36) in
the NOg form. Thus, N0£ losses can be safely neglected.
93
-------
ABSORPTION
Absorption of gas phase species by components of the interface does
not generally represent a loss mechanism. The interface component must even-
tually become saturated, at which point the true steady-state concentration is
measured by the analyzer. This type of interaction may cause a response lag
and will be considered in the following section.
Sample loss may occur by solution of gas phase species in conden-
sate, which is collected and removed from the system. Maximum losses will
occur when the gas comes to equilibrium with the condensate at the point of
removal.
SO? Solubility Losses
The vapor-liquid equilibria involved in the solution of S02 in
water are considered in detail in Appendix I (33). Losses are determined
for condensate removal at 25°C from a sample stream containing 6% water vapor
and various SOp concentrations. As shown in Table 1-2, 6% moisture is typical
of coal-fired and oil-fired power plants (SOp monitoring is not required for
gas-fired power plants). The results of Appendix I indicate no significant
loss of S02 at levels above 10 ppm. In fact, at levels of about 10 ppm S02
and higher, the S02 partial pressure leaving the condenser is actually
greater. This is due to the fact that the reduction in gas phase volume
caused by removal of water vapor outweighs the solubility loss of SO^.
Solubility losses of S02 were measured experimentally in the
test system shown in Figure 1-18. The impinger was filled with deionized
water, and the test gas, containing 1200 ppm S02> was metered through the
impinger at a flow rate of 3.3 slpm. The gas then passed through a 15.2 m
coil of polyethylene tubing, a water trap, and an S02 analyzer.
The analyzer response along with the impinger water temperature
ard coil inlet and outlet temperatures is plotted vs experiment time in
Figure 1-19. Initially, it takes about 20 minutes to saturate the water in
the impinger at 20-25°C. At saturation, a steady-state SCn concentration
of 1160 ppm is measured. The temperature of the impinger is then increased
94
-------
HEATED
IMPINGER
OF WATER
Tl
FLOW METER
s
MANOMETER
MANOMETER
50'
POLYETHYLENE
ANALYZ E R
•VENT
U
MANOMETER
WATER
TRAP
Figure 1-18. Experimental Apparatus for Solubility Loss Tests.
-------
IDESORBT10N FROM IMPINGER
i-' TEMP
INCREASE-
STEADY STATE
AT ING
ER ®
Vy
i*?l
4
,
flr • * ?
*&®
7\
5)
• « •
• • B • "
P A A A A
-• PPM READING
® IMPINGER WATER TEMP~
• INLET TEMP Tl
A OUTLET TEMPT2
1,1,11
® Qp
• •
A A
— ^
- . L . .„.
-60
-55
-50
-45
-40
-35
-30
- 25
- 20
- 15
- 10
-b
D
0
a
LJ
CL
r>
o:
i ii
iii
10 20 30 40 50 60 70 60 vO IOC MO 120 130
EXPERIMENT TIME MIN
Figure 1-19. Measured S02 Level and Temperature Conditions
-------
from 20°C to 60°C. This increases the amount of water vapor in the sample
gas from 2.3% to 20%. If S02 solubility losses are important, more SCL
should be removed in the water trap and a lower steady-state S(L concentra-
tion should be measured. As shown in Figure 1-19 when the impinger tempera-
ture is increased S02 desorbs, but as the impinger is held at 60°C the
steady-state S(L concentration approaches the same value, 1160 ppm. This
indicates that there is no significant loss of S02 in the water trap.
It may also be noted that the reason the S(L reading never
reaches 1200 ppm is that the 1200 ppm is for dry span gas while the''gas
passing through the analyzer is saturated with water vapor at the trap
temperature. Taking the dewpoint at the analyzer as 25°C, then the vapor
pressure in the sample stream is ^3.07 k N/m and the span gas is diluted to:
+ 760 mm Hg
which is exactly what was measured.
NO Solubility Losses
The solubility of nitric oxide is less than sulfur dioxide, and
NO follows Henry's Law. At 0°C the relation between liquid mole fraction,
X, and partial pressure, PNQ, is (34):
XNQ= PNO/H = 1.923xlO-5PNO
where PNQ is in atmospheres. The solubility is less at higher temperatures.
A simple calculation for 10% water vapor indicates a solubility loss of only
0.0002% on a dry basis. Therefore, NO solubility losses are entirely
negligible.
NOo Absorption Losses
The absorption of N02 by water proceeds by the following reaction:
3N02(g) + H2OU) * 2HN03(soTn) + N0(g)
-------
Absorption presumably takes place through the Intermediate formation of
nitrous acid, but equilibrium concentrations of nitrous acid are small and
the overall reaction Is as given above. There Is considerable disagreement
on equilibrium data (32) as well as rate data (35) for the N(L-water system.
The equilibrium data Indicate that the uptake of ML can be very
great. Concentrated solutions of HML can be easily formed by absorption of
ML. Essentially all ML will be absorbed under practical equilibrium con-
ditions.
Loss of NOg can be avoided If the rate of approach to equi-
librium 1s sufficiently slow. The rate of ML absorption is given in
Appendix J. It is assumed that the entire Inside surface area of the tube
is wetted with condensate. For a sample flow rate of 2 slpm through 6.35 mm
o.d. x 1.016 mm wall tubing, 50% of the sample 1s lost in only one foot of
tubing. For unheated sample lines of any practical length, complete loss
of N02 can be expected.
A number of commercially available sampling systems keep the
sample above its dewpoint until it reaches the condenser. It is then
rapidly cooled to prevent solubility losses. The effectiveness of this
procedure depends in large part on the condenser design. An ordinary 6.35
mm diameter coil of tubing in 1.67°C surroundings would require at least a
foot length to cool 2 slpm of flow. A more elaborate design would have to
be devised in order to prevent significant NOg losses.
An attempt was made to experimentally measure ML losses by the
following procedure. NO,, was doped Into flue gas at a rate sufficient to
give 500 ppm at the sampling point downstream. The flow setting was deter-
mined by doping with NO to give a 500 ppm concentration and using the same
setting for ML. The experimental precision was not very good because of
r obi ems with flow stabilization. However, when doping with ML, an NO
£ A
Ci.i :entration of 350 ppm was measured rather than 500 ppm, and more than 70%
o" the N0x was in the NO form. This indicates a significant ML loss.
Loss of SOp and NO by reaction and absorption have been shown
to be negligible under practical sampling conditions. The absorptive loss
98
-------
of N02 represents the only real problem. However, the samll concentration
of NOp in combustion stack gases would allow the interface to completely
remove N0? and still remain within the 10% tolerance limit on NO . Stack
b A
gases from nitric acid manufacture may contain significant amounts of N02-
For this case, a permeation dryer is recommended to remove water vapor with-
out removing N0~. Other alternatives are to keep the sample above its dew-
point from stack to analyzer, or, if feasible, to use heated sample lines
and a low residence time condenser.
ADSORPTION
Adsorption of gas species onto interface components does not affect
the steady-state concentration measured at the analyzer but may result in an
unacceptably long response time. Adsorption interactions are best determined
experimentally since they are difficult to quantify theoretically.
Tests were conducted with both NO and SO,, using 15.2 m lengths of
six different sample line materials. Nitrogen containing 1200 ppm of either
NO or SOp was passed through the selected sample line, and the concentration
was measured and recorded as a function of time. At time zero, the sample
flow was discontinued, and room air was passed through the sampling system
at the same flow rate (1 slpm). The system response to room air was
measured for each sample line.
The results are shown in Figure 1-20. Desorption from sample line
walls results in a tailing effect and longer response times. The wall ef-
fects can be compared more easily by determining the time from the point
where the concentration first begins to fall to the point at which it
reaches 5% of the initial value. The response times for SO^ are given in
Table 1-12. The "blank" gives the response time of the system without sample
lines. Teflon, polypropylene, and 316 SS are all quite close to the "blank"
response and therefore exhibit negligible adsorption effects. Nylon ex-
hibits a slightly longer response time indicating a small adsorption effect.
Polyethylene and tygon give response times which are, respectively, three
99
-------
-Introduction of Zero Air-
SYSTEM BLANK
316 SS
YPROPYLENE
NYLON
EFLON
OLYETMYLENE
YGON
I I I ! I I I I I I I I I 1 I I I I I I
o
o
NO
LINES LEFT TO RIGHT!
SYSTEM BLANK
POLYPROPYLENE
NYLON
2500
PPM
2000-
I50O-
1000-
TYGON
TEFLON
POLYETHY_EME
r | i i I r [ r i i i |
1.5 2D 2.5 min
-
__ Q 0
0 0.5 1.0 1.5 2.0 Z.5mn
Figure 1-20. Absorptive Interactions with Various Sample Line Materials.
-------
TABLE 1-12
RESPONSE TIMES FOR VARIOUS SAMPLING LINE MATERIALS
Material
Blank
316 SS
Polypropylene
Teflon
Nylon
Polyethylene
Tygon
5. 95% Response Time for S(L
No sample line
6.35 x 0.889 x 15.24 m
6.35 x 0.762 x 15.24 m
9.525 od. x 6.35 i.d. x 15.24 m
6.35 x 0.762 x 15.24 m
9.525 o.d. x 6.35 i.d. x 15.24 m
11.113 o.d. x 6.35 i.d. x 15.24 m
0.54
0.44
0.62
0.63
0.75
1.58
2.05
Measured from first observed concentration change to 5% of initial con-
centration
101
-------
and four times the "blank" response. Adsorption effects for these materials
are therefore significant. Hence, polyethylene and tygon are not recommended
for extensive use in S02 monitoring systems.
The results for NO indicate that none of the materials tested ex-
hibit significant adsorption effects with this gas. Therefore, for NO moni-
toring any of the materials tested can be used.
Based .on the above results and the results of the section entitled
Design of Sampling Interface, subsection Materials of Construction, polypropylene
appears to be the material that is most suitable from the standpoint of a
minimum design. Its chemical resistance is satisfactory for all except con-
centrated sulfuric and nitric acids; its adsorptive interactions are negligi-
ble; and it is very inexpensive.
102
-------
SPECIFIC REFERENCES
1. "Standards of Performance for New Stationary Sources," Federal Register.
36 (247), Dec. 23, 1971, 24879.
2. Cuffe, S. T. and Gerstle, R. W., Emissions from Coal-Fired Power Plants:
A Comprehensive Summary, PHS Publ. No. 999-AP-35, Public Health Service,
Cincinnati, Ohio (1967).
3. Smith, W. S., Atmospheric Emissions from Fuel Oil Combustion: An In-
ventory Guide, PHS Publ. No. 999-AP-2, Public Health Service,
Cincinnati, Ohio (1962).
4. Data obtained from U.S. Department of Commerce, National Oceanic and
Atmospheric Administration, Environmental Data Service, National
Climatic Center, Federal Building, Asheville, North Carolina 28801.
5. Personal Communication, Dr. William Zolner, Thermo Electron Corp.,
Waltham, Mass. (November 1973).
6. Houser, E. A., "Beckman Analysis Systems for S02/NOX in Power Plant
Stack Gases," from Beckman Representatives' Memorandum (November 1971).
7. "Stationary Source Monitoring for S02 on Fossil Fuel-Fired Combustion
Processes," L&N Application Bulletin No. E1.1301-AB.
8. Perry, J. H., editor, Chemical Engineers' Handbook, 4th ed., New York:
McGraw-Hill (1963), pp. 23-13 to 23-30.
9. Gelber Pump Company Chemical Resistance Chart, Gelber and Sons, Inc.,
Chicago, Illinois (1972).
10. Plastiline Inc. Chemical Resistance Handbook, Plastiline Inc., Pompano
Beach, Fla. (1970).
11. Demco Plastics Application Guide, Demco Plastics Inc., Temple, Texas
(1971).
12. Chemical Resistance Characteristics of Standard Tygon Tubing Formula-
tions, from Norton Plastics and Synthetics Division, Akron, Ohio.
13. Jacquot, R. D. and Houser, E. A., "Qualification Testing of an Infra-
red Analyzer System for S02 and NO in Power Plant Stack Gas," Proc. of
the 27th Annual Conference of the Instrument Society of America, Instru-
ment Society of America Publications, Pittsburgh, Pa. (1972).
14. Stairmand, C. J., "The Sampling of Dust-Laden Gases," Trans. Instn.
Chem. Engrs.. 21, 15 (1951).
15. Driscoll, J., et_al_., "Improved Chemical Methods for Sampling and
Analysis of Gaseous Pollutants from the Combustion of Fossil Fuels,"
Final Report, Part 1, Contract CPA 22-69-95 (Sept. 1970).
103
-------
REFERENCES (continued)
16. Atmospheric Emissions from Nitric Acid Manufacturing Processes, PHS
Publ. No. 999-AP-27, Public Health Service, Cincinnati, Ohio (1966).
17. Atmospheric Emissions from Sulfuric Acid Manufacturing Processes, PHS
Publ. No. 999-AP-13, Public Health Service, Durham, N.C. (1965).
18. Personal communication with Mr. Jack Kertzman, Perma Pure Products,
Inc., Oceanport, N.J. (Nov. 1973).
19. Perry, op. cit.. p. 9-8.
20. Ibid., p. 9-14.
21. Ibid., p. 9-3.
22. Ibid., p. 5-21.
23. McCabe, W. L. and Smith, J. C., Unit Operations of Chemical Engineering.
New York: McGraw-Hill (1956), p. 436.
24. Eckert, E. R. and Drake, R. M., Heat and Mass Transfer. 2nd edition,
New York: McGraw-Hill (1959), p. 197.
25. Ibid., p. 315.
26. Ibid., p. 242.
27. Ibid., p. 91.
28. Carslaw, H. S. and Jaeger, J. C., Conduction of Heat in Solids. 2nd ed.,
London: Oxford University Press (1959), pp. 58-62.
29. Perry, op. cit., p. 5-62.
30. Cooper, H. B. H. and Rossano, A. T., "Source Testing for Air Pollution
Control," Environmental Science Services, 24 Danbury Road, Wilton,
Conn. 06897.
31. Olson, R. W., et al., Chem. Eng. Progr.. 46_, 614. (1950).
."'. Stevenson, R. M., Introduction into the Chemical Process Industries,
New York: Reinhold (1966), pp. 140-145.
?3. Calculations performed by Prof. Michael Modell, Dept. of Chemical Engi-
neering, MIT, Cambridge, Mass.
104
-------
REFERENCES (continued)
34. Perry, op. cit.. p. 14-6.
35. Margolis, G. and Driscoll, J., "Critical Evaluation of Rate-Controlling
Processes in Manual Determination of Nitrogen Oxides in Flue Gas,"
Envir. Sci. Tech.. 6, 727 (1972).
36. Bartok, W., Crawford, A., and Piegari, G., "Systematic Field Study
of NOX Emission Control Methods for Utility Boilers, EPA Report
No. APTD1163, (NTIS No. PB210-739) (Esso Contract CPA 70-90)
December 1971.
37. Personal communication, James B. Homolya, Project Officer, EPA,
Research Triangle Park, North Carolina, 27711 (August, 1974).
GENERAL REFERENCES
1. Houser, E.A., Principles of Sample Handling and Sampling Systems Design
for Process Analysis, Instrument Society of America, Pittsburgh, Pa.,
T97T:
2. Verdin, A., Gas Analysis Instrumentation, John Wiley & Sons, New York,
1974.
105
-------
APPENDIX A
CALCULATION OF STACK GAS COMPOSITIONS
The approximate composition of combustion stack gases can be easily
calculated from the composition of the fuel and from the amount of air used
for combustion. The use of 20% excess air wW be assumed throughout.
A. NATURAL GAS
For gas-fired power plants, the composition of Texarkana natural gas
was assumed (19):
CH4 = 96% vol
N2 = 3.2%
C02 = 0.8%
The heat of combustion for methane (20) is 37.73 k joule/1, and the combustion
reaction is:
CH4 + 202 -»• C02 + 2H20
Basis: 2832 si of fuel
Theoretical 02 = .96(2832)(2) = 5437.4 si 02
Theoretical air = 5437.4/0.21 = 25893 si air
Actual air = 25893 + (0.20)(25893) = 31070 si air
N2 in stack = (0.79)(31070) = 24545 si N?
02 in stack = (0.20)(0.21)(25893) = 1087.5 si 02
C02 in stack = 2718.7 si C02
H20 in stack = (.96)(2832)(2) = 5437.4 si
107
-------
The stack gas composition is:
Constituent sl %
co2
H20
N2
°2
TOTAL
33,902.3 si „
2741.4
5437.4
24636.0
1087.5
33902.3
1 si fuel
8.09
16.04
72.67
3.20
,,io6
_ 330.5 ski
73(0.96) k joules 6" ~ G-joule
_ . ... , . . . 86. gm „ 22.4 1 v 1 G-joule , 07 in-4 107
Fed N0x 1imit = G^jblne- x WW x 330.550 si = ]'27 x 10 = 127
nnm
ppm
_ . ... , . ,. .. 43 gm „ 1 G-joule „ 1000 mg n ,, . ,
Fed pamculate limit = 6-jou1e x 330.500 si x gm °'13 mg/sl
B. FUEL OIL
The approximate composition of residual fuel oil (3) is:
C = 86% wt
H = 10%
H20 = 1%
N = 0.5%
S = 1.6%
Inerts =0.9%
The heating value (3) is 42.5 k joule/gm. The combustion reactions are:
C + 02 -*• C02
4H + 02 - 2H20
108
-------
2N- N2
S + 02 * S02
Basis: 45.36 kg of fuel
Constituent kg g-moles g-moles 00 Required
c
H
H20
N
S
39.04
4.536
4.536
.227
.726
3250.9
4536
24.95
16.33
22.68
3250.9
1134
22.61
4407.6
Theoretical 02 =4407.6 g-moles
Theoretical air = 4407.6/0.21= 20988.6 g-moles air
Actual air = 20988.6 + (0.20)(20988.6) = 25186.3 g-moles air
N2 to stack = (0.79)(25186.3) = 19897.2 g-moles N?
02 to stack = (0.20)(0.21)(20988.6) = 881.5 g-moles 02
The stack gas composition is:
Constituent
co2
N2
°2
so2
26360.4
Hue gas volume = 453i5'0"
g-moles
3250.9
2292.9
19912.6
881.3
22.68
26360.4
g-moles 22.4 si
g 1 g-mole
12.33
8.70
75.54
3.34
0.086
1 gm
42.5 k joule
x1"6
"^
= 306 skl/G-joule
109
-------
Fed S02 limit = 344 9>" S°2 22.4 si 1 G-.loule = 3 94 x 10-4 _ 3g4 D_
2 G-joule x 64 gm x 306,000 si J>y4 x 10 " 394 ppm
129 9"" N0« 0
o /I
Fed MOX IMt - Tcjsn * T51i * 3ooCO si ' 2"05 * 10 ' 205
Fed peculate IMt - . = x . c.,« n,9/sl
C. COAL
The composition of Pittsburgh coal 1s assumed (21):
C = 76.6% wt
H = 5.2%
0 » 6.2%
S = 1.3%
N = 1.6%
Ash = 9.1%
The heating value is 31.65 kjoule/gm (21). The combustion reactions are:
C + 02 * C02
4H + 02 * 2H20
S02
Basis: 100 1b fuel
110
-------
Constituent kg g-moles g-moles 00 Required
c
H
0
S
N
34.746
2.359
2.812
0.590
0.726
2895.3
2358.7
175.8
18.4
51.7
2895.3
589.7
-88.0
18.4
3415.4
Theoretical 02 » 3415.4 g-moles 02
Theoretical air = 3415.4/0.21 = 16,263.8 g-moles air
Actual Air = 16,263.8 + 0.20 (16,263.8) = 19,516.6 g-moles air
N2 to Stack = (0.79)(19,516.6) = 15,418.1 g-moles N?
02 to Stack = (0.21) (0.20) (16,263.8) = 683.1 g-moles 02
The stack gas composition is:
Constituent
co2
HgO
N2
°2
so2
20224 g-moles „
g-moles
2895.3
1179.3
15447.Q
683.1
18.4
20224.0
1 gm fuel
X
14.32
5.83
76.38
3.38
0.09
x ?2.4 si xl£.
Flue «aa * «.«...* - 45359 gm fuel
Fed SO, Limit = 516 gm S02 22.4 si 1 G-joule _ 5 72 x lfl-
2 G-joule x 64 gm x 315,660 si
-4
OOuTe " 64 gm " '" cnn -1 " 'J = 57Z ppm
Fed N0¥ Limit = 301 gm N°2 22.4 si 1 G-joule = 4 64 x 10-4 . 464 pm
x G-joule X 46 gm X 315,600 si q>M IU ™ pp
Fed Particulates limit = x 1 • 1-36 x 10'4 gm/sl = 13.6 mg/1
111
-------
APPENDIX B
CALCULATION OF FLOW RATE VS PRESSURE DROP
When the pressure drop is less than about 10% of the downstream pres-
sure and when flow is laminar, the pressure drop through a circular tube is
given by the Poiseuille equation (22):
P P - M L V
H ~ —~
where: u = absolute viscosity (kg/m-sec)
L = length of tube (m)
V = average linear velocity (m/sec)
D = inside diameter of tube (m)
2
P, = upstream pressure (N/m)
2
P, = downstream pressure (N/m )
The following assumptions are made:
(1) The gas temperature is 25°C along the entire length of tubing.
Heat transfer calculations show that the sample quickly comes to the temper-
ature of the surroundings.
(2) For the purpose of calculating fluid properties, the average pres-
sure is taken as 1 atm, and the sample is assumed to be air.
(3) Flow is assumed to be laminar (an assumption which will be checked),
(4) The pressure drop is assumed to be less than 10% of the downstream
pressure (which can be checked by the results).
For air at 1 atm and 25°C:
P = 1.838 x 10~5 kg/m-sec.
0 = 28.9 x 10"3kg ¥ 273 _ , ,R .
P - * J x -KKfr - I .lo kg
0.0224 nT Z98 jjjt
113
-------
The volumetric flow rate 1s given by:
F = VS
Where: F = volumetric flow rate (I/sec)
2
S = cross sectional area (m )
V = average linear velocity (m/sec)
For a flow rate, Fg, given in standard liters per minute, the average
linear velocity is:
V = 1.667 x 10"5 -4 m/sec (2)
Substitution into Equation (1) with L = 30.48 m gives:
P, - P, = 9.356 x 10"9 -4 N/m2
e. i Df
where F~ is the sample flow in standard liters/minute and D is the inside
line diameter in m. For the diameter given in cm, D', the corresponding
equation is:
P9 - P, = 9.356 x 10"17—^T- N/m2
2 ] (D1)4
This equation was used to calculate the results presented in Figure 1-11.
The assumption of laminar flow can be checked by calculating the
Reynolds number:
D Vp
Re -
114
-------
Or, substituting Equation (2):
667 x 10"5 FS p
The higher FS/D, the higher the Reynolds number. For the range of variables
covered in Figure 11, the maximum F^/D occurs at 100 slpm through .1143 m i.d,
tubing.
Re = 1303
The onset of turbulence does not occur until a Reynolds number of 2100 is
reached.
115
-------
APPENDIX C
HEAT TRANSFER FROM SAMPLE LINES
A. DERIVATION OF EQUATIONS
Twl
1
r
i
i
x=o
.
Vj
t
LAt*
\
^ Sample
' *~ now
\
' )
1
[ X+AX
Ta
The temperature of the sample gas, T, can be determined as a function of
position, X, in the sample line by making a heat balance on a differential
segment of line between X and X + AX. The rate of heat flow through the
walls of the segment is given by:
where: q = local rate of heat transfer (kjoules/hr)
o
A = outside tube diameter (m )
U - local overall heat transfer coefficient based on the outside
tube diameter
T = average bulk gas temperature within the differential element
T = ambient temperature distant from the tube
a
The heat balance then becomes:
Heat Input = W C T|x
Heat Output = W Cp T|X+AX + UQ TT DQ AX(T - Tfl)|
*"^~
Accumulation = 0 (steady state)
117
-------
where: W = mass flow rate of sample (kg/hr)
C = heat capacity of sample (kjoule/kg-°C)
T|x = average gas temperature at X (°c)
T|X+AX = average gas temperature at X + AX (°C)
DQ = outside tube diameter (m)
These terms lead to the following differential equation:
- «cp 37 ' uo " Do
-------
where: h, = heat transfer coefficient from sample, gas to tube wall
based on Inside diameter (kjoule/hr-mz-°C)
h = heat transfer coefficient from outside tube wall to ambient
based on outside diameter (kjoule/hr-mz-°C)
k = thermal conductivity of tube wall material (kjoule/hr-m-°C)
D.J = Inside wall diameter (m)
Y = wall thickness (m)
D~ • average tube diameter, ^ (D^ + DQ) (m)
It remains to determine h, and h .
1 o
As shown in Appendix B, flow within the sample lines is laminar under all
reasonable sampling conditions, i.e., the Reynolds number is less than 2100.
The average Nusselt number from 0 to X for laminar flow inside tubes is given
by (24):
0.0668 (D./X) Re Pr
Nu = 3.65 + 1 m (5)
1 + 0.04 [(D./X) Re Pr]^/J
where the dimensionless numbers are:
hi Di
Average Nusselt number = Nu = .
D, V p
Reynolds number = Re = —
C y
Prandtl number = Pr = -*?-—
and where: k = thermal conductivity of sample gas (kjoule/hr-m-°C)
7= average linear velocity (m/sec)
o
p = density of sample gas (kg/m )
y = viscosity of sample gas (kg/m-sec)
Under all practical sampling conditions the second term of Equation (5),
which gives the entrance effects at small X, is entirely negligible. The
average Nusselt number is then constant at 3.65. This value is actually
119
-------
based on a constant wall temperature along the length of the tube. However,
it should not be greatly affected by a varying wall temperature along the
sample line length.
Heat transfer from the outside tube wall takes place by natural convec-
tion and depends on the Grashof and Prandtl numbers. The Mussel t number at
a particular X location averaged over the circumference of the tube is (25):
Pr1/2 Gr1/4
(0.952 +
The Prandtl number for gases is very insensitive to temperature. For air
the Prandtl number varies from 0.70 to only 0.68 over the temperature range
of 65.56°C to 326. 7°C. Using the value of 0.70 in Equation (6) gives:
Mu -Ufi^ » 0.40 Gr1/4 (7)
The Grashof number is given by:
6r . °.3 4 »f « "„
where: pf = density of air at mean film temperature (kg/m )
Bf = coefficient of thermal expansion at mean film temperature =
reciprocal of absolute temperature for ideal gases (°k)
p
g = acceleration of gravity * 9.80 m/sec
\i, = viscosity of air at mean film temperature (kg/m-sec)
AT = average difference in temperature between outside pipe wall
0 and air distant from tube (°C)
Properties are determined at the mean film temperature which is the
arithmetic average of the ambient temperature and the wall temperature at
position X. Since the wall temperature is not constant along the length of
the sample line, the Grashof number will vary and the Mussel t number will
also vary. It is not a simple matter to account for this changing wall
temperature. One approach is to assume a constant Grashof number over
short lengths of the sample line and to make a calculation for each
120
-------
segment. A more approximate approach is to assume a constant Grashof number
despite the variation 1n wall temperature. The latter approach will be
followed.
C. CALCULATION OF TUBING LENGTH FOR COOLING TO 200°F
Plastic materials such as polypropylene cannot be placed directly in the
stack of combustion sources because of their heat limitations (approximately
93.33°C). It is of interest to determine the length of heat resistant material
(e.g., stainless steel) that must be used before a transition to lower tem-
perature plastics can be made.
The following conditions are used:
DQ = 6.35 mm
D. = 4.318 mm
F = 2 st. Iiters/m1n
Ta = 37.78°C
T. = 260°C
TQ = 93.33°C (outlet temperature from heat resistant tube)
Sample - air
Tube - stainless steel
The arithmetic average temperature of the sample flowing through the
tube is 176.7°C. At the point where the gas sample is at a temperature of
176.7°C the outside wall temperature may be determined as follows. Due to
the high thermal conductivity of the tube, assume that the inside and out-
side wall temperatures are the same. The inside resistance to heat flow 1s
1/hj D.J and the outside resistance is l/hQ DQ. The temperature of the tube
wall will be intermediate between 176. 6°C and 25°C and will be determined by
the relative magnitudes of the inside and outside resistance. That is, the
inside temperature drop is proportional to the inside resistance, etc.
T - T
As a first approximation, hQ D^h^ D^ = 0.5 is assumed and TW = 130.56°C. The
average film temperature outside the sample line is 1/2 (130.56 + 37.78) = 84.16°C,
121
-------
The following properties were determined for air at 84.16°C and 1 atm
pressure:
3
P - 0.980 kg/nT
,-3 <,„-!
6 = 1/357.8°k = 2.795 x 10 * °K
w = 2.107 x 10 kg/m-sec
ATQ = (84.16 - 37.78) = 46.38°C
k - 0.11 kjoule/hr-m-°C
2 -3
(6.350 x 10'3m)3 (0.980 (L-79 10 )(9.8-) 46.38°C
Gr = ^-F 5 ^ = 709
(2.107 x 10"" kg/m-secT N
Substituting in Equation (7) gives:
h D
Nu = -TT^- = 2.06
and
hn = ' - = 35.75
0 (6.35 x 10"J)
For flow inside the sample tube, fluid properties are calculated at the
average temperature 176.7°C.
k . .1334 kjoule/hr-m-°C
Cp = 1.0205 kjoule/kg-°C
The inside heat transfer coefficient is:
hi Pi
Nu = - = 3.65
h , _(3.65) (.1334) = n2 76 kjoule/hr-m2-°C
1 (4.318 x 10"J)
The actual value of h D /h. D. is 0.47, close to the assumed value of 0.50.
The overall heat transfer coefficient is calculated from Equation (4),
taking the thermal conductivity of stainless steel as 39.346 kjoule/hr-m-°C.
122
-------
J_. 6.35 x IP"3 + (1.016 x 10"3)(6.35 x IP"3) ,
Uo (112.76) (4.318 x 10"3) (39.346)(5.334 x 10"3)
24.36 kjoule/hr-m2-°C
For a flow of 2 slpm
Substituting in Equation (3) gives:
, (.1548) (1.0205) 260 - 37.78 „ 45
L = — - — - o - ltn QO oo -44 "jo '^3 '"
(TT) (6.35 x 1(T3) (24.36). 93'33 ' 37'78
Therefore, in a length of about .5 m, stack gases at an initial temperature of
260°C would be cooled to the point where low temperature plastics could be
used for the rest of the sample line. This calculation is conservative on
the basis that a 37.78°C ambient is assumed and a stagnant air mass is assumed.
D. LENGTH REQUIREMENT FOR UNHEATED SAMPLE LINES
A closely related calculation to the previous one is a determination of
the maximum allowable sample-line length before condensate freeze-up can be
expected in subfreezing surroundings.' In order to make a conservative esti-
mate in this case a different set of assumptions must be made. It is assumed
that the ambient temperature is -6.67°C and a continuous 16.09 kph wind is blow-
ing normal to the sample line.
It is assumed as a first approximation that the resistance to heat trans-
fer is approximately the same inside and outside the tube. The tube wall
will then be at a temperature midway between the sample gas and the ambient.
The Reynolds number outside the tube is:
DQ Vp
Re = -£- —
123
-------
Properties are determined at the mean film temperature. The mean sample gas
temperature is 1/2 (260 + 0) = 130°C. The average wall temperature is
1/2 (130 - 6.667) = 61.67°C. The average film temperature is 1/2 (61.67 - 6.67)
= 27.5°C. At this temperature
P = 1.177 kg/m3
Vs 1.847 x 10"5 kg/m-sec
k - 0.0944 kjoule/hr-m-°C
The linear velocity corresponding to 16.09 kph is 4.48 m/sec.
Re = (6.35 x IP"3) (4.48) (1.77) = 18]3
1.847 x 10"5
For the range of Reynolds numbers between 1 and 4000, the Mussel t number is
given by (26):
Nu = -2p2. = 0.43 + 0.48 (Re)0'5
Nu = 20.87
or
h m (20.87) (0.0944) = 31Q k.joule/nr.m2.oc
0 (6.35 x 10"J)
For the internal heat transfer coefficient, the average sample temperature
is 13Q°C at which temperature the thermal conductivity of air is 0.1219
kjoule/hr-ft-°C, and the heat capacity is 1.0205 kjoule/kg-°C.
- hi Di
Nu = -^-^ = 3.65
h = (3.65) (.1219) = 1Q3 Q4 kjoule
1 " 4.318 x 10~3 ' hr-m2-°C
By Equation (4):
U = 57.02 kjoule/hr-m2-°C
o
124
-------
and from Equation (3):
L . _ - An 260 . Q
(*) (6.35 x 10"J) (57.02) 0 + 6.667 = °'508 m
Therefore, at a sampling rate of 2 slpm through a 6.35 mm o.d. tube
freeze-up may occur within 0.51m.
E. CALCULATION OF TUBE WALL RESISTANCE
It is of interest to calculate the magnitude of the terms making up the
overall heat transfer coefficient. From Equation (4), the ratio of wall
resistance to internal film resistance is given by:
wall resistance Y D/k ff
internal film resistance " D /hj D^
For the conditions of the two previous calculations, this ratio is approxi-
mately 1.5 x 10" . Thus, the resistance due to the stainless steel wall is
only about 0.15% of the internal film resistance. If teflon were used for
the tube wall at a thermal conductivity of 0.872 kjoule/hr-m-°C, the resistance
of the wall would be only about 10% of the internal film resistance. Thus,
the tube material has relatively little effect on the rate of heat transfer.
125
-------
APPENDIX D
EFFECT OF DEAD-END VOLUMES ON RESPONSE TIME
The effect of dead-end volumes will be illustrated by calculating the
rate of diffusion of span gas (SOg) from a calibration line of infinite
length into the sample line. In the diagram shown below the concentration
Sample
Line
Zero
Concentration
Calibration Line
Diffusion
I I
X X+AX
Sample
Flow
at time zero is C.., the span gas concentration, throughout the calibration
line, and is zero, the sample concentration, throughout the sample line. At
times greater than zero, diffusion of span gas into the sample occurs. The
diffusive flux is given by Pick's Law (for diffusion in the negative X
direction):
g-moles/cm sec
p
where: N = flux of SOg span gas (g-moles/cm sec)
diffusion coefficient for S02 (cm2/sec)
concentration of SO, (g-moles/cc)
(1)
N
D
C
Y = mole fraction of SO
X =
molar density of gas (g-moles/cc)
linear distance (cm)
A material balance on a differential slice of calibration line gives:
Input = N|X+AX A At(g-moles) = D pm A At(3Y/3X)|x+AX
Output = N| A At(g-moles) = D pm A At(3Y/3X)|x
127
-------
Accumulation = A AX pm Y|t+At - A AX pm Y| (g-moles)
where: A = cross-sectional area of calibration line (cm)
t = time from introduction of sample (sec)
These terms combine to give the equation:
3Y _ n 32Y
" ~ D
- 3X
The problem is simplified somewhat by using the dimensionless parameter
6 = Y/Y.. , where Y.. is the S(L mole fraction at t = 0 throughout the calibra-
tion line. With this substitution, Equation (2) becomes:
36 _ n 329
3t " ax2
The boundary conditions are:
(1) At t = 0 6 = 1 for all X > 0
(2) At X = 0 9=0 for all t > 0
(3) At X • - 9 = 1 for all finite t
The above equation and boundary conditions have been previously solved
by a number of authors (e.g., 27 and 28) in relation to unsteady-state heat
transfer in an infinite slab. The solution for 8 in terms of the error
function is:
6 = erf
To determine the flux into the sample stream, the concentration gradient
must be determined at X = 0:
f = _i_exp(.x2/4Dt)
128
-------
and
36
3X
x=o
By Equation (1):
Dp Y.
Flux into sample stream = —!D__L Si^i
/irDt cm sec
To determine the response effect of this diffusive flux, a calculation
will be made of the time required for the concentration of S02 in the sample
to drop to 5% of the span gas concentration (95% response time). The fol-
lowing is assumed:
Sample flow rate =1 slpm
Temperature = 0°C
Pressure = 101.3 N/m2
D = 0.104 cm2/sec for S02
Tube i.d. = 4.318 mm (6.35 o.d. x 1.016 wall)
A material balance on SO^ around the mixing point gives:
Input due to flow = 0
D p Y. A
Input from calibration line = q"
sec
Output to analyzer = m Y
where: m = sample flow rate (g-moles/sec) = 7.436 x 10"
Equating input to output and solving for t gives:
tl/2 ""h* Y1
=
For 95* response, Y^V = 20.
129
-------
tl/2 = (0.104)(4.461 x 10"5)(0.1464)(20)
(7.436 x 10"4)(0.5716)
t = 1.02 x 10"3 sec
This calculation Indicates that under all practical conditions, dead
volumes have a negligible effect on response time.
130
-------
APPENDIX E
RESPONSE CHARACTERISTICS OF BACK-MIXING VOLUMES
The various mixing volumes can be considered to be connected end to end
with negligible transport time between consecutive volumes. Initially, the
F
v°
Yl
"l '1
concentration is uniform at a mole fraction of Y.. At time zero, a sample
containing zero concentration flows into the first volume at a rate F. Per-
fect mixing is assumed in each volume.
The general equation relating mole fraction to time can be derived by
considering volume 2.
Input: F YI At liters
Output: F Y2 At liters
Accumulation: " Y^ liters
where: F =
v
V
V
t =
volumetric flow rate (1pm)
mole fraction in and leaving volume 1
mole fraction in and leaving volume 2
volume of volume 2 (liters)
time (min)
These terms lead to the following differential equation;
Y2 ' Yl = ' T ~dt~
131
-------
The time constant for volume 2 is defined as:
V2
T2if (1)
and the differential equation becomes:
d Y2 Y2 Y
This can be solved by introducing the integrating factor exp (t/T2):
Y Y
exp(t/T2) dY2 + ^exp(t/T2) dt = ^-exp(t/T2) dt
The left-hand side is recognized as the derivative of Y2 exp(t/T2) so:
Y2 exp(t/T2) = ^-fri exp(t/T2) dt + C
or in general terms* where n is the volume number:
exp(-t/T ) r
Yn = - - — — / Vl MP(t/Tn) dt + C exp(-t/Tn) (3)
Except where n = 1, Y , is a function of t and cannot be removed from the
integral. The boundary condition for evaluation of the integration constant
is:
At t = 0 Yn = Yi for all n
A. UNEQUAL TIME CONSTANTS
1. One Volume
For one volume, n = 1 , Y , = Y = 0, C = Y. and
Y,
/= exp(-t/T,) (4)
Yi '
132
-------
2. Two Volumes
For two volumes 1n sequence, n = 2, Y j = YI given above and
Equation (3) becomes:
exp(-t/T?) r r .. , vi
Y2 = T— Yi J expKT T ) dt + C exP(-t/T2)
after integration and rearrangement:
C =
and
Y2 Tl r
«-- exp(-t/T9) - ' exp(-t/i,) - exp(-t/r,) (5)
Yi ^ T1"T2 ' z ' '
3. Three Volumes
For three volumes, n = 3, and Yn-1 is given by Equation (5). Sub-
stituting in Equation (3), integrating, and rearranging gives:
C - 1 + V2 T]2 T2
and
Y T 2
exp(-t/T3) + (Ti.T2f(T2.T3) [exp(-t/T3) - exp(-t/T2)|
T 2
) - exp(-t/T1)| (6)
Equations (4), (5), and (6) give, respectively, the output as a
function of time from one, two, and three consecutive volumes.
B. EQUAL TIME CONSTANTS
For the special case in which the time constant is the same for each
volume, a much simpler result is obtained.
133
-------
1. One Volume
For one volume, the result is identical to Equation (4):
Y,
Y^-= exp(-t/T) (7)
2. Two Volumes
For two volumes, Equation (3) gives
YZ • exp(-t/T)
which gives C = Y.
and £ = (l +^exp(-t/i) (8)
3. Three Volumes
For three volumes, the use of Equation (8) in Equation (3) gives
C = Y. and
Y7" (] +T+rexp(-t/T) (9)
4. General Result
Continuation of the above procedure leads to the following general
equation:
YM / N *n \
/= ] + £ -Sr exp(-t/r) (10)
Yi \ n=l n! in/
where N is the number of consecutive volumes with equal time constants.
134
-------
C. SAMPLE CALCULATION OF SYSTEM RESPONSE TIME
To illustrate the above equations, the 95% response time will be cal-
culated for the following hypothetical sampling system:
Plug-flow volume (sample lines and interconnecting tubing): 60.96 m of 6.35
o.d. x 1.016 mm tubing
Back-mixing volumes: pump = 200 cm3
trap = 500 cm
analyzer cell = 400 cm3
Sample flow rate = 1 1pm
First, the response time for the mixing volumes will be determined by
Equation (6). The time constants are:
TI = 0.20 liters/1 1pm = 0.20 min
TZ = 0.50 min
T3 = 0.40 min
For 95% response, Y^Y^ =0.05. The coefficients for Equation (6) are:
= -8.333
h2
(TTT2MTTT3)
= 0.6667
Substituting into Equation (6) gives:
0.05 = exp(-2.5t) - 8.333[exp(-2.5t) - exp(-2t)]
- 0.6667 [exp(-2.5t) - exp(-5t)]
This equation is solved by trial and error to give:
t = 2.39 min
135
-------
For the plug-flow volume, the total volume is:
V = .. . 0.893 11ters
(4)(103)
The lag time for this volume is:
t . V . 0.8« liters . „.„„ „.„
The total 95% response time of the sampling system is 3.28 minutes. Of the
total response time, 27% is due to plug flow and 73% due to back mixing.
136
-------
APPENDIX F
CONDENSER COOLING REQUIREMENT
The feed is assumed to be saturated at 37.78°C. If the total moisture
content is 15%, then a large fraction of the moisture entering the condenser
is in the liquid phase. The cooling requirement for this liquid will be
added onto the cooling requirement for the saturated gas stream determined
below.
NF
\0 - °-°6«
Condenser
1.67°C
NV = 2 slpm = 5.386 g-moles/hr
YH20 ' 6'82 x 10'3
37.78°C
Mater balance on condenser:
,-3
0.0647 NF = 6.82 x 10 " NV + NL
Substituting NV and solving simultaneously gives:
NF = 5.719 g-moles/hr =0.165 kg/hr
NL = 0.333 g-moles/hr =' 5.99 x 10"3 kg/hr
Ny = 5.386 g-moles/hr =0.155 kg/hr
Heat capacity for air = 1.495 kjoules/kg-°C
Heat capacity for water =4.186 kjoules/kg-°C
Latent heat of vaporization = 2511.4 kjoules/kg
Air: qa = W C AT = (0.155^) (la
(36.1TC) = 8.367 kjoules/hr
137
-------
Condensation: q = W A = (5.99 x 10'3 {&) (2511r4 kJou1e) = 15.04 kJPule
nr kg hr
'3 4'186
Liquid water: q = W C AT = (5.99 x 10' L)(4'186 M^\) N7.78°C) = .446 kjoule
9 hr
Total cooling requirement =23.8 kjoule/hr
If the feed contains 16% moisture:
(0.0647)(5.-719) + -(1.0) NpL = 0.16(5.719 + NpL)
where NpL is the moles per hour of liquid condensate fed to the condenser.
NpL = 0.649 g-moles/hr = 0.0116 kg/hr
The cooling requirement for this liquid flow is:
q = W C AT = (0.0116 ) (4.186 e ) (36.11'C) = 1.76
p nr Kg - i.
The total cooling requirement is about 25.56 kjoules/hr.
The total condensate to be removed is about 18 ml/hr.
138
-------
APPENDIX G
SPECIFICATIONS FOR SAMPLING INTERFACE
The specifications for a sampling interface following the flow diagram
of Figure 1-1 are given below.
A. SAMPLE PROBE/COARSE FILTER
Probe tip to be 316 SS tubing 6.35 mm o.d. open to downstream direction.
Probe tip to be connected via 316 SS compression fittings to 316 SS check
valve. Cracking spring pressure in check valve to be as small as possible
o
(less than 3.45 N/m preferred). Coarse particulate filter to be alundum
thimble in stainless steel housing. Provisions must be made for admitting
calibration gas upstream of the coarse filter. Calibration gases to be ad-
mitted via 6.35 o.d. 316 SS tube and compression fitting tee. Entire
sampling probe assembly (including probe tip, check valve, calibration tee,
and alundum filter) to be mounted inside stack with feedthroughs provided
for calibration gas inlet and sample outlet. All materials in contact with
sample or calibration gas to be 316 SS, teflon or viton.
Fabricate from following materials:
Probe tip - 6.35 mm o.d. 316 SS x 0.889 mm wall tube
Connection to check valve - 6.35 mm tube x 6.35 mm pipe female con-
nector compression fitting - 316 SS
Check valve - 6.35 mm NPT male connections. Lowest possible
spring pressure. 316 SS. Viton seat.
Calibration tee - 6.35 mm NPT Tee, 316 SS
Connection to calibration line - 6.35 mm tube x 6.35 mm pipe male
angle connector. Compression fitting - 316 SS
Calibration line - 6.35 mm o.d. - 316 SS x 0.762 mm wall tube
Connection to filter - 6.35 mm close nipple 6.35 mm x 12.7 mm red. bush.
Both in 316 SS.
Filter - alundum particulate sampling thimble; SS holder with
12.75 mm NPT female ports
Sampling line connection - 6.35 mm x 12.75 mm NPT red. bush, 6.35 mm tube
x 6.35 NPT male connector compression fitting,both 316 SS.
139
-------
Probe assembly to be Inserted.through 7.62 cm pipe stack sampling port with
lines exiting through 7.62 cm NPT cap (or plug).
B. SAMPLE LINES
Outdoor service*where freezing 1s possible,on moisture-containing stacks:
Dekoron electrically traced teflon -6.35 mm.o.d. tube x 0.762 mm wall. Variac to
be used as power supply and power adjusted to prevent temperature drop below
freezing point. Outdoor service on dry stacks and Indoor service on all
stacks. 6.35 o.d. x 1.016 mm wall polypropylene tubing. Maximum temperature
for pp = 107.2°C.
C. CALIBRATION GAS LINES
Calibration gases to be dry. Use unheated polypropylene with nylon com-
pression fittings. 6.35 mm o.d. x 1.016 mm wall.
D. VENT LINES
Use 6.35 mm o.d. polyethylene x 1.016 mm wall. If line extends more than
several inches outside, use heat traced Teflon to prevent constriction due to
freezing where subfreezing temperatures are possible.
E. SAMPLING PUMP
Sampling pump to be a diaphragm type. Internal parts to be Teflon
Required flow = 2 slpm (4.24 scfh) into 20.68 k N/m with suction at -13.7g
k N/m2 (4" Hg Vac). Buy Thomas Industries 107CA110.
Pump must be piped with discharge-to-suct1on bypass. Make loop volume
small as practically possible. Use WMtey needle valve with 6.35 mm Swagelok
2
tube connections. Maximum flow = 14.16 sl/min. Take maximum aP = 6.895 k N/m
.'. min. Cy factor = 0.130. Buy Whitey 1RS4-316 with 4.369 mm orifice. Also
use two nylon tees, 6.35 mm tube, Swagelok.
F. PRESSURE GAUGES
Both gauges to be 316 SS tip and tube. Standard accuracy (+ 2%) PI-1
-34.47 to +34.47 kN/m2, PI-2 0-68.95 kN/m2. Prefer 6.35 mm NPT lower male connection.
140
-------
G. LIQUID TRAP
To be a ball float trap of stainless steel construction. Liquid removal
required: 20 ml/hour. Buy Armstrong 11-LD. Constructed of 304 and 440
stainless steel. Top port 3/4 NPT female. Bottom port 1/2 NPT female.
Top connections:
19.05 x 6.35 NPT red. bush - 316
6.35 NPT x close nipple - 316
6.35 NPT tee - 316
s
2 - 6.35 tube x 6.35 NPT male connectors - one bored through nylon
Bottom connections:
12.7 x 6.35 NPT red. bush - 316
6.35 tube x 6.35 NPT male connector - nylon
H. FLOW METERS
Nominal flow for all instruments = 1 slpm = 2 scfh. Flow meter to in-
clude integral SS metering valve. Buy Dwyer RMS-3-SSV.
I. FINE FILTER
Fabricate from glass wool in a glass tube. Use approximately 12.7 mm
diameter glass tube. Pack with glass wool until a "reasonable" flow re-
sistance is obtained. Butt-connect to system with tygon tubing.
J. REFRIGERATED CONDENSER
To be of 316 SS construction. Cooling requirement very small = 23 kjoules/
hr. Buy smallest unit available. Buy Hsnkison E-3GSS with three cooling
coils for contingencies.
K. AUTOMATIC CALIBRATION SYSTEM
Automatic calibration to occur every 24 hours ± 15 minutes. Each of
three calibration gases is to be admitted to analyzer in turn for a period
of 10 minutes each. System schematic shown in Figure 1-10.
141
-------
Parts List
(1) Zenith Model 2400-1 24-hour timer.
(2) Eagle Model 340 IMP-4-A6-22 with cycle gear MP5-44 and
#HN320 enclosure. 33-minute timer.
(3) Three toggle switches DPST - contacts rated at 1 amp.
(4) Three pilot lights 115 VAC. High Impedance neon.
(5) Two conductor cables needed. 18 gauge 300 volt. Insulation
suitable for outdoors.
(6) Solenoids - stainless steel
(a) 2-way #826207 ASCO
(b) 3-way #832061 ASCO
(7) 1/4" compression fittings
8 - 1/4" tube x 1/4 NPT male connectors - nylon
1 - 1/4" tube union - 316
1 - 1/4" tube x 1/4 NPT male connector - 316
(8) Pipe fittings 316 SS
1-1/4 NPT cross
3 - 1/4 NPT x close nipple
142
-------
APPENDIX H
SOURCES OF SAMPLING SYSTEM COMPONENTS
A. PROBE-TIP FILTERS
1. ASCO Sintering Co.
3000 S. Vail Ave.
Los Angeles, Calif.
2. Leeds & Northrup Co.
North Wales, Pa. 19454
3. Mott Metallurgical Corp.
Spring Lane
Farmington Industrial Park
Farmington, Conn. 06032
4. Pall Trinity Micro Corp.
Cortland, N. Y. 13045
INTERNAL TYPE (ALUNDUM THIMBLES)
1. American Hospital Supply Corp.
Scientific Products Division
1210 Leon Place
Evanston, Illinois 60201
2. Balston, Inc.
703 Mass. Ave.
Lexington, Mass. 02173
3. Research-Cottrell, Inc.
P. 0. Box 750
Bound Brook, N. J. 08805
4. Van Maters & Rogers, Inc.
P. 0. Box 3200
Rincon Annex
San Francisco, Calif. 94119
5. Western Precipitation Co.
1000 W. Ninth Street
Los Angeles, California 90015
INTERNAL TYPE (OTHER)
1. BGI Inc.
1254 Main Street
Waltham, Mass. 02154
2. Flotronics Div. of Selas Corp.
Spring House, Pa. 19477
3. Gelman Instrument Co.
600 Wagner Rd.
Ann Arbor, Mich. 48106
143
-------
B. FINE FILTERS
1. Balston, Inc.
P. 0. Box C
703 Mass. Ave.
Lexington, Mass. 02173
2. Creative Scientific Equipment Corp.
2305 Cherry Industrial Circle
Long Beach, Calif. 90805
3. Flotronics Div. of Selas Corp.
Spring House, Pa. 19477
4. Gelman Instrument Co.
600 S. Wagner Rd.
Ann Arbor, Mich. 48106
5. Millipore Corp.
Bedford, Mass. 01730
6. Mine Safety Appliance Co.
201 N. Braddock Ave.
Pittsburgh, Pa. 15208
7. Pall Trinity Micro Corp.
Cortland, N. Y. 13045
C. REFRIGERATION UNITS
1. Cole-Partner
7425 N. Oak Park Ave.
Chicago, 111. 60648
2. FTS Systems, Inc.
P. 0. Box 158
Stone Ridge, N. Y. 12484
3. Hankinson Corp.
Cannonsburg, Pa. 15317
4. Markson Science, Inc.
P. 0. Box 767
Del Mar, Calif. 92014
5. Neslab Instruments, Inc.
871 Islington St.
Portsmouth, N. H. 03801
6. Tecumsch Products Co.
Refrigeration Division
Tecumsch, Mich. 49286
144
-------
D. PERMEATION DRYERS
1. Perma Pure Products, Inc.
P. 0. Box 70
Oceanport, N. J. 07757
E. SAMPLE PUMPS
1. (Aspirator type)
Air-Vac Engineering Co., Inc.
100 Gulf St.
Mil ford, Conn. 06460
2. Gelber Pumps
5806 N. Lincoln Ave.
Chicago, 111. 60659
3. Metal Bellows Corp.
1075 Providence Hwy.
Sharon, Mass. 02067
4. Thomas Industries
1419 Illinois Ave.
Sheboygan, Wis. 53081
F. HEAT-TRACED SAMPLE LINES
1. Samuel Moore & Co.
Dekoron Division
Industrial Park
Mantua, Ohio 44255
G. LIQUID CONDENSATE TRAPS
1. Armstrong Machine Works
Three Rivers, Mich. 49093
2. Nicholson Div. of Electronic Specialty Co.
12 Oregon St.
Wilkes-Barre, Pa. 18702
3. Sarco Co., Inc.
1951 26th St., S.W.
All entown, Pa. 18105
H. PRESSURE GAUGES AND MANOMETERS
1. Alnor Instrument Co.
420 N. LaSalle St.
Chicago, 111. 60610
145
-------
2. F. W. Dwyer Mfg. Co.
P. 0. Box 373
Michigan City, Ind. 46360
3. Meriam Instrument Co.
10920 Madison Ave.
Cleveland, Ohio 44102
4. Research-Cottrell, Inc.
P. 0. Box 750
Bound Brook, N. J. 08805
5. Western Precipitation Co.
1000 W. Ninth St.
Los Angeles, Calif. 90015
I. NEEDLE VALVES
1. American Instrument Co.
8030 Georgia Ave.
Silver Spring, Md. 20901
2. Cajon Co.
32550 Old South Miles Rd.
Salon, Ohio 44139
3. Hoke Manufacturing Co.
P. 0. Box 501
Tenafly, N. J. 07670
4. Nupro Company
15635 Saranac Rd.
Cleveland, Ohio 44110
5. Whitey Research Tool Co.
5679 Landregan St.
Emeryville, Calif. 94608
J. ROTAMETERS
1. Brooks Instrument Co.
407 W. Vine St.
Hatfield, Pa. 19440
2. F. W. Dwyer Mfg. Co.
P. 0. Box 373
Michigan City, Ind. 46360
3. Fischer & Porter, Inc.
County Line Rd.
War-minster, Pa. 18974
146
-------
4. Ideal Precision Glass Co.
Manostat Division
P. 0. Box 287
Carlstadt, N. J. 07072
5. Men am Instrument Co.
10920 Madison Ave.
Cleveland, Ohio 44102
6. Schutte & Koerting, Inc.
2239 State Rd.
Cornwells Heights
Bucks County, Pa. 19020
K. TUBING FITTINGS
1. Cajon Co.
32550 Old South Miles Rd.
Solon, Ohio 44139
2. Crawford Fittings Co.
29500 Solon Rd.
Solon, Ohio 44139
3. D & G Plastics Co.
P. 0. Box 209
Kent, Ohio 44240
4. Hoke Manufacturing Co.
P. 0. Box 501
Tenafly, N. J. 07670
147
-------
APPENDIX I
CALCULATION OF S02 LOSS AND pH OF CONDENSATE
The following calculation (33) gives condenser losses and pH of the
condensate for various gas phase concentrations of SO^. A constant water
content of 6% is assumed.
°2
co2
Condenser Feed
(all vapor) ^
NF = 1 mol
Pp = 1 atm
Typical Analysis
(20% excess air)
= 76.2 mol %
= 3.4
= 14.2
Condi
i
N
n
C
t N
MSO
pH
to Analyzer .
v (mol)
so2 (atm)
ondensate
(mol)
(mol/liter)
2
H20 = 6.0
SO, = 0.2 (from 3%S in coal)
so.
trace
Given: PS« p = partial pressure (atm) of S02 in feed
PU c = partial pressure (atm) of water in feed
T = condenser temperature
Determine: PSQ = partial pressure (atm) of S02 in off-gas
M
SO
pH - pH of condensate
in condensate
(Note: MSQ is a unique function of pH of condensate.
Therefore, pH can potentially be used to cor-
rect gas analysis for S02 loss in condensate.)
149
-------
Worst Case: (1) Assume all water 1s condensed.
(2) Assume condensate and off-gas reach equilibrium with
respect to dissolved S02.
A. SOg-HpO VAPOR-LIQUID EQUILIBRIUM
T = °K
[SO (aq)] iH2l ,_.,
S02(v) = S02(aq) K, = * (J } log^K, =1^1- 7.3756 (1)
[H+][HSO:]
- 4.7150 (2)
[S02(aq)J
]
loginK,/i =-7-2(0 25°C) (3)
[HS03]
H,0 = H* + OH" 1C. = [H+][OH"] loginK,, = -14(@ 25°C) (4)
(results not sensitive to
2
(results not sensitive to
[S02] = total dissolved S02 = [S02(aq)] + [HSO~] + [SOj]
Unknowns: [H*]. [OH'], [S02(aq)], [HSO"]f [SO*] = 5
Use four equations above plus:
Electrical Neutrality: [H+] = [K03] + 2[S03] + [OH"] (5)
Method of Solution: Solve for one of the sensitive variables as a function
of Pcn . Insensitive variables (species present in minor amounts) are OH",
_ 5U2
SOI. Eliminate these from Equation (5) using Equations (4) and (3), (5) be-
•J
comes:
150
-------
Use Equations (1) and (2) in (6) to eliminate [HSOg] and [S02(aq)], thereby
obtaining [H+] = f(Pcn ).
[H+] = [HSOI] 1 + - +
3 + +
K1KH
SO
2K, 1C,
[H
or [H+]2 = K,K PQ 1 + - + - (7)
1 H S02 [H+] [H+.]
Results of solving Equation (7) shown in Figure 1-21.
Using [H ] found from solving Equation (7):
Calc [SO(aq)] from Equation (1)
[HSOg] from Equation (2)
[SO^] from Equation (3)
then [S02] = [S02(aq)] + [HSO^] + [SOJj]
Results shown in Figure 1-22 ([$02] vs pH). These are consistent with tabular
data for S02 in water given in Perry's Handbook.
Therefore, calculations! procedure was incorporated as a subroutine in
subsequent calculations.
B. MATERIAL BALANCE AROUND CONDENSER
PSO ,F(atm) = PSO (atm) NV(mol) + [S02] ('ir)
/mol H90\
55\-^-J
151
-------
1000
100 -
CM
o
00
n.
10
1.0
25'C
Figure 1-21. Partial Pressure S02 vs pH.
2 3
PH
152
-------
Figure 1-22. Liquid Concentration of S02 vs pH.
1.0
10
-1
-------
„
/N»
p
[SO.]NL
SO,/1""1 ny' Mv 750 55~~
7gn „ i
/•MM !_!*• 1 / W P CA 1M"
en rVmm Hg; - -?TS- I 5CLJN
•'up MY oUp»r 33 f.
Equation (8) is solved simultaneously with vapor-liquid equilibrium in
attached Fortran program.
Fortran program, which can be used for any dibasic acid, is attached.
Requires data of first and second ionization constants and Henry's Law con-
stants (as f(T) if available).
Runs for S02 were made from PSQ F of 10"6 to 10"2 atm (1 to 10,000 ppm)
at 25°C and 50°C, assuming 6% H20 in feed and all water condenses.
Results indicate negligible loss of S02 in condensate.
NOTE: S02 is more soluble than nitric oxide and carbon dioxide.
Therefore, these should not be effected by condenser.
However, N02 is more soluble, but dissolution more com-
plex.
(2N09 + H90 = HNO, + H+ + NOl)
£. (. i. 3
[3HN09 = H+ + NO! + 2NO + H90 (slow in cold)]
b «3 £
154
-------
S02 CONDENSER LOSS
tn
tn
T(°C)
25.0
25.0
25.0
25.0
25.0
25.0
25.0
25.0
25.0
25.0
s,<->
0.999E-06
0.199E-05
0.300E-05
0.399E-05
0.499E-05
0.600E-05
0.700E-05
0.799E-05
0.900E-05
0.999E-05
Nv
0.940E 00
0.940E 00
0.940E 00
0.940E 00
0.940E 00
0.940E 00
0.940E 00
0.940E 00
0.940E 00
0.940E 00
NL
0.600E-01
0.600E-01
0.600E-01
0.600E-01
0.600E-01
0.600E-01
0.600E-01
0.600E-01
0.600E-01
0.600E-01
[SO 1 mo^
2 a
0.122E-03
0.178E-03
0.221E-03
0.257E-03
0.289E-03
0.318E-03
0.345E-03
0.371E-03
0.394E-03
0.417E-03
PH
0.391E 01
0.375E 01
0.366E 01
0.359E 01
0.354E 01
0.350E 01
0.347E 01
0.344E 01
0.341E 01
0.339E 01
Pso2
0.920E-06
0.192E-05
0.293E-05
0.395E-05
0.498E-05
0.601E-05
0.704E-05
0.808E-05
0.912E-05
0.101E-04
Pso2
PS02,F
0.920E 00
0.960E 00
0.978E 00
0.989E 00
0.996E 00
0.100E 01
0.100E 01
0.101E 01
0.101E 01
0.101E 01
-------
APPENDIX J
CALCULATION OF N02 ABSORPTION RATE
The rate of absorption of N02 from the sample stream can be calculated
If the rate of mass transfer is known. It Is assumed that the absorption
is klnetically controlled with flux (35):
N = 25 x ID'6 P
2
where: N = flux in g-moles/cnr sec
PNO = Part^ Pressure N02 i
In the diagram below, the interior walls are assumed to be totally wetted
with condensate.
m
g-moles/sec
X X+AX
A material balance on the differential element gives:
Input = m Y.,n i
MU2 X
Output = m Ywn | + N TT D AX
WU2 X+AX
Accumulation = 0
These terms combine to give:
dPNO, , * D P
_2 m 25 1(J-6
m
157
-------
where: P = total pressure (atm)
D = tube Inside diameter (cm)
m = flow rate (g-moles/sec)
Y = mole fraction
This equation Integrates to:
(PNO >
_
N02 initial "
For a flow rate of 2 slpm through a tube of 1/4 o.d. x 0.040 wall, the ML
concentration decreases by 50% in a tubing length of about 1 ft:
x- - ,n0.50
(25xlO~b)(Tr)(0. 4318)(1)
X = 30.4 cm
158
-------
INVESTIGATION
OF
EXTRACTIVE SAMPLING INTERFACE PARAMETERS
PART II
FIELD DEMONSTRATION
159
-------
INTRODUCTION.
The third phase of this program was a field demonstration of interface
systems designed according to the data and information presented in Part
One of this report. Initially it was intended to demonstrate systems on the
following Category I sources:
a. coal-fired power plant
b. oil-fired power plant
c. sulfuric acid power plant
d. nitric acid power plant
Due to lack of program resources only the two power plant systems were inpleme
In keeping with the principles described in Part One, the systems im-
plemented were designed to demonstrate adequate interface systems in a minimum
sense. Particularly, it was intended to demonstrate that probe filter back-
flushing is not necessary, that sample lines heated above the sample dewpoint
are not necessary and that expensive samples line materials are not required.
Also, it was intended to demonstrate a through-the-probe-tip-fliter calibratio,
system.
The results of the demonstration program are inconclusive. However,
there is no evidence which would indicate that the assumptions underlying the
major items mentioned above are in error. The inconclusive results stem from
the difficulties of operating complex field programs with experimental equip-
ment under difficult-to-impossible operating conditions. The latter included
extremely inclement winter weather and frequent boiler shut-downs resulting
from the energy crisis of the winter of 1973-1974.
Lessons from the experience of this field program included under-estimati
the difficulties involved in the start-up of a newly designed system and the
problems associated with the simultaneous demonstration of two systems using
time-shared field crews.
161
-------
DESCRIPTION OF DEMONSTRATION SYSTEM
BACKGROUND
The measurement systems were designed according to the design
information presented in our Interim Report. At the time of writing the
Interim Report two general arrangements for pump and moisture removal
systems were identified. These systems are depicted in Figure II-1. The
system depicted in Figure 11-1 a was selected because removing condensate
from a pressurized water knockout trap is considerably less difficult than
from an evacuated trap. Furthermore, with the condensation taking place at
the high pressure side of the pump, it is not subject to additional conden-
sation forming before reaching the analyzers, whereas, this condition is to
be expected for the case shown in Figure Il-lb if the sample temperature
conditions are about equal before the pump and before the analyzer.
The pump configuration chosen lead to problems with pump life
which were not entirely unanticipated; however, the circumstances of seeking
a minimum system favored the simpler instrumental trade-off for the potentially
lower reliability of the pump.
All data collected were with a system of the type shown in Figure
Il-la. However just before the end of the field program, an attempt was made to
convert a system to one of the type shown in Figure II-Ib, but no program re-
sources remained for operation of the system. The detailed description of the
fabrication requirements for the demonstration system is presented in Part
One, Appendix G.
SAMPLING AND INSTRUMENTAL LOCATIONS
The demonstration program took place at two sites:
1. Boston Edison's Edgar Station at Weymouth, Massachusetts
(Oil-fired plant)
2. Public Service Company of New Hampshire's Merrimack Station
at Bow, New Hampshire (Coal-fired plant)
163
-------
V
PROBE
FILTER
PUMP
DEMISTED
FINE FILTER
WATER
KNOCKOUT
TRAP
Figure Il-la
ANALYZER
>VENT
^
VpRoe
FILTE
< . 1 — J 1
£ DEMISTER ( ^ ~ '
O ^- r-i n r- f-i i -rr- o V / A Kl A 1 Y 7 C"
' - FINE FILTER ^A-—^ AiNAi_T^.t
PUMP
E
WATER
KNOCKOUT
TRAP
R
Figure Il-lb
Figure II-l. General Arrangement of Pump in Sampling Systems.
-------
These plants were of different designs. The oil-fired plant
was the stack through the roof design while the coal-fired plant was an
inline design with the stack located outdoors behind the boiler. These
power plant differences affected the sampling interface design only in the
amounts of sampling line located outdoors and indoors. This line length
was significant because electrically heated sampling lines were required
and used for all outdoor exposures since the ambient temperatures during the
demonstration periods were normally sub-freezing. Where exposed to the
weather, sampling lines were insulated with electrically-traced Teflon
(Dekoron ). Sampling lines used indoors were simply polypr'oplene tubing.
The sampling line temperatures were not maintained above the
sample dew point- Therefore, it was necessary to pitch the lines from the
sampling probe back to the condensate knockout trap in order to drain con-
densate from the sampling lines.
Figures II-2 and II-3 depict the power plant configuration and
interface installations for the oil-fired and coal-fired sites respectively.
FABRICATION OF SAMPLING SYSTEM
Probe Assembly
The probe assemblies were identical for both demonstration
sites;only the depth of probe insertion varied. Insertion depths were 0.92 m
and 0.76 m for the oil-fired and coal-fired sites respectively. The probe
assembly incorporated the following features:
a. sample nozzle
b. coarse particulate filters
c. introduction point for zero and span gases
d. mechanical support for the above
165
-------
ROOF FEED THRU
DEKORON
f
-/
ROOf
STACK
GAS
FLOW
nHOm DEKORON
^\\\\\ \-TVV Roo7 L iNT
ROOF FEED THRU
LEVEL 7
POLYPROPYLENE LINES
LEVEL 6
PROBE
UNIT 4
PRECIPITATOR
CONTROL
RM
RUMENT
RACK
Figure I1-2. Oil-Fired Site.
166
-------
INSTRUMENT RACK
ELECTROSTATIC
' i
PRECIPATATOR
PLAN VIEW
DUCT
DUCT
•- PORTS
0 /-PROBE
2* ^ FLOW
0^^^
'• 20m "•
DEKERON
«u f~
20m T
I/u
I0m
STACK || |
X
a
ELECTROSTATIC
' PRECI PATATOR
HOPPER ROOM
H~1 INSTRUMENT
I I RACK
FAN ROOM
DUCT
\\\\\\v\\\\
ELEVAT ION
Figure II-3. Coal-Fired Site.
\ \
-------
The arrangement of components is depicted In Figure II-4. During sampling,
sample gas is drawn in through the downstream-facing nozzle, passes through
the check valve in the forward direction and is filtered by a medium porosity
(~5ym) alundum thimble. The filtered sample gas is then sucked through a 6.35
mm 316 stainless steel tube (located inside the black iron pipe support) to
outside the duct wall; where a connection is made with tne sampling line.
Sample gas is brought into the duct through a 6.35 mm, 316
stainless steel tube also located inside the black iron pipe support. Inside
the duct the calibration line is brought out through the black iron pipe and
connects with a 316 stainless steel tee located between the duct valve and the
filter holder. When calibration gas is introduced under pressure, the check
valve is closed and the calibration gas passes through the filter (the check
valve poppet was removed from both systems in the course of the program, see
Discussion section).
The mechanical support for the probe is taken through a pipe
cap modified into a port connector and probe clamp. This connector was
screwed into the existing port bushing in the duct wall.
CALIBRATION AND SAMPLING LINES
Calibration lines from gas tank regulators to the sample probe
assembly were 6.35 mm o.d. x 0.076 mm wall polypropylene. Fittings were
nylon compression type.
All indoor sampling lines were 6.35 mm, o.d. x 0.076 mm wall
polypropylene tubing. All outdoor sampling lines were electrically traced
TM
Teflon tubing Dekoron type 2150-045. This tubing was used since it is
less expensive than fabricating long sections of heated polypropylene tubing.
Heating was used only to prevent the sample from freezing in the
sample line. Power to heat the lines was controlled by a variable transformer,
TM
Voltage settings were determined from the Dekoron technical literature on
168
-------
6AS
FLOW
NUPRO
-CHECK VALVE
WESTERN PRECIPATION
-ALUMDUM Fl LTER
AND HOLDER
SCREW IN
5 (f^Etffl
SUPPORT OF 1/2
BLACK I RON PI PE
r— PORT CONNECTOR
/ AND CLAMP
r
2"
t~ — ^^- -t^-^_— ~ /JT.">
^
-------
temperature rise above ambient for power input. The voltage settings were
calculated to provide about a 5°C temperature rise above minimum expected
ambient temperatures. An attempt was made to monitor the temperature of
the sample gas exiting from the heated tubing using thermocouples and a
recorder. However, recorder difficulties precluded these measurements.
The approximate length of each type of tubing for the two sites
are shown in Table II-l.
TABLE II-l.
APPROXIMATE LENGTHS OF SAMPLING LINE
FROM PROBE TO INSTRUMENT RACK
LINE OIL-FIRED SITE COAL-FIRED SITE
TM
Dekoron 9m 20m
Polypropylene 20m 1.5m
INSTRUMENT RACKS
Other than the probe assembly and sampling lines the monitoring
systems were packaged into instrument racks. The locations of the instrument
racks at the test sites are shown in Figures II-2 and II-3. These racks in-
cluded the following:
1. analyzers
2. sample pump
3. moisture removal equipment
4. recorders
5. automatic and manual calibration control
6. calibration gases
170
-------
The scheme was similar for both the oil-fired and coal-fired
sites. However, the choice of analyzer dictated differences in the moisture
removal equipment. The oil-fired site was equipped with a NDUV (DuPont
Model 400) S02 analyzer and a chemiluminescence (Thermoelectron Model 10A)
NO-NO analyzer. As discussed in Part One these analysis methods required
A -.-___
only the removal of liquid water from the sample stream. Figure II-5 shows
schematically the sample and calibration gas connections. The tubing lengths
are shown. All tubing was 6.35 mm o.d. x 0.076 mm wall polypropylene. Figure
11-6 shows pictorially the mechanical arrangement of the instrument rack.
The coal-fired site was equipped with a NDIR (INTERTECH) NO Analyzer
and an electrochemical cell (Dynasciences) S02 Analyzer. The electrochemical
cell analyzer requires only the removal of liquid water from the sample stream,
but the NDIR Analyzer also requires that the water vapor in the sample stream by
held at a constantly low level. For this reason, the sample stream to the NDIR
analyzer was passed through a Hankisons refrigerated condenser upstream of the
analyzer. The sample and calibration gas connection for the instrument racks
is shown schematically in Figure II-7. Tubing was 6.35 mm o.d. x 0.076 mm wall
polypropylene. Figure II-8 shows pictorially the front and control panels of
the instrument rack.
Components common to both systems were the following:
1. sample pump - Thomas Industries, 107CA110
2. ball float traps - Armstrong, 11-LD
3. flow meters with metering valve - Dwyer, RMS-3-SSV
4. fine filter - glass wool plug in a -1 cm x 12 cm long diameter
glass tube
The calibration system was either automatically or manually con-
trolled. On automatic control, the monitoring system was zeroed and spanned
on S02 and NO in each 24-hour period. Each calibration gas was passed through
the system for approximately 10 minutes. The circuit diagram for the control
system is presented in Part I. The calibrations were prepared and analyzed by
Matheson Gas products and supplied in 1A cylinders. The zero gas was nitrogen
and the S02 and NO preparations were in nitrogen.
171
-------
BYPASS CONTROL
VA LV E r=,
PRESSyRE GM/GE
MANOMETER
FROM PROBE
TO PROBEo
PO
SAMPLE
.SAMPLE PUMP
BALL
FLOAT
TRAPS
IN OUT
DUPONT
S02
ANALYZE R
RECORDER
RECORDER
N
TECO
NO-MOg
A
N
A
L
OUT6 OVENT
I
GAS CALIBRATION
SYSTEM SOLENOID
- , VALVES
IN FROM GAS
CYLINDERS
& 30 cm
GAS FLOW VALVES
Figure 2-5
Figure II-5. Schematic of Sample and Calibration Gas Connections for Analyzers.
-------
NO
MIXED
so2
MIXED
OUPONT
PHOTOMETRIC
ANALYZER
TECO
READOUT
RECORDER
NITROGEN
O O G O
VALVES *!L*
<9 Q 0 Bl
CLOCK
TIMER
DU PONT
RECORDER
CONTBOL
•UNIT
CONVE »TER
TECO
CHEMILUM-
INESCENT
NQa
ANALYZED
PUM P
AC
I NPUT
PUMP
Figure I1-6. Instrument Rack
-------
PRESSURE
GUAGE
F ROM PRCBE
--
-oe>-
CALIBRATJCN
GASES
DfAPH RAGM
SAM PLE PUK D
TC STACK
54cm
WATER
MANQMETE R
FINE
l|iLTER FLOWMETERS
/
1
&cm
1
r'l
/
£/ .
IN OUT
HANK ISDN
EVAPORATOR
90cm
I8crn
BALL
FLOAT
TRAP
LIQUID
DRAIN
90cm
IN FRO'-
CYLINrERL;
0-30
MA
RUSTRAK
RECORDERS
36cm
72cm
22cm
FINE
FILTER
INTERTECH
NO ANALYZE R
SAMPLE IN
VENT
-o
SAMPLE
IN
DYNASCIENCE
so2
ANALYZER
VENT
-O
BALL
FLOAT
FILTER
L IQUID
DRAIN
Figure II-7. Schematic-of Sample and Calibration Gas Connections
for Instrument Racks.
174
-------
_1
MV
MA
PROGRAM
Tl M E ^
n
n
HANKISON
EVAPORATOR
o
o o oo
VALVES
9 Q 9
[
PWR
CLOCK
D
TIMER
IN TERTECH
N Dl R
M ON I TOR
AI R
0
PC L U T I 0 N
MONITOR
O O Q O O
DYNASCIENCES
1J
Figure I1-8. Front and Control Panels of Instrument Rack,
175
-------
OPERATION
GENERAL
In preparation for the field demonstrations, the instrument racks
were fabricated in the Maiden shops. While fabrication was underway, a par-
ticulate survey was made of both sites by EPA Method 5, and the sampling and
calibration lines installed between the intended probe and instrument locations.
After fabrication of the racks, operability was tested in the lab-
oratory for sample flow, automatic calibration sequence, recorder operation
and moisture removal from sample stream. Then the instrument racks and probes
were installed: the first at the oil-fired site, and the second at the coal-
fired site.
The problems of start-up, i.e., setting flow rates, setting pressures,
setting calibration timers, checking for leaks, setting power for heated sample
lines, etc., required more attention and time than was initially anticipated.
OPERATIONAL DATA
Table II-2 is a compilation of the operating data of the monitoring
system by test site.
OPERATIONAL PROBLEMS
The following is a discussion of problems that were encountered
during the operation of the system.
During the initial operation of the system at the oil-fired
site beads of water in the polypropylene tubing were noticed downstream of the
traps and the glass wool of the fine filters was soaked. It was suspected that
condensate flowing into the pump was being re-evaporated as the sample was
heated up in the pump. Indeed to the touch, the sample line into the pump was
noticably cooler than the outlet line. Evidently condensation in the sample
gas was still occurring after the ball-float trap. To correct this situation
177
-------
TABLE II-2
OPERATIONAL DATA
FUNCTION OIL-FIRED SITE COAL-FIRED SITE
Sample flow
Pump suction
Pump pressure*
Heated tubing setting
SCL span gas
NO span gas
Calibration Gas Pressure**
Room temp, at instrument rack
* 2 1/min
3.0 kN-m"2
20-40 kN-m"3
35 volts
381 ppm
490 ppm
- 40 kN-m"2
- 33°C
- 2 1/min
3.0 kN-m"2
20-40 kN-m"2
30 volts
1070 ppm
790 ppm
- 40 kN-m"2
- 28°C
Pulsations in the flow at the exit of the pump made the pressure guage
needle fluctuate between the readings.
Pressure was set as low as regulator would permit
178
-------
a cooling coil of about 1 m of polypropylene material was made and Inserted
between the pump and the ball-float trap. Then the fine filter which was
connected with a short line to the ball-float trap was relocated close to the
flow meter (about 1 m of sampling line down-stream) and the sample line pitched
so as to drain any condensete in this line back into the trap. The cooling
coil alteration worked well enough so that no condensate was noticed forming
after the traps.
During most of the demonstration period the boiler at the oil-
fired site was shut down. These shut downs were a result of a fuel saving
scheme throughout the Northeast during the energy crisis of the winter of 1973-
1974. While the coal-fired site also had shutdowns during this period, these
were not as frequent or for as long a duration as the oil-fired unit. (See Table
II-3). The shutdowns of the coal-fired plant were the results of equipment break-
downs. These shutdowns made the continuous operation of the equipment difficult
(coal-fired) to Impossible (oil-fired).
At the oil-fired plant, it was found that the check valve at
the probe tip seemed to work well until the boiler was shut down and then re-
started. After this cycle the check valve was found stuck in the closed position
after calibration. The probe was removed and the valve examined visually. There
was an apparent caking of particulate matter around the poppet of the check valve.
It was reinstalled and it operated correctly until the next boiler shutdown-startup
cycle then the problem re-occured. At this time, the check valve was removed
from the system and calibration was performed by gas flooding (injecting more
calibration gas at the probe tip than was sampled). Later at the coal-fired
site the check valve was observed to stick open. This was at the conclusion of
the demonstration. Hence, the expense of gas flooding is preferred over the
low reliability of the check valve approach, at least for the particular valve
chosen.
179
-------
TABLE II-3
PLANT OPERATION
1973
DATE
12/12
12/13
12/14
12/15
12/16
12/17
12/18
12/19
12/20
12/21
12/22
12/23
12/24
12/25
12/26
12/27
12/28
12/29
12/30
12/31
PLANTS
OIL
On
Off
Off
Off
Off
Off
Off
Off
On
On
On
On
On
On
On
Off
Off
On
Off
On
COAL
On
On
On
Off
Off
Off
On
On
On
On
On
On
On
On
On
On
On
On
On
On
1974
DATE
1/1
1/2
1/3
1/4
1/5
1/6
1/7
1/8
1/9
1/10
1/11
1/12
1/13
1/14
1/15
1/16
1/17
1/18
1/19 '
1/20
1/21
1/22
PLANTS
OIL
Off
Off
Off
Off
Off
Off
Off
Off
Off
Off
*
COAL
On
On
On
Cn
On
Off
Off
Off
Off
Off
Off
Off
Off
Off
Off
On
On
On
On
On
On
*
* Instruments removed from test site
180
-------
Shortly after beginning the demonstration program the
chemiluminescent analyzer failed due to an Insulation breakdown 1n the high
voltage power supply for the ozone generator. This failure was not a result
of the sampling Interface system.
During a time when the oil-fired site was not attended the
manometer fluid upstream of the pump was sucked Into the system. It Is
believed that this problem was caused by the check valve sticking closed.
As a precautionary rough measure the manometers were removed from both
systems.
Almost from the onset there was difficulty maintaining
constant flow through the system at the coal-fired site. The flow rate
tended to decrease steadily. Originally it was thought that the pump bypass
control valve setting was changing due to vibration. The valve was replaced,
but the problem persisted. Then the pump was removed and the head dis-
assembled. Inspection showed that the read valves were not operating properly
because of the collection of a dark, ash-appear1ng crust with a slight greenish
tinge. This pump had about 500 hours of sampling operation. The replacement
pump failed completely by the end of the program with less operation time.
At the coal-fired site condensate flooded the system due to
Inadequate gas-liquid separation at both the ambient temperature trap and at
the refrigerated trap. Inspection of the refrigerated trap following the
condensor revealed inadequate designing. The sample flow was over the trap
inlet but not through the ball-float trap itself. Visual Inspection showed
no reason why the subsequent ambient level trap failed to separate, and in a
test of pouring water into the sample Intube, the trap functioned properly.
The program ended as system modifications were being implemented to remedy
this problem.
181
-------
The weather caused operational and safety problems principally
by limiting the field crew activities. During the course of the field program,
ambient temperatures were frequently below freezing with snow and freezing rain
common. These conditions interferred to a major degree with the manual sampling
of the flue gas, and somewhat with the operation of the equipment. A weather
related problem which had not been accounted for in the sampling interface design,
is when a boiler.shuts down with a below freezing temperature. At the coal-
fired site there was such a shutdown and condensate in the probe external to the
stack froze and blocked the sample, requiring probe removal for thawing. During
operation this point of the probe was maintained above freezing by the hot sample
gas flow. Installation of electric heaters on the probe which could be turned on
during a re-start of the system after shut down will correct this difficulty.
182
-------
QAI&
Instrumental and manual data were taken on both the oil-fired and
coal-fired flue gases. Due to the difficulties described in the previous
section, the volume of data collected is limited. Background data on
particulate loadings (EPA Method 5) and gas composition were taken before
installing the sampling systems. These data are shown in Table II-4. From
»
the monitoring systems, data were taken on response times. These data are
shown in Table I1-5. Simultaneous manual and instrumental measurements were
made on the flue gases. The manual measurements were EPA Method 6 for SCL and
EPA Method 7 for N0tf.
n
The manual and instrumental data are shown in Tables II-6, II-7, and II-8.
The agreement of the instrumental methods with the manual methods is
fairly good, confidence 1n the values for the NDIR instrument is poor. The
range of the manual NO data seems large for this type of process stream. How-
A
ever the reproducibility of Method 7 results has been reported as 7% of value
for a one standard deviation interval*. This explains in part variations in
the manual data. The small mean difference for the widely scattered results in
Table II-8 is explicable as chance.
Two of the Method 6 results in Table II-7 were excluded from the analysis
because the very low Method 6 results occurred for samples for which ice was
observed in the peroxide collection solutions in the impinger train.
The zero and span data are variable. It is suspected that flow changes,
due to the pump problem discussed before, affected the response. These problems
precluded the use of these data to establish instrumental performance changes
due to operation life. In fact, the strip chart data is in a state which pre-
cluded data reduction.
Hamil, H. F. and D. E. Camann, "Collaborative Study of Methods for the
Determination of Nitrogen Oxide Emission from Stationary Sources (Fossil
Fuel Fired Stream Generator): EPA Contract No. 68-02-0623, Southwest
Research Institute, San Antonio, Texas (December 10, 1973).
183
-------
TABLE II-4
BACKGROUND DATA
COAL-FIRED
TEST
Mass loading
Average
Gas Moisture
cc2
°2
N2
Gas Velocity
Date
1
95.9
7.24
14
5
81
11.1
11/30
2
203
153
6.32
14
5
81
11.1
12/3
3
161
5.17
14
5
81
11.1
12/3
1
99.
10.
8
9.
82.
9.
OIL-FIRED
3
66
5
5
14
11/24
62
81
10
8
9
82
9
1
2
.4
.2
.03
.5
.5
.13
1/24
81
11
8
Q
82
9
1
3
.7
.94
.5
.5
.17
1/24
(UNITS)
mg-m"
mg-nf
%
%
%
%
m-S'1
184
-------
TABLE II-5
RESPONSE TIME TO SPAN GAS FOR
INTERFACE SYSTEM PLUS ANALYZERS
OIL-FIRED COAL-FIRED
S02 1 min. 3/4 min.
NOX 1/2 min. 2 min.
185
-------
TABLE II-6
OIL-FIRED PLANT
DATA
DATE
12/20/73
12/20/73
12/20/73
12/20/73
12/20/73
12/20/73
INSTRUMENTAL
NDUV ANALYZER
(ppm)
210
180
240
250
230
270
MANUAL
EPA METHOD 6
(ppm)
239
311
297
340
323
327
DIFFERENCE
INST. - MANUAL
(ppm)
- 30
- 130
- 60
- 90
- 90
- 60
Span = 1000 ppm
Mean = - 80O-8% of span
o • 30
100 x
3.92a
span
12% (95% confidence Interval)
186
-------
TABLE II-7
COAL-FIRED PLANT SOg DATA
DATE
12/18/73
12/18/73
12/18/73
12/18/73
12/18/73
12/18/73
12/18/73
12/26/73
12/26/73
12/26/73
12/26/73
12/26/73
12/26/73
12/26/73
Span = 5000 ppm
INSTRUMENTAL
ELECTROCHEMICAL
CELL ANALYZER
(ppm)
1300
1300
1200
*
*
*
1400
1500
1400
1400
1400
*
1500
1000
100
MANUAL
EPA METHOD 6
(ppm)
1470
164
227
227
2025
1476
1443
1605
1788
1777
1734
1680
1886
1734
Mean = -300
a = 200
Y 3.920 _ lfi«
x span" - 16%
DIFFERENCE
INST. - MANUAL
(ppm)
- 200
**
**
-
-
-
0
- 100
- 400
- 400
- 300
-
- 400
- 700
O-6% of span
(95/8 confidence Interv.
* Manual Calibrate Interference
** Manual Data Excluded
187
-------
TABLE II-8
COAL-FIRED PLANT N0y DATA
DATE
12/18/73
12/18/73
12/18/73
12/18/73
12/18/73
12/18/73
12/18/73
12/26/73
12/26/73
12/26/73
12/26/73
12/26/73
12/26/73
12/26/73
12/26/73
Span = 1600 ppm
•
INSTRUMENTAL
NDIR ANALYZER
(ppm)
1100
1000
900
1080
1000
900
960
1000
*
870
1000
920
920
870
*
MANUAL
EPA METHOD 7
(ppm)
455
1124
1092
1156
1295
1177
962
757
624
792
1498
1063
1017
1102
831
Mean = -703?
o = 280
100x3.92o _ EJ., /„
DIFFERENCE
INST. - MANUAL
(ppm)
650
- 120
- 190
- 80
- 290
- 280
0
240
-
80
- 500
- 100
- 100
- 230
-
-4% of span
IM rnnfiHonre •fn+opu:
span
188
-------
The system response times which were taken, manually are consistent with
the value predicted by the design data in Part One, viz., 3 1/4 minutes.
The example used a flow rate of one liter per minute whereas the demonstration
system operated at about two liters per minute. The demonstration system also
had somewhat shorter sampling and connection lines and smaller back mixing
volumes. These differences resulted in faster response times. For the case
of the electrochemical cell Instrument, the analyzer and not the interface
system was the response time limiting element of the monitoring system.
There were some semi-quantitative data collected on the particulate
loading on the coarse filters. The alundum thimbles from the oil-fired
and coal-fired plants collected 0.33 grams and 0.48 grams of particulate
respectively. Laboratory tests indicated that the filter should work with coal
fly ash collection as high as about 70 grams in the thimble. Due to down times
for both the demonstration system and the boiler, it is not possible to specify
with accuracy the cumulative flow through these filters, hence filter life.
However, assuming that the systems operated for a total of fifteen days, which
is a reasonable estimate, then an operation time of one year on the coal-fired
plant without removing the filter for cleaning is to be expected.
189
-------
DISCUSSION
From the results of this field program many of the elements of a
minimum type sampling system were Identified by the design data, 1n particular
the following:
• A. a simple probe filter without back-flushing
B. low cost polypropylene sample lines (for temperatures
above freezing)
C. moisture removal from the sample stream to ambient level dew
points for the analysis methods other than NDIR*
Additionally the system can be calibrated by introduction of span and
zero gases into the probe tip filter. Gas flooding of the probe upstream of
filter with the calibration gases was found to be perferred over the originially
designed system (check-valve) on the basis of reliability.
Failure of the sampling systems was associated with inadequate moisture,
mist, and fine particulate removal,' and sample pump failures. Both demonstra-
tion systems were flooded with liquid, the coal-fired plant systems twice.
Three sample pumps were available to this program and all three failed. All
pumps failed while Installed on the coal-fired system; although one pump,which
was removed from the oil-fired demonstration system,failed within four hours
after installation on the coal-fired system probably from exposure effects from
the oil-fired plant sample gases. Since the demonstration system had the
moisture removal system after the pump*it is concluded that the sample pump
requires the protection of the moisture removal system and probably the protectic
of a fine particle filter. This arrangement requires a somewhat more complicatec
approach as indicated In the discussion in Part I. An alternative approach
may be to continue to use a post pump moisture removal system with a more ex-
pensive sample handling pump; e.g., the type the Metal Bellows Company is develo
ing for handling raw flue gas. These pumps are under development and the field
data are not yet available .
* This was not field demonstrated for the chemiluminescent analyzer due to
instrument failure not related to the interface system.
191
-------
Considering the system failure 1n this field program, one is tempted
to use a ne plus ultra moisture removal system ahead of the most robust
sampling pump available. This approach would not be representative of a "minimum"
type system. A "minimum" type system would be some compromise between this
over designed approach and the system which was implemented in this demonstra-
tion program. The determination of the lowest-cost suitable approach must be
established through further field tests. This program has not resolved this
issue, Hence, it can only offer the experience of this demonstration as a
caveat to the would-be designer that the reliability of any interface system
depends upon protection of the sampling pump and protection of the sampling
system from flooding.
The approach of this program made a principal departure from the approach
of many other practicioners in the requirements for removing condensate from the
sample stream. Many designers have indicated that efforts have to be made to
reduce the absorption of the sample gases, i.e., S09 and NO from the sample
b n
stream. These efforts included sample lines heated above the sample dew point
and condensers designed to remove condensate with a minimum of exposure time
to the sample stream. For the reasons described in Part I, this demonstration
/'
program used a system which took no precautions against sample absorption in
the condensate as the sample and connection lines were operated below the sample
dew point. The results indicate that this approach is sufficient. Interface
systems for coal- and oil-fired effluents need not go to the expense of trying
to reduce absorption losses from the sample stream.
The demonstration of the non-backflushing probe was successful. As
indicated in the previous section the filters used in this demonstration could
be expected to operate for a year or longer without maintenance. This inline
filter design easily permits a through the probe-tip-fliter calibration design.
The check-valve arrangement was not successful on the through probe tip
filter calibration system. Not unexpectedly, the check-valve proved to be a
high maintenance item. However, the gas flooding approach works well. The
costs of the higher calibration gas consumption for gas flooding is deemed
worthwhile for the reliability gained.
192
-------
In conclusion this demonstration program vouched for many^of the
"minimum" type interface system components identified in Part I. However,
*•-»--•" *
the reliabilities of the "minimum" type sample pump and condensate removal
system were poor. It is not clear at this time just what improvement of these
components is required in order to achieve a sufficient system without over
design. However, improvements of this nature do not lend themselves to
engineering design calculations; rather, they are evolved as a result of trial
and error from field experience. Therefore, further field work is required
\
before the complete "minimum" type system can be specified.
193
-------
TECHNICAL REPORT DATA
(I'll asc re ad laaric nuns vn tlic rcursc before completing)
1 REPORT NO.
EPA-650/2-74-089
4. TITLE ANDSUBTITLE
Investigation of Extractive Sampling Interface
Parameters
7. AUTHOH(S)
K. J. McNulty, J. F. McCoy, J. H. Becker, J. R.
Ehrenfeld and R. L. Goldsmith
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Walden Research Division of Abcor, Inc.
201 Vassar Street
Cambridge, Massachusetts 02139
3 RECIPIENT'S ACCESSION NO.
5 REPORT DATE
October 1974
6. PERFORMING ORGANIZATION CODE
8. PERFORMING ORGANIZATION REPORT NO
10. PROGRAM ELEMENT NO.
1AA010
11. CONTRACT/GRANT NO.
68-02-0742
12. SPONSORING AGENCY NAME AND ADORtSS
Environmental Protection Agency
Office of Research and Development
Washington, D. C. 20460
13. TYPE OF REPORT AND PERIOD COVERED
Final 12 month
14. SPONSORING AGENCY CODE
IS. SUPPLEMENTARY NOTES
16. ABSTRACT
This document is the result of a twelve month, three-phase investigative program
with the intent of providing EPA with sufficient information to permit the
establishment of minimum specifications for the design of continuous extractive
sample interface systems. An extractive sampling interface system is the
equipment associated with an instrumental soure measurement system which extracts,
transports, and conditions a sample of the source effluent. The work in this
program was directed toward an investigation of interface systems for use on
Category I sources for the following instrumental techniques: 1) non-dispersive
infrared analyzers, 2) ultra-violet analyzers, 3) electrochemical sensors, and
4) chemiluminescent analyzers. Both laboratory and field results are reported.
7.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS
C. COSATI I:icld/Gl0llp
Mori toring, Stack gas
SOp - emission measurement
NO - emission measurement
rt
interface systems
sampling systems
stack sampling
13. DISTRIBUTION STATEMENT
Release Unlimited
19 SECURITY CLASS (ThisReport)
21 NO. OF PAGES
207
20 SECURITY CLASS (Thispage)
22. PRICE
EPA Form 22200 (9-73)
194
------- |