EPA-650/2-75-051-a

June 1975         Environmental Protection  Technology Series
                                 EVALUATION
                          &TIONARY  SOURCE
               PARTICULATE MEASUREMENT
                                    M

                          L PO
Pi
; T t
                                   «c
                                   55
  o
  UJ
  a
                                J.S. Environmental Protection Agency
                                Office of Research and Development
                                     Washington, D.C.;

-------
                              EPA-650/2-75-051-Q
           EVALUATION

   OF  STATIONARY  SOURCE

PARTICULATE MEASUREMENT

             METHODS
   VOLUME I, PORTLAND CEMENT PLANTS
                  by
    I.E. Howes, Jr . , R. N .  Pesut, and W . M . Henry
         Battelle, Columbus Laboratories
              505 King Avenue
            Columbus,  Ohio 43201
            Contract No. 68-02-0609
           ROAP No. 26AAM, Task 41
          Program Element No. 1AA010
     EPA Project Officer: Dr. Kenneth T. Knapp

        Chemistry and Physics Laboratory
       National Environmental Research Center
     Research Triangle Park, North Carolina 27711
               Prepared for

     U.S. ENVIRONMENTAL PROTECTION AGENCY
      OFFICE OF RESEARCH AND DEVELOPMENT
           WASHINGTON, D.  C. 20460

                June 1975

-------
                        EPA REVIEW NOTICE

This report has been reviewed by the National Environmental Research
Center - Research Triangle Park, Office of Research and Development,
EPA, and approved for publication.  Approval does not signify that the
contents necessarily reflect the views and policies of the Environmental
Protection Agency, nor does mention of trade names or commercial
products constitute endorsement or recommendation for use.
                    RESEARCH REPORTING SERIES

Research reports of the Office of Research and Development, U.S. Environ-
mental Protection Agency, have been grouped into series. These broad
categories were established to facilitate further development and applica-
tion of environmental technology.  Elimination of traditional grouping was
consciously planned to foster technology transfer and maximum interface
in related fields.  These series are:

          1. ENVIRONMENTAL HEALTH EFFECTS RESEARCH

          2. ENVIRONMENTAL PROTECTION TECHNOLOGY

          3. ECOLOGICAL RESEARCH

          4. ENVIRONMENTAL MONITORING

          5. SOCIOECONOMIC ENVIRONMENTAL STUDIES
          6. SCIENTIFIC AND TECHNICAL ASSESSMENT REPORTS

          9. MISCELLANEOUS

This report has been assigned to the ENVIRONMENTAL PROTECTION
TECHNOLOGY series.  This series describes  research performed to
develop and demonstrate instrumentation,  equipment and methodology
to repair or prevent environmental degradation from point and non-
point sources of pollution.  This work provides the new or improved
technology required for the control and treatment of pollution sources
to meet environmental quality standards.
This document is available to the public for sale through the National
Technical Information Service, Springfield,  Virginia 22161.

                 Publication No. EPA-650/2-75-051-a
                                 11

-------
                              TABLE OF CONTENTS
ABSTRACT	    1

ACKNOWLEDGEMENTS	    2

CONCLUSIONS 	    3

     Sampling System Temperature	    3
     Anisokinetic Sampling	    4
     Filter Size	    4
     Nozzle Configuration 	    5
     Filter, Nozzle, and Temperature Interactions  	    5
     Filter Type and Location	    5
     Method 5 Precision 	    6

RECOMMENDATIONS 	    7

     Sampling System Temperature	    7
     Isokinetic Sampling	    7
     Filter Size	    8
     Nozzle Configuration 	    8
     Particulate Filter 	 	    9
     Method 5 Precision	    9

INTRODUCTION	10

OBJECTIVE	11

EXPERIMENTAL WORK AND RESULTS	12

     Experimental Approach	12
     Process and Sampling Site Description	14
           Cement Manufacturing Process	14
           Sampling Site Description  	   16

     Sampling Equipment 	   20
           Particulates	*  .  .   20
           Gas Analysis	28
     Sample Collection and Analysis  Procedures	33
           Particulate Sampling	33
           Sample Recovery and Analysis	35
           Gas Analyses	37

     Test  Descriptions and Results.	39
     Anisokinetic Sampling Experiments	41
     Filter Size, Temperature, and Nozzle Configuration Experiments  .  .   48
     Filter Box/Probe Temperature Experiments  	   55
     In-Stack-Filter Experiments	61

-------
                              TABLE OF CONTENT
                                  (Continued)                              Page

      Characterization of Kiln Emissions  and Particulate Collection. ...  67
           General Chemical Composition of Participates. 	  67
           Selected Chemical Analysis  of  Particulate Collections 	  67
           Impinger Analysis 	  79
           Particle Size Measurements	79
           Gas Composition of Kiln Emissions	87

DISCUSSION	89

REFERENCES	91


                                  APPENDIX A

METHOD-5  -  DETERMINATION OF PARTICULATE  EMISSIONS  FROM STATIONARY SOURCES  A-l

                                  APPENDIX B

STACK GAS MEASUREMENT DATA	B-l

                                  APPENDIX C

SAMPLING  SYSTEM OPERATION DATA	C-l

                                  APPENDIX D

STATISTICAL DESIGN OF EXPERIMENTS TO  STUDY SAMPLING METHODOLOGY 	  D-l

                                  APPENDIX E

CONDENSATION OF SULFURIC ACID AND AMMONIUM SULFATE UNDER VARIOUS
   CONDITIONS ENCOUNTERED IN CEMENT PLANT EMISSIONS	E-l

-------
                              LIST OF FIGURES

                                                                         Page
FIGURE  1.  KILN OPERATION OF PORTLAND CEMENT PRODUCTION	15

FIGURE  2.  SAMPLING PORT LOCATION IN DUCT LEADING FROM PRECIPITATOR
            TO STACK	17

FIGURE  3.  VELOCITY PRESSURE AND TEMPERATURE PROFILE OF CEMENT KILN
            EMISSION DUCT AT SAMPLING LOCATION	18

FIGURE  4.  SCHEMATIC DIAGRAM OF METHOD 5 TRAIN USED FOR SAMPLING CEMENT
            KILN EMISSIONS	21

FIGURE  5.  EQUIPMENT USED FOR SAMPLING PARTICULATES IN CEMENT PLANT
            EMISSION	22

FIGURE  6.  PITOT TUBE HEAD AND THERMOCOUPLE USED FOR VELOCITY HEAD AND
            STACK TEMPERATURE MEASUREMENTS	23

FIGURE  7.  THERMOCOUPLE ASSEMBLY USED TO MEASURE GAS TEMPERATURE AT
            PROBE OUTLET	25

FIGURE  8.  THERMOMETER ASSEMBLY USED TO MEASURE GAS TEMPERATURE AT THE
            BOX FILTER OUTLET	25

FIGURE  9.  SAMPLING NOZZLES	26

FIGURE 10.  OUT-OF-STACK FILTER HOLDER FOR 6.25 -CM-DIAMETER  FILTERS ...  27

FIGURE 11.  OUT-OF-STACK FILTER, HOLDER FOR 10.2-CM-DIAMETER FILTERS ...  27

FIGURE 12.  IN-STACK FILTER HOLDER FOR 6.25- CM-DIAMETER FILTERS	29

FIGURE 13.  IN-STACK FILTER HOLDER FOR GLASS FIBER THIMBLES 	  29

FIGURE 14.  CONCURRENT SAMPLING AT TEST SITE	34

FIGURE 15.  TEMPERATURE MEASUREMENT POINTS	36

FIGURE 16.  PROBE RESIDUES SHOWING CRYSTALLINE MATERIAL OBTAINED
            FROM SAMPLING WITH METHOD 5	70

FIGURE 17.  INFRARED SPECTRA OF PROBE RESIDUE AND AMMONIUM SULFATE
            REFERENCE SAMPLE	71

FIGURE 18.  PARTICLE SIZE DISTRIBUTION OF FILTER CATCH FROM METHOD 5
            SAMPLING OF CEMENT KILN EMISSIONS (RUN 49A)	81

FIGURE 19.  PARTICLE SIZE DISTRIBUTION OF PROBE WASH RESIDUE FROM METHOD
            5 SAMPLING OF CEMENT KILN EMISSIONS (RUN 49A)	82

FIGURE 20.  PARTICLE SIZE DISTRIBUTION OF FILTER CATCH FROM IN-STACK
            SAMPLING OF CEMENT KILN EMISSIONS (RUN 49B)	83

-------
                              LIST  OF  FIGURES
                                 (Continued)
                                                                         Page

FIGURE 21.  PARTICULATES COLLECTED ON METHOD 5 FILTER (RUN 49A) .....   84

FIGURE 22.  PARTICULATES RETAINED IN METHOD 5 PROBE (RUN 49A) . ......   85

FIGURE 23.  PARTICULATES COLLECTED ON IN- STACK FLAT FILTER (RUN 49 B) . .   86

FIGURE E-l.  TEMPERATURE DEPENDENCE OF EQUILIBRIUM CONSTANT .......   E-3

FIGURE E-2.  DEWPOINT AS A FUNCTION OF SULFURIC ACID AND WATER VAPOR
             CONCENTRATION .......................   Er2
FIGURE E-3.  NOMOGRAPH SHOWING TEMPERATURE OF AMMONIUM SULFATE FORMATION
             FOR VARIOUS AMMONIA AND SULFURIC ACID CONCENTRATIONS. ...   E-3

FIGURE E-4.  NOMOGRAPH SHOWING TEMPERATURE OF AMMONIUM BISULFATE FORMATION
             FOR VARIOUS AMMONIA AND SULFURIC ACID CONCENTRATIONS. ...   E-4

-------
                               LIST OF TABLES

                                                                         Page

TABLE 1.  EMISSION SOURCE CHARACTERISTICS	   19

TABLE 2.  COMPOSITION OF FILTER MATERIALS	   30

TABLE 3.  ANALYSIS OF AVAILABLE ELEMENTS	   31

TABLE 4.  pH AND ALKALINITY OF FILTERS	   31

TABLE 5.  RANDOMIZED TEST DESIGN FOR ANISOKINETIC SAMPLING STUDY ....   42

TABLE 6.  SAMPLE WEIGHT DATA-ANISOKINETIC SAMPLING EXPERIMENTS 	   44

TABLE 7.  STATISTICAL ANALYSIS DATA - ANISOKINETIC SAMPLING STUDY
          ISOKINETIC VERSUS HYPERISOKINETIC CONDITIONS 	   45

TABLE 8.  STATISTICAL ANALYSIS DATA ANISOKINETIC SAMPLING STUDY
          ISOKINETIC VERSUS SUBISOKINETIC CONDITIONS 	   45

TABLE 9.  ANALYSES OF VARIANCE-ANISOKINETIC SAMPLING EXPERIMENT
          (ISOKINETIC VERSUS HYPERISOKINETIC CONDITIONS)  	   47

TABLE 10. ANALYSES OF VARIANCE-ANISOKINETIC SAMPLING EXPERIMENT ISOKINETIC
          CONDITIONS VERSUS SUBISOKINETIC CONDITIONS 	   47

TABLE 11. RANDOMIZED TEST  PATTERN  FOR STUDY OF FILTER SIZE, TEMPERATURE,
          AND NOZZLE CONFIGURATION   	   49

TABLE 12. SAMPLE  WEIGHT  DATA-FILTER SIZE,  TEMPERATURE, AND NOZZLE
          CONFIGURATION  STUDY  	   50

TABLE 13. DATA USED IN STATISTICAL ANALYSIS OF FILTER SIZE, TEMPERATURE,
          & NOZZLE CONFIGURATION EFFECTS 	   51

TABLE 14. ANALYSES OF VARIANCE FILTER SIZE, TEMPERATURE,  & NOZZLE
          CONFIGURATION STUDY	   52

TABLE 15. COMPARISON OF LEVEL OF SIGNIFICANT FACTORS - FILTER SIZE,
          TEMPERATURE, AND NOZZLE CONFIGURATION STUDY	   53

TABLE 16. RANDOMIZED EXPERIMENTAL DESIGN TO STUDY THE EFFECT OF PROBE
          AND FILTER TEMPERATURE COMBINATIONS	   56

TABLE 17. SAMPLE WEIGHT DATA-FILTER BOX/PROBE TEMPERATURE EXPERIMENTS. .   57

TABLE 18. DATA FOR STATISTICAL ANALYSIS --.FILTER BOX/PROBE TEMPERATURE
          EXPERIMENT	   58

-------
                               LIST  OF  TABLES
                                 (Continued)

                                                                         Page

TABLE 19. ANALYSIS OF VARIANCE -- FILTER BOX/PROBE TEMPERATURE EXPERIMENT. 59

TABLE 20. RANDOMIZED STATISTICAL DESIGN FOR COMPARISON OF PARTICULATE
          MEASUREMENTS WITH METHOD 5 AND TWO TYPES OF IN-STACK FILTERS . . 62

TABLE 21. SAMPLE WEIGHT AND DISTRIBUTION DATA-IN-STACK FILTER EXPERIMENTS. 63

TABLE 22. DATA FOR STATISTICAL ANALYSIS — IN-STACK FILTER EXPERIMENTS . . 64

TABLE 23. ANALYSIS OF VARIANCE -- IN-STACK FILTER EXPERIMENTS	65

TABLE 24. ANALYSIS OF GRAB SAMPLE AND METHOD 5 PARTICULATE SAMPLES .... 68

TABLE 25. ANALYSIS OF PARTICULATE COLLECTIONS FROM METHOD 5 AND IN-STACK
          SAMPLING	69

TABLE 26. ANALYSIS OF FILTER AND PROBE SAMPLES FROM EXPERIMENTS USING
          VARIOUS FILTER/PRO BE TEMPERATURE COMBINATIONS	74

TABLE 27. CHEMICAL ANALYSES OF SAMPLES COLLECTED IN VARIOUS METHOD 5 AND
          IN-STACK SAMPLING TRAIN COMPONENTS 	 76

TABLE 28. SULFATE IN FILTERS AND PARTICULATE COLLECTIONS 	 77

TABLE 29. IMPINGER COLLECTION DATA	80

TABLE 30. SUMMARY OF MONITORING FOR GASEOUS COMPONENTS OF CEMENT KILN
          EMISSIONS	'	88

TABLE B-l. STACK GAS DATA - ANISOKINETIC STUDY	B-l

TABLE B-2. STACK GAS DATA-FILTER SIZE, TEMPERATURE, AND NOZZLE
           CONFIGURATION EXPERIMENTS 	 B-2

TABLE B-3. STACK GAS DATA-PROBE/BOX FILTER TEMPERATURE EXPERIMENTS . . . . B-3

TABLE B-4. STACK GAS DATA-IN-STACK FILTER EXPERIMENTS	B-4

TABLE C-l. SAMPLING DATA - ANISOKINETIC SAMPLING EXPERIMENTS 	 C-l

TABLE C-2. SAMPLING DATA - FILTER SIZE, TEMPERATURE, AND NOZZLE CON-
           FIGURATION EXPERIMENTS	C-2

TABLE C-3. SAMPLING DATA - PROBE/BOX FILTER TEMPERATURE EXPERIMENTS. . . . C-3

TABLE C-4. SAMPLING DATA-IN-STACK FILTER EXPERIMENTS 	 C-4

TABLE D-l. BALANCED INCOMPLETE BLOCK DESIGN FOR THREE REPLICATIONS OF
           AN EXPERIMENT WITH THREE TREATMENT LEVELS OCCURRING IN BLOCKS
           OF TWO	D-4

TABLE D-2. ANALYSES OF VARIANCE FOR BALANCED INCOMPLETE BLOCK DESIGN OF
           EXPERIMENTS	D-5

-------
                                  ABSTRACT
          A study was performed to evaluate the EPA Method 5 procedure for
measurement of particulate emissions from a cement kiln.  The program included
three series of experiments to study the effects of anisokinetic sampling,
sampling system temperature, filter size, and nozzle configuration on particulate
mass emission determinations.  Method 5 was compared with two types of in-stack
filters in a fourth set of experiments.  The results of the experiments were
analyzed by statistical analysis to assess the significance of the sampling
system variables on observed differences in collected particulate mass.
Chemical and physical characterizations were performed to evaluate the
representativeness of collected particulates and to elucidate the cause of
mass differences introduced by various sampling system operating parameters.
          The data show that sampling system temperature is probably the most
critical factor in application of the current Method 5 procedure (December 23,
1971) to particulate emission measurements from cement kilns.  With the
probe and filter temperatures specified in Method 5, ammonium sulfate and/or ammonium
bisulfate precipitated in the probe giving high and variable weights of collected
particulates.  Operation of the sampling probe with a gas outlet temperature of
about 205C (400 F) and the filter box of 121C (250F) eliminated the formation
of ammonium sulfates in the sampling system.
          Anisokinetic sampling at 0.7 and 1.3 times the isokinetic rate and
the use of two different nozzle configurations, goose neck and 90°, did not
produce statistically significant differences in particulate mass measurements.
Filter size variation produced a significant effect, possibly due to reaction
with sulfuric acid in the emissions.
          Tests with in-stack flat filters yielded results which were not
significantly different from values obtained with the Method 5 procedure
 modified by using 205 C (400 F)  probe and 121 C (250 F)  box filter  temperature.
Results obtained with Munktell in-stack glass fiber  thimbles  (marketed by
Carborundum Co.) were about 18 percent higher than Method 5 values.  The
difference is shown to result from extraneous sulfate formation in the Munktell
thimble.  Analyses of Method 5 filters and particulate catches also indicate
some reaction of sulfur oxides with MSA 1106BH filter medium.
          This report is submitted to the Environmental Protection Agency by
Battelle Columbus Laboratories as partial fulfillment of Contract No.  68-02-0609.
The work reported in this document was completed on February 28,  1975.

-------
                              ACKNOWLEDGEMENTS


          Our appreciation is extended to the Columbia Cement Corporation
for permitting the use of their facilities at the Zanesville Plant to conduct
the experimental sampling program.  The excellent cooperation throughout the
study of Messrs.  H. J. Ivie, C. G. Osberg, Dan Wren, and John Kortz of Columbia
Cement is gratefully acknowledged.
          Several Battelle-Columbus staff members assisted in the field
sampling efforts and they are acknowledged for their dedicated efforts.  The
group includes: Messrs. D. L. Sgontz, R. L. Livingston, W. C. Baytos, E. J.
Schulz, and D. F. Kohler.
          Dr. Ken Knapp, EPA Project Officer is acknowledged for his guidance
and direction of this program.

-------
                                 CONCLUSIONS

          The following conclusions regarding methodology for determining
particulate emissions from cement kilns may be drawn from this study:

                         Sampling System Temperature

          •  The operation of the sampling system at temperatures above the
             currently specified minimum has a significant effect on the
             particulate mass data.  Operation of the Method 5 sampling train
             with a filter box temperature of 121 C (250 F) and the minimum
             specified gas temperature at the probe exit, 121 C (250 F),
             yielded mean particulate emission values in two test series
             which were 23 and 43 percent higher than values obtained with
             the sampling system maintained at the same temperature as the
             stack gas, about 205 C (400 F).  Analysis of the two statis-
             tically designed test series shows that the differences are due
             to system operating temperature and that the effect is sig-
             nificant at confidence levels as high as 98 percent.  The "
             difference in results is attributed to formation of ammonium
             sulfate and/or ammonium bisulfate in the probe operated at a
             gas exit temperature of 121 C (250. F), from ammonia and
             sulfuric acid in the source emissions.
          •  Operation of the Method 5 sampling train with a filter box
             temperature of about 121 C  (250 F) and the gas temperature
             at the probe exit of about  200 C  (392 F)  (-80 degrees Centi-
             grade higher than the above test) gave particulate mass
             emission results which were the same as results obtained
             with the box filter and probe outlet gas temperature of
             about 200 C  (392 F).  The high formation of ammonium
             sulfate or ammonium bisulfate in  the probe or other com-
             ponents of the system was not detected at either of the
             latter temperature conditions.

-------
                  Anisokinetic Sampling

•  Particulate sampling at about 0.7 and 1.3 times isokinetic
   flow rate produced particulate mass emission results which
   were an average 5.9 percent higher and 11.5 percent lower,
   respectively, than results obtained under isokinetic sampling
   conditions.  The mean mass diameter (MMS) of the cement kiln
                                                V
   emissions (after precipitator) was about 0.6 micrometer as
   determined from in-stack collections analyzed by electron
   microscopy.•  Although the observed differences result in
   estimated confidence levels as high as 70 and 85% in the
   statistical analysis, usual statistical interpretations of
   these types of analysis do not declare significant differences
   until the confidence level exceed a much higher specified
   level - usually 95 or 99%.  Thus, based on the higher con-
   fidence level criteria, it is concluded that the observed
   differences are not statistically significant.
                       Filter Size

•  The particulate mass emission measurements obtained with a
   nominal 6.4-cm (2.5-in.) filter in the Method 5 train are
   an average of 8.5 percent lower than results obtained using
   a nominal 10.2-cm (4-in.)  filter.  Based on a statistical
   test of the data, it is concluded with confidence levels
   as high as 94 percent that the difference which resulted
   from the two filter sizes is significant.  The specific
   reason(s) for the effect of filter size was not identified
   but may result from reaction of SO  (different filter area),
   filtration efficiency, or difference in performance due to
   filter holder design.

-------
                  Nozzle Configuration

•  The use of a goose-neck and smooth 90  nozzle in a Method
   5 sampling train did not produce a statistically significant
   difference in particulate mass emission results.  The mean
   difference of particulate measurements made with the two
   nozzle configurations is about 1.6 percent.
      Filter, Nozzle, and Temperature Interactions
•  Statistical analysis of data derived from a confounded factorial
   designed experiment indicates that interaction effects between
   temperature (121 and 205 C), filter size (6.4 and 10.2 cm), and
   nozzle configuration (goose-neck and smooth 90 ) did not
   produce significant differences in particulate mass measurements.

                Filter Type and Location
•  Sampling with Munktell in-stack glass fiber thimbles gives
   particulate mass emission values which are an average of 17
   and 18 percent higher than values obtained with in-stack, flat,
   glass fiber filters and with Method 5 (operated with gas temperature
   of ~205C), respectively.  The in-stack, flat, glass fiber filter system
   (filter and nozzle catch only) yields particulate mass measurements
   which are an average 0.8 percent higher than Method 5 results.  The
   difference between results obtained with the Munktel thimble
   and with the in-stack flat filter or Method 5 is statistically
   significant at a confidence level as high as 99 percent.  The
   confidence level for the in-stack flat filter and Method 5 difference
   is not sufficient to declare that the two methods give significantly
   different particulate mass emission results.
•  The reaction of SO- or H2SO,(g) with the Munktell in-stack glass fiber
   thimble (marketed by Carborundum) yields sulfates which accounts
   for the differences between the thimble and the in-stack flat
   filter and Method 5 results.

-------
Chemical analyses of filters and filter catches indicate that
S03 or H2SO,(g) may also react with MSA 1106 BH glass fiber
filter medium when used both in-stack and in the Method 5
configuration.  In-stack and Method 5 particulate mass results
may be from 10 to 16 percent high due to the sulfate formation.

               Method 5 Precision

The within-laboratory precision (repeatability) of particulate
mass emission measurements by Method 5 when performed with a
filter box temperature of 121 C (250 F) and a probe outlet
temperature of about 205 C (400 F) is estimated to be about
5.5 percent.  The estimate was derived from fairly steady
state, fixed point measurements; therefore, any errors due to
stack traversing and its associated variables (stack temperature,
velocity pressure, and movement of equipment) would not be
included in the precision estimate.

-------
                                 RECOMMENDATIONS

             Based  on  the  conclusions given  in the preceding section, the following
  modifications are recommended to specifically adapt the Method 5 procedure (and
 similar procedures) to determination of particulate mass emissions from cement
 plants (kilns).    (Many of these recommendations may also be applicable
 to particulate mass measurements from other types of stationary sources.)

                          Sampling System Temperature

           In order to prevent ammonium sulfate and/or sulfuric acid condensation
           •   Operate the filter box at a minimum temperature of 121 C (250 F).
           •   Heat the probe to maintain the gas sample above the temperature
               of ammonium sulfate and/or sulfuric acid condensation throughout
               the  entire  system up to  the box  filter outlet.   Based on
               this study  a minimum gas sample  temperature of  191 C (375 F)
               is suggested for general application  of Method  5 to cement
               plants.
           o   Sample  gas  temperature  should be measured on the outlet  side
               of the  box  filter frit with a thermocouple or metal  thermometer
               immersed in the gas stream.   Probe heating should be regulated
               to give recommended operating temperature.

                             Isokinetic Sampling

          Sampling data on a well-controlled source  show that  deviations of 0.7
and 1.3 times isokinetic sampling rate  do not produce a statistically significant
difference in mass measurements.  The current Method 5 requirement for a valid
performance test limits the maximum deviation to the range 0.9 to 1.1 times
isokinetic sampling rate.   For current and future well-controlled cement kilns,
emissions, isokinetic sampling deviations within the specified range should not
produce a significant error in particulate mass measurements.
          The study results suggest that the permissible range of isokinetic
sampling deviation could be extended without adverse effects.   However, normal
cement kiln operating characteristics do not impose any problems in maintaining
sampling rates within the specified range of ± 10 percent.

-------
o   Consequently, no changes are recommended in the acceptable
    isokinetic sampling rate deviation range set forth in
    Section 6.8 when Method 5 is used for cement plants.

                       Filter Size

•   Additional investigation is suggested to study the effect
    of filter size (area) on particulate mass data.  A study
    of SO  or H2SO^(g) reaction with the MSA 1106 BH filter
    medium (also suggested in a subsequent section) may provide
    an explanation.
•   From a performance standpoint,  a nominal 7.6-cm (3-in.)
    filter holder of the type used in this study is recommended
    for Method 5 sampling.  The filter area is sufficient to
    accept anticipated loadings and ease of handling (assembly,
    disassembly, and placement in heated filter box) is
    superior to a similar 10.2-cm (4-in.)  filter holder
«•  The 6.2-cm (2.5-in.)  filter holder of  the' design used in this
    study is not recommended for use in Method 5.  Filter assembly
    and recovery in a quantitative manner is difficult to perform
    without extreme care.  Damage to the filter can occur from
    misalignment of the sintered glass frit and its retaining ring.
    Pressure drop across the smaller filter could also be a problem
    at higher particulate concentrations.

                  Nozzle Configuration

•   A goose neck nozzle is recommended for the Method 5 sampling
    train.  Separate wind tunnel studies have shown that relative
    nozzle and pitot tube location is more critical with the 90°
    nozzle.

-------
               Particulate Filter

The Munktell glass fiber filter thimble currently marketed
by Carborundum is not recommended for particulate sampling
of cement kilns.
                                                          2
The use of an in-stack flat filter with at least a 20.3-cm
        2
(3.14-in.) area is recommended as a suitable alternate to
the Method 5 filter configuration.
Additional study of the reaction of sulfur oxides with
filter media is recommended to quantify the extent of
extraneous sulfate formation and to select materials which
minimize this source of error.

               Method 5 Precision

Specific equipment and procedures for recovery of particulates
from the probe should be given in Method 5.  The inherent
and potential error in this step is probably the most
significant in determining the precision of the method.

-------
                                     10
                                INTRODUCTION

          The Clean Air Act as amended in 1970 provides the impetus for
programs to improve the air quality in the U.S. through research to broaden
the understanding of the effects of air pollutants, research and development
of techniques to control emissions, and the enactment of air quality regulations
to protect the public welfare.  Pursuant to Section 111 of the Act, the Environ-
mental Protection Agency (EPA) on December 23, 1971, promulgated Standards of
Performance for New Stationary Sources (amended) for fossil fuel-fired steam
generators, incinerators, portland cement plants, and nitric and sulfuric acid
      (] )*
plants  '  .  On March 8, 1974, similar performance standards were issued for
asphalt concrete plants, petroleum refineries, storage vessels for petroleum
liquids, secondary lead smelters, secondary brass and bronze ingot production
                                                         (2)
plants, iron and steel plants and sewage treatment plants   .  All new
and modified sources in the preceding categories are required to demonstrate
compliance with the standards of performance.
          The performance standards are intended to reflect "the degree of
emission limitation achievable through the application of the best system of
emission reduction which (taking into account the cost of achieving such
                                                                          (3)
reduction) the Administrator determines has been adequately demonstrated"
          Compliance with required performance is determined by testing proce-
dures specified with the standards.  The use of a procedure called "Method 5
                                                                (4)
Determination of Particulate Emissions from Stationary Sources"    is specified
in all instances where particulate mass emission measurements must be made.
A copy of the Method as promulgated is given in Appendix A.  The Method 5 pro-
cedure consists of isokinetic extraction of a sample from the emission stream
with a heated probe and collection of the particulates on a heated filter.
With the recent exception of  fossil fuel-fired power plants   , the same
sampling system operating parameters have been adopted for all stationary sources.
          The source categories subject to Method 5 particulate measurements
include diverse processes which encompass a wide range of the following emission
characteristics; moisture content, gas temperature, gas composition, particulate
concentration and composition, and flow dynamics.  Interaction of these emission
properties with the Method 5  sampling technique can produce significant variations
in the results of particulate emission measurements.  The following are examples
of some of the reactions which may affect particulate measurements.
*  References are given on page 91.

-------
                                     11
          (1)  SCL or H-SO,  in emissions can condense to form sulfates
               which increase the mass of collected "particulates".   The
               SCL-HpSO,  dew point is dependent on SO  concentration and
               moisture content of the emissions.
          (2)  The filter particulate catch may present a surface for
               reactions with gaseous emission components such as SO
                                                                    X
               and NO .  Reactivity would be dependent on particulate
               loading and composition and on gas  composition of the emissions.
          (3)  Changes in gas temperature in the sampling system may alter the
               apparent particulate concentration  through  condensation or
               volatilization .
          Such interactions with the sampling process must be recognized and
controlled if Method 5 is expected to yield reliable particulate measurements
for individual source categories.
          The work presented in this report which  was performed as part of an
EPA program to study the applicability of the Method 5 procedure to  measurement
of particulate emissions from a variety of stationary sources.  Specifically,
this work addresses the question of whether Method 5 provides an accurate,
reliable measurement of particulate emissions from cement plants.  Earlier, an
interim report    was issued on the preliminary study results.  Some data from
this earlier report has been included here to present a complete document on the
use of Method 5 for sampling cement plant emissions.
                                 OBJECTIVES

          The objective of the overall EPA program is the evaluation of the
applicability and reliability of Method 5 (conducted as specified in the Federal
Register, December 23, 1971) for the determination of particulate mass emissions
from stationary sources for which performance standards have been promulgated.
The portion of the overall program covered by this report is aimed at evaluation
of Method 5 performance when the procedure is applied to cement plants.  The
study sought to identify any characteristics of the sampling method or unique
properties of cement kiln emissions which would adversely effect particulate
measurements and, if possible, recommend appropriate corrective measures in
sampling methodology.

-------
                                     12
                         EXPERIMENTAL WORK AND RESULTS
                            Experimental Approach

          In accord with the program objectives, an experimental plan to
evaluate the performance of Method 5 was formulated which included determination
of the general emission characteristics of a cement kiln, a study of the parameters
of the Method 5 sampling procedure which might affect particulate measurements,
and comparision of Method 5 with in-stack sampling techniques.  All experiments
were performed in statistically designed series to provide the maximum information
with a minimum number of experiments and to permit assessment of the statistical
significance of observed differences from the factors treated as independent
variables.
          Initially, the effects of Method 5 sampling system parameters on
particulate measurements was studied.  These parameters included
          e  Anisokinetic sampling
          o  Probe and filter temperature
          o  Filter size
          «  Nozzle configuration.
          The experiments were intended to reveal the sensitivity of particulate
measurements to the sampling variables and to determine if current Method 5
operating parameters are within a range which will produce accurate, reliable
results.
          Two types of in-stack filters (glass fiber flat and thimble) were
compared with Method 5.  Particulates collected under in-stack conditions might
be considered less subject to compositional alterations, especially by condensation
products and reactions which may occur upon cooling below stack temperature
assuming stack temperature is greater than 121C(250F).  Accordingly, comparison
of the in-stack and Method 5 particulate collections provides an approach to the
study of sample alterations which may be introduced by the Method 5 sampling
procedure.  In addition,  in-stack collection essentially eliminates deposition
of the particulates in probe and facilitates the study of reactions which may
occur in the probe, e.g., condensation of H?SO,.
          Another consideration behind the methods  comparison is that EPA is

-------
                                     13
considering adoption of the in-stack technique as an optional performance test
method.  Equivalency of Method 5 and the in-stack method must be demonstrated
to maintain consistency with established performance standards.
          The approach selected to conduct the experimental study consisted
of concurrent sampling at approximately the same point in the cement kiln
stack with two sampling systems operated under the various conditions under
study.  Filter and probe collections were analyzed gravimetrically and chemically
to detect differences resulting from various sampling parameters.
Analyses of the gas composition of the kiln emissions were performed to identify
components which might interact with the sampling process.  The mass emission
data  from the experiments were analyzed statistically to determine the
significance of observed differences.
          The following sections present detailed descriptions of the emission
source, sampling equipment^ and sampling and analysis procedures.  In sub-
sequent sections, the experimental test results are then presented and discussed.

-------
                                      14
                    Process and Sampling Site Description

Cement Manufacturing Process

          The manufacture of port land cement is a basic industry in the United
States with a current annual production of about 80,000,000 metric tons.
Based on the past few years, production is expected to grow at the rate of
about 5 percent per year.
          Portland cement is produced from the following raw materials  (listed
in decreasing order of consumption); limestone (including oyster and clam
shell), cement rock, clay and shale , gypsum, marl, sand and sandstone, and
blast furnace slag.  The production process involves four major steps;
quarrying and crushing, grinding and blending, clinker formation, and finish
grinding.  In production by the dry process, the raw materials are ground,
blended, and fed into the kiln (clinker formation) in dry form.  In the wet
process, water is used in the grinding and blending operations and the raw
materials are fed into the kiln as a slurry containing as much as 30 to 40
percent water.  In 1972, 62 percent of the 175 plants in the United States
manufactured cement by the wet process and 38 percent employed the dry
process.
          The kiln operation, used in the clinker formation step, is the
major source of emissions (both gaseous and particulate) in  cement  production.
The kiln is a rotating steel cylinder, typically 10 to 20 feet in diameter
and 200 to 400 feet long, lined with refractory brick.  It is mounted
horizontally with a slight slope from the upper end at which raw materials
are introduced to the lower, end through which the clinker exits and the kiln
is fired.  Coal, fuel oil, or natural gas may be used as fuel for the kiln.
          The drawing in Figure 1 illustrates the main features of the kiln
operation.  The raw materials are fed into the upper end of the gently
sloped, rotary kiln and flow slowly toward the lower, increasingly hotter
end of  the kiln.  In passage, the charge is dried, calcined and finally at
a point of incipient fusion forms a substance called clinker.  The clinker
is cooled, ground, intermixed with gypsum to produce the finished product.

-------
Sampling;
 point
                   Raw
                 materials

              Dust   Jr
             collector
Raw mix is kiln burned
to partial fusion at 2700 F
                                                               Coal
                                                               fuel
                                                            Clinker
                                                            cooler
                                          Materials
                                        stored separately
Clinker
4r=^
11 ' -'

V
:V^'---.''-- .
•*••.;•;»••; ••••'•• '-.••..
-V^"* '*:*"''•.• VJ""':.*.~*. "'.'•
fi~"-""5 ^
f
\
4
I
*%$&
s:;.--:\*
i
i
i
i
i
:tGyp
i
1 >2C?
' 71
ix
• sum
                                        Clinker and gypsum conveyed
                                        to grinding  mills 	*-
                     FIGURE  i.   KILN  OPERATION OF PORTLAND CEMENT PRODUCTION

-------
                                      16
The combustion products and excess air pass through the kiln countercurrent
to the raw material flow carrying with them reaction gases from the calcina-
tion process (principally CO.), moisture, and dust particles from throughout
the kiln.  The dust loadings generated by the kiln operation,which are
reported to range from 1 to 33 percent of the finished product   ..require
 a  control  device  such  as an electrostatic  precipitator or  baghouse
to comply with current emission regulations.

Sampling Site Description

          All sampling experiments conducted as p'art of this study were per-
formed at the Columbia Cement Corporation, Zanesville, Ohio.  This plant
operates two rotary kilns and produces in excess of 3,000,000 barrels of  cement
per year by the wet process.  The principle raw feed materials are limestone
and shale and the kilns are fired with pulverized coal which contains about
three percent sulfur.  The emission of particulate material from the kilns
is controlled by recently installed electrostatic precipitators.
          Sampling was performed in a vertical run of duct carrying emissions
from kiln #5 precipitator outlet to the stack discharge blower.  The point
of sampling is indicated on the drawing presented in Figure 1.  The duct
at the sampling location, a large platform beneath the precipitators, contains
six sampling ports arranged as shown in Figure 2.  A temperature and velocity
head (AP) profile of the stack cross section at the sampling ports is pre-
sented in Figure 3.
          The general emission characteristics of the kiln are summarized in
Table 1.  Carbon dioxide in the emissions results from decarbonation of the
limestone as well as from combustion.  Lower S00 levels had been anticipated
                             (7)               2-
based on previous information   .  Significant variations were observed
in S02 emission levels as will be noted from continuious monitoring data
presented later in this report.  The flow through the kiln was relatively con-
stant.  Gas temperatures variations of about 22 degrees centigrade (40 F)
were observed during some of the two hour testing periods.   Both gas tempera-
ture and SO- variations are associated with coal firing rate changes which
are required to maintain the charge (fused mass) in the proper zone of the kiln.

-------
^-  ••"
       FIGURE 2.  SAMPLING PORT LOCATION  IN
                  DUCT LEADING FROM  PRECIPITATOR
                  TO STACK

-------
                                                                        Top View
Distance,
meters



i.i

\
i

i
52
i.:
\
1

1
30
0.
I }
t
*-
93
0.!
t 1


57
1.83 1.78 1.65 1
70 1.83 1.93
226 225 225 228 235 242
1.85 1.63 1.32 1.35 1.65 .93
225 221 218 7ube 22° 23° 243
1
Nozzles
1.83 1.65 1.30 1.35 1.73 1.88
-t- + f 4- -t- -t-
229 222 217 220 232 244
1.49 1.73 1.35 1.60 1.80 1.50
-t 1- -t- H- -t- -t
222 220 2I8 229 237 243
Upper values are velocity pressure, cm
of 1120
Lower values are gas temperature, C
l_ U U
U
123456
                                     Port
FIGURE 3.  VELOCITY PRESSURE AND TEMPERATURE PROFILE OF CEMENT KILN
           EMISSION DUCT AT SAMPLING LOCATION

-------
                              19
          TABLE 1.  EMISSION SOURCE CHARACTERISTICS
Composition of Kiln Emissions
            Particulates
                     co2
                      °2
                      N2
                     so2
                     NO
                       x
                Moisture
Flue Gas Conditions
            Temperature
        Volumetric flow
             Average /\P
        Static pressure
Flue and Port Dimensions
             Flue Size
          Port diameter
 ~23 mg/Nm3  (-0.01 grain/SCFD)
 ~19 percent, dry basis
  ~8 percent, dry basis
 ~73 percent, dry basis
~450 ppm  (Average)
~535 ppm  (Average)
~30 percent
188 to 236 C  (370 to 457 F)
1,473 m3/min, Dry   (~52,000 DSCFM)
1.65 cm  (0.65 itu) H20
2.1 cm (0.81  in) Hg, negative
1.83 tn x 2.13 m  (6  ft x  7  ft)
7.62 cm  (3 in^

-------
                                       20

                             Sampling Equipment

Particulates

          The particulate sampling was performed with two identical Method 5
trains comprised of components connected as shown schematically in Figure 4.
The control module and impinger box assembly of the systems, shown in Figure
5, are commercially-built units.  The sampling probes, which were constructed
at Battelle, were 2.06 m (6.8 ft) in length and were constructed from
approximately 11-mm-ID x 16-mm-OD glass tubing.  Heating was provided by a
3.05-m (10-ft) glass fiber-insulated heating tape wrapped around the glass
probe.  A thermocouple junction was taped to the outer surface of the glass tubing
at a point between the heater windings and at the midpoint of the: probe.  The
glass probe and heating tape assembly was insulated by a wrapping of asbestos
tape.  A stainless steel sheath was used to protect the glass probe and to hold
the fitting to attach the nozzle.  The seal between the glass probe and the 1.6
cm (5/8 in.) Swagelok® nozzle connection was made with a silicone 0-ring.
          A single type "S" pitot tube attached to one of the sampling probes
was used for velocity head readings.  The tube was constructed at Battelle
from approximately 7.5 mm I.D. x 9.5 mm O.D. stainless steel tubing.  A
stainless steel sheathed thermocouple was attached to the pitot tube to provide
stack temperature measurements.  A photograph showing the pitot tube sensing
head and the position of the thermocouple is presented in Figure 6.  The type
"S" pitot tube was calibrated against a standard pitot tube over the velocity
range of 10.7 to 29.6 m/sec (35 to 96 ft/sec) in the Battelle wind tunnel
facility.
          Modifications and variations to the "normal" Method 5 sampling train
were made for several test series to study sampling methodology.  These are
described in the following sections.

-------
                                        12
                                                                                       K>
Stainless steel nozzle
Glass-lined probe
7.62 cm (3 in.) filter
Heated box for filter
Ice bath for impingers
                      8
 6. Modified G-S impinger with 100 ml water
 7. Greenburg-Smith impinger with 100 ml water
 8. Modified G-S impinger
 9. Silica gel trap
10. Thermometers or Thermocouples
         13
11. Vacuum gauge
                                                                                               15
                                    17
                        16. Calibrated orifice
                                       12.  Flow control valve  17. Manometer (AH)
13.  Pump
                                                               18. "S" type pilot tube
                                       14.  Flow control valve  19. Manometer (AP)
                                       15.  Dry test meter
FIGURE 4.  SCHEMATIC DIAGRAM OF METHOD 5 TRAIN
           USED FOR SAMPLING CEMENT KILN EMISSIONS

-------
FIGURE 5.  EQUIPMENT USED FOR SAMPLING PARTICULATES IN CEMENT PLANT EMISSIONS

-------
FIGURE 6.   PITOT TUBE HEAD AND THERMO-
            COUPLE USED FOR VELOCITY
            HEAD AND STACK TEMPERATURE
            MEASUREMENTS

-------
                                     24
          Temperature Measurements.  The glass connectors from the probe
outlet to the filter and the filter outlet to the first impinger were
modified to permit additional measurements of gas sample temperature.
Figure 7 shows the modified connector used in all tests to measure and control
gas temperature at the probe outlet.  The connector contains a thin-wall
thermocouple well which extends about 5.1 cm  (2 in.) into the outlet end of
the probe.  Figure 8;shows the modified connector used in some tests to
obtain gas temperature measurements at the filter outlet.  The tip of the
metal thermometer was positioned about 1.3 cm  (0.5 in.) from the frit in the
filterholder.

          Nozzles.  The two types of nozzles shown in Figure 9 were used
in a study of the effect of nozzle configuration.  The nozzles were formed from
approximately 7.5-mm-ID x 9.5-mm-OD stainless  steel tubing and the
sampling tips were machined to a knife edge with a 7.7" mm (0.303 in.) diameter
opening.  The goose neck type nozzles were used on both sampling systems
in all tests with the exception of the experiments to study nozzle
configuration.

          Filterholders.  The 6.25 cm  (2.5 in.) and 10.2 cm  (4 in.) filter-
holders shown in Figures 10 and 11, respectively, were used in the experiments
to study the effect of filter size.  The silicone rubber ring shown with
the 6.25 cm filter was not used because it stuck to the glass fiber filters.

          A 7.6 cm (3 in.) filterholder of the design shown in Figure 11 was
used in all other tests with the Method 5 train and as the back up filter
in tests in which an in-stack filter was used.

-------
      FIGURE 7.  THERMOCOUPLE ASSEMBLY USED TO
                  MEASURE GAS TEMPERATURE AT
                  PROBE OUTLET
FIGURE 8.  THERMOMETER ASSEMBLY USED TO MEASURE GAS
           TEMPERATURE AT THE BOX FILTER OUTLET

-------
                                         30  140
FIGURE 9.  SAMPLING NOZZLES

-------
 FIGURE 10.  OUT-OF-STACK FILTER HOLDER FOR
             6.25-CM-DIAMETER FILTERS
FIGURE 11.  OUT-OF-STACK FILTER HOLDER FOR
            10.2-CM-DIAMETER FILTERS

-------
                                       28
           The two types of filter holders used for in-stack sampling are shown
 in Figures 12 and 13.  The 6.25 cm (2.5 in) holder for flat filters shown in
 Figure 12 is a commercially-available unit.  The actual filtration area is
              2        9
 about 25.7 cm  (4.0 in ).
           The in-stack filterholder shown in Figure 13 was  constructed at
 Battelle  and was  designed to  use  Munktell glass  fiber thimbles which are
 marketed  by  Carborundum Co.*   The thimble was sealed to the inlet with the
 spring-loaded, stainless steel collar and a Teflon® gasket  was used to seal
 the filterholder  body.
           Both in-stack  filterholders were  equipped with fittings  to  connect
 them between  the  probe and the goose  neck  nozzle described  previously.
           Filter  Materials.  Mine  Safety Appliance  (MSA) 1106  BH  glass  fiber
 filter medium was used in all  flat  filter  applications, both out-of-stack  and
 in-stack,  and  liunktell glass fiber  thimbles were used  for some  in-stack
 particulate  collections.  An analysis of these filter  materials obtained by
 optical emission  spectroscopy  is  given  in  Table .2.  Analyses of some  alkali
 metal and  alkaline earth  elements which can be removed  from the filter media
by leaching with hot 0.1 N HCl are presented in Table 3.  Table 4 gives the results
                                        T4:
                                        (9)
                                                  /Q N
of pH analyses obtained by TAPPI Method T435 su-68  _  and alkalinity analyses
 performed according to ASTM Method D202

Gas Analysis

          Continuous monitoring for S0« and NO  was performed during the
                                      tL       X
preliminary cement plant sampling study by an Environmetrics Model NS 300
AC Faristor unit.  The gas sample was extracted from the duct through a
0.63 cm  (0.25 in) diameter stainless steel tube and passed through a moisture
trap prior to the Faristor unit.  Teflon® lines were used to connect the monitor
to the sampling probe.
*Munktell's Swedish filter paper is made by Grycksbo Pappersbruk,  AB,  Sweden.

-------
FIGURE 12.  IN-STACK FILTER HOLDER FOR
             6.25-CM-DIAMETER FILTERS
FIGURE 13.  IN-STACK FILTER HOLDER FOR
            GLASS FIBER THIMBLES

-------
TABLE 2.  COMPOSITION OF FILTER MATERIALS
Element Composition, weight percent
Filter Type
MSA 1106 BH
Thimble
MSA 1106 BH
Thimble
Si
Major
Major
Mn
0.002
0.005
Na
3-6
3-6
Pb
< 0.005
0.02
Ca
4-7
1-2
Cu
< 0.001
0.005
K
1-2
2-3
Zn
< 0.02
2-4
Ba
0.02
2-4
Ti
0.005
0.01
Al
2-4
2-4
Zr
0.005
0.005
B
1-2
1-2
Ni
< 0.001
< 0.001
Fe
0.1
0.05
Sr
0.02
0.1
Mg
1.0
0.05
Cr
0.005
0.003
                                                                              to
                                                                              O

-------
                                31
               TABLE 3.   ANALYSIS OF AVAILABLE ELEMENTS
Element


Ba
Ca
K
Na
Zn
Availability
Milligrams /Filter
MSA 1106 BH Thimble
<0.02 5.8
0.12 1.6
0.1 3
1.5 13
0.006 4.1
of Elements (b) (c)
Approx % of Total
MSA 1106 BH
--
0.7
3
12
--

in Filter
Thimble
12
6
8
17
8
  (a)  Leached  from filter  materials  with  hot  O.ln HCl.

  (b) MSA 1106 BH  -  8.26 cm  (3.25  in) disc  -  _   , „      ...
  ^ '                                         i. Results are the average
      Thimble  -  Munktell  (Carborundum)       /of two determinations

  (c)  Analyses performed by atomic absorption
                TABLE 4.   pH AND ALKALINITY OF  FILTERS
Filter Type (a)
MSA 1106 BH - A
- B
Munktell - A
(Carborundum)
~ B
PH
9.40
9.35
9.35
9.10
Alkalinity1^
0.088
0.077
0.117
0.123
(a)   A and B represent analyses of two different filters
     of each type.  MSA filter  size -  8.26 cm  (3.25  in.) diameter,
(b)   Milliequivalent  of acid required to neutralize the
     alkalinity of  leach solution.

-------
                                     32
          Analyses for CO,., and 0~ were determined with Orsat and Fyrite
equipment.  Nitrogen was assumed to comprise the balance of the emission
gas mixture.
          Ammonia in the kiln emissions was determined by sampling with a
Method 5 train in which 150 ml 1M sulfuric acid was used in the first and
second impingers in place of water.
          Sulfur oxides were determined by the Goksoyr-Ross method
The sampling system consisted of the particulate sampling probe with a
modified nozzle to hold a quartz wook filter, a temperature controlled
condenser to collect sulfuric acid, and two serially connected Greenburg-
Smith impingers containing hydrogen peroxide to trap S02-  The system is
similar to the equipment described in ASTM Method D 3226-73T^   .

-------
                                     33


                Sample Collection and Analysis Procedures

Particulate Sampling

          In all tests, particulate sampling was performed concurrently
with two systems (designated A and B), each with a separate operator.
The systems as employed for sampling at the test site are shown in Figure
14.  Sampling system equipment and operating conditions used in the tests
were varied in accordance with statistically designed experimental patterns.
          All sampling was performed at a fixed-point at the center of the
duct in an area of nearly uniform velocity.  Sampling probes of the two
systems were inserted at a slight angle into the duct through two adjacent
ports so that the pitot tube attached to one of the probes was positioned
equidistance between the sampling nozzles.  The separation between the
pitot tube and each nozzle was about 2.5 cm (1 in.).  The relative nozzle-
pitot tube position and the point of sampling within the duct are indicated
in Figure 3.
          At the start of each test day, the laboratory calibration of the
gas metering components of both sampling systems was checked by setting the
orifice manometer (AH) to the meter box calibration factor  (AH@) and
measuring the flow rate through the dry gas meter over a 5-minute period.
                      3
A flow rate of 0.021 m /min (0.75 cfm) confirmed that the gas metering
system remained in calibration.
          The preparation of the particulate collection trains for all tests
was performed as specified in Paragraph 4.1.2 of Method 5 with the following
modifications:
          (1)  The entire sampling train was leak checked by plugging
               the sampling nozzle inlet and evacuating to 38.1 cm
               (15 in.) of Hg.  Leak rates did not exeed 566 cm /min
               (0.02 cfm).  This procedure was used for both the regular
               Method 5 train and the in-stack filter train.  Prior to
               leak testing, the in-stack filter assembly was heated
               with a 260-399 C (500-750 F) heat gun.  Heating was con-
               tinued until the probe was inserted into the duct to
               initiate sampling.

-------
                   Impinger Boxes
                    Meter Boxes
FIGURE 14.  CONCURRENT SAMPLING AT TEST SITE

-------
                                     35
           (2)  The probe was heated until the thermocouple at the
               outlet indicated that the desired operating temperature
               was achieved prior to initiation of sampling.
           (3)  Heat guns (260-399 C) were used for supplemental heat-
               ing when the box filter was operated at temperatures
               above 121 C (250 F).
           (4)  An additional modified Greenburg-Smith impinger (without
               water) was required to collect condensed moisture when
               sampling rates of 1.3 times isokinetic were used.
           In performance of the tests, sampling trains were operated as
described  in Paragraph 4.1.3 of Method 5 with the exception that readings
of AP, AH, stack temperature and sampling system temperatures were recorded
at 10-minute intervals.  The velocity head (AP) for both systems was deter-
mined from one pitot tube and nomographs were used to obtain the proper
sampling rate (AH).  Temperature measurements were obtained at the points
shown in Figure 15.
           The sampling period for each test was 2 hours and total dry gas
sample volumes at isokinetic sampling rates ranged from 2.1 to 2.4 m
(74 to 85  ft3).
           After completion of the tests, the trains were again leak checked,
sealed to  prevent contamination, and transferred to the sample recovery area.

Sample Recovery and Analysis

           Filters were removed from holders, sealed in Petri dishes or
glass jars (thimbles) and immediately placed in a desiccator.  In Method 5-
type tests, the probe and nozzle were disassembled and washed separately.
First, the probe was first  rinsed with acetone  without brushing,  then rinsed
with acetone while slowly inserting and removing a Nylorr^ brush in a
rotating fashion.  The acetone wash and brushing were continued until visual
inspection indicated that all particulates were removed.  The brush was
thoroughly flushed with acetone prior to removal from the probe.  The
probe wash (usually about 100 to 150 ml) was collected in an Erlenmeyer

-------
                                    36
Probe
Mid-point
                                                     Probe outlet
Filter outlet
                                                                     Impinger
               FIGURE 15.  TEMPERATURE MEASUREMENT POINTS

-------
                                     37
flask sealed onto the probe outlet ball joint.  Particulates were recovered
from the nozzle and the inlet half of the filter holder by alternately brushing
and rinsing with acetone.  The wash solutions from all three components
(probe, nozzle, and filterholder) were combined for analysis.
          In tests with in-stack filters, the in-stack filter was removed
from the probe and particulates were recovered from the probe as described
in the previous paragraph.  The outlet side of the in-stack filterholder
and in inlet side of the back-up filterholder (box filter) were separately
brushed and washed with acetone and the solutions were combined with the
probe wash.  The nozzle and inlet side of the in-stack filterholder, as
one unit, was alternately brushed and washed with acetone.  This wash
solution was not combined with the probe wash, but analyzed separately.
          At least one 200 ml acetone blank was obtained each day from
the wash bottle dispenser.  All acetone wash solutions and blanks were
                                  ®
stored in glass bottles with Teflon  -lined cpas for transfer to the
laboratory for analysis.
          The MSA 1106 BH filters and particulate catch were desiccated at
least 24 hours (usually longer) prior to weighing.  It was found necessary
to desiccate the Munktell (Carborundum) thimbles at least 72 hours prior
to weighing (both tare and final) to achieve a constant weight.
          The acetone wash solutions were evaporated to dryness in a reverse
airflow, clean hood and the residues were desiccated to a constant weight
(usually 24 to 48 hours).  Residues and filters were weighed to the nearest
0.1 mg.
          Calculations were performed as described in Section 6 of Method 5.

Gas Analyses

          Continuous analyses for S09 and NO  were performed concurrently
                                    ^       X
with all tests in the preliminary series of particulate sampling experiments.
A gas sample was withdrawn from the kiln emissions at the rate of about
1.5 liters/minute passed through an ice-cooled moisture trap and analyzed
with an Environmetrics NS300 AC monitor.  Periodic calibrations were

-------
                                     38
performed on-site by passing gas mixtures containing known concentrations
of SCL and NO through the same intake system.
          (X>2 and 02 analyses by Fyrite or Orsat were performed during
each particulate test.
          Two ammonia determinations were performed using a Method 5
particulate train operated with a probe outlet gas temperature of 213 C
(416 F) and a box filter temperature of 121 C (250 F).  Sampling was
conducted at the rate of about two liters/minute for 30 minutes in one
test and 60 minutes in another.  The ammonia was reacted with 1 M sulfuric
acid contained in the impinger train and subsequently determined by micro
Kjeldahl analysis.  Samples of the sulfuric acid solutions used in the
impingers were analyzed for ammonia blank.
          A controlled SO, condensation procedure similar to ASTM D 3226-73T^  '
was used to sample for SO- and S00 (SO ) in the kiln emission.  Sampling was
                         j       Z.    X
performed for 60 minutes at a flow rate of 2.8 liters/minute with the probe
operated at about 205 C (400 F).   The sulfuric acid condensation coil was
maintained at 82 C (180 F).  SO- was collected in 3% hydrogen peroxide
which oxidized it to sulfate.  After sampling, the probe and condensation
coil were rinsed with water and the combined solution was analysed for
sulfate resulting from S0».  The hydrogen peroxide trap solutions were
analyzed for sulfate formed from the S0~.  Sulfate analyses were performed
by the barium perchlorate-thorin titration method.

-------
                                      39
                        Test Descriptions and Results

Four separate test series were conducted  in  this  study to  investigate  the  effects
of the following sampling variations  on the  particulate collection process:

          1.  Isokinetic versus anisokinetic sampling,
          2.  Filter size, temperature and nozzle configuration,
          3.  Filter box and probe temperatures, and
          4.  Filter types  (Method 5  and  in-stack).
In designing the statistical plan for each test series, it was necessary
to select appropriate patterns in light of the number  of simultaneous  samples
which could be  collected and the number of treatment combinations  (La  combinations
of varying levels of experimental conditions) being studied.  The  sampling
environment allowed for  the collection of only two simultaneous samples while
the various test series  involved more than two treatment combinations.
Consequently not all treatment combinations  could be evaluated simultaneously.
Because of the  probability that stack conditions would vary over  time, experimental
designs were selected which permitted statistical adjustments in  the collection
of the data,  so  that effects which may have  been  attributable  to differences
in controllable factors  (i.e., the factors being  studied) would not be con-
founded with variations  in the uncontrollable factors  (i.e., varying emission
rates in  the source) over time.  Two  statistical designs were used.
          A Balanced Incomplete Block Design was used for two of the four test
series conducted — the series on effect of filter type and the series on the
effect of probe  and filter temperature combinations.   In each of these series,
three treatment  combinations were studied in all possible simultaneous pairs.
Thus, if the treatment combination levels are denoted by "A",  "B", and "C", the
statistical pattern for the series  required that three pairs of simultaneous
samples - (A,B), (A,C),  and (B,C) -  be collected, with appropriate adjustments
in the data for  temporal variations  in the source emissions.

-------
                                       40
          A Confounded Factorial Design was used for the test series on effect
of anisokinetic sampling conditions and for the series on the effects of filter
size, sampling system temperature, and nozzle configurations.  In this design,
treatment combination were assigned to the simultaneous pairs which were collected
in such a way that the experimental factors which were of primary interest were
not  confounded with temporal source variations.  This condition was achieved by
deliberate confounding of less important factors with the source variations over
time, or  through partially balanced confounding over replications.  That is,
with replications of  the experimental  design, different effects were  confounded
during each replication so that partial information was available on  these  factors,
          The statistical approach used in analysis of data  in this study
is the commonly used  "analyses of variance", which partitions  the total variation
observed  in the data  (as measured by the sum of squares of deviations  about
the  mean) into components which are attributable to each variable.  An F-ratio
is calculated as the  ratio of the mean square of the variables to the mean
square error term.  As this F-ratio becomes progressively larger, the statistical
significance of the variable becomes greater.  Ordinarily a  significance level
in the range of five  percent or less,  which is associated with at least a 95
percent confidence level, is established as the criterion which must be met
before it is concluded that there is a statistically significant effect.
          More detailed discussions of the statistical designs and analysis of
variance  procedure used in the experiments are presented in  Appendix  D of this
report.
          The following sections present for each of the experiments, the
statistical design features, the sampling system operating parameters, sample
weights and final test results, and the statistical analysis of the test
data.  Tables which give the stack conditions and the sampling system operational
data for  each tests are presented in Appendices B and C, respectively.

-------
                                       41
                      Anisokinetic Sampling Experiments

          Sixteen tests were performed to investigate the magnitude of anisokinetic
sampling effects on the particulate collection process.  The statistical pattern
for the tests is given in Table 5.  This pattern was derived from the use of
confounded factorial experimental design which permitted the investigation of
effects due to anisokinetic sampling, sampling systems, and any interaction effect
betxreen anisokinetic sampling and sampling system.  Replication of each pair of
test  conditions was performed to increase the power of the statistical tests
for detecting significant differences due to anisokinetic sampling.  In performance
of the test series, operators were randomly assigned to the control modules to
minimize any bias due to sampling system operation.
          Sampling was performed to compare results obtained at isokinetic rates  and
the following anisokinetic conditions:
          1.  0.7 times isokinetic rate
          2.  1.3 times isokinetic rate.
These anisokinetic conditions,  outside the acceptable  range for Method 5
results,  were selected to accentuate any observable effects on particulate
collections.
          The results of the anisokinetic sampling experiments are presented
in Table 6.   The table gives the filter and probe sample weights and the
particulate concentrations reported in mg/Nm .   Analysis of the probe
residues from all tests showed that variable quantities of ammonium sulfate
and/or ammonium bisulfate were formed in the probe when the temperature was
that specified in Method 5,  i.e.,  121 C (250 F)  at the probe outlet (neither ammonium
sulfate nor ammonium bisulfate were not detected in the filter catch).
Consequently, any effects resulting from anisokinetic  sampling were inadvertently
confounded with mass variations due to ammonium sulfate formation.   In
order to resolve the test data, the following procedure was used to correct
for the extraneous ammonium sulfate in each probe residue.

-------
                              42
TABLE 5.  RANDOMIZED TEST DESIGN FOR ANISOKINETIC SAMPLING STUDY
Test
No.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Statistical
Design
Block No.
16
14
15
3
12
1
6
9
13
2
7
4
5
8
10
11
SYSTEM A
Isokinetic
Condition, %
100
100
70
100
100
100
130
70
70
130
100
130
100
130
100
70
SYSTEM B
Isokinetic
Condition, %
70
100
100
130
70
100
130
70
70
130
130
100
100
100
100
100

-------
                                       43
          The corrections were made from the ratio of calcium in the probe and
filter catches and the filter catch by using the equation:
          PW (corrected)  =
Ca fprobe)
                            _Ca (filter)J
where
          FW (corrected)  =  probe residue weight corrected for ammonium
                             sulfate
          FW  =  weight of filter catch
          Ca(probe)  =  weight of calcium in probe residue
          Ca(filter) =  weight of calcium in filter catch.
This approach assumes that calcium, the most abundant element in the particulate
emissions, is present in the same weight percentage in both the filter and probe
catches.  Chemical analysis data presented a following section corroborate that
this is a reasonable assumption.
          The differences between the actual and corrected probe weights provide
a measure of the quantity of ammonium sulfate formed which in the extreme cases
accounts for almost 90 percent of the probe residue weight.
          The particulate concentrations, which were used in the statistical
analysis, were calculated  from  the corrected probe residue weight data.
          Table 7 and 8 present the particulate emission data in the confounded
factorial design format.  The statistical design blocks numbered one to eight
were used to compare the results obtained at isokinetic and hyperisokinetic
conditions.  The remaining statistical design blocks numbered nine through sixteen
were used to compare the results obtained at isokinetic and subisokinetic
conditions.
          In Tables 7 and 8, the data from Blocks 1, 2, 5 and 6 and Blocks 9,
10, 13 and 14 have effects of isokinetic sampling conditions confounded with
blocks (i.e., with source variations over time), and were used to measure the
effects of systems and the interaction effect of systems with isokinetic sampling

-------
                          44
TABLE  6.   SAMPLE WEIGHT DATA-ANISOKINETIC SAMPLING EXPERIMENTS
Sample Weights, mg
Run No
1A
B
2A
B
3A
B
4A
B
5A
B
6A
B
7A
B
8A
B
9A
B
10A
B
11A
B
12A
B
13A
B
14A
B
ISA
B
16A
B
Nominal
Percent
Isokinetic
100
70
100
100
70
100
100
130
100
70
100
100
130
130
70
70
70
70
130
130
100
130
130
100
100
100
130
100
100
100
70
100
Filter
41.8
30.7
61.3
64.4
51.7
71.2
55.2
71.7
58.9
42.0
59.3
58.7.
70.7
68.5
29.0
29.6
30.5
30.2
75.1
80.2
54.7
67.3
94.6
69.5
64.7
64.5
123.1
110.7
71.2
72.8
48.4
66.7
Probe factual)
47.2
43.1
44.3
14.9
34.0
44.2
28.3
17.8
42.1
16.2
39.3
36.0
20.4
13.0
20.7
16.4
32.5
7.7
18.0
11.7
63.7
11.9
23.0
37.4
46.8
47.1
61.1
85.8
80.5
79.0
62.5
85.2
Probe (Corrected)
9.4
9.7
11.5
7.9
16.2
18.4
9.3
9.8
11.5
7.4
15.9
13.7
9.4
7.2
10.5
7.0
10.3
5.0
13.7
7.1
9.7
8.2
14.7
14.3
12.0
11.7
62.0
61.2
34.8
16.5
17.5
18.0
Total
51.2
40.4
72.8
72.3
67.9
89.6
64.5
81.5
70.4
49.4
75.2
72.4
80.1
75.7
39.5
36.6
40.8
35.2
88.8
87.3
64.4
75.5
109.3
83.8
76.7
76.2
184.2
171.9
106.0
89.3
65.9
84.7
mg/Nm
22.8
25.4
31.4
31.6
43.1
41.8
30.4
28.9
32.2
31.9
34.7
33.2
28.7
26.8
26.1
23.7
27.2
22.4
32.9
31.2
30.2
27.4
38.8
38.3
34.9
33.7
64.7
81.9
49.1
41.2
42.8
38.3

-------
                                             45
              TABLE 7.  STATISTICAL ANALYSIS DATA - ANISOKINETIC SAMPLING STUDY
                        ISOKINETIC VERSUS HYPERISOKINETIC CONDITIONS

                                     (Measurements are in mg/Nm )
Sampling
Condition
Isokinetic
1.3 x
Isokinetic
Blocks 1, 2,
Blocks 3, 4,
Sampling
Condition
Isokinetic
0.7 x
Isokinetic
System
A
B
A
B
Replication"* 1 2
Block -* 1 2 3 4 5 6 7 8
34.7 — 30.4 — 34.9 — 30.2
33.2 — — 38.3 33.7 -- — 81.9
32.9 — 38.8 — 28.7 — 64.7
31.3 28.9 -- — 26.8 27.4
5, 6 are confounded with isokinetic conditions effect
7, 8 are confounded with -isokinetic conditions x systems interaction effect
TABLE 8. STATISTICAL ANALYSIS DATA ANISOKINETIC SAMPLING STUDY
ISOKINETIC VERSUS SUBISOKINETIC CONDITIONS
(Measurements are in mg/Nm )
System
A
B
A
B
Replication"* 1 2
Block -9 10 11 12 13 14 15 16
49.1 — 32.2 — 31.4 — 22.8
41.2 38.3 — — 31.6 41.8
26.1 — 42.8 « 27.2 — 43.1
23.7 — — 31.9 22.4 — — 25.4
•Blocks  1,  2,  5,  6  are  confounded  with isokinetic  conditions  effect
 Blocks  3,  4,  7,  8  are  confounded  with isokinetic  conditions  x systems  interaction effect.

-------
                                     46
conditions.  Blocks 3, 4, 7, and 8 in Table 7 and Blocks 11, 12, 15, and 16 in
Table 8 do not have isokinetic sampling conditions confounded and are suitable
for estimating the magnitude of the effects due to hyperisokinetic and
subisokinetic sampling conditions.
          In Tables 9 and 10, the statistical significance of these differences
are tested.  The statistical test used is an accepted analysis of variance procedure,
In Table 9, comparing isokinetic and hyperisokinetic sampling conditions data,
the F-ratio of 2.74 is associated with an 84 percent confidence level, or a 16
percent significance level.  In Table 10, the F-ratio of 1.34 determined from
analysis of the isokinetic conditions is associated with a 70 percent confidence
level, or a 30 percent significance level.  In neither case is the usual criterion
for statistical significance met i.e. a 95 percent confidence level or a 5
percent significance level.  Consequently, the conclusions drawn from Tables 9
and 10, are that variations in particulate collections due to anisokinetic
sampling, or to interaction effects of systems and conditions are not significant
based on the normal criteria i.e. significance at the 95 or 99 percent confidence
level.

-------
                               47
 TABLE 9.  ANALYSES OF VARIANCE-ANISOKINETIC SAMPLING EXPERIMENT
           (ISOKINETIC VERSUS HYPERISOKINETIC CONDITIONS)
Source
of
Variation
Systems (S)
Isokinetic Conditions (C)
CxS Interaction
Replications
Blocks Within Replications
Error (Remainder)
Total
Sum
of
Squares
2.40
55.12
0.08
223.50
2910.10
100.41
3291.62
Degrees
of
Freedom
1
1
1
1
6
5
' 15
Mean
Square
2.40
55.12
0.08
223.50
485.02
20.08
—
F
Ratio Conclusion
0.12 Not Significant
2.74 Not Significant
0.01 Not Significant
—
--
—
--
TABLE 10.  ANALYSES OF VARIANCE-ANISOKINETIC SAMPLING EXPERIMENT
            ISOKINETIC CONDITIONS VERSUS SUBISOKINETIC CONDITIONS
Source Sum
of of
Variation Squares
Systems (S)
Isokinetic Conditions (C)
CxS Interaction
Replications
Blocks Within Replica-
tions
Error (Remainder)
21.16
8.20
0.03
98.01
919.71
30.63
Total 1077.74
Degrees
of
Freedom
1
1
1
1
6
5
15
Mean
Square
21.16
8.20
0.03
98.01
153.28
6.13
_-
F
Ratio Conclusion
3.45 Not Significant
1.34 Not Significant
0.01 Not Significant
—
—
--


-------
                                      48
Filter Size, Temperature, and Nozzle
Configuration Experiments

           Three experimental factors were varied in this test series,-each
at two levels:
           (1)  Filter size:  6.4cm (2.5  in.)  versus  10.2cm  (4 in.)
           (2) Temperature: Stack  ~ 205 C  versus 121 C; and
           (3) Nozzle type: Goose neck nozzle versus 90° nozzle.
With two levels for each of the three factors, eight combinations of experimental
conditions were possible.  Since only two ports were suitable for sampling,
only two of these combinations could be run simultaneously.   Thus it was necessary
to use an experimental design which would selectively confound information on
some of the effects with the variation which might be expected from varying
stack conditions over time.  The design used for this experiment was the partially
confounded factorial design.  Each combination of the three factors was performed
three times.  The combinations were assigned to simultaneous pairs (statistical
blocks) in such a way that varying stack conditions over time would be
confounded in a balanced manner allowing partial information on the effects of
each of the three factors being studied as well as information on the effects of
first order interaction of these factors (i.e., their effects in combination which
may be more than the sum of their individual effects).  The statistical blocks
were randomly ordered over time and the two Method 5 trains were randomly assigned
within each block.  The randomized test pattern used for the experiment is shown
in Table 11.  Operators were also randomly assigned to sampling systems.
           The data collected for this experiment are shown in Table 12.  In
Table 13, the data are shown in the format of the confounded factorial design.
The statistical analyses of these data are present in Table 14.   With the same
criterion as before, the F-ratios in the analysis of variance indicate that
there is a significant difference in sample weights which is due to sampling sys-
tem temperature.  The confidence level associated with this F-ratio of 29.36 is
greater than 99 percent.  A significant effect due to filter size is also noted.
The F-ratio of 5.37, indicates at a confidence level of approximately 94 percent,
that differences between filters are significant.  The effect of nozzle configuration,
or of any interaction effects are not statistically significant.
           Further examination of the significant factors are presented in Table 15.

-------
                                       49
          TABLE 11.   RANDOMIZED TEST PATTERN FOR STUDY OF FILTER SIZE,
                     TEMPERATURE,  AND NOZZLE CONFIGURATION
Order
of
Test:
1
2
3
4
5
6
7
8
9
10
11
12

Filter
Size, cm
10.2
6.4
6.4
6.4
10.2
10.2
10.2
6.4
10.2
10.2
6.4
10.2
System A '
Temp , C
Stack(c)
121
Stack
Stack
121
Stack
121
Stack
121
Stack
121
121

Nozzle
G
G
" 90°
G
90°
90°
G
G
G
90°
G
90°

Filter
Size, cm
6.4
6.4
10.2
10.2
6.4
10.2
6.4
6.4
6.4
6.4
10.2
10.2
(a)
Svstemv '
Temp , C
Stack
Stack
121
Stack
121
121
Stack
121
121
121
Stack
Stack

Type
Nozzle
90°
90°
90°
90°
G
G
G
90°
90°
90°
G
G
Block No,
In Stat'l
Design
8
1
12
7
5
2
10
3
6 •
11
9
4
(a)   Systems A and B are both Method 5 trains with variations as shown.
(b)   G =  goose  neck nozzle,  90°=90°  nozzle.
(c)   Stack  temperature was  about  205C.

-------
                                        50
             TABLE  12.   SAMPLE WEIGHT DATA-FILTER SIZE, TEMPERATURE,
                         AND NOZZLE CONFIGURATION STUDY

Run
No.
17A
B
18A
B
19A
B
21A
B
22A
B
23A
B
24A
B
25A
B
27A
B
28A
B
29A
B
31A
B
Nominal
Filter Diam, cm
10.2 (4in.)
6.4 (2. Sin.)
6.4
6.4
6.4
10.2
6.4
10.2
10.2
6.4
10.2
10.2
10.2
6.4
6.4
6.4
10.2
6.4
10.2
6.4
6.4
10.2
10.2
10.2
Test Conditions
Temperature , C
Stack
.121(250 F)
Stack
Stack
121
Stack
Stack
121
121
Stack
121 '
121
Stack
Stack
121
121
121
Stack
121
121
Stack
121
Stack

Nozzle^
G
S
G
S
S
S
G
S
S
G
S
G
G
G
G
S
G
S
S
S
G
G
S
G
Sample
Filter
51.9
42.8
67.1
74.7
59.0
50.8
60.1
61.3
53.9
50.7
36.9
28.2
24.5
23.7
36.9
31.9
43.8
39.3
63.1
49.7
40.2
46.4
60.8
69.7
Weights ,
Probe
9.9
14.3
39.5
12.9
12.2
40.3
9.3
10.0
26.3
22.9
4.5
16.7
29.5
6.1
14.5
25.5
31.4
25.4
15.1
43.4
24.4
9.2
32.6
10.4
mg
Total
61.8
57.1
106.6
87.6.
71.2
91.1
69.4
71.3
80.2
73.6
41.4
44.9
54.0
29.8
51.4
57.4
75.2
64.7
78.7
93.1
64.6
55.6
93.4
80.1
2
tng/Nm
27.2
24.7
46.1
37.5
34.6
44.3
30.1
30.9
34.2
31.1
17.3
18.4
24.6
13.0
22.1
24.3
34.7
29.9
34.8
41.5
28.4
24.2
40.3
33.9
(a)   G -  Goose  neck nozzle,  S  -  90  nozzle.
(b)   Stack temperature was about 205C.

-------
                                       51
                 TABLE 13.  DATA USED IN STATISTICAL ANALYSIS OF
                            FILTER SIZE, TEMPERATURE, & NOZZLE
                            CONFIGURATION EFFECTS
                          (Measurements- are in mg/Nm )
Filter Size, F
Temperature, T

55 Goose
„," Neck
B

-------
          TABLE 14.  ANALYSES OF VARIANCE FILTER SIZE, TEMPERATURE, AND NOZZLE CONFIGURATION STUDY
Source of Deg
Variation Kr
F, Filter Size
T, Temperature
N, Nozzle
FXT
FXN

TXN
FxTxN
Reps
Blocks Within Reps
Remainder
Total
rees of
eedom
1
1
1
'1
1

1
-
2
9
6
23
Sum of
Squares
29.16
159.39
1.27
0.11
3.33

0.02
-
1.89
1410.04
32.57
1637.78
Mean
Square
29.16
159.39
1.27
0.11
3.33

0.02
-
0.95
156.67
5.43
•
p(a) Information
Ratio from Blocks
5.37 5,6,7,8,9,10,11
29.36 1,2,3,4,9,10,11,
12
0.23 1,2,3,4,5,6,7,8
0.02 1,2,3,4,5,6,7,8
0.61 1,2,3,4,9,10,11,
12
0.01 5,6,7,8,9,10,11
12
None
All
All
All
All
Conclusion
Significant at the
94% Confidence Level
Significant at the
99% Confidence Level
Not Significant
Not Significant
Not Significant

Not Significant





(a) For  an  F-ratio with  1  and  6  degrees  of  freedom,  as  in  this  table,  any  calculated  value  of  F exceeding 3.78
   is significant at  the  90 percent  confidence  level,  exceeding  5.99  is significant  at  the 95 percent confidence
   level,  or  exceeding  13.70  is significant  at  the  99  percent  confidence  level.
                                                                                                                    Ul
                                                                                                                    KJ

-------
                    TABLE 15.  COMPARISON OF LEVEL OF SIGNIFICANT FACTORS - FILTER SIZE,
                               TEMPERATURE, AND NOZZLE CONFIGURATION STUDY
Factor
Filter Size
Temperature
Level
Description
6.4
121
1

Mean
29.
33.
2
5

Effect^
3
ng/Nm
3
mg/Nm
Level
Description
10.2
Stack
2

Mean
31
27
.9
.2

Effect(d)
3
mg/Nm
mg/Nm
Difference In Mean
Effect ("Level 1 - Level
-2.7
+6.3
3
mg/Nm
3
mg/Nm
Percentage
2) Difference
-8.5
+23.2
(a) Using only data from those statistical blocks identified in Table 14.

-------
                                      54
The effect of using the smaller 6.4cm filter is a 8.5 percent decrease the in
average sample weight collected as compared with the 10.2cm filter.  This
difference was shown to be statistically significant at the 94 percent confidence
level.  The most statistically significant result is due to sampling system
temperature.  Operation of the Method 5 train at the normal probe and filter box
temperatures yield particulate results with average sample weights which are 23.2
percent higher than results obtained when the sampling system tempe rature is
maintained at about 205C.  This result is significant at a confidence level of
greater than 99 percent.
            In  summary,  the results  indicate  that  significant  particulate mass
  differences  resulted  from variation  of  the  filter  size  and the  sampling system
  temperature.  Significant differences due  to  nozzle  configuration or  to any
  interaction  of  the  three factors,  filter size,  temperature,  and nozzle design
  were  not  observed.

-------
                                      55
 Filter  Box/Probe  Temperature  Experiments

          Three combinations  of probe  and  filter  temperatures were  studied  in
 this  experiment:
           (1)  Both at  stack  temperature  (~205  C)..
           (2)  Both at  121~C;  and
           (3)  Probe at  stack  (205  C),  filter box  at 121 C.
 Since only  two temperature combinations could  be  run  concurrently,  three
 separate  collection runs were necessary so that every possible  pair of
 temperature  combinations could be  operated simultaneously.   Each of these
 pairs was repeated three times to  increase the power  of the  statistical
 analysis.   The randomized experimental design  for these nine statistical blocks
 is  shown  in  Table 16.   The temperature  combinations and operators were randomly
 assigned  to  the two sampling  systems.
          Table 17 summarizes the  data collected  for  this experiment.   In Table
 18, data  are given in  the incomplete randomized block pattern in which the
 experiment was performed.  The adjusted treatment means shown at the bottom
 of  Table  18  have  the effect of varying stack conditions removed by  the
 standard  statistical procedure described in Appendix  D. Comparisons among  these
 adjusted  treatment means represent the best estimate  of the  differences in  the
 sampling  methods  due to the  temperature operating conditions of the filter  box
 and probe.
          Table 19 presents  the statistical analysis  of the  test data.  The F-
.ratio,  6.77  indicates,  at the 98 percent confidence level, that there is a
 statistically  significant difference in  results  obtained with the three sampling
 system  temperature operating  conditions.
          The  significance of differences  between the various pairs of sampling
 temperature  conditions  can be evaluated by examination of the adjusted treatment
 means given  in Table 18.  The adjustment is necessary to avoid  contaminating
 differences  in means with temporal variations  in  the  source  emissions.   Signi-
 ficances  at confidence levels of  95 and  99 percent are associated with adjusted
 mean  differences  of 7.49 and  11.09 mg/Nm^,  respectively.  The method of
 determining  these critical values  is described in  Appendix D.   Thus,  the
 respective differences  of 10.0 and 10.1 mg/Nm3, (about  43 percent difference)
 indicate with  greater than 95  percent  confidence  that  results obtained with both

-------
TABLE 16.  RANDOMIZED EXPERIMENTAL DESIGN TO STUDY THE EFFECT
           OF PROBE AND FILTER TEMPERATURE COMBINATIONS
Test No.
1
2
3
4
5
6
7
8
9
Statistical
Block No.
1
8
4
9
2
7
6
3
5
System A Temperatures, C
Probe Outlet
121
121
Stack
Stack
Stack
121
Stack
Stack
121
Filter Box
121
121
Stack
Stack
121
121
121
Stack
121
System B Temperatures, C
Probe Outlet
Stack
Stack
121
Stack
121
Stack
Stack
Stack
Stack
Filter Box
Stack
121
121
121
121
Stack
Stack
121
121

-------
                      57
TABLE   17.  SAMPLE WEIGHT DATA-FILTER BOX/PROBE
            TEMPERATURE EXPERIMENTS

Run
No.
32A
B
33A
B
34A
B
35A
B
36A
B
37A
B
38A
B
39A
B
40A
B
Nominal Temperatures, C
Filter Box
121 (250 F)
Stack (-400)
121
121
Stack
121
Stack
121
121
121
121
Stack
121
Stack
Stack
121
121
121
Probe
121
Stack
121
Stack
Stack
121
Stack
Stack
Stack
121
121
Stack
Stack
Stack
Stack
Stack
121
Stack
Sample Weights,
Filter
40.8
41.5
39.7
38.7
25.9
20.5
32.7
30.0
27.1
24.3
28.1
29.6
30.9
32.2
30.4
28.2
27.9
27.8
. Probe
76.5
36.0
80.7
41.2
9.8
25.1
11.7
11.7
7.8
14.9
23.4
8.1
16.2
8.5
14.1
13.4
40.1
18.4
mg
Total
117.3
77.5
120.4
79.9
35.7
45.6
44.4
41.7
34.9
39.2
51.5
37.7
47.1
40.7
44.5
41.6
68.0
46.2
mg/Nnr
56.2
36.8
58.5
38.5
16.5
20.4
21.4
20.1
16.4
18.3
23.4
17.3
21.5
18.9
20.1
19.1
30.6
21.3

-------
                                      58
             TABLE 18.  DATA FOR STATISTICAL ANALYSIS
                        TEMPERATURE EXPERIMENTS

                             (Results are in mg/Nm )
— FILTER BOX/PROBE

Statistical Probe
Block No. Filter Box:
1
2
3
4
5
6
7
8
9
Treatment Means
Adjusted for Block
Effects
Temperature
121 C
Combinations
Stack (~205 C)
121 C Stack (--205 C)
(A) CB)
56.2
18.3
—
20.4
30.6
—
23.4
58.5
--
33.1 mg/Nm3
36.8
--
20.1
16.5
--
18.9
17.3
--
21.4
23.1 mg/Nm3

Stack (-^205 C)
121 F
(C)
—
16.4
19.1
--
21.3
21.5
--
38.5
20.1
23.0 mg/Nm3
Differences in adjusted treatment means;
                       3
      A-B  =10.0 mg/Nm
      A-C  =10.1 mg/Nm3
      B-C  =  0.1 mg/Nm3

-------
                                        59
                TABLE 19.   ANALYSIS OF VARIANCE  —  FILTER BOX/PROBE
                           TEMPERATURE EXPERIMENT
  Source
    of
 Variation
  Sum
   of
Squares
Degrees                .  .
   of      Mean F-ratia^  '
Freedom	Squares	Conclusion
 Between Filter Box/
 Probe Temperatures
 Combinations               306.07
 Between actual
 Block Totals              2381.86

 Remainder                 158.10
 (Error Term)
                    8

                    7
           153.04   6.77  Significant
                          at the 95%
                          Confidence Level


           (discard)

            22.59
 Total
2846.03
 17
(a)  For an F-ratio with 2  and 7 degrees  of freedom,  as  in this table,  any
    calculated value of F  exceeding 5.59 is significant at the 95 percent
    confidence level, or exceeding 9.55  is significant  at the 99 percent
    confidence level.

-------
                                     60
the 205C probe/205C filter and 205C probe/121 C filter conditions are significantly
lower than results obtained with the 121 C probe/121 C filter.   Results
obtained with the 205 C probe/205 C filter temperatures and with 205 C probe/
121 C filter temperatures are not significantly different.

-------
                                     61

In-Stack-Filter Experiments

          Experiments were performed to compare particulate measurements made
with the following three types of filters;
          (1)  In-stack flat filter,
          (2)  In-stack thimble filter, and
          (3)  Method 5 box filter.
The probes and filter boxes were operated at about 205  C and  121  C, respectively,
for tests with all three filter types.
          Since only two filter types could be evaluated simultaneously in  the
stack, three separate collection runs were necessary in order  to  operate every
possible pair of filter types concurrently.  To increase the  power  of  the
statistical analysis, the use of each pair of filter types was  repeated three
times, resulting in nine statistical blocks of data collection.   These nine
statistical blocks were randomly ordered over time and  filter  types were randomly
assigned within blocks as shown in Table 20.  The two operators were also
randomly assigned to sampling systems.
          The data collected for this experiment are displayed  in Table 21.
These data fit the incomplete randomized block design as shown  in Table 22.
The adjusted treatment means shown have had the effects of varying  stack
conditions removed, and comparisons  between these means reflect best
estimates of the differences in the  sample system due to the  type of
filter used.  Data are given for nozzle and filter catch and  for  total system
catch  (nozzle, filter, probe, and backup filter) in the in-stack  filter runs.
          The analyses of variance of the test data is  summarized in Table  23.
Using the data based on the nozzle and in-stack filter  collections, an F-Ratio
of 12.78 obtained from this analysis.  This F-ratio indicates that there are
high significant differences in results obtained with the various types of  filters.
          The differences in adjusted treatment means shown in Table 22 can be
examined to reveal significant differences in results obtained with the various
filter types.  An observed difference in adjusted treatment means which exceeds 3.7
     3                                               3
mg/Nm  an observed difference which exceeds 5.5 mg/Nm   is statistically
significant at the 99 percent confidence level.  Therefore, the respective differences
                    3
of 6.7 and 7.0 mg/Nm  between the in-stack thimble and  the in-stack flat and the
Method 5 results are statistically significant at a confidence levels greater than 99
                                     3
percent.  The difference of 0.3 mg/Nm  between the in-stack flat  filter and Method
5 results is not statistically significant.   These conclusions are based on only
the nozzle and in-stack filter collections in the in-stack filter runs.  Use of
the total system collections for the in-stack filter experiments  does not change
these conclusions.

-------
                                       62
           TABLE 20.  RANDOMIZED STATISTICAL DESIGN FOR COMPARISON OF
                     PARTICULATE MEASUREMENTS WITH METHOD 5 AND TITO
                     TYPES OF IN-STACK FILTERS
Test             Statistical         System A                  System B
 No.	Block No.	Filter Type	Filter Type
 1                    6            In-stack thimble         In-stack flat
 2                    3            In-stack flat            In-stack thimble
 3                    1            Method 5                 In-stack flat
 4                    5            In-stack thimble         Method 5
 5                    4            In-stack flat            Method 5
 6                    8            In-stack thimble         Method 5
 7                    2            Method 5                 In-stack thimble
 8                    9            In-stack flat            In-stack thimble
 9                    7            Method 5                 In-stack flat

-------
                                            TABLE 21.  SAMPLE WEIGHT AND DISTRIBUTION DATA-IN-STACK FILTER EXPERIMENTS
Sample Weights,
Run No
41 A
B
42 A
B
43 A
B
44 A
B
45 A
B
46 A
B
47 A
B
48 A
B
49 A
B
. . In-Stack
Filter TvpetaJ Box Filter Filter
Thimble
Flat
Flat
Thimble
Method 5
Flat
Thimble
Method 5
Flat
Method 5
Thimble
Method 5
Method 5
Thimble
Flat
Thimble
Method 5
Flat
2.6
3.6
3.6
3.9
77.0
1.6
3.5
100.6
3.8
76.3
2.9
50.5
57.4
1.9
5.1
5.2
127.7
4.2
87.4
80.4
88.8
107.8
79.8
127.5
77.7
73.2
79.0
93.6
111.6
120.7
Probe
1.2
1.9
1.5
1.8
5.6
1.6
0.6
6.1
0.6
10.0
1.8
4.0
6.3
1.3
1.1
-0.3
9.4
2.3
Nozzle
1.3
4.7
2.2
0.7
4.7
4.0
3.2
0.9
2.1
. 6.6
4.4
9.0
mg
Total
92.5
90.6
96.1
114.2
82.6
87.7
135.6
106.7
85.3
86.3
78.8
54.5
63.7
84.3
106.4
121.2
137.1
137.1
mg/Nm3(b) Sample Distribution, Percent of Total Collection

37.5
36.0
40.1
48.5
36.1
36.9
56.8
46.6
35.7
37.4
31.6
23.1
26.9
34.9
42.9
48.7
58.7
55.3

(39.1)
(38.3)
(42.4)
(51.1)
(38.3)
(58.6)
(37.6)
(33.6)
(36.3)
(45.6)
(50.9)
(58.1)
Box Filter
2.8
4.0
3.7
3.4
93.2
1.8
2.6
94.3
4.4
88.4
3.7
92.7
90.1
2.3
4.8
4.3
93.1
3.1
In-Stack
Filter
94.5
88.7
92.4
94.4
91.0
94.0
91.1
92.9
93.7
88.0
92.1
88.0
Probe
1.3
2.1
1.6
1.6
6.8
1.8
0.4
5.7
0.7
11.6
2.3
7.3
9.9
1.5
1.0
0.0
6.9
1.7
Nozzle Filter & Nozzle
1.4
5.2
2.3
0.6
5.4
2.9
3.8
1.1
2.5
6.2
3.6
7.2
95.9
93.9
94.7
95.0
96.4
96.9
94.9
94.0
96.2
94.2
95.7
95.2
Behind Filter
4.1
6.1
5.3
5.0
3.6 cr.
3.1
5.1
6.0
3.8
5.8
4.3
4.8
(a)   Thimble-Instack,  Munktell glass fiber thimble
     Flat-Instack,  6.4cm MSA 1106BH glass fiber filter
     Method 5- 7.6cm MSA 1106BII glass fiber filter
(b)   Based on nozzle and in-stack filter catch.  Values in parenthesis are calculated from total system catch.

-------
                                       64
                 TABLE 2Z  DATA FOR STATISTICAL ANALYSIS — IN-STACK
                           FILTER EXPERIMENTS(a)
Statistical
Block No.
1
2
3
4
5
6
7
8
9
Treatment Means
Adjusted For Block
Effect


In- stack thimble
(A)

34.9
48.5
-
56.8
37.1
-
31.6
48.7
45.4
(47.3)
--
(36.3)
(51.1)
-
(58.6)
(39.1)
-
(33.6)
(50.9)
mg/Nm3
Filter Type
In-stack
(B)
36.9
--
40.1
35.7
--
36.9
55.3
—
42.9
38.7 mg/:
(40.7)

Flat
(38.3)

(42.4)
(37.5)

(38.3)
(58.1)

(45.6)
Nm

Method 5
(C)
36.1
26.9
—
37.4
46.6
.
58.7
23.1
—
38.4 mg/N
m
Difference in adjusted treatment means:
                              3
      A-B   =  6.7 (6.6) mg/Nm3
      A-C   =  7.0 (8.9) mg/Nm::
      .B-C   =  0.3 (2.3) mg/Nm

(a)  Analysis based on nozzle and filter catch.
     calculated from total system catch.
Values in parenthesis are

-------
                                               65
                TABLE 23.  ANALYSIS OF VARIANCE -- IN-STACK FILTER EXPERIMENTS*-
Source
of
Variation
Between Corrected
Treatment (Filter-
types) totals
Sum
of
Squares
141.15 (193.64)
Degree
of
Freedom
2
Mean
Square
70.57 (16.82)
Conclusion of
Statistical
F-ratio(b) Test
12.78 (20.87) Significant
at the 99%
Confidence
Level
Between actual
Block totals

Remainder
(Error term)
1496.81 (1582.81)    8     (discard)


  38.66 (32.49)      7     5.52 (4.64)
Total
1676.62 (1808.94)   17
(a)  Analysis based on nozzle and filter catch.  Values in parenthesis are based
     total system catch.
(b)  For an F-ratio with 2 and 7 degrees of freedom, as in this table, any
     calculated value of F exceeding 5.59 is statistically significant at the 95
     percent confidence level, or exceeding 9.55 is statistically significant at the
     99 percent confidence level.

-------
                                      66
          In summary, the results using only the nozzle and filter catch in
in-stack filter experiments indicate that the use of the in-stack thimble filter
yields particulate concentrations which are 17 and 18 percent higher than the
respective in-stack flat and Method 5 filter results.  Respective differences
based on total catch of the in-stack systems are 16 and 23 percent.  Statistically
comparable results were obtained by Method 5 and in-stack flat filter sampling.

-------
                                       67
          Character"! zaM nn r»f Kiln PVn-i gg-i nng anH  Pa rf i' nil a t~p  foil pqt" inns

General Chemical Composition of Particulates

            Extensive chemical analyses were performed on a grab  sample and
filter and probe samples obtained in the earlier  sampling effort^   .  The
results of the analyses are summarized in Table 24.  Within the limits of
sampling and analytical error, the composition of collections with  the Method
5 sampling system appear the same as the grab sample, which should  be representive
of the particulate emissions.
            Analyses of particulate samples obtained during concurrent sampling
with Method 5 (206 C probe and 129 C filter box)  and an in-stack  flat filter
system are presented in Table 25.  The particulate  samples were removed from the
filter  medium before analysis.  The probe and nozzle  samples were  residues
from an acetone rinse of these components.  With  the exception of Na and K,,
there appear  to be  no  significant differences  in  the elemental composition of  the
samples.  Sodium and potasium concentrations in the filters are significantly
higher than concentrations in the Method 5 probe  and in-stack system nozzle
samples.  However, total Na and K in samples collected with the two types  of
trains is approximately the same.

Selected  Chemical Analysis of  Particulate  Collections

            A white crystalline material was found, in varying quantities, in
all Method 5 probe residue samples when sampling  was performed using filter box
and probe outlet gas temperatures of 121 C  (250 F).  A pair of probe residue
samples (Runs 15A and  B) containing the material  are shown in Figure 16.
            Infrared spectrometry  (IR), X-ray diffraction (XRD),  and classical
chemical analysis were employed to identify the extraneous material in the probe
collections.  Results  from these analyses confirmed that  the  substance was
ammonium  sulfate, (NH,),>SO, .
            The  IR  spectrum  of  a  probe  residue  sample  is  shown  in Figure  17,   A
reference spectrum of  ammonium sulfate obtained under  the same conditions  (KBr pellet)
is presented  for comparison.  The ammonium  ion  is indicated by the  spectral  structure
centering near 3150 cm   (N-H stretch) and by absorption  at 1400  cm   (N-H
deformation).   The asymetric stretching frequency of the  sulfate  absorbs at 1100
cm  .  The band a 612  cm   is also indicative  of  sulfate  structure. Weak  bands
near 650  cm"  may indicate the presence of  small  amounts  of other sulfates.

-------
                      TABLE 24.  ANALYSIS OF GRAB SAMPLE AND METHOD 5 PARTICULATE SAMPLES
                                                     Composition, weight percent
Sample                Al     C     Ca   Cl     Fe     H     K      Mg    N     Na     S      Si     C03     SO,
Grab Sample           2.7   7.2   26.2  0.4   1.3    0.3   3.5    0.7   0.1   0.5    3.2   5.3    36.6     9.4         <»
Method 5 (Average)
Filter catch          1-5         20-40       1-5          1-5    .5-2         1-5          5-10   17-46    8-15
Probe catch           3-10        20-50       1-10         5-20   1-5         1-10         3-10           7-15

-------
                     TABLE  25.  ANALYSIS OF PARTICULATE COLLECTIONS FROM METHOD 5 AND IN-STACK SAMPLING
                                                                                       (a)
Sample         	Elemental Composition, weight percent
Method 5
        	Al   Ba     Ca    Cr    Cu    Fe     K     Me   Mn    Na    Ni   Pb   Si   Sn    Sr    Ti    V     Zn
        (a)
Filter catch   3.   .03   15.25  <.01  .005  3.    15-25  1.    .1    3.   <.01  .04  3.   <.01   .04    .3     .01  <.l

Probe catch    3.   .02   10-15   .02  .03   1.     2.    1.    .05     .5    .01  .02  3.     .05   .02    .1   <.01    .2

In-Stack Flat Filter^ '

Filter catch   3.   .02   15-25  <.01  .005  2.    15r25    .8   .1    4.   <.01  .04  3.   <.01   .04    .3     .01  <.l

Nozzle catch   2.   .01   10-15   .07  .02   1.     3.      .5   .1    1.     .05  .02  3.     .03   .02    .1   <.01    .5


(a)  Analyzed by optical emission spectroscopy
(b)  Samples from Run No. 43A
(c)  Samples from Run No. 43B

-------
FIGURE 16.  PROBE RESIDUES SHOWING CRYSTALLINE
            MATERIAL OBTAINED FROM SAMPLING
            WITH METHOD 5

-------
                                                WAVtlfNGTH (MIOIONS)
                                                                                                        30 40
4000       3500      3000      2500       2000    1800    1600     1400     1200    1000     800     600     400      200
                                                 FREQUENCY (CM1)
                                                WAVELENGTH (MICRONS)
                                                                     8    9   10     12     15     20     30 40
  Reference  Sample
4000       3500      3000      2500       2000
                                               1800    1600     1400
                                                 FREQUENCY (CM1)
                                                                      1200    1000
          FIGURE 17.   INFRARED SPECTRA  OF  PROBE RESIDUE AND
                             AMMONIUM SULFATE REFERENCE  SAMPLE

-------
                                       72
A trace of an organic carboxylic acid salt could be indicated by absorption at
1620 cm   and the very weak shoulders at 2920 and 2840 cm" .
          XRD patterns were obtained by placing a small amount of the probe residue
on a glass fiber and exposing the sample to FeKQ1 x-rays in a 57.3 mm diameter
Debye-Scherrer Camera.  The pattern obtained matched that for Mascagnite  (NH,)p
SO,, (Powder Diffraction File Number 10-343).  The following lines attributable
to ammonium hydrogen sulfates were not present in the sample pattern;
            NH.HSO.                        4.75 and 3.69 A
              4   4
            (NH4)3H(S04)2                  4.95, 4.98 and 3.77 A
The following lines were present in the sample pattern were not identified:

            Line, A                        Intensity
            5.69                            Weak
            3.55                            Very weak
            2.87                            Very weak
            2.734                           Weak
            2.553                           Very, very weak
            2.242                           Very, very weak
            Classical chemical analysis for NH, and SO, in the crystalline probe
material were performed by Kjeldahl and barium perchlorate-thorin titration
procedures, respectively.  The results confirm the presence of NH, and SO,  in
a ratio consistent with the stoichiometry of  (NH,)2SO,.

-------
                                     73
          Potential interactions during sampling include condensation of
ammonium sulfates and/or sulfuric acid and reaction of sulfur oxides with the
filtermedium and/or collected particulates.  Accordingly, selected samples
were analyzed for NH,  and SOT ions to investigate the formation of extraneous
sulfate during the sampling process.  Calcium was also determined to provide
a relative measure of cementitious material in the samples.
          Analyses of filter samples were performed by digestion of the filter
and particulate catch in hot, dilute hydrochloric acid.  Ammonium determinations
were performed by the Kjeldahl method; sulfate by barium perchlorate -- thorin
titration; and, calcium by atomic absorption spectrometry.  Corrections were
applied for background (blank) levels of Ca, NH,, and SO, found in unused
filters.
          The data in Table 26 were obtained from analyses of filter and probe
samples collected with various probe and filter box operating temperatures.
Filter analysis was performed by digestion of sample and the filter medium.
Corrections were applied for filter blanks.

-------
TABLE 26.   ANALYSIS OF FILTER AND PROBE SAMPLES FROM EXPERIMENTS
            USING VARIOUS FILTER/PROBE TEMPERATURE COMBINATIONS
Run Operating Temperature, C Fnter
Weight Percent in Filter
Catch
No. Probe Outlet Filter Box Catch mg Ca
34A
B
36A
B
37A
B
38A
B
39A

40A
B
201
124
198
122
122
189
200
189
201
188
126
191
205
122
123
125
126
204
123
203
196
123
123
123
25.9
20.5
27.1
24.3
--

—


—
27.9
27.8
17.1
18.6
14.5
14.7
--

--


—
18.4
16.9
NH^ 804
<.4 23.9
<.4 25.4
<.4 28.4
<.4 29.2
-_ -_

-_ __


—
<.4 27.2
<.4 24.8
Probe
Weight Percent in Probe
Catch mg Ca
9.8
25.1
7.8
14.9
23.4
8.1
16.2
8.5
14.1
13.4
40.1
18.4
16.6
9.6
17.9
12.8
6.0
14.8
16.0
13.4
19.1
19.0
8.9
19.6
NH, SO,
4 4
<1 7.1
6.4 29.5
<1 10.3
4.7 29.5
9.0 28.6
<1 21.0
<1 8.0
<1 10.6
<1 7.8
<1 11.2
9.5 36.7
<1 8.2
Ratio Probe
NH , / SO,
_.
.22
—
.16
.31
--
__
--
__
—
.26
—

-------
                                     75

The high NH, and SO, concentrations  in the probe samples from Runs 34B, 36B,
37A, and 40A indicate  the  formation  of ammonium  sulfates when  the  probe is
maintained  at about  121  C.  The  stoichiometric ratios  of NH,  and SO,  suggest
                                                            4        4
that bisulfate or mixtures  of bisulfate  and  sulfate  are  formed.  Ammonium
sulfates were not detected in the filter catch when the lower probe temperature
of 121 C (250F)  was used.
          The probe  and  filter samples obtained  with  a  probe  temperature of
about 200 C and  filter box  temperature of  either 121 C or  200  C do not  show
the formation of ammonium  sulfates.  These samples,  however,  show  significantly
higher  percentages of  sulfate in the filter  analyses than  in  the probe  analyses.
          Analytical results of  particulate  samples  from the  Method 5 and in-
stack filter comparison  test are presented inTable 27.   In concurrent tests,
Method  5 and in-stack  flat  filter sampling gave  similar  results while the in-
stack thimble yielded  higher particulate mass results.   The data in Table 27
show that calcium-to-sulfate ratios  are  the  same for the Method 5  and in-stack
flat filter catches.   However, ratios  are  much lower for the  in-stack thimbles
due to  much higher quantities of sulfate.   If the weight of the thimble catch
is corrected for "excess sulfate" by using the Ca/SO,  ratio of the Method 5
or in-stack flat filter  catch obtained concurrently, the results 'of the three
configurations are in  good  ag-reement.  The results indicate that the difference
between the results  obtained with the  in-stack thimble and Method  5 and the
in-stack flat filter can be attributed to  extraneous sulfate  formation  on
the thimble.
          The backup filters used behind the in-stack  filters  show generally
                             +
lower quantities of  Ca and  NH, and higher  levels of  sulfate which  may indicate
a reaction  with  sulfur oxides.   However, the very low  sample x^eights preclude
any firm conclusions regarding these results.  As noted  in the previous data,
the sulfate concentration in Method 5 filters is  significantly  higher than the
probe sulfate content.
          The high filter-to-probe ratio of  sulfate  (coupled with  approximately
equal Ca concentrations) suggest that  extraneous sulfate may  also  be formed by
reaction of sulfur oxides with the MSA 1106BH filter medium.   In order  to
investigate this possibility, filters from  selected runs were  halved (<5 mg
difference  in halves)  and  sulfate analyses were performed on one-half  (filter
and particulate) and on  particulate  samples  (about 50  percent) catch carefully
removed from the other half.  The results  showing distribution of  the sulfate
between the particulates and the filter  (by  difference)  are given  in Table  28.
The results indicate that  sulfur oxides  also react to  a  lesser extent with  MSA
1106 BH filter  material when used in  either Method  5  train or in-stack.

-------
                           76
TABLE 27.   CHEMICAL ANALYSES  OF  SAMPLES  COLLECTED  IN
            VARIOUS METHOD  5 AND  IN-STACK SAMPLING  TRAIN
            COMPONENTS
Run
No.
42A
42 B
45A
45B
47A
47B
Weight of
Sample Catch, mg
In-stack flat filter
Backup Filter
In-stack thimble
Backup Filter
In-stack flat filter
Backup filter
Nozzle
Method 5 filter
Probe
Method 5 filter
Probe
In-stack thimble
88.8
3.3
107.8
3.9
77.7
3.8
4.9
76.3
14.0
57.4
10.5
79.0
Particulate
Ca
18.2
2.2
12.3
2.6
16.0
3.2
10.8
16.3
17.6
18.8
13.6
12.8
Composition, Weight Percent
NH/ S07
4 4
0.02
3.3
0.02
0.8
0.03
0.8
<0.4
<0. 02
<0.1
<0.03
<0.2
<0.02
26.5
41.7
39.8
23.0
29.3
50.0
6.1
30.1
6.4
20.9
5.7
52.0

-------
                          TABLE 28.   SULFATE IN FILTERS AND PARTICULATE COLLECTIONS
Run
No.
41B
44 B
48A
50A
Filter(a)
Configuration
In-stack flat
Method 5
In-stack flat
Method 5
Weight of
Filter Catch, mg
90.6
106.7
106.4
108.0
Distribution of Sulfate, mg
Filter and Catch
14.1
31.6
21.9
34.9
Catch
6.1
14.9
11.2
17.2
Filter (difference)
8.0
16.7
10.7
17.7
Filter Sulfate
Contribution.
Percent
8.8
15.7
10.1
16.4
(a)   Filter sizes:  Method 5 ~ 8.3 cm (diameter) In-stack ~ 6.4 cm

-------
                                      78
Inaccuracies in division of the filters would produce errors in the exact
distributions, but the consistency of the results tends to verify that some
sulfate is formed on the MSA filter.  Assuming the data in Table 28 is
reasonably accurate, the extraneous sulfate formation in the filter medium
could produce significant errors, in this case about 16 percent.
          In  summary, the chemical  analyses of samples obtained in this study
support the following observations.
          (1)  The gross composition of cementitious particulate collections
               do not appear to be  altered by the Method 5  sampling procedure
          (2)  Sampling with Method 5 as currently promulgated results in
               in formation of ammonium sulfates in  the probe
          (3)  Precipitation of ammonium sulfates in Method 5 sampling can
               be eliminated by operation of the probe at about 205 C.
          (4)  Sulfur oxides, presumably sulfuric acid vapor, react with the
               Munktell in-stack filter thimble.
          (5)  Reaction of sulfur oxides with MSA 1106 BH filter material is
               also indicated.

-------
                                        79
 Impinger Analysis

             The  impinger  samples  from four  pairs  of  tests  in which sampling was
 performed with Method  5 trains were  analyzed  as described  in the  Federal  Register,
 August  21,  1971.   The  results are presented in Table 29.   Ammonium and sulfate
 analyses of the  samples from Runs 23A and 23B show that  ammonium  sulfate  accounts
 for  91  and  95 percent, respectively,  of  the water residue  weight.

 Particle Size Measurements

             Particle size determinations were performed  to characterize the
 particulate emissions  from  the cement kiln  (after precipitator) and to aid in
 interpretation of  the  anisokinetic sampling data.  Aerodynamic  sizing  was attempted
 with an Andersen in-stack impactor,  but  questionable results were  obtained.  A
 large fraction  (about  40  percent) of the total catch was deposited on  the impactor
 wall prior  to the  first stage.  In addition,  the  glass fiber mats  used on the
 stages  to retain particles  could  not be  quantitatively recovered  because  small
 slivers were cut from  the edges by the spacing rings.  Negative  weights were
 obtained from stages with visible deposits.
             Particle size data obtained  by  electron  microscopy  is  presented in
 Figures 18,  19, and20.   Figures 18 and  19 show the size distribution  of particulate
 material collected on  the filter  and in  the probe during sampling  with Method 5.
 The  filter  box and probe  temperatures during  the  test (49A)  were maintained at
 132  and 206 C, respectively.  The mean mass diameter of  the  filter and probe
 particulate catches are about 0.95 and 1.3  micrometers,  respectively.
             Figure 20  gives the distribution of particulate  material collected on
 an in-stack flat filter.  The sample was collected in Run  49B which was performed
 concurrently with  the  Method 5 sampling.  The mean mass  diameter  of the particulates
 is about 0.6 micrometers.
             Photographs of  the particulate  collections are presented in Figures
21 ,  22, and 23.  The particles collected on both  filters,  Method  5 and in-stack,
 have about  the same size  and morphology.  The probe  catch  is comprised of larger
 particles and agglomerations of small particles which probably  deposited  under
 the  influence of gravity.

-------
                                       80
                       TABLE 29.   IMPINGER COLLECTION DATA
System Temperatures, C
Run No. Filter Box Probe
21A
21B
23A
23B 121 121
32A 121 121
32 B
35A
35B
Sample Residue Weights, mg
Water (extracted)
116.5
277.4
109.8
95.9
78.4
243.5
116.0
108.8
Chloroform/ Ether
15.8
29.4
43.0
8.4
13.2
11.5
20.9
14.5
Acetone
1.9
3.0
2.6
2.0
(a)
(a)
(a)
(a)
(a)   Sample  not  analyzed.

-------
                                 81
99.99
QQ Q
99
98
95
90
.c
'55 so
£ *>
"c 60
1 »
CL

^ 30
ij 20
CJ
10
5
2
1
0.5
0.2
O.I
0.05
0.01
0
^~
—
;
;
-
5
^
^
I
;
=
^
~
^
~
;
^
-
-
=
1 * ' ' ' \ 1 II



















>
.




















t
C





















till
















t







































,




















I


















X
X
X






1 1 1 1 1 1 1 1 1












X ^







t 1 1 1 1 1 1 11











^








f II 111(1 1











t
s





























Illl




















Illl










>









Illl




















Itll





















.02 0.03. 0.04 0.06 0.08 O.I 0.2 0.3 0.4 0.6 0.8







X












1 1 1 1 1 1 1 I 1
2
                      Equivalent Particle Diameter, micrometers
FIGURE 18.  PARTICLE SIZE DISTRIBUTION OF FILTER CATCH FROM METHOD 5
            SAMPLING OF CEMENT KILN EMISSIONS (RUN 49A)

-------
                                       82
.c
g>

I

>.
GO
c
0>
o

CD
CL


<.






Illl
0












X
X







Mlf
e











x
<














X
X
X
^y
X
X
X
X
<*








1 1 1 1 1 1 1 1 1
2
                            Equivalent Particle Diameter, micrometers
        FIGURE 19.   PARTICLE SIZE DISTRIBUTION OF PROBE WASH RESIDUE  FROM METHOD

                    5  SAMPLING OF CEMENT KILN EMISSIONS (RUN 49A)

-------
                                  83
99.99
QQ Q


98
95
£
0>
j 00
£D TO
(U
" 50
£ 50
Q.
>
••— *n
0 30
3
g
3
0
IU


n *^
Qo

.1
u.uo
n ni<
^>
•—
I
;
-
E
^
;
5
=
5
^
=
-
-
^
5
-
-
™
-



















/







































'




















^






































j




















c





tilt











X
X
X
X






^ 1 1 1 1 1 1 1 1










xx x




















X









r n 1 1^1 1 1









>































i in









































jiii







x












(III









































1 1 1 1 1 1 1 1 1
0.02   0.03 0.04   0.06  0.08  O.I          02     0.3   0.4     0.6  0.8   I
                      Equivalent Particle Diameter, micrometers

FIGURE 20.   PARTICLE SIZE DISTRIBUTION OF  FILTER CATCH FROM IN-STACK SAMPLING
             OF CEMENT KILN EMISSIONS  (RUN  49B)

-------
       *...
                  •'        4      *
v
                              .
                                •
     V   \
                     %
                        *
  15,OOOX
 15,OOOX
FIGURE 21.  PARTICULATES COLLECTED ON

            METHOD 5 FILTER (RUN 49A)

-------
                 15,OOOX
15,OOOX
15,OOOX
    FIGURE 22.  PARTICULATES RETAINED IN METHOD 5 PROBE (RUN 49A)

-------
             t
                        (*
                         *
                                   l»     *
                               • •.*

       15,OOOX
                               *•.
                                    *
*<
      15,OOOX
FIGURE 23.  PARTICULATES COLLECTED ON IN-STACK
           FLAT FILTER (RUN 49B)

-------
                                       87

Gas Composition of Kiln Emissions

             During initial experiments,  continuous  monitoring was performed for
 S09  and NO  (NO and N09  combined)  to establish concentration ranges in the
 kiln emissions.  A summary of the  data are presented in Table 30.  The table
 shows minimum,  maximum,  and estimated average S09 and NO  concentrations over
                                                 ^        X
 11 different 30 to 60-minute sampling periods covering a 3-day interval.  The
 table also includes the  results  of single  C09 and 09 analyses performed during
 the  sampling periods.
             The grand  average S09  and NO  concentrations are 450 and 535 ppm,
                                 *•        X
 respectively.   The S09 concentration in  the emissions exhibits large fluctuations;
 presumably associated  with coal  firing rate.   Less  flucuation was observed in
 oxides of nitrogen concentration.
             Analyses for ammonia and sulfur oxides  in the kiln emissions were
 performed to obtain correlative  data on  the potential for sulfate formation during
 particulate sampling.   Two measurements  of ammonia  in the kiln emission obtained
 on different days gave concentrations of 11 and 18  ppm.
             Two S09 measurements performed during the same day gave con-
centrations of 450 and  765 ppm-.  Both the corresponding sulfuric acid  (SO,)
measurements produced results indicating a level less than the minimum
detectable concentration, in this case about 0.1 ppm.  However, there is a
possibility that the sulfuric acid reacted with the quartz wool filter used
in the sampling.  If the level is similar to that of other coal-fired com-
bustion sources, a sulfuric acid concentration of about 1 percent of the S0»
concentration would be expected, i.e., about 4 to 8 ppm.

-------
                            88
TABLE 30.  SUMMARY OF MONITORING FOR GASEOUS  COMPONENTS  OF
           CEIIENT KILN EMISSIONS
Sampling
Period
I
2
3
4
5
6
7
8
9
10
11

Minimum
30
45
33
~ 0
25
260
~ 0
360
120
* 0
960
S09 (PPM)
Maximum
415
132
78
1000
280
2340
170
712
870
840
2260
NOV (PPM")
Average
176
89
55
500
173
952
81
466
301
425
1715
Minimum
635
490
460
285
615
420
430
310
400
435
240
Maximum
1556
600
580
570
730
780
570
260
663
630
340
Average
961
553
515
414
669
579
495
339
549
501
315
Per Cent
co2 o2
23.3
23.7
24.4
25.5
23.3
26.0
23.0
25.9
24.1
24.6
25.6
4.8
5.2
4.7
4.1
5.4
4.6
5.3
4.1
4.9
4.5
3.9

-------
                                     89

                                 DISCUSSION
          Of the sampling system operating parameters investigated in this
work, sampling temperature has the most significant effect on mass measurements.
Results obtained with Method 5 temperature conditions were up to 43 percent
higher than results obtained using a sampling temperature near stack conditions,
205C (400F).  Chemical analysis clearly shows that the difference results
from ammonium sulfate and/or ammonium bisulfate formation in probe when sampling
is performed at the minimum temperature specified by Method 5.
          The data presented in Appendix E, Figure E-2 indicates that the
minimum sampling temperature specified by Method 5 may not be above the dew
point of sulfuric acid, particularly for wet process cement plants which have
high moisture content in the emissions.  Figures E-3 and E-4 show that if about
2 ppm H2S04 and 15 ppm ammonia are present in the emissions, the sampling
temperature must be held above 190C (375F) to prevent formation of ammonium
sulfate in the sampling train.
          It is apparent that depending on the sampling temperature, different
mass emission results can be obtained when sampling the same cement plant..
Consequently, Method 5 when applied to cement plants must specify sampling
temperature conditions more rigorously.  In addition, gas temperature measurements
should be made during sampling and care should be exercised in regulating the
temperature in order to produce results with satisfactory precision.
          Selection of filter materials appears to be another important
consideration in measuring particulate emissions from cement plants (and probably
other sources with sulfur oxides emissions).  The Munktell glass fiber thimble which
has a relatively high tare weight compared to flat filters (about 2 grams versus
0.3 grams) also has a correspondingly high content of cations which readily react
to form sulfates.  Mass emission values obtained with the thimbles averaged 18
percent higher than Method 5 operated at elevated temperature and using MSA
1106 BH filters.  However, the results indicate that extraneous sulfate may be
formed on the MSA filter medium and may yield mass data which is high by 10
to 16 percent.  It is important to note that the sulfate formation occurs well
above the dew point of sulfuric acid.  Extraneous sulfate formation observed
                            (12  13)
in this study and by othersv   '   ' suggests the need for an alternate filter
material which is unreactive with sulfur oxides.

-------
                                      90
          The direction and magnitude of errors due to anisokinetic sampling
were on the order predicted by previous experimental and theoretical work       .
At 0.7 and 1.3 times isokinetic rate the results deviated from isokinetic mass
collections by +5.9 percent and -11.5 percent, respectively.  These deviations
are not declared highly statistically significant when compared to the normal
measurement errors.  Sampling within the ^ 10 percent range specified by Method
5 should restrict the magnitude of anisokinetic sampling error to an acceptable
level.
          The comparison of in-stack sampling with MSA 1106 BH filter medium
and Method 5 operated at a gas sample temperature of ~205C  (401F) and using
MSA filters shows that the two methods give equivalent results.  In-stack sampling
results including only the nozzle and in-stack filter catches were an average of  0.8
percent higher than Method 5 results, a difference which is not statistically
significant.  Analyses confirm that particulate collections made by the two
methods are also equivalent in chemical composition.
          In-stack sampling was not compared experimentlly with Method 5 operated
at the specified minimum gas temperature, 121C (250F).  However, it is evident
from other experiments which were performed that in this situation Method 5
would give significantly higher mass results.
          One advantage of the in-stack technique is that it greatly facilitates
sample recovery.  The Method 5 probe wash, during which there is potential for
sample loss or contamination may be eliminated.  In addition, if a backup filter
is used, a check is provided on any leakage of the in-stack filter.  Use of a
backup filter may also prevent invalidation of a source test due to leakage
of the primary filter.
          Overall, Method 5 appears to be an acceptable procedure to determine
mass emissions from cement plants with the exception of the problem of ammonium
sulfate formation discussed previously.  When sampling is performed by maintaining
an elevated gas temperature, the precision of the method appears acceptable,
5.5 percent (repeatability) and the chemical composition of the collections
appear to be representative of the particulate material at  the sampling location
in the stack.

-------
                                       91
                                  REFERENCES
  (1).  Federal Register, Volume  36, No. 247,  pgs 24876-24895,  Thursday,  December
       23, 1971.

  (2).  Federal Register, Volume  39, No. 47, pgs 9308-9323,  Friday,  March 8,  1974.

  (3).  Federal Register, Volume  39, No. 177,  pg 32852,  Wednesday,  September  11,  1974.

  (4).  Reference  1, pgs 24888-24890.

  (5).  Reference  3, page 32856.

  (6).  Henry, W.  M., Howes, J. E., Jr., and Engdahl,  R.  B.,  "Evaluation  of
       Particulate  Sampling and  Analysis Techniques  for  Cement Plant  (Kiln)
       Emissions",  Interim Report No  1, March 8, 1974.

  (7).  Kreichelt, T. E., Kemnitz, D.  A., and  Cuffe,  S.  T.,  "Atmospheric  Emission
       from  the Manufacture of Portland Cement", Public  Health Service Publication
       No. 999-AP-17  (Clearing house  Report No PB  190236)  1967.

  (8).  "Hydrogen  Ion Concentration  (pH) of Paper Extracts  -- Hot  Extraction  Method",
       TAPPI Method, T 435 Su-68, Technical Association  of  the Pulp and  Paper
       Industry,  360 Lexington Avenue, New York, N.Y.  10017.

  (9).  ASTM  Method  D 202, 1974 Annual Book of ASTM Standards,  Part 20, American
       Society for  Testing and Materials, 1916 Race  Street,  Philadelphia,  Pa. 19103.

 (10).   Goksdyr,  H.  and Ross,  K., "The Determination of Sulfur Trioxide in Flue
       Gases", J. Inst. Fuel, 35., pg  177  (1962).

 (11).  ASTM  Method  D 3226-73T, 1974 Annual Book of ASTM Standards,  Part  26,  American
       Society for  Testing and Materials, 1916 Race  Street,  Philadelphia,  Pa. 19103.

 (12).  Hemeon, W. C. L., and  Black, A. W., "Stack  Dust  Sampling;  In-Stack Filter  or
       EPA Train",  Journal of the Air Pollution Control  Association,  22, pgs. 516-
       519  (July  1972).

 (13).  Baum, F. and Riechardt, I., "Measuring Errors  occurring during Determination
       of Trioxide  on Measuring  Filters" Strub-Reinhalt  Luft,  27.  pgs 18-22  (September
       1967).

 (14).  Schmel, G. A., "Particle  Sampling Bias Introduced by Anisokinetic Sampling
       and Deposition within  the Sample Line", American Industrial Hygiene Association
       Journal, 31, pgs. 758-771,  (November-December  1970).

 (15).  Vitols, V. "Theoretical Limits of Errors due  to  Anisokinetic Sampling of
       Particulate  Matter", Journal of the Air Pollution Control  Association, 16.
       pgs.  79-84 (February 1966).

 (16).  Hemeon, W. C. L., and  Haines,  G. F., "The Magnitude  of  Errors  in  Stack Dust
       Sampling", Air Repair, 4,  pg 159  (1954).


S

-------
                                       92
(17).  Badzioch,  "Correction of Anisokinetic Sampling of Gas-Borne  Dust  Particles",
      Journal of the Institute of Fuels, .33., pg 106  (1960).

-------
            APPENDIX A
    METHOD 5 - DETERMINATION OF
    PARTICULATE EMISSIONS FROM
        STATIONARY SOURCES
Federal Register, Volume 36, No 247
    Thursday, December 23, 1971

-------
                                                  APPEND LX-A
                                                  RULES  AND  REGULATIONS
                                              2.1.4  Filter  Holder—Pvrex1  glass  with
                                            heating sysjem capable or maintaining mini-
                                            mum temperature of 225* F.
                                              2.1.5   Implngers / Condenser—Four impin-
                                            gers connected In series with glass ball Joint
                                            fittings. The  first, third, and fourth Impin-
                                            gers  are of  the Greenburg-Sinith  design,
                                            modified by replacing the tip with a '/a-Inch
                                            ID glass tube  extending to one-hair Inch
                                            rrotn the bottom or the flask. The second im-
                                            ptnger  Is  or the  Greenburg-Smlth design
                                            with  the standard  tip. A condenser may be
                                            •used in place or the impingers provided that
                                            the moisture content or the stack gas can
                                            still be determined.
                                              2.1.6   Metering  system—Vacuum  gauge,
                                            leak-free pump, thermometers  capable or
                                            measuring temperature  to within 5' F., dry
                                            gas meter  with. 2%  accuracy, and  related
                                            equipment,  or equivalent,  as required to
                                            maintain on, Isoklnetlc sampling rate and to
                                            determine sample volume.
                                              2.1.7   Barometer—To measure atmospheric
                                            pressure to  ± 0.1 Inches Hg.
                                              2.2  Sample recovery.
                                              2.2.1
                                            probe.
                                              2.2.2
                                              2.2.3
                                              2.2.4
          Probe  brush—At  least as  long as
          Glass wash bottles—Two.
          Glass sample storage containers.
          Graduated cylinder—250 ml.
    2.3  Analysis.
    2.3.1  Glass weighing dishes.
    2.3.2  Desiccator.
    2.3.3  Analytical balance—To measure to
   ±0.1 mg.
    2.3.4  Trip  balance—300 g. capacity,  to
   measure to ±0.05 g.
    3. Reagents.
    3.1  Sampling.
    3.1.1  Filters—Glass fiber. MSA 1106  BHi.
   or  equivalent, numbered  for identification
   and preweighed.
    3.1.2  Silica  gol—Indicating  type,  6-16
   mesh, dried ac 175* C. (350* F.) tor 2 hours.
    3.1.3  Water.
    3.1.4  Crushed Ice.
    3.2  Sample recovery.
    3.2.1  Acetone—Reagent grade.-
    3.3  Analysis.
    3.3.1  Water.
                                                                                                  cS TflA:;i OPTIONAL MAY BE REPLACED
                                                                                                   BY AH EQUIVALENT CONOE«SEJ?
                                                                       HEATED AREA  FILTER HOLDER / THERMOMETER   CHECK
                                                   PROBE    -fT  STACK
                                                     \      | I—WALL
                                                       (r~
                                             REVERSE-TYPE
                                              PITOT TU3E
                                3^£MMM
                                                                                                                   ^VACUUM
                                                                                                                      LINE
METHOD  5—DsrzsiirNATiotr or
   EMISSIONS FKOM  STATIONARY SOUBCES

   1. Principle and applicability.
   1.1  Principle. Particular matter is with-
drawn isckinstlcally from the source and its
weight Is determined gravimetrlcally after re-
moral or uncombined water.
   1.2  Applicability. This method is applica-
ble for the determination or paniculate emis-
sions from stationary  sources only when
cpecified by the test procedures ror determin-
ing  compliance  with New Source Perform-
ance Standards.
  2. Apparatus.
   2.1  Sampling train. The design specifica-
tions or the paniculate sampling train  used
by EPA (Figure 5-1) ore described in APTD-
0581. Commercial  models  of  this train are
available.
   2.1.1  Nozzle—Stainless  steel (316)  with
sharp, tapered leading edge.
   2.1.2  Probe—Pyrex1 glass  with a heating
system capable of maintaining a minimum
f.is  temperature or  250' F. at the exit end
d'irtng sampling  to prevent  condensation
frc*m  occurring.  When  length  limitations
{£rv.\ter than about S it.) tire encountered at
temperatures less than 600' F., Incoloy  825 >,
or equivalent, may be used. Probes ror  sam-
pling gas streams  at temperatures In. excess
of 600* F. must have been approved  by the
Administrator.
   2.1.3  Pltot tube—Type  S,  or equivalent,
attached  to probe  to  monitor stack gas
velocity.
                                                         PIT01 MANOMETER

                                                                  ORIFICE
            THERMOMETEI
                                     IMPiNGERS  .          ICE BATH
                                            BY-PASS VALVE
                                                                                                       VACUUM
                                                                                                        GAUGE
                                                                                                MAIN VALVE
                        DRY TEST METER
AIR-TIGHT
  PUMP
                         Figure 5-1. Particulate-sampling train.
  3.3.3  Desiccant—Drierlte,' indicating. - •
  4. Procedure.
  4.1  Sampling
  4.1.1  After selecting the sampling site and
the minimum number of sampling points,
determine the stack pressure,  temperature,
moisture, and range of velocity head.
  4.1.2  Preparation  or  collection  train.
Weigh to the nearest gram approximately 200
g. of silica gel. Label a filter or  proper diam-
eter, desiccate'  ror at least 24  hours and
weigh to the nearest 0.5 mg. in a room where
the relative humidity is less than 50%. Place
100 ml. or water  In each or the first two
implngcrs, leave the third implnger empty,
and place approximately 200 g. or preweighed
silica  gel In  the  fourth  implngsr. Se: up the
train  without the  probe  as in Figure 5—1.
Leak check  the  sampling  train at the sam-
pling  site by plugging up the inlet to the lli-
ter holder and pulling a 15 in. Kg vacuum. A
leakage rate  not in excess of 0.02 c-f jn. at a
vacuum of  15 in. Hg  is acceptable. Attach
the probe and adjust the heater to provide a
gas temperature  of about 250° F. at the probe
outlet. Turn on the filter  heating  system.
Place crushed ice around the implngers. Add
  i Trade nama.
  »Trade name.
  'Dry1 using Drterit*' at 70' F.±10' F.
  more Ice during the run to keep the temper-
  ature of the gases leaving the last impinger
  as low as possible  ar.d preferably at 70*  F..
  or less. Temperatures above 70° F. may result
  in damage to the dry  gas meter from either
  moisture condensation or excessive heat.
    4.1.3  Partionlnte train operation. For each
  run, record the dcvta required on the example
  sheet shown In Figure 5-2. Take readings at
  each sampling point, at least every 5 minutes,
  and when significant changes in stack con-
  ditions  necessitate  additional  adjustments
  in flow rate. To  begin sampling, position the
  nozzle at the first traverse  point  with the
  tip pointir.g  directly  into the gas stream.
  Immediately start the pump and adjust the
  Sow to isckir.etic  conditions. Sample  for at
  ioast  5 minuics  ac each traverse point; sam-
  pling tiir.i must be the same ror tarh point.
  Maintain isokinetlc sampling throughout the
  sampling period. Nomographs  are available
  which aid  in the  rapid adjustment of the .
  sampling raio without other computations.
  APTD-0573  details the procedure  ror using
  these nomographs. Turn off the pump at the
  conclusion of each run and record the final
  readings. Remove the probe and nozzle from
  the stack and handle in accordance with the
  sample recovery process described in section
  4.2.
                                FEDERAL REGISTER,  VOt.  36,  NO. 247—THURSDAY, DECEMBER 23, 1971

-------
                                                                 A-2
                                                  RULES AND  REGULATIONS
                                                                                24S89
t-ua 	


CKUTO
Dill

•mWV

iAUU IOI NO.
UHIQKUNO. 	 ..__. . ..
MTMiHj. .. . ..

tUVTBIKXm
Huvm












tOTIL
SAwtma
tine
!«.•*»
June
PWSSUKX
sua
rtjtffRAIURK
i !
1
1
















i









S
-------
 24890
                                                                  A-3

                                                    RULES  AND  REGULATION?
                               PLANT.

                               DATE
                               RUN NO.
CONTAINER
NUMBER
1
2
TOTAL.
WEIGHT OF PARTICULATE COLLECTED,
mg
FINAL WEIGHT



TARE WEIGHT


WEIGHT GAIN




FINAL
INITIAL
LIQUID COLLECTED
TOTAL VOLUME COLLECTED
VOLUME OF LIQUID
WATER COLLECTED
IMPINGER
VOLUME.
. ml




SILICA GEL
WEIGHT.
g



g» ml
  CONVERT WEIGHT OF WATER TO VOLUME BY DIVIDING TOTAL WEIGHT
  INCREASE BY DENSITY OF WATER.  (1 g. ml):


                                        INyE?Sf; 9   = VOLUME WATER, ml
                                          U g/ml)
                                                                                           whore.
                                                                                                I = Pi>rrcnt of teoklnnlc sampling.
                                                                                              V|f—Tula! volume of liquid collected In imp tag era
                                                                                                    and silli-a col (See Fig. 5-3), mL
                                                                                             PH.O = Density of water, 1 s./mJ.
                                                                                               R = Idral iris  constant. 21.S3 Inches ITg-«u. ft./lb.
                                                                                                    mi\e-°n.
                                                                                            Mnto "Molecular welt-lit of water, IS Ib./Ib.-mole.
                                                                                              Vo, a Volume ofu^s sample through the dry gas meter
                                                                                                    (nH'tur conditions), cu. ft.
                                                                                              Tis«A'»S'.>lutc  avrrx:e dry  gas meter temperature
                                                                                                    isco l'i-.'ure*--j).aR.
                                                                                             Fb«r = B;u-omciric pressure at sampling site, Inches
                                                                                                    lie.
                                                                                              <}II=AviT;ii:e pressure drop across the  orifice (se«
                                                                                                    rt. 5-y. iudu-s H;O.
                                                                                               T. = A!j«c'iute  aTprace stack gas  temperature (see
                                                                                                    l-itf.4-'J).«H?
                                                                                                0—T(.'t:il s:unplinc: time, mln.
                                                                                               V. = .-'t:i'.-k r:is velocity calculated by  Method  1,
                                                                                                    Initiation 2-2,  ft./scc.
                                                                                               r,<»AI)sf-1 *** ba3i3'
    '    '
                                                  M'~T°r^1 amount °' P^1"1*'0 maUor ca^tid-
                                                V.rtj-Volnme ol fas sample through dry gas motor
                                                       (standard conditions), ou. ft.
                                             6.7  Isoklnetlc variation.
.
1=
                  9V.P.A.
                                        X100
                              *V'P'A»                                     Equation 5-6


                                 FEDERAL  REGISTER, VOL. 36, NO. 247—THURSDAY,  DECEMBER 23,  1971

-------
        APPENDIX B
STACK GAS MEASUREMENT DATA

-------
               APPENDIX B
TABLE B-l.  STACK GAS DATA - ANISOKINETIC STUDY
Run No .
1A
3
2A
B
3A
B
4A
B
5A
B
6A
B
7A
B
8A
B
9A
B
10A
B
11A
B
12A
B
13A
B
14A
B
ISA
B
16A
B
VAP (avs),
cm H2Q3
1.12
1.13
1.13
1.12
1.10
1.08
1.08
1.07
1.08
1.05
1.07
1.10
1.12
1.12
1.08
1.10
Ts (avg)
C
203
212
213
222
223
,222
218
223
220
220
221
221
222
217
223
222
' Ps>
mm Hg
722.9
722.6
725.6
'724.9
725.2
724.4
723.6
721.9
722.1
720.6
723.4
723.1
722.1
722.9
721.4
720.1
o2, %
8.5
8.5
8.5
8.5
8.5
8.5
8.5
8.5
8.5
8.5
8.5
8.5
8.5
8.5
8.5
8.5
C02, 7o
18.5
18.5
18.5
18.5
18.5
18.5
18.5
18.5
18.5
18.5
18.5
18.5
18.5
18.5
18.5
18.5
w v
7 Ib/lo-mole
/o
32.94
32.86
31.87
31.77
32.20
32.00
30.56
30.97
30.54
30.23
31.09
30.03
30.61
30.38
31.37
31.07
32.70
31.17
31.03
30.21
32.66
32.81
31.22
30.72
30.21
29.40
32.92
32.48
29.71
29.81
30.05
29.48
31.3
31.3
31.3
31.3
31.3
31.3
31.3
31.3
31.3
31.3
31.3
31.3
31.3
31.3
31.3
31.3
Vo (avg),
& /
m/s
16.0
16.3
16.3
16.3
15.9
15.9
15.8
15.6
15.8
15.4
15.5
15.9
16.3
16.3
15.7
16.1

-------
                          B-2
TABLE B-2.  STACK GAS DATA-FILTER SIZE. TEMPERATURE. AND
         NOZZLE CONFIGURATION EXPERIMENTS
Run No.
17A
B
ISA
B
19A
B
21A
B
22A
B
23A
B
24A
B
25A
B
27A
B
28A
B •
29A
B
31A
B
V£p (avg) ,
cm K70*
1.13
1.15
1.04
1.13
1.16
1.15
1.08
1.13
1.07
1.10
1.10
1.1^
T (Avg),
O
c
209
211
214
214
211
198
205
203
221
213
207
206
Ps>
ram HG
717.3
716.5
715.5
713.7
712.7
714.2
714.2
713.5
717.6
716.3
718.6
718.1
o2, %
7.0
7.0
7.0
7.0
7.0
7.0
7.0
7.0
7.0
7.0
7.0
7.0
C02, Z
18.0
18.0
18.0
18.0
18.0
18.0
18.0
18.0
18.0
18.0
18.0
18.0
3 M,,
wo d
7o Ib/lb-mole
31.84
31.92
32.46
31.94
31.12
31.18
32.36
32.31
32.24
32.35
34.10
32.88
31.97
31.49
33.49
32.62
29.83
29.65
31.82
31.58
31.82
31.15
31.76
32.00
31.2
31.2
31.2
31.2
31.2
31.2
31.2
31.2
31.2
31.2
31.2
31.2
Vs (avg),
m/s
16.4
16.6
15.2
16.6
16.9
16.6
15.6
16.3
15.5
15.9
15.8
16.2

-------
                               B-3
TABLE B-3.  STACK GAS DATA-PROBE/BOX FILTER TEMPERATURE EXPERIMENTS
Run No.
32A
B
33A
B
34A
B
35A
B
36A
B
37A
B
38A
B
39A
B
40A
B
VAP (av§),
cm H2Q3
1.08

1.00

1.08

1.00

1.08

1.08

1.04

1.08

1.08

Ts (avg), Pg,
C mm Hg
194

204

189

204

188

190

199

189

188

714.2

713.7

713.7

713.0

713.0

712.7

712.5

713.7

714.2

o2, %
8.0

8.0

8.0

8.0

8.0

8.0

8.0

8.0

8.0

C02, %
18.0

18.0

18.0

18.0

18.0

18.0

18.0

18.0

18.0

B , M,,
%° Ib/lg-mole
31.49
30.27
30.27
28.91
30.17
29.41
28.23
27.76
32.01
31.50
30.50
30.72
30.24
30.74
30.75
30.98
29.51
30.40
31.2

31.2

31.2

31.2

31.2

31.2

31.2

31.2

31.2

Vs (avg),
m/s
15.0

14.4

14.8

14.3

14.8

14.8

14.9

14.9

14.8


-------
                         B-4
TABLE B-4.  STACK GAS DATA-IN-STACK FILTER EXPERIMENTS
Run No.
41A
B
42A
B
43A
B
44A
B
45A
B
46A
B
47A
B
48A
B
49A
B
VAP (avgi, Ts (avg), Pg,
cm H-O 3F c ram Hg
1.18
1.13
1.15
1.13
1.13
1.16
1.16
1.16
1.16
232
236
226
230
228
225
222
226
223
720.9
720.3
718.6
720.3
721.1
720.6
723.4
723.4
721.9
02, %
8.5
8.5
8.5
7.0
7.0
6.0
6.0
6.0
5.5
C02, %
19.0
19.0
19.0
19.0
19.0
19.5
20.5
20.5
22.5
B , M, ,
wo d
% lb/16-mole
31.01
31.38
31.34
31.36
31.33
31.71
32.00
32.62
32.10
32.33
31.95
32.27
32.37
33.05
32.18
32.29
31.68
32.08
31.4
31.4
31.4
31.3
31.3
31.4
31.5
31.5
31.8
Vs (avg),
m/s
17.4
16.8
16.8
16.7
16.7
17.1
17.0
16.9
16.9

-------
          APPENDIX C
SAMPLING SYSTEM OPERATION DATA

-------
TABLE C-l. SAMPLING DATA - ANISOKINETIC SAMPLING EXPERIMENTS
Average System Temperatures, C
Run No
1A
B
2A
B
3A
B
4A
B
5A
B
6A
B
7A
B
8A
B
9A
B
10A
B
11A
B
12A
B
13A
B
14A
B
ISA
B
16A
B
Meter Volume
(Vm), 0,3
2.37
1.67
2.45
2.42
1.67
2.29
2.28
3.04
2.31
1.62
2.31
2.30
2.98
2.98
1.61
1.59
1.62
1.63
2.93
2.95
2.23
2.91
2.97
2.32
2.35
2.41
3.00
2.37
2.30
2.34
1,65
2.38
Barometer,
mm He
741.9
741.9
741.7
741.7
744.7
744.7
744.0
744.0
744.2
744.2
743.5
743.5
742.7
742.7
740.9
740.9
741.2
741.2
739.7
739.7
742.4
742.4
742.2
742.2
741.2
741.2
741.9
741.9
740.4
740.4
739.1
739.1
AH.
mm H30
41.4
18.3
41.9
38.1
19.3
35.1
38.4
60.5
37.9
16.3
37.1
33.3
62.7
55.9
17.8
16.0
18.0
16.3
59.9
53.9
35.6
54.6
63.5
33.8
39.1
35.8
66.6
35.8
36.8
33.0
18.8
34.8
Avg. Meter
Temp (TmX C
32
29
32
33
34
37
38
39
33
28
34
32
37
32
33
24
38
26
41
32
29
32
33
33
35
33
33
31
33
38
34
36
Dry Gas Sampled
Std Cond. (Vjpstd), m3
2.24
1.59
2.31
2.28
1.57
2.14
2.12
2.82
2.18
1.55
' 2.16
2.18
2.78
2.82
1.51
1.54
1.50
1.57
2.69
2.78
2.12
2.76
2.81
2.18
2.20
2.26
2.84
2.24
2.15
2.16
1.54
2.21
Percent
Isokinetic
106
75
107
106
73
99
98
131
103
73
104
103
132
134
74
75
74
76
133
136
107
139
135
104
102
104
135
106
103
103
72
103
Filter Box
127
123
122
129
126
120
124.
127
126
122
124
128
128
123
122
120
124
124
123
127
126
121
126
124
123
122
123
124
125
122
122
124
Gas @ Probe
Outlet
129
126
124
126
127
124
133
127
134
122
136
126
121
122
123
122
121
124
124
124
129
124
123
124
123
125
129
130
129
128
124
128
Probe
Mid -Point
79
52
77
116
99
72
102
92
98
103
104
76
74 "•
81
95
77

-------
TABLE C-2. SAMPLING DATA - FILTER SIZE, TEMPERATURE, AND NOZZLE CONFIGURATION EXPERIMENTS
Average System Temperatures,
Run No
17A
B
ISA
B
19A
B
21A
B
22A
B
23A
B
24A
B
25A
B
27A
B
28A
B
29A
B
31A
B
Meter Volume
(VnX u.3
2.41
2.49
2.52
2.54
2.21
2.22
2.46
2.48
2.51
2.54
2.48
2.52
2.30
2.40
2.46
2.47
2.27
2.29
2.38
2.42
2.34
2.37
2.43
2.47
Barometer,
mm Hg
742.7
742.7
741.9
741.9
740.9
740.9
739.1
739.1
738.1
738.1
739.7
739.7
739.7
739.7
738.9
738.9
743.0
743.0
741.7
741.7
744.0
744.0
743.5
743.5
All, Avg. Meter
mm 1I20 TempCQ, C
42.2
39.6
43.4
40.9
36.1
32.5
42.4
38.1
44.7
40.1
45.0
40.1
38.4
35.1
40.6
38.4
36.8
33.0
40.6
36.6
39.9
36.1
42.7
38.4
33
38
34
41
37
38
34
35
35
36
26
23
29
28
31
27
29
33
31
38
25
26
31
29
Dry Gas Sampled
Std Cond. 
-------
TABLE C-3.  SAMPLING DATA - PROBE/BOX FILTER TEMPERATURE EXPERIMENTS
Average System Temperatures. C
Run No
32A
B
33A
B
34A
B
35A
B
36A
B
37A
B
38A
B
39A
B
40A
B
Meter Volume
(VmX m3
2.23
2.27
2.19
2.17
2.29
2.33
2.21
2.21
2.25
2.27
2.31
2.31
2.31
2.28
2.32
2.32
2.33
2.33
Barometer,
mm Hg
739.7
739.7
739.1
739.1
739.1
739.1
738.4
738.4
738.4
738.4
738.1
738.1
737.9
737.9
739.1
739.1
739.7
739.7
AH
mm H20
35.6
30.0
33.8
27.9
36.6
3Z.3
34.3
35.1
36.8
37.9
38.1
38.4
38.1
38.4
38.6
38.9
38.6
39.1
Avg. Meter
Temp (T,^ C
35
37
32
27
31
27
32
32
32
32
29
32
29
31
28
33
28
36
Dry Gas
Std Cond
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
Sampled
• 
-------
TABLE C-4.  SAMPLING DATA-IN-STACK FILTER EXPERIMENTS
Run No.
41A
B
42A
B
43A
B
44A
B
45A
B
46A
B
47A
B
48A
B
49A
B
Meter Volume
<"„?> M
2,48
2.49
2.41
2.38
2.44
2.44
2.41
2.43
2.41
2.44
2.48
2.54
2.45
2.47
2.47
2.51
2.48
2.53
Barometer,
mm Hg
741.2
741.2
740.7
740.7
738.9
738.9
740.7
740.7
741.4
741.4
740.9
740.9
743.5
743.5
743.7
743.7
742.2
742.2
An.
mm H20
45.5
46.7
41.9
41.9
42.2
43.9
42.7
43.9
42.4
43.9
44.7
46.0
44.7
45.7
43.7
45.7
44.7
46.2
Meter Temp.
Tm. C (Ave)
29
31
33
34
33
33
27
33
33
32
32
37
27
35
33
33
33
37
Dry Gas Volume ,
Std cond. 
-------
           APPENDIX D
STATISTICAL DESIGN OF EXPERIMENTS
  TO STUDY SAMPLING METHODOLOGY

-------
                                 APPENDIX D
                      Experimental Design For The Study
          The sampling location used in the field studies permitted the use
of only two ports at which sampling nozzles could be located in close proximity.
Thus, the number of samples which could be collected concurrently was limited
to two.  On the other hand, the number of combinations of experimental conditions
to be investigated was always greater than two.  Because stack conditions can
vary substantially over time, experimental combinations effects which may be
significant would be confounded with the variations in the stack conditions.
One method of avoiding this problem is to use statistical designs for the
experiment which selectively confound particular factor effects with the
variations in stack conditions, or which handle data where not all experimental
combinations are tested concurrently.  Two such designs were used for this
study — an incomplete randomized block design and a confounded factorial
design.  The purpose of this appendix is to describe these designs and to
discuss the manner in which their calculational methods differ from the
standard analysis of variance methods.

Incomplete Randomized Block Design

          In experimental design discussions, it is common practice to speak
of "treatment" combination.  The term "treatment" is intended to mean any
controlled variation in experimental conditions.  (In the present study, for
example, the use of in-stack flat filter versus the use of an in-stack thimble
filter may be thought of as two treatments).  Usually, experiments are designed
so that data on all treatment levels are collected within one statistical block
where the term "block" indicates simultaneous collection of data or some such
arrangement for keeping error variance as small as possible.
 (1)  Davies, Owen L., Editor, The Design and Analysis of Industrial Experiments.
 Hafner Publishing Company, New York (1960), Chapter 6.

-------
                                     D-2
          When there is only one experimental factor being investigated (e.g.,
type of filter), and there are more treatment levels of that factor than can
be compared at one time, and all comparisons are equally important, an incomplete
randomized block design can be used to plan the experiment.  For this study,
several experiments involved only one factor (such as filter type) with three
levels of treatment which could only be employed two at a time.  In these
experiments, every pair of treatment combinations was repeated three times.
Since there were three pairs of treatment combinations, nine statistical blocks
of data were collected.  These blocks were structured so that every pair of
treatment combinations appeared an equal number of times (thus assuring balance)
and then the order in which the blocks were run was randomized.  This produces
the incomplete randomized block design needed.  As an example, the data
collected from such a design can be represented in the form shown in Table
D-l, where a dash indicates no data for one of the treatment levels within
the block and the X's indicate data.  Row sums and column sums of this table
represent totals of observations for particular blocks (B) or treatments (T),
while G indicates the grand total of all 18 observations.
          In order to present the calculational methods for this design, the
following notation is necessaryr
          t = the number of treatment levels
          r = the number of replications of each treatment level
          b = the number of statistical blocks
          k = the number of experimental units per block
          N = the number of observations.
For this study, t=3, r=6, b=9, k=2, and N = tr = bk = 18,
          To calculate the treatment effects Davies indicates that one must
          (1) first calculate, for each treatment,
              Q. = kT^ - (Sum of totals for all blocks containing the i—
                          treatment); then
          (2) obtain P. =  N(k^/(t_^	    '
where the P^ terms are the estimated treatment effects, or more correctly the
estimated deviations of particular treatment effects from the average effect of
all treatments used in the experiment.  Thus the corrected or adjusted mean for
the i—  treatment can be calculated as the grand mean (G/N) plus P..  This
                                   t*Vi                              ^"
sum represents an estimate of the i— treatment mean one would observe if all
treatments had been used in every block.

-------
                                D-3
TABLE D-l.  BALANCED INCOMPLETE BLOCK DESIGN FOR THREE REPLICATIONS
            OF AN EXPERIMENT WITH THREE TREATMENT LEVELS OCCURRING IN
            BLOCKS OF TWO.

Block
1
2
3
4
5
6
7
8
9
Total T.
i
Intermediate
Adjustment Factor
Estimated Deviations
of Treatment Effects
From Grand Average

1
-
X
X
-
X
X
-
X
X
Tl

QI


pl
Treatment
2
X
-
X
X
-
X
X
-
X
T2

Q2


P2

3
X
X
-
X
X
-
X
X
-
T3

^3


P3

Total
Bl
B2
B3
B4
B5
B6
B7
B8
B9
G

0.0


0.0

-------
                                     D-4
          The calculational formulae for the Analysis of Variance Table
resulting from this design are presented in Table D-2 .  The interpretation of
this table follows that for the usual analysis of variance except that one
is usually not interested in testing for significance between blocks.
          Comparisons between any two particular treatment means can be made
by calculating the standard deviation of the difference between the corrected
means of any two treatments as
       N9k-l)
multiplying this standard deviation by the critical value of the Student's
t - distribution for (N-t-b+1) degrees of freedom and a chosen confidence
level, and comparing the result to the difference in the treatment means.
Any difference which exceeds this result would be declared statistically
significant at the chosen confidence level.

                           (2)
Confounded Factorial Design

          This type of design represents the most efficient way of dealing
with a situation where several experimental factors are to be investigated
simultaneously and the resulting number of combinations of treatment levels
of these factors is larger than the number of combinations which can be
carried out under uniform test conditions.  For example such a problem existed
in this study in investigating the effect of filter size on the sampling system
while at the same time testing for significant effects due to temperature and
nozzle type.  With each of these factors having two treatment levels, there were
eight treatment combinations which could only be run in blocks of two at a time.
          The confounded factorial design divides the experiment into smaller
blocks in a particular manner, such that the more important effects of the factors
can be investigated under uniform test conditions (i.e. through concurrent
measurement) while the less important effects are confounded with variations
between blocks.
          A discussion of the procedures used to develop the confounded designs
is based upon mathematical concepts of modulus arithemetic.  The interested
reader is referred to Davies.  The experimental designs which are presented
(2)  See Davies, Chapter 9.

-------
                           D-5
TABLE D-2.  ANALYSES OF VARIANCE FOR BALANCED INCOMPLETE
            BLQCK DESIGN OF EXPERIMENTS
Source
of
Variation
Between Corrected
Treatment Totals
Between Actual
Block Totals
Remainder
Total
Sum of
Squares
of Deviations
ft-n 1 -9
Nk(k-l) 1=1 Qi
b o
A •$ G*
k N
Difference
t b 2 02
Degrees
of
Freedom
(t-1)
(b-1)
(N-t-b+1)

Mean
Squares
Sum of Squares
Degrees of Freedom
(Discard)
Sum of Squares
Degrees of Freedom

.  -  £-
                                 (N-l)

-------
                                     D-6
in this report were developed using such concepts.  Because the number of
statistical blocks of data which were collected was sufficiently large, it was
possible to partially confound selected factors rather than completely confound
factors.  By this method, the confounded factors were only confounded in
selected blocks of data, and their effects could still be estimated from the
remaining blocks in which they were not confounded.  The factors and the blocks
in which they were confounded have been identified in the tables in the main
body of this report.  The calculational methods for this partially confounded
factorial design are performed in the same manner as is done for the standard
analyses of variance design with one exception.  The exception is that any
data from a block which is confounded with a particular factor are excluded
from the calculation of the sums of squares of deviations for that factor.

-------
           APPENDIX E
CONDENSATION OF SULFURIC ACID AND
 AMMONIUM SULFATE UNDER VARIOUS
CONDITIONS ENCOUNTERED IN CEMENT
         PLANT EMISSIONS

-------
                              APPENDIX  E
                           Sulfuric Acid Dewpoint

          The dewpoint of a vapor is defined as that temperature at which
vapor of a given composition is in equilibrium with a condensed liquid phase.
In the case of the condensation of sulfuric acid from flue gas, this involves
both water vapor and sulfuric acid vapor in equilibrium with sulfuric acid
containing water, viz.,

          H2S04(aq)  - H20(v) + H2S04(v).                         (1)

Both the composition of the liquid phase and the dewpoint in such a system
are functions of the overall initial composition of the vapor, and the
equilibrium constant can be expressed as

          K'= ( P   P
               H2°H2S04)/(NH20NH2S04)                           (2)

where P. and N. represent the partial pressure and mole fraction in the
liquid phase for each component.  Inasmuch as sulfuric acid compositions are
usually expressed in terms of weight fractions rather than mole fractions, it
is convenient at this point to express the equilibrium constant in terms of C,
the weight fraction of sulfuric acid in the liquid phase.
This equilibrium constant can then be related to the dewpoint of the
vapor by
          In K = A - AR/RT                                       (4)
where AH is the enthalpy change for the process and A is a constant that is
related to the entropy change.  Muller^   has examined in detail  the thermodynamics
of the sulfur acid/water system, and has derived a rigorous approach to
calculation of dewpoints based on the temperature dependencies of various components
of A and AH in Equation (4).  However, it can be shown that existing vapor
pressure and dewpoint data can be adequately predicted with such a correlation
over a wide range of sulfuric acid/water compositions.
          While a number of reports of dewpoints and total vapor pressures over
(1)  Muller, P.E., "A Contribution to the Problem of Sulfuric Acid on the Dew Point
     Temperature of Flue Gases", Chem. Ing.  Tech. 31, pgs 345-351 (1959).

-------
                                     E-2
sulfuric acid solutions are in the literature, only in a few cases is it
possible to either cite or infer the individual partial pressures of water and
                                                                                 (2 )
sulfuric acid.  The following is therefore based primarily on data given by Abel
                            (3)
and by Lisle and Sensenbaugh   .  Figure E-l shows a semilogrithmic plot values
of K calculated from these two sources as a function of the inverse of the
temperature.  Units employed in this plot are torr, weight percent, and degrees
Kelvin.  It can be seen that the simple form of Equation (4) does represent
these data over approximately an eight-decade range of the equilibrium constants.
A least-squares fit of Equation (4) to these data yields
          log A = 19.580 ± 0.18 and AH = 35,500 ± 390 cal/mole
Expressing the partial pressures in terms of volume fractions and rearranging
the form of Equation (4), the dewpoint is given by
          T,=Kx =        -35,500 _
          U K;   R[ln(v  nV  qn /C(l-C)] - 31.819                     (5)
                        H2° H2b°4
          In order to apply Equation (5) to the prediction of dewpoints it is
necessary to know the appropriate values of C as a function of vapor phase
composition.  Examination of the plots given by Muller indicates that the
following relationship can be used at partial pressures of water vapor typical
for most flue gases (ca. 0.1 atm)
          C  -  1 + 0.05 log(V/V)                               (6)
As indicated by this equation, and as noted by Lisle and Sensenbaugh, small
variations in water vapor content do not strongly influence either the
composition of the liquid phase formed or the dewpoint.  This is illustrated
further in Figure E-2, where the dewpoint is plotted as a function of sulfuric
acid content for several water vapor concentrations.
          Thus, the correlation represented by Equation (5) appears to be accurate
over a wide range of vapor compositions.  However, some care must be exercised
in the use of Equation (6) to calculate liquid phase compositions.  Equation (6}
is reasonably accurate in the range of normal flue gas compositions, but does
not apply at very low water contents; as the water content is lowered, additional
equilibria such as the dissociation of sulfuric acid to form S0_ must be considered,
          The data presented in Figure E-2 can be used to predict the sulfuric
(2)  Abel, E. "The Vapor Phase above the System Sulfuric Acid-Water", J. Phys.
     Chem., 50, pgs 260-283 (1946).
(3) Lisle, E. S. and Sensenbaugh J. D., "The Determination of Sulfur Trioxide
    and Acid Dewpoint in Flue Gases", Combustion, pgs 12-16 (January 1965)

-------
                                  E-3
     10°
            I     I    I    I     I    I     I    I    T
I    I    I
     io-
c
o
(/>
c
o
o
cr
UJ
     IO1
    10"
    ,0-2|	L
                             K  =
                                 PH20PH2S04
                              o  Abel  (Reference 2)

                              A  Lisle and Sensabaugh (Reference 3)
       1.8       2.0       2.2      2.4      2.6      2.8


                                    I03/°K


               FIGURE E-l.  TEMPERATURE DEPENDENCE OF

                            EQUILIBRIUM CONSTANT
    3.0      3.2

-------
                              E-4
    t\i
    ro
    o
no
 CO
    03
    ro
    CM
    QD
    01
                       Temperature, F
                    03
                    o
           ro
           ro
           O
             r   i
    I    I
               CD
               O
               o
               o
              OJ
              A
              O
                                         ui O  o
                                                   ro
                                                    O
         I
I
I    I
I
               l
I
I    i
I	l
         FIGURE E-2.  DEWPOINT AS A FUNCTION OF

                      SULFURIC ACID AND WATER

                      VAPOR  CONCENTRATION

-------
                                     E-5
acid dewpoint in cement kiln emissions.  For example, the dewpoint will be about
121 C (250 F) for a sulfuric acid concentration of 4 ppm and a moisture content
of 5 percent; which is typical of dry process, cement kiln emission.  The dewpoint
for the same sulfuric acid concentration in wet process emission with about
30 percent moisture will be about 138 C (280 F).

                         Ammonium Sulfate Formation

          In flue gas systems containing ammonia and sulfur oxides, there is
a possibility for precipitation of sulfites and/or sulfates during the sampling
process.  Possible compounds include (NH4)2S03, (NH^SCyHjO, (NH4)9S04>
and NH,HSOX.  Within this group of compounds, the sulfates are much more
stable than the sulfites, and the sulfites are not expected to form under
normal sampling conditions unless the partial pressure of ammonia is very high.
For example, (NH.KSO. can be expected to form at 66 C (150 F) in flue glases
containing 10  ppm S02 and 10  ppm NH_.  However, ammonium sulfates can form
at considerably lower concentrations and higher temperatures.  Figures E-3 and
E-4 are nomographs showing the temperatures of precipitation of ammonium
sulfate and ammonium bisulfate.  Construction of these nomographs was based on.
                                                       (4)
thermodynamic data for the sulfates as given by Latimer   .  As an example of
the use of these plots, it can be seen that both the sulfate and the bisulfate
are probable reaction products at 188-191 C  (370-375 F) for flue gases contain-
ing 10 ppm NH_ and 4 ppm H_SO,.
 (4)  Latimer, W. M., "The Oxidation States of the Elements and Their Potentials
     in Aqueous Solutions", 2nd Edition, Prentice-Hall, Inc., Englewood Cliffs,
     New Jersey (1952).

-------
300

*
200
100
80
60


40
CO
I 30
Z
"o
2 20
CL
Of.

*
10
8
6
4

3

2

i
» 1 « • * £L i O ^3 V^^ t *• \ ^^^_^—_ Cv 1 NJ 1 I TI i /+ \ r ( I ^^^\^A t n\
• &• « i w / ^ j * y / « ^^T v y ^
Temperature,
C

I 232-
-
_
218-

—
204-

"
190-

-
I I77~
-
_.
163-
—

-
149-

F

-450

^—
-425
•
-
-400


-375


-350


-325



-300





100

40

20
15
^
"0 °
8 °
6 M_
O
4 ^
3 DL
J CL
2

1



.,
S
6
4
3
2
in'2
FIGURE E-3.
NOMOGRAPH SHOWING TEMPERATURE OF
AMMONIUM SULFATE FORMATION FOR
VARIOUS AMMONIA AND SULFUEIC ACID
CONCENTRATIONS

-------
                                               E-7
CL
a.
300

200
100
80
60
40
30
20

10
8
6
4
3
2
n
NH4HS045^H3(g).H2S04,,, -
Temperature,
C

-
204-
199-
193-
'188-
182-
~ 177-
•*
-
F


-400
-390
-380
-370
-360
-350


-
-
-

30

20
10
8
6
4
3
2

1






                                                                                              C?
Q_
Q.
                           FIGURE E-4.  NOMOGRAPH SHOTING TEMPERATURE OF
                                        AMMONIUM BISULFATE FORMATION FOR
                                        VARIOUS AMMONIA AND SULFURIC ACID

-------
                                   TECHNICAL REPORT DATA
                            (Please read Instructions on the reverse before completing)
1. REPORT NO.
 EPA-650/2-75-051a
                                                           3. RECIPIENT'S ACCESSION NO.
4. TITLE AND SUBTITLE

 Evaluation of Stationary  Source Particulate Measurement
 Methods.   Volume 1, Portland Cement Plants
                               5. REPORT DATE
                                  June 1975
                               6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)

 J.E.  Howes, Jr., R.N.  Pesut,  and W.M. Henry
                                                           8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS

  Battelle
  Columbus Laboratories
  505 King Avenue
  Columbus, Ohio	43201	   	
                               10. PROGRAM ELEMENT NO.

                                  1 AA010
                               11. CONTRACT/GRANT NO.
                                  68-02-0609
 12. SPONSORING AGENCY NAME AND ADDRESS

  Environmental Protection  Agency
  Chemistry and Physics  Laboratory
  NERC,  Research Triangle Park,  N.C.
                               13. TYPE OF REPORT AND PERIOD COVERED
                                  Interim
                               14. SPONSORING AGENCY CODE
           27711
15. SUPPLEMENTARY NOTES
16. ABSTRACT
  A study was performed to evaluate the EPA Method 5 procedure for measurement  of
  particulate emissions from  a  cement kiln.  The program  included three series  of
  experiments to study the effects  of anisokinetic sampling,  sampling system  temper-
  ature,  filter size; and nozzle  configuration on particulate mass emission deter-
  minations.   Method 5 was compared with two types of  in-stack filters in a fourth
  set  of  experiments.  The results  of the experiments  were  analyzed by statistical
  analysis to assess the significance of the sampling  system  variables on observed
  difference  in collected particulate mass.  Chemical  and physical characterizations
  were performed to evaluate  the  representativeness of collected particulates and
  to elucidate the cause of mass  differences introduced by  various sampling system
  operating parameters.
17.
                                KEY WORDS AND DOCUMENT ANALYSIS
                  DESCRIPTORS
                                              b.lDENTIFIERS/OPEN ENDED TERMS
                                             c. COSATI Field/Group
       EPA Method  5
       Source Sampling
       Particulate sampling
       Emission measurements
       Portland cement  plants
       Emission characteristics
       Air Pollution
Gas Sampling
Air pollution  control
Stationary  sources
EPA Sampling methods
13. DISTRIBUTION STATEMENT
  Release  Unlimited
                                              19. SECURITY CLASS (TMs Report)
                                                                         21. NO. OF PAGES
                                              20. SECURITY CLASS (Thispage)
                                                                         22. PRICE
EPA Form 2220-1 (9-73)

-------
                                                         INSTRUCTIONS

    1.   REPORT NUMBER
        Insert the EPA report number as it appears on the cover of the publication.

    2.   LEAVE BLANK

    3.   RECIPIENTS ACCESSION NUMBER
        Reserved for use by each report recipient.

    4.   TITLE AND SUBTITLE
        Title should indicate clearly and briefly the subject coverage of the report, and be displayed prominently. Set subtitle, if used, in smaller
        type or otherwise subordinate it to main title. When a report is prepared in more than one volume, repeat the primary title, add volume
        number and include subtitle for the specific title.

    5.   REPORT DATE
        Each report shall carry a date indicating at least month and year. Indicate the basis on which it was selected (e.g., date of issue, date of
        approval, date of preparation, etc.).

    6.   PERFORMING ORGANIZATION CODE
        Leave blank.

    7.   AUTHOR(S)
        Give name(j) in conventional order (John R.  Doe, J. Robert Doe, etc.).  List author's affiliation if it differs from the performing organi-
        zation.

    8.   PERFORMING ORGANIZATION REPORT NUMBER
        Insert if performing organization wishes to assign this number.

    9.   PERFORMING ORGANIZATION NAME AND ADDRESS
        Give name, street, city, state, and ZIP code.  List no more than two levels of an organizational hirearchy.

    10.  PROGRAM ELEMENT NUMBER
        Use the program element number under which the report was prepared. Subordinate numbers may be included in parentheses.

    11.  CONTRACT/GRANT NUMBER
        Insert contract or grant number under which report was prepared.

    12.  SPONSORING AGENCY NAME AND ADDRESS
        Include ZIP code.

    13.  TYPE OF REPORT AND PERIOD COVERED
        Indicate interim final, etc., and if applicable, dates covered.

    14.  SPONSORING AGENCY CODE
        Leave blank.

    15.  SUPPLEMENTARY NOTES
        Enter information not included elsewhere but useful, such as:  Prepared in cooperation with, Translation of. Presented at conference of,
        To be published in, Supersedes, Supplements, etc.

    16.  ABSTRACT
        Include a brief (200 words or less) factual summary of the most significant information contained in the report. If the report contains a
        significant bibliography or literature survey, mention it here.

    17.  KEY WORDS AND DOCUMENT ANALYSIS
        (a) DESCRIPTORS - Select from the Thesaurus of Engineering and Scientific Terms the proper authorized terms that identify the major
        concept of the research and are  sufficiently specific and precise to be used as index entries for cataloging.

        (b) IDENTIFIERS AND OPEN-ENDED TERMS - Use identifiers for project names, code names, equipment designators, etc. Use open-
        ended terms written in descriptor form for those subjects for which no descriptor exists.

        (c) COSATI FIELD GROUP - Field and group assignments are to be taken from the 1965 COSATI Subject Category List. Since the ma-
        jority of documents are multidisciplinary in nature, the Primary Field/Group  assignment(s) will be specific discipline, area of human
        endeavor, or type of physical object. The application(s) will be cross-referenced with secondary Field/Group assignments that will follow
        the primary posting(s).

    18.  DISTRIBUTION STATEMENT
        Denote releasability to the public or limitation for reasons other than security for example "Release Unlimited." Cite any availability to
        the public, with address and price.

    19. & 20.  SECURITY CLASSIFICATION
        DO NOT submit classified reports to the National Technical Information service.

    21.  NUMBER OF PAGES
        Insert the total number of pages, including this one and unnumbered pages, but exclude distribution list, if any.

    22.  PRICE
        Insert the price set by the National Technical Information Service or the Government Printing Office, if known.
EPA Form 2220-1 (9-73) (Reverse)

-------