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

-------