NTROPY
NVIRONMEIMTAL.IST8, INC.
SPECIALISTS IN AIR POLLUTION MEASUREMENT & MANAGEMENT
CONDENSIBLE PARTICULATE
AND ITS IMPACT ON
PARTICULATE MEASUREMENTS
Guy B. Oldaker, Ph.D.
MAY 1980
P.O. Box 12291. Resseanch "Triangle Park, North Carolina 27~7O9
Dhone SIS-TEH-355O
-------
DRAFT
CONDENSIBLE PARTICULATE
AND ITS IMPACT ON
PARTICULATE MEASUREMENTS
May 1980
Prepared by:
Guy B. Oldaker, Ph.D.
Entropy Environmentalists, Inc,
Research Triangle Park
. North Carolina
Prepared for:
Division of Stationary Source Enforcement
United States Environmental Protection Agency
Project Officer: Kirk Foster
Contract Number: 68-01-4148
Task Number: 69
INTROPY
^mmtm
IIMVIRONMENTAUSTB, INC.
-------
This document has not been reviewed by the U. S. EPA, and is
not intended to reflect official policy or standards. The
opinions, suggestions, and conclusions expressed herein are
those of the author, and do not necessarily represent those
of the United States Environmental Protection Agency.
-------
TABLE OF CONTENTS
Page
Introduction 1
Historical
The Effects of Condensible Particulate
on the Development of NSPS Particulate
Sampling Methodology 5
Physicochemical Effects on the Condensible
Particulate Loading 25
Physical Changes 26
Chemical Reactions 46
Summary 64
Assumptions and Limitations of Reference
Method 5 Sampling 66
Recommendations 73
IIMTROPY
••^H
IIMVIRONMEIMTAUBTB. INC.
-------
INTRODUCTION
For sources subject to New Source Performance Standards
(NSPS), compliance with particulate standards is most often
determined through the use of EPA Reference Method 5. Put
simply, this method entails sampling a metered volume of
effluent with a heated probe and then collecting the
particulate matter on a heated filter. The particulate
catch is currently interpreted by the EPA as being the
material caught in the probe and on the filter.
The use of Reference Method 5 produces acceptable
particulate emissions data for most NSPS source categories.
However, when the method is extended to non-NSPS or novel
source categories, the particulate emissions data sometimes
suffer from imprecision and positive biases. These problems
with the Reference Method 5 data arise from the fact that
some particulate-forming reactions occur in the effluent
stream: before, during, and even after sampling.
The positive bias, or extra particulate, which is
measured is termed "condensible particulate," or sometimes,
"pseudoparticulate." As the term "condensible particulate"
implies, the extra particulate originates from condensation
processes. Most condensible particulates are formed by
gases condensing in the effluent, and the particulate
IISTTROPY
•••••
INVmOIMMENTAUSTajNC.
-------
control equipment has no effect upon gaseous substances.
Since the focus of particulate testing for many compliance
determinations is on the performance of the particulats
control equipment, if gaseous substances form condensible
particulate in the effluent downstream of the control
equipment but upstream of the sampling probe, the resultant
particulate measurements will provide a poor indication of
control equipment performance. Indeed, such measurements
would be biased against the source.
The imprecision observed in Reference Method 5 data
results from the fact that the formation of condensible
particulate is usually strongly dependent on temperature.
Thus, the appearance of condensible particulate depends on
the temperatures of the effluent and of the probe and filter
of the Reference Method 5 sampling system. Depending on the
chemical properties of the condensible particulate, it is
sometimes possible for a difference in temperature of a few
degrees to determine whether or not condensible particulate
is formed.
The subject of condensible particulate is important,
because it bears directly on determinations of source
compliance. In fact, condensible particulate, because it
contributes a positive bias to Reference Method 5
measurements, can be the determinant of compliance or
non-compliance. In addition, the issue of condensible
INTROPY
•••••
IIMVIRONMEIMTAUBTS, INC.
-------
participate raises questions regarding: (1) the use of
Reference Method 5 at sources which are not subject to NSPS;
(2) the precision required for controlling the temperature
of filtration during Reference Method 5 sampling; (3) the
interactions which occur between particulate and condensible
particulate (is it always possible to distinguish the two?);
(4) the relation between opacity measurements obtained by
transmissometry and by visual methods (Reference Method 9);
and (5) the interpretation of "particulate" itself.
This paper addresses the subject of condensible
particulate from several directions. The first section of
the paper views condensible particulate from the historical
perspective. The impact of the condensible particulate
problem was recognized when the New Source Performance
Standards and their associated testing methods first
appeared in the Federal Register in 1971. Since that time,
numerous references to the condensibles problem have been
made in the Federal Register and also in EPA documents for
NSPS review and development.
The section which follows the historical discusses
those physicochemical factors which determine the identity
and loading of condensible particulate. Throughout the
section, general principles are stressed, which can be
applied to interpreting or predicting the effects of
condensibles on the Reference Method particulate
measurements. In a sense, this section can stand alone, and
IIMTROPY
•••••§
INVIRONMEIMTAUBTa.llMC.
-------
is intended to serve as a brief guide for understanding the
formation of condensible particulate.
The following section, "Assumptions and Limitations of
Reference Method 5 Sampling," views the impact of
condensible particulate on the interpretation of current
Reference Method results. This section leads to the
"Recommendations" section, which addresses solutions to the
problems created by condensible particulate.
The use and interpretation of particulate data obtained
from including the back-half catch (i.e., particulate
measured in the water filled impingers which follow the
heated filter in the Reference Method 5 train) have been
controversial issues, because the impact of condensible
particulate appears to be greater in the back-half. Limited
data exist which correlate back-half particulate data to
front-half (probe and filter) data, mostly because back-half
data are not required for NSPS sources. In addition, few
state and local agencies require the inclusion of back-half
particulate results. Because the regulatory status of the
back-half particulate catch is variable, and also because of
the limited amount of quality data from the back-half, this
paper will not specifically address the effects of
condensible particulate on back-half particulate
measurements.
NTROPV
•••••
IMVIRONMEISrrAU8TS,INC.
-------
HISTORICAL
The Effects of Condensible Particulate
on the Development of
NSPS Particulate Sampling Methodology
The subject of condensible particulate and its effect
on particulate emissions measurements is best approached by
reviewing the literature pertaining to New Source
Performance Standards (NSPS). The following discussion is a
chronological review of the pertinent material from the
Federal Register. The reader should note that the focus of
the review is on NSPS sources rather than on existing
sources subject to State Implementation Plans (SIPs). This
distinction must be made, since the statements which appear
in the Federal Register apply to NSPS sources and are not
always extendable to SIP sources, which represent an
extremely varied population in terms of process operation,
control systems, and emissions.
Several reoccurring themes, with respect to condensible
particulates, appear in a review, of the NSPS developments.
Among these are: the importance of defining the particulate
state, the potential for the formation of particulate by
physicochemical mechanisms, and the use of measurement
methodology for evaluating control system performance.
Sources subject to New Source Performance Standards
(NSPS) are required to measure emissions of particulate
matter in order to determine compliance with emission
IIMTROPY
•••••
INVIRONMENTAUBTa.lNC.
-------
standards. The methodology used in measuring particulate
emissions is stated within 40 CFR 60, Appendix B. Currently
applicable methods are Reference Methods 5 and 17.
Illustrations of the sampling equipment used in the
reference methods are presented below.
Reference Method 5 requires that a sample be withdrawn
from the effluent stream via a heated sampling probe, and
that the sample stream be subsequently filtered at an
elevated temperature. The filter is located outside of the
stack, and its temperature is controlled by locating it
within a thermostated filter oven.
2
As Reference Method 5 was originally proposed in 1971,
the impingers following the filter were considered part of
the measurement system. Material which passed through the
filter was collected in the impingers. The particulate
catch, using the originally proposed methodology, consisted
of the sum of the filter and probe catches (front-half) and
the impingers1 catch (back-half) . At this time, "parti-
culate" was defined in terms of the state of the material
which was collected, rather than in terms of the measurement
methodology used. Thus, particulate was defined as "any
material except uncombined water, which exists in a finely
divided form as a liquid or solid at standard conditions."
Including the back-half catch and employing standard
conditions in the definition of particulate provide a
iisrraopv
m^mmt
|NVIRaNMEIMTALJBTa,IIMC.
-------
CZ
ib
n
TEMPERATURE SENSOR
- PROBE
TEMPERATURE
SENSOR
IMPiNGER TRAIN OPTIONAL.MAY BE REPLACED
BY AN EQUIVALENT CONDENSER
HEATED AREA
THERMOMETER
PITOTTUBE
PROBE
REVERSE-TYPE
PITOTTUBE
PITOT MANOMETER IMPINGERS ICE BATH
) BY-PASS VALVE
VACUUM
GAUGE
THERMOMETERS
DRY GAS METER
AIRTIGHT
PUMP
CHECK
VALVE
VACUUM
LINE
../
-------
m
\
3
TEMPERATURE
SENSOR
SAMPLING
NOZZLE
IN-STACK
FILTER
HOLDER
REVERSE-TYPE
PITOTTUBE
IMPINGER TRAIN OPTIONAL. MAY BE REPLACED
BY AM EQUIVALENT CONDENSER
THER.V.OV.ETER
CHECK
VALVE
AIR-TIGHT
PUMP
ORIFICE MANOMETER
VACUUV.
LINE
DRY GAS METER
oo
-------
reference point for particulate measurements, and make it
possible to relate particulate measurements to the state of
the particulate near ambient conditions. Thus, in
principle, it would be possible to relate the measured
particulate to the nature and quantity of particulate which
would exist after dilution and cooling of the effluent
stream to ambient temperature and pressure.
The inclusion of the impingers' or back-half catch into
the determination of particulate measurements was a
4
controversial point. Some critics maintained that the
back-half catch was biased because of chemical reactions
r C T Q Q
which could occur within the impingers. ' ' ' ' Their
arguments focused on the formation of "pseudoparticulate,"
2-
specifically, particulate sulfate (SO. ) formed by the
oxidation of dissolved gaseous sulfur dioxide (S02(aq)).
[Q2]
S02(aq)—> S042~(aq)
For this case, when the contents of the impingers are
evaporated, the sulfate will remain as a weighable residue.
Put simply, their argument was that a substance which
was normally a gas at standard conditions was being
converted by the test method to a substance which would
ultimately be measured as a solid. Thus, this particulate
was not included within the definition of 1971; it was not
true particulate, it was "pseudoparticulate.n It was also
INVIRONMENTAUBTa,INC.
-------
10
argued that other reactions could occur within the proposed
sampling train (including the impingers) which have no
counterparts in the effluent stream. Measuring the products
of these reactions gives results that are non-representative
of the source's actual emissions; they are, instead,
artifacts created by the test method.
The "pseudoparticulate" argument led ultimately to the
exclusion of the back-half catch from Reference Method 5
measurements. Introductory statements accompanied the
10
promulgated method , which addressed the omission of the
back-half catch:
Particulate matter performance testing
procedures have been revised to
eliminate the requirement for impingers
in the sampling train. Compliance will
be based only on the material collected
in the dry filter and the probe
preceding the filter. Emission limits
have been adjusted as appropriate to
reflect the change in the test methods.
The adjusted standards require the same
degree of particulate control as the
originally proposed standards.
Thus, the change in the test methodology and the
revision of the emission standards focused on the
evaluation/regulation of control equipment performance. EPA
concluded its introductory remarks by citing Section III of
the Clean Air Act, which requires that the standards of
performance "reflect the degree of emission reduction which
(taking into account the cost of achieving such reduction)
the Administrator determines has been adequately
IIMTROPY
•••••I
INVIPONMENTAU8TB,INC.
-------
11
demonstrated."
In 1972 the EPA published supplementary statements
which pertained to the final promulgation of the sampling
methodology, and which detailed the background for the
omission of the back-half particulate catch. The EPA
responded to suggestions that particulate standards should
"be based either on the 'front-half (probe and filter) of
the EPA sampling train or on the American Society of
Mechanical Engineers' test procedure. Both of these methods
measure only those materials that are solids or liquids at
250 F and greater temperatures." (One of the differences
between the two methods lies in the location of the filter:
the EPA method employs an out-of-stack filter; the ASME
method uses an in-stack filter.) The EPA opined that,
"particulate standards based either on the front-half or the
full EPA sampling train will require the same degree of
control if appropriate limits are applied." They stated
further that their "analyses show that the material
collected in the impingers of the sampling train is usually,
although not in every case, a consistent fraction of the
total particulate loading." This statement was the apparent
basis for the subsequent omission of the back-half catch
from the sampling procedure and the concommitant reduction
of the some of the particulate emission standards (see
Table I). Thus, the back-half catch was apparently assumed
to contribute a consistent fraction of the total particulate
IMTROPY
•••••
IWIRONMEIMTAU8T8,INC.
-------
TABLE I
Proposed and Promulgated
Particulate Standards
for NSPS Sources
12
Originally proposed
particulate standards
(full EPA train)
Recommended
particulate standards
revised sample method
(front half only)
Steam generators -
lbs/106 Btu heat input
0.20
0.10
Incinerators -
gr/scf at 12% C02
0.10
0.08
Cement Kilns -
Ibs/ton feed
0.30
0.30
Cement Coolers -
Ibs/ton feed
0.10
0.10
IMTROPY
•••••
I IMVIROIMMENTAUSTS,IIMC.
-------
13
catch, and the standards were scaled down accordingly. An
obvious impetus for omitting the back-half was the resultant
simplification of sampling and analysis procedures.
Table I shows that two of the four source categories
were affected by the omission of the back-half catch from
the sampling methodology. The values of the affected
standards suggest that the back-half catch may contribute up
to 50% of the mass of particulate measured at fossil fuel
fired steam generators, and up to 20% of the measured
particulate mass at incinerators. The standards for cement
kilns and cement coolers indicate that the material
condensed in the impingers does not make a significant
contribution to the total particulate catch.
It is noteworthy that the following statement was also
included within the EPA opinion: "There has been only
limited sampling with the full EPA train such that the
occasional anomalies cannot be explained fully at this
time."
The change in the particulate measurement methodology
was reflected in the revised definition of "particulate" in
40 CFR 60.2:
"Particulate matter" means any finely
divided solid or liquid material, other
than uncombined water, as measured by
method 5 of the appendix.
IIMTROPY
^•••B
IIMVIRONMEIMTAUSTB,IIMC.
-------
14
This definition does not include an explicit statement
regarding the important physical parameters of the
particulate, but instead defines particulate in terms of the
measurement methodology. The essence of this change is that
the methodology is no longer directed toward measuring a
sample which is directly representative of source emissions
on a defined absolute basis. Instead, the focus is on
evaluating source operation and control system performance.
Again, this change is in keeping with the provisions of
Section III of the Clean Air Act. Thus, the fate of the
emitted pollutants is not the primary issue. Of greater
importance is the degree to which the affected facilities
control their pollutant emissions, relative to the magnitude
of emissions which would result if no control was present.
12
Critics , nevertheless, maintained that the revised
Method 5 sampling train was still subject to biases caused
by the condensation of gases within the probe and on the
filter. It was argued that condensible particulate was not
restricted to the back-half. Emphasis was placed on
condensation processes involving sulfur oxides: sulfur
dioxide and sulfur trioxide. Data were presented which were
interpreted as indicating the extent of these condensation
effects. (See Table II.) The data were obtained from
simultaneous sampling of the effluent from a fossil-fuel
fired steam generator with a Reference Method 5 sampling
train and a train incorporating an in-stack filter. In
IIMTROPY
I^MMMM
INVIRONMENTAUBTe.llMC.
-------
15
TABLE II
Weights of "Particulate" Matter
Measured During Simultaneous Sampling -
EPA vs. In-Stack Sampling Apparatus
EPA train Alundum thimble
"front half," filter inside stack,
Sample No. mg mg
1
2
3
4
5
6
7
8
9
10
1139
1168
1097
1159
944
1111
171
200
252
196
233
460
483
750
325
785
140
107
245
36
IIMTROPY
•••••
INVIROIMMEIMTAUSTB, INC.
-------
16
almost all cases, the Reference Method 5 train yielded a
greater amount of measured particulate.
In October 1975 the EPA published revisions to the
performance test procedures for fossil-fuel fired steam
generators. " These revisions addressed the biases caused by
condensation reactions involving sulfur oxides. With the
filter temperature maintained at 120 C, gaseous sulfur
oxides were condensing within the probe and on the filter.
The EPA stated that, "the inclusion of this condensible
matter would not be indicative of the control system
performance." In addition, studies were cited which
o
indicated that sampling at 120 C produced variable biases.
The magnitudes of the biases caused by the increased
particulate loading from the condensing sulfur oxides were
recognized to be unpredictable because of their apparent
dependence on "total sulfur oxide concentration, boiler
design and operation, and fuel additives." The EPA stated
that the particulate mass contributed by the condensed
sulfur oxides was not a serious problem, since studies had
shown the contribution to range from 0.001 to 0.008 grains
per standard cubic foot, which is relatively insignificant
when compared to the then current standard of 0.07 grains
per standard cubic foot. Nevertheless, a higher sampling
o
temperature of 160 C was accepted for testing at fossil-fuel
fired steam generators, "to insure that an unusual case will
not occur where a high concentration of condensible matter,
IIMTROPY
INVIRONMEIMTAUSTB,INC.
-------
17
not controllable with an ESP felectrostatic precipitator]
would prevent attainment of the particulate standard."
The EPA also discussed the temperature dependence of
particulate matter within the comments accompanying the
revisions. The use of sampling methodology incorporating an
out of stack filter — as opposed to an in-stack filter —
was supported because of the necessity to control and
measure the temperature during the determination of
particulate. It was stated that "[temperature control] is
needed to define particulate matter on a common basis, since
it is a function of temperature and is not an absolute
quantity." Continuing along these lines, the EPA said:
If temperature is not controlled,
and/or if the effect of temperature upon
particulate formation is unknown, the
effect on an emission control limitation
for particulate matter may be variable
and unpredictable.
This statement contains the crux of the difficulties
which stem from condensible particulate: specifically, that
the effects of temperature upon particulate formation must
be known before one can understand the relationship between
the sampling methodology and what is actually measured.
Also in their comments of October 1975, the EPA cited
the results of tests which indicated that S0_ does not react
to a significant degree to yield condensed particulate
within the front half of the Method 5 train. The EPA
IMTROPY
mumm
NVmaNMEIMTAUBTajNC.
-------
18
emphasized again that Reference Method 5 was intended for
evaluating the control system performance of stationary
sources. They concluded with a statement pointing out that
the application of sampling methodology may require
flexibility, depending on what is to be measured. As an
example, they cited the condensation of sulfur trioxide and
water to form sulfuric acid mist. If control performance is
to be evaluated, particulate would be best measured at a
o
temperature of 160 C. This temperature would prevent the
condensation of the SO-,, the product of which is not
controlled. If the "applicable standards are based upon
emission reduction to achieve ambient air quality standards
rather than on control technology, a lower sampling
temperature would be appropriate." Put simply, what is to
be measured determines how the measurement will be made.
Implicit within this argument is the fact that the
applicability of Reference Method 5 is not necessarily
general.
A recent addition to EPA particulate measurements
methodology is "Reference Method 17: Determination of
Particulate Emissions from Stationary Sources (In-Stack
Filtration Method)."14'15 The difference between Method 5
and Method 17 involves the location of the filter.
Reference Method 17 employs an in-stack filter and is thus
similar to the ASME method. It was this method which was
used in the arguments discussed earlier, which showed that
IIMTROPY
•••••
IIMVIRONMEIMTAUSTSJIMC.
-------
19
the Reference Method 5 train, even without the inclusion of
the back-half, was subject to condensation processes. The
introduction to the method bears on the problems caused by
condensible material:
Particulate material is not an ab-
solute quantity; rather it is a function
of temperature and pressure. Therefore,
to prevent variability in particulate
matter, emission regulations, and/or as-
sociated test methods, the temperature
and pressure at which particulate matter
is to be measured must be carefully de-
fined. Of the two variables (i.e., tem-
perature and pressure), temperature has
the greater effect upon the amount of
particulate matter in an effluent gas
stream; in most stationary source cate-
gories, the effect of pressure appears
to be negligible.
This paragraph is then followed by the criterion which
underlies the applicability of the method:
Therefore, where particulate matter
concentrations (over the normal range of
temperature associated with a specified
source category) are known to be inde-
pendent of temperature, it is desirable
to eliminate the glass probe and heating
systems, and sample at stack tempera-
ture.
Where particulate matter concentrations are independent
of temperature, Reference Method 5 and Reference Method 17
should give identical results. Reference Method 17 would be
preferred in those situations, because the sampling
procedure is easier and requires less equipment.
Again the issue of the temperature dependence of
particulate matter concentration is stated, and again the
INTROPY
•••••
INVIRONMEIMTAUBTS, INC.
-------
20
focus is on the filtration temperature. The application of
this methodology is important, because it shows that the
problem of condensible material is not general, i.e., then-
are sources and situations where all filtration methods will
give equivalent results, and there are sources and
situations where different filtration methods will
significantly affect the results.
More recently, the issue of condensible particulate was
raised during the review of performance standards (NSPS) for
14 15
petroleum refineries. ' Performance tests of fluidized
catalytic cracking (FCC) units appeared to be biased by the
condensation of sulfuric acid mist in the Reference Method 5
probe and filter. Table III shows the results of a
performance test cited within the background information.
It is significant that approximately 50% of the catch can be
ascribed to sulfur oxides, e.g., sulfuric acid (H-SO.) and
2-
sulfate (SO. ). (These compounds are also primarily
responsible for the condensibles1 interference in tests at
fossil-fuel fired steam generators.) Currently, neither a
higher Reference Method 5 filter temperature nor an in-stack
filtration method (Reference Method 17) are applicable to
fluidized catalytic cracking units. Thus, the bias
associated with condensible sulfates is included in
particulate sampling results at petroleum refineries.
IIMTROPY
•••••
INVIRONMEIMTALJaTSJIMC.
-------
21
TABLE III
Results for Evaluation of
Condensible Particulate Loading
at FCC Unit Regenerators
Test
Result
ASME instack filter
NaOH titration of
Method 5 catch for
H2S04
89% less particulate matter
than Method 5
50% H2S04
Thermal analysis of
Method 5 catch
60% weight loss
Sulfate analysis of
Method 5 catch
64% sulfate
X-ray spectrograph of
Method 5 catch
27%
in probe wash
IMTROPY
•••M
IMVIROIMMEIMTAU8TS,INC.
-------
22
In summary, when the literature pertaining to NSPS
source sampling is reviewed with regard to the subject of
interferences from condensible particulate matter, several
ideas and concepts are repeatedly emphasized. These
include: (1) the strong dependence of the particulate catch
on the filtration temperature, (2) the variability of the
particulate catch as a result of the condensation of sulfur
oxides, (3) the variability of the particulate catch due to
particulate forming reactions, and (4) the Reference Method
5 particulate catch as an indicator of control system
performance.
iisrmoPY
••••i
IIMVIRONMEIsrTAUBTSJIMC.
-------
23
FOOTNOTES
Standards £f Performance for New Stationary Sources.
A Compilation. November 1977. EPA-340/1-77-015.
2
Federal Register, Vol. 36, p. 15704, August 17,
1971.
3 Ibid.
4
"Response," J. Air Pollut. Control Assoc., 22, 726
(1972).
W. S. Smith and R. A. Estes, "Condensibles, Reactive
Compounds, and Effect of Sampling Train Configuration,"
Source Sampling Reference Manual, Part III. Supplemental
Training Material for Technical Workshop on Evaluating
Performance Tests. U. S. EPA, November 1977.
D. R. Kendall, "Recommendations on a Preferred
Procedure for the Determination of Particulate in Gaseous
Emissions," J^ Air Pollut. Control Assoc., 26, 871 (1976) .
W. C. L. Hemeon and A. W. Black, "Stack Dust
Sampling: In-Stack Filter or EPA Train," J_._ Air Pollut.
Control Assoc., 22, 516 (1972).
o
L. J. Hillenbrand, R. B. Engdahl, and R. E. Barrett,
"Chemical Composition of Particulate Air Pollutants from
Fossil-Fuel Combustion Sources," U. S. EPA Report, March 1,
1973.
q
J. Kowalczyk, et al., "Source Test Procedure for
Determination of Particulate Emissions from Veneer Driers,"
Publication of the Control Agency Directors - 8 Source Test
Committee, Pacific Northwest International Section, Air
Pollution Control Association, September 1972.
10
Federal Register, Vol. 36, No. 247 - Thursday,
December 23, 1971.
Federal Register, Vol. 37, No. 55 - Tuesday, March
21, 1972.
12 Kendall.
Federal Register, Vol. 40, No. 194 - Monday, October
6, 1975.
IIMTROPY
•••^
INVIROIMMEMTAUSTS,INC.
-------
24
14 Federal Register, Vol. 41, No. 187 - Friday,
September 24, 1976.
15 Federal Register, Vol. 43, No. 37 - Thursday,
February 23, 1978.
"Determining Dust Concentration in a Gas Stream,"
Performance Test Code 27-1957. American Society of
Mechanical Engineers, New York, New York.
Federal Register, Vol. 44, No. 205 - Monday, October
22, 1979.
18
K. Barrett and A. Goldfarb, "A Review of Standards
of Performance for New Stationary Sources - Petroleum
Refineries," March 1979, EPA-450/3-79-008.
IIMTROPY
•MIM
INVIRONMEIvrAU8TS,INC.
-------
25
PHYSICOCHEMICAL EFFECTS
ON THE
CONDENSIBLE PARTICULATE LOADING
The loading of condensible particulate within an
effluent stream depends primarily on the chemical identity
of the condensible material. The chemical identity of the
condensible material will in turn determine the importance
of chemical and physical changes in affecting the observed
loading of condensible particulate. When viewed together,
chemical and physical changes may be termed physicochemical
changes.
The discussions which follow briefly describe the
operation of physicochemical changes and attempt to show how
the observed loading of condensible particulate can be
subsequently rationalized. These discussions are quite
general, because the operation of physicochemical changes is
not trivial. Many of the arguments have been simplified;
nevertheless, the ideas presented in the discussions can be
applied to understanding and interpreting measurements which
suggest interferences from condensible particulate.
IIMTROPY
IIMVIRaNMEIMTAUST8,INC.
-------
26
PHYSICAL CHANGES:
Definitions and Examples
Physical changes are familiar as "changes of state."
For example, water changing from liquid to gas (evapora-
tion) , and dry ice changing from solid to gas (sublimation),
are examples of physical changes. Those physical changes
which can potentially affect the amount of condensible
material measured are condensation, evaporation, sublima-
tion, absorption, adsorption, and desorption.
The term condensation describes the process during
which material in a gaseous state (or phase) changes to
either the liquid or the solid phase. Familiar examples of
these condensation processes are the formation of rain and
snow from water vapor. Condensation processes increase the
particulate mass loading, and as a result, may contribute a
positive bias to such measurement.
Evaporation and sublimation are physical processes
which are essentially the reverse of the condensation
processes described above. Evaporation occurs when a
substance in the liquid phase changes to a gaseous phase,
and sublimation occurs when a substance in the solid phase
passes directly to the gaseous phase. The biases of these
processes operate in reverse of those associated with
condensation; evaporation and sublimation reduce the
measured particulate mass loading.
INTFtOPY
•••M
IIWIRONMEIMTAUBTS, INC.
-------
27
Adsorption and absorption are special condensation
processes in which a substance in the gaseous phase
associates itself with the surface of a solid material
(adsorption), or with the bulk (i.e., both the surface and
the interior) of a liquid or solid material (absorption) .
The two processes are easily confused, and are often
difficult to distinguish; therefore, the two processes are
often jointly termed "sorption." Because sorption is a
condensation process, it can contribute a positive bias.
However, in general, the effects of sorption are of lesser
magnitude than the other condensation processes.
Gaseous sulfur dioxide dissolving in liquid water is an
example of absorption. Sulfur dioxide can adsorb on
materials- used for sampling gases (e.g., Tygon tubing and
surgical tubing). In effect, the material becomes coated
with a molecular film of sulfur dioxide. (Adsorbed
substances can exist in layers much thicker than
monomolecular films.)
The exact reverse of the sorption processes described
above is termed desorption; material adsorbed on a solid
surface or absorbed in liquids returns to the gas phase.
The biases contributed by desorption are analogous to those
of evaporation and sublimation.
IMTOOPY
•••••
IIWIRONMEIMTAUBTS,INC.
-------
28
Physical Parameters
At the point of generation, the physical state of the
effluent is different than it is at the point of exhaust.
If all of the effluent components are in the gas phase, the
state of the effluent is described by the following
parameters: (1) the chemical composition, as determined by
the identities and associated concentrations of all
components; (2) the pressure; and (3) the temperature. If
non-gaseous components (solids or liquids) are also present,
then the ratio(s) of the solid/liquid to gaseous phases must
be considered for each component.
Of the parameters above, the temperature is the most
important with respect to potential changes of state. Many
of the equations which are used in describing physical
changes have a logarithmic temperature dependence. This is
an important fact, because it predicts that phase changes
will be very sensitive to temperature changes.
The pressure of the effluent is ordinarily relatively
close to ambient pressure, and as a result, pressure changes
usually have a lesser effect on phase changes than
temperature does. The effluent composition pressure
independence, however, does not necessarily extend to the
effluent as defined at the Reference Method 5 filter.
Across the filter there often exists a significant pressure
INTROPY
•••••i
IIMVIRONMEIMTAU8TB,INC.
-------
29
differential, which may affect the concentrations of
volatile compounds on the filter and the rates of sorption,
desorption, condensation, and/or sublimation. In general, a
decrease in pressure will push the equilibrium of gas-liquid
systems and gas-solid systems toward the vapor state.
If the composition of the effluent is known, it is
often possible to predict the physical states of the
components within the effluent. For example, if one knows
that the effluent contains sulfur trioxide and water vapor
in certain concentrations, it is possible to predict when
sulfuric acid mist will condense, if the effluent
temperature is known. The ability to predict the acid
dewpoint is strictly analogous to determining the water
dewpoint; both require two important pieces of information:
(1) the identity of the substance (e.g., sulfur trioxide),
and (2) the concentration of the substance expressed in
units of mass per unit volume (e.g., g/L; gr/ft , mg/m ,
etc.) .
For the majority of effluent systems, most of the mass
can be chemically identified. Unfortunately, the small
amount of mass which resists complete characterization is
often that same mass associated with the condensible
particulate. A complete characterization of some source
effluents is often impossible,.
NTROPY
••••I
NVIRONMENTAU8TB.INC.
-------
30
The effluent stream of a coal-fired boiler provides a
good example of why effluent streams resist complete
characterization. The identity of the effluent is dependent
on the particular fuel characteristics, the source
operation, the control equipment operation, and any chemical
reactions occurring within the effluent stream. All of
these parameters may change with time, and some of these
parameters are interrelated. Thus, a thorough effluent
characterization can be miserably compromised by temporal
variability.
If the chemical identity of the effluent is not
adequately known, it may be difficult to interpret
particulate measurements obtained using standard
methodology. Ignorance of effluent composition may even
further compromise any measurements obtained using modified
sampling procedures. Put simply, the prerequisite for
meaningful measurements is a knowledge of what is being
measured and an understanding of the effects and/or
limitations of the measurement technique on the parameters
of interest.
INTOOPV
•••••I
I NVIRONMENTALJSTSt INC.
-------
31
The Operation of Physical Changes
Within an Effluent
The discussion which follows describes the physical
changes that can occur within an effluent stream. A
hypothetical effluent generated by the combustion of
fossil-fuel at a utility boiler is used as an example system
here, because the complexity of such an effluent provides a
most general case, since all the physical processes
described above are operating. The physical changes
occurring in this effluent will be discussed from generation
to collection on the reference method filter. Where
necessary, digressions will be made in order to elaborate on
the descriptions of the physical processes.
In this section, chemical reaction pathways will be
pointed out, but will not be discussed. The chemical
reactions that accompany the physical changes will be the
subject of the following section.
Initial Conditions
At the point of generation, the effluent is quite hot
O o 2
[1000 C (1800 F)] and still chemically reactive, even
though the major chemical reaction, oxidation, has gone
essentially to completion. The particulate can be described
as a mixture of unburned fuel and metal and non-metal
oxides. At this high temperature, many compounds and some
elements will be in a vapor state.
IIMTROPY
•••••
IIMVlRONMENTALJBTa.llMC.
-------
32
As used here, the word "vapor" implies that these
species may condense by the time the effluent reaches the
filter of the sampling train. As such, all of these
elements and compounds may be termed condensible
particulate.
Recent investigations at fossil fuel fired steam
generators have dealt with the chemical composition of
particulate matter as a function of depth into the particles
(chemical depth profiles), and as a function of particle
4
size. The results indicated that the more volatile elements
are associated with the particle surfaces and those same
volatile elements are preferentially adsorbed on the smaller
particles. Complementary investigations focused on
particulate chemical depth profiles as a function of
distance traveled within the effluent stream. It was found
that elements within the effluent were fractionated with
respect to volatility, i.e., the more volatile elements were
found associated with the particulate obtained at the
greater distances (and cooler temperatures) within the
effluent. All of the studies indicate: (1) that
condensation processes are in continuous operation
throughout the effluent stream, and (2) that particulate
matter serve as nuclei for the condensation processes.
It is difficult to quantify the contribution these
particle-surface condensations make to the condensible
IIMTROPY
•^••M
INVIRONMENTAUSTS.INC.
-------
33
particulate loadinq, because no studies have been conducted
which focus on such condensations in the Reference Method 5
train. Because this particular formation mode is a surface
phenomenon, and because it can be argued that the surface is
but a small part of the entire particle, it can also be
argued that particle-surface condensations contribute little
to the condensible particulate loading. Further study into
this particle formation mode is needed before truly accurate
rationalizations and/or predictions can be made.'
The Temperature Profile
The extent to which condensation processes — and by
extension, physical changes — occur, is dependent on the
effluent temperature and on the amount of time the effluent
spends at that temperature. This idea is illustrated by the
effluent temperature profile. The temperature profile, as
used here, is the relation between the change in effluent
temperature with time. (See Figure 3.) (The effluent flow
rate defines the relationship between the effluent stream
temperature and the distance traveled along the effluent
path.)
INTROPY
•••m
IIMVIFIOIMMENTAUSTStlNC.
-------
34
In general, the extent of a chemical reaction increases
with time. Indeed, this is also true when extended to
physical changes, since the same basic principles apply; at
least a finite amount of time is required for a system to
reach an equilibrium state. Thus, the extent of a physical
change is dependent upon the temperature and the time the
effluent spends within that temperature region.
In the effluent at generation, some condensible
material exists which may be described as condensible metal
oxides [e.g., calcium oxide (CaO), sodium oxide (Na-0),
etc.]. Only at very high temperatures are these metal
oxides stable as vapors. When the effluent temperature
drops as the effluent leaves the boiler, the conditions
favorable for condensation are produced, and the metal
oxides quickly condense. The condensations which occur
c
within boilers produce scale. Since the condensation of the
metal oxides occurs upstream of the control equipment (e.g.,
an electrostatic precipitator [ESP]), the condensation
products, particulate, can be collected by the control
equipment. These condensible components will have completed
condensation well ahead of the sampling and filtration
points. Therefore, the measurement of these metal oxides
provides an accurate gauge of control equipment performance.
Other classes of condensible compounds still exist in
vapor form at the high temperatures at the exit of the
boiler. If any of these compounds condenses before the ESP,
INTROPY
i^HMIM
INVIRaNMENTAUBT8,IIMC.
-------
35
measuring that compound as particulate would contribute to
an accurate evaluation of control equipment performance.
Whether these compounds condense, and the extent to which
they do condense, will be determined to a large extent by
the remainder of the effluent temperature profile.
In addition to the effluent temperature profile, the
extent of physical changes is intimately associated with the
chemical identity of the condensible substance within the
effluent stream. Each substance is characterized by its
boiling point and vapor pressure. The concept of vapor
pressure is fundamental for an understanding of the
magnitude of the condensible particulate loading.
The discussions of vapor pressure which follow will
assume equilibrium conditions. This assumption ignores the
fact that a temperature gradient exists within the effluent
stream. The reader should understand that the vapor
pressure of a substance will be variable, and will show a
dependence on the effluent temperature and the time within a
temperature domain.
The vapor pressure of a substance is a measure of the
amount of material in the gas phase at equilibrium with the
amount of material in the condensed phase. Water provides
an example of this physical property. If water is placed in
a closed container, the water will evaporate until enough
water exists in the gas phase to provide a balance between
IIMTROPY
^MHMB
IIMVIRONMENTAUSTB, INC.
-------
36
the evaporation and condensation processes. At equilibrium,
water vapor will be condensing at the same rate as liquid
water will be evaporating. All condensed phases display
this behavior to varying extents; thus, any condensed phase
will be associated with some vapor.
The vapor pressure of a substance is described by the
following equation:
in
\
P
-
^Pb/
-H i
—
R '
f 1
( T
1
-
Tb
H is the heat (enthalpy) of vaporization or
sublimation,
P. is the vapor pressure of the substance
at its normal boiling point, usually
29.92 in. Hg (760 mm Hg),
P is the vapor pressure of the substance at
temperature T,
R is the gas constant,
T. is the substance's normal boiling point
temperature, in absolute units,
T is the temperature of the substance when
its vapor pressure is P.
The significance of this equation lies with the
logarithmic dependence of vapor pressure on temperature. In
simple terms, small changes in temperature may produce large
changes in vapor pressure. Thus, small changes in
temperature may produce large changes in the distribution of
NTOOPY
•••••
NVIFIOIMMEIMTALJ8T8,INC.
-------
37
mass between the condensed phase and the vapor (condensible)
phase.
The vapor pressure curve for water is presented in
Figure 4. (Vapor pressure curves for other substances are
similar.) Two features of the curve are of special
importance. Firstly, as the system approaches the boiling
point, the vapor pressure increases at an exponential rate,
and secondly, with decreasing temperature, the vapor
pressure decreases and approaches zero asymptotically.
These properties of vapor pressure afford the following
generalizations:
(1) The amount of mass in the vapor phase increases
significantly near the boiling point. The
magnitude of the effects of condensible
particulates will be of greatest importance for
those substances which have boiling points near
the filtration temperature. In addition, the
magnitude of the effect will be very sensitive to
temperature.
(2) Substances with boiling points well removed from
the filtration temperature will not interfere (as
"condensible particulate") with the measured
particulate.
INTROPY
••••••
IIMVIRONMENTAIJBTSJNC.
-------
30
1.00
20 40 CO gO 100
YaporPressure Curve-for
Water
J^iqur-e v
IIMTROPY
INVIRaiMMENTAUBT8,INC.
-------
39
These ideas are graphically illustrated with the
hypothetical example presented in Figure 5. The temperature
dependence of the vapor pressure has been interpreted in a
different way. Here, the percentage of mass which is
uncondensed (i.e., in the vapor phase) is plotted versus
temperature. The percentage of uncondensed mass is
proportional to the vapor pressure; thus, the curves reflect
the logarithmic temperature dependence of vapor pressure.
-H a R~
A multicomponent effluent is represented here. A
filtration temperature range is shown by the rectangular
region within the graph. Relative to the filtration
temperature, substance C would be totally in the form of
particulate: i.e., substance C would have totally condensed
before collection within the Reference Method 5 train.
Substance A represents the opposite case. As indicated in
the figure, substance A would be totally in the gas/vapor
phase at the specified filtration temperature range. Thus,
substance A would either condense in the back-half or pass
IIMTPOPY
•••••
INVmONMENTAUSTBtlNC.
-------
40
through the sampling system without condensing.
Substance B displays intermediate behavior, since it
condenses within the filtration temperature range. The mass
of B collected on the filter will depend on the temperature
at the moment of filtration and on the variation in filter
temperature after collection. Thus, substance B can be
collected on the filter, but may subsequently evaporate off
if the temperature of the filter increases. The collection
of substance B at the indicated filtration temperature range
varies roughly from 10 to 60 percent.
Each substance, A, B, or C, will contribute to the
particulate catch in proportion to its mass loading in the
effluent. In a real life situation, the contribution of
substance B to the measurable particulate mass loading could
be insignificant. Therefore, the variability caused by the
condensation of substance B would not impact on the measured
particulate, and no condensible particulate problem would be
observed.
The essence of this graph is that as the effluent moves
through the ductwork, a potential multitude of condensation
processes can ensue. As the effluent cools, those compounds
with high boiling points will condense first, followed by
the compounds with lower boiling points.
IIMTROPY
mmi^m
INVIRONMersrTAUBTa.lNC.
-------
41
These phase changes are not strictly abrupt. In the
neighborhood of its phase change temperature, each component
will be partitioned between the condensed phase and the
vapor phase, Thus, when Reference Method 5 is applied, clear
demarcations may not exist between particulate and
condensible particulate at any particular point along the
effluent pathway.
Vapor pressure can also play a significant role in
affecting the measured efficiency of particulate control
equipment. Most control equipment operates by
discriminating between condensed material and gases/vapors.
Such is the case for electrostatic precipitators, cyclones,
and fabric filtration units. Particulate material entering
these devices can be removed from the effluent, and thus,
can be controlled. Obviously, particulate material which
results from the condensation of vapors after the control
device cannot be affected. The potential control, however,
is not necessarily a clear cut issue. In those cases where
vapors condense while passing through the control equipment,
the degree of control is potentially variable. This will be
true for those effluent components which have boiling points
within the temperature profile of the control equipment.
Figure 6 illustrates these concepts.
The graph shows the temperature dependence of the
percent uncondensed mass of three components. The behavior
IIMTROPY
•••M
IIMVIRaiMMEIMTAU8T8,INC.
-------
42
of component C in this diagram indicates that all the mass
will be in the condensed phase at the temperature of the
control equipment. The curve for component B passes wi';hin
the temperature range of the control equipment. For this
case, the collection efficiency of the component will be
variable. Over the temperature range of the control
equipment, between 50% and 10% of the mass of this component
will be in the gas phase and wil] not be controlled. The
behavior of component A in this scenario will be, even more
extreme. Between 45% and 100% of the mass of component A
will be uncontrollable. The magnitude of this effect will
be dependent mainly upon the relative mass loadings of the
affected components, and the temperature of the control
equipment. Again, in real life situations, the magnitude of
the effect may or may not be significant.
M
ass
IMTHOPY
••••••
INVIRaiMMEIMTAU8Ta,IIMC.
-------
43
Physical Changes in the Sampling System
In most respects, the physical environment inside the
sampling probe is quite similar to the physical environment
the particulate matter views within the effluent stream.
Indeed, If the temperature of the probe is maintained
identical to the effluent temperature, and if isokinetic
sampling is conducted, the physical properties of the
effluent sample should differ little from those of the
effluent itself. Particulate matter will collect in the
probe as a result of gravitational settling and impaction
with the walls of the probe.
The effect of temperature on the particulate measured
in the probe can operate in two directions, depending on
whether the probe temperature is greater than or less than
the temperature of the effluent. In addition, this measured
particulate will reflect the relative temperature difference
which exists during the entire sampling operation.
If the probe temperature is less than the effluent
temperature, condensation reactions can occur on the cooler
probe walls. Similarly, condensible substances may adsorb on
the walls of the probe. On the other hand, if the
temperature of the probe is greater than the temperature of
the effluent, not only will condensation and adsorption
reactions be prevented, but material already condensed and
IMTROPY
•••••
IMVIROIMMEIMTAUSTB,irMC.
-------
44
adsorbed may be volatilized. Particulate can be caught in
the probe, but also can be subsequently removed by
volatilization. Thus, if condensible particulate is
sampled, a variable probe temperature can cause a variable
probe particulate catch. However, the operation of the
Reference Method 5 sampling train is intended to account for
this potential variability, through the provision of the
heated filter which follows the probe and is maintained at a
known temperature.
After the effluent sample exits the probe, it passes
through the filter. Particulate matter will be impacted on
the filter and may accumulate to form a filter cake if the
loading is sufficiently high. The filter cake will provide
an additional site for physical reactions, depending on the
temperature of the filter relative to the temperature of the
effluent sample. Indeed, the potential physical changes
will parallel those which were associated with the probe.
The temperature of the filter can affect the state of any
condensible particulate which is formed or is collected on
the filter. However, if particulate is vaporized from the
filter, it will pass on to the impingers, and may not be
quantified. Therefore, if only the front-half of the train
is used for the particulate determination, the filter
temperature is the last major physical parameter which
determines the measured particulate.
IIMTROPY
•••••
IIMVtRONMEMTAUBTS,INC.
-------
45
The pressure difference across the filter and filter
cake makes a lesser contribution to the potential
variability of the filter particulate catch. If the filter
cake contains compounds having have vapor pressures which
are significant at the filtration temperature, these
compounds may evaporate as a result of the pressure
difference. The evaporation reflects the shift in
equilibrium toward the gas phase, brought about by the lower
relative pressure on the other side of the filter. The
evaporation of volatile compounds on the filter during
sampling is analogous to the lower boiling point of water
observed at higher altitudes, and lower atmospheric
pressures.
Since the filter marks the final point in sampling an
effluent, it is extremely important that any material
passing through is physically well characterized; otherwise,
the characterization of the material collected on the filter
will be compromised. If condensible particulate is present
within an effluent stream, it may pass through the filter.
The amount of uncharacterized and unmeasured condensible
particulate which passes through will depend on the
temperature history of the filter. The observed variability
in the measured particulate will be strongly dependent on
the variability of the filtration temperature. Finally, the
effect of the filtration temperature on the measured
particulate can be profound.
IIMTROPY
wmmmtm
IIMV1RONMEMTAUSTB, INC.
-------
46
Chemical Reactions
The occurrence of chemical reactions within an effluent
can greatly affect the observed loading of condensible
particulate. The reactions that may occur are not
necessarily independent of physical changes; thus, chemical
reactions may lead to physical changes, and vice versa. In
the discussions that follow, the reader should be mindful of
the complicated interplay of chemical reactions and physical
changes.
Definitions and Examples
The operations of chemical reactions within effluent
streams will be addressed from three general subject areas.
The first area, chemical reactivity, is concerned with the
ability of substances to react to form products. The
reactivity of an air-gasoline mixture serves as a simple
example. Experience tells us that this mixture is quite
reactive, if the conditions are right.
The second subject area, chemical thermodynamics, deals
with the stabilities of the products and reactants of
chemical reactions. The reaction of ammonia, sulfur dioxide,
and water vapor provides an example. Reactivity arguments
predict that the mixture is reactive, with the product being
INTROPV
•••••
INVIRONMENTALJBTB.INC.
-------
47
ammonium bisulfite. Thermodynamics, however, shows that the
O o
product of the reaction is unstable above 400 C (750 F);
therefore, if ammonium bisulfite is formed above 400 C, it
will instantaneously decompose back to the original
reactants, ammonia, sulfur dioxide, and water. For this
chemical system, thermodynamic arguments determine the
reaction actually observed.
Chemical kinetics, the third subject area, deals with
how fast a reaction occurs and, in a sense, how far a
reaction goes to completion. Time is the parameter of
interest here. An example which shows the interplay of
thermodynamics and kinetics involves the reaction of
nitrogen, oxygen, and water to form nitric acid.
nitrogen + oxygen + water > nitric acid
(gas) (gas) (liquid) (liquid)
Thermodynamics predicts that the reaction can occur at
ambient temperatures. Chemical kinetics, however, shows
that the reaction is infinitely slow. Experience supports
this argument, for if the rate of reaction were significant,
the atmosphere would dissolve in the oceans to form nitric
acid.
IIMVIRONMENTAUSTB,INC.
-------
48
Reactivity and Reactions
Most effluent streams are still chemically reactive at
the moment of generation. This is especially true for
effluents generated by combustion. Thus, even though
oxidation, the primary reaction, has occurred, many other
reactions can still occur.
The combustion of a fossil-fuel, coal, can serve as an
example of how reactivity arguments can be used to
rationalize the formation of condensible particulate by
Q
chemical reactions. During combustion, the primary reaction
is oxidation; the principle products are carbon dioxide,
CO-, and water, H?0. Impurities associated with the coal
are also oxidized. If the oxidized impurities are gases,
they will mix with the excess air, carbon dioxide, and
water. On the other hand, if the oxidized impurities are
solids, they will either be collected in the ash pit or be
entrained in the effluent as fly ash (particulate).
Using simple acid/base theory, general statements can
be made regarding the relative reactivities of the solid and
gaseous oxidized impurities. According to simple acid/base
9
theory, the oxides of metallic elements are classified as
bases, and the oxides of non-metallic elements are
classified as acids. Using this classification scheme, the
typical fly ash constituents, sodium oxide (Na_0), calcium
oxide (CaO), iron oxide (Fe203), and aluminum oxide (A1-0-)
IMTOOPY
NVIRaNMENTAUBTB.IIMC.
-------
49
are classified as basic substances. The gaseous compounds,
sulfur dioxide (S0_) , sulfur trioxide (SO,) , carbon dioxide
(C02) , and nitrogen dioxide (N02) • are consequently
classified as acidic substances. According to simple
acid/base theory, acids react with bases; thus, if a process
feedstock contains metallic and non-metallic elements, and
if this feedstock is decomposed in an oxidizing environment,
potentially reactive acidic and basic compounds will occur
together in the effluent. An example of an acid/base
reaction that occurs in the effluent of a coal-fired boiler
is given by the reaction of sulfur trioxide and iron oxide
to yield iron sulfate, a component of boiler scale.
+
Sulfur trioxide iron oxide iron sulfate
(gaseous) (solid) (solid)
"acid" "base"
In the course of the reaction, a solid and a gas
combine to give more solid. The reaction is rather general
with respect to the physical states represented here; thus,
the acid is a gas, and the base and products are solids.
Many chemical reactions similar to the example above
occur in the effluent stream. These reactions may continue
to the moment when the particulate is weighed in the
laboratory, or viewed from a different perspective, to when
the particulate becomes dispersed in the atmosphere. It
should be emphasized that chemical condensation reactions
IIMTROPY
••^•B
INVIRaNMENTAUSTS,INC.
-------
50
tend to produce positive biases to particulate measurements,
if such reactions occur within the sampling probe or on the
heated filter of the Reference Method 5 train. Using the
reaction above as an example of the potential impact of the
bias, the solid phase increases its relative mass by a
factor of 2.5, more than doubling the mass which would be
termed particulate.
The effect of chemical condensation reactions on the
measured particulate is dependent on where these reactions
occur within the effluent. If a condensation reaction
occurred before a control device, then the particulate
collected in the probe and on the filter during emissions
tests will reflect control equipment performance. Obviously,
reactions occurring after the control equipment would not
reflect control equipment performance. Finally, reactions
occurring on the reference method filter may lead to
erroneous interpretations of source performance, especially
if these reactions have no counterparts within the effluent
stream. The potential occurrence of particulate-forming
reactions on the filter has been the subject of
investigations of sampling at fossil-fuel fired steam
generators.
The limited amount of data currently available suggests
that chemical reactions occurring at the filter do not
contribute significantly to particulate measured at
I NVIRONMEIMTALJBTB.IIMC.
-------
51
fossil-fuel fired steam generators. Further studies at
other source categories are needed to assess the importance
of particulate forming reactions at the filter.
For oil-fired boilers, one chemical reaction which can
make a significant contribution to the condensible
particulate loading is that between sulfur trioxide and
water vapor to form sulfuric acid mist.
sulfur
trioxide water sulfuric acid
(gas) (gas) (liquid)
As was discussed in the Historical section, this
reaction was responsible for the higher filtering
temperature which is permitted at affected fossil fuel fired
steam generators. With regard to the formation of
condensible particulate, the result of the reaction of SO,
and H20 is essentially the same as the one presented earlier
for S02 and Fe203. The difference is that instead of the
mass of the solid phase increasing, particulate matter is,
in effect, appearing "spontaneously" in the effluent.
The consequence of the reaction between SO., and H,,0
extends beyond the formation of the sulfuric acid mist. The
sulfuric acid can react with additional water to form an
aqueous (water) solution of sulfuric acid.
IMTROPY
•••M
IIMVIRONMENTAU8T8,IIMC.
-------
52
sulfuric acid water sulfuric acid
(liquid) (gas) (aqueous)
[a solution of
sulfuric acid
and liquid water]
Thus, the reaction of gases to yield particulate matter
can ultimately result in further condensations which will
increase the observed mass loading.
The problem here transcends the increased particulate
loading contributed by the reaction. The fact that water
contributes to the condensation reaction is of greater
significance. Water, unlike the sulfur trioxide, is not
classified as a pollutant; thus, condensation reactions
which involve water will doubly bias the observed
particulate loading.
Water can contribute to the condensible particulate
loading in yet another fashion. Compounds can be chemically
associated with water molecules. These associations are
called hydrates and are often in the form of solids.
Examples of hydrates and their formulations are given by
iron sulfite trihydrate (FeS03 • 3H?0), magnesium sulfate
heptahydrate (MgS04 « 7Ho°^ » and magnesium chloride
hexahydrate (MgCl2 * 6H20).
•V
I
IIMVIPONMBNTAUSTBtllMC.
-------
53
The formulations presented here illustrate two
important points. The first is that the number of
associated water molecules depends on the chemical identity
of the compound of interest. Secondly, the mass that the
water contributes to the total compound is also variable.
The percentages of water by weight for the preceding
compounds are tabulated below.
Compound % Water by Mass
FeS03 • 3H20(s) 28%
MgS04 • 7H20(s) 51%
MgCl2 •6H20(s) 53%
It is apparent from the examples that the contribution
made to the particulate mass' by the water can be
significant. As a prelude to discussions which appear in
the section dealing with the stability of compounds, it
should be added that the number of water molecules
associated with a specific compound may be variable. For
example, magnesium chloride hexahydrate (MgCl2 •» 6H20)
represents the maximum hydration observed for this compound.
A lower formulation exists: magnesium chloride dihydrate
(MgCl2 • 2H20), which is the more stable hydrate form at
higher temperatures.
Not all compounds form hydrates; nevertheless, because
moisture is a common effluent component, the potential for
iixrraopv
•••••
• NVIRaNMENTAUSTS, INC.
-------
54
hydrate formation should be recognized. In addition,
hydrates are often stable above the boiling point of water,
so that heated probes and filters will not necessarily cause
their decomposition or prevent their, formation. Finally,
data showing the contribution of water of hydration to
particulate measurements are lacking; thus, it is currently
not possible to assess the real impact of hydrates on such
measurements.
The discussions above treated effluent reactivity.
Only two simple chemical reaction types were discussed:
reactions of acids and bases, and reactions involving water
vapor. Many other reaction types exist, but it would be
beyond the scope of this paper to discuss them all.
Finally, it must be emphasized again that a limited amount
of data exists for combustion processes — even on a simple
level.
IIMTROPY
•••••
IIMV1ROIMMEMTAIJST8.IIMC.
-------
55
Stability Considerations/Chemical Thermodynamics
Many reactions can be postulated as occurring within an
effluent stream. Whether the reactions actually occur is
dependent in part on thermodynamic considerations.
Thermodynamics involves the stability of chemical systems,
either pure compounds or elements, or mixtures thereof.
Within the scope of thermodynamics the important parameters
are: the chemical identity and concentrations of all the
substances in the system, the pressure of the system, and
the temperature of the system. The examples which follow
illustrate how thermodynamic concepts can be applied to
reactions within effluent streams.
12
The results of recent investigations into the
anomalous behavior of a cement kiln plume have suggested the
occurrence of the following condensation reaction:
+
ammonia sulfur water ammonium
(gas) dioxide (gas) sulfite
(gas) (solid)
Ammonia, sulfur dioxide, and water are all stable up to
very high temperatures. This is not the case, however, for
o 013
the product, ammonium sulfite. Between 60 C and 70 C,
ammonium sulfite decomposes to the reactants, NH.., SO,,, and
H-0. This fact dictates that if the reaction occurs above
this temperature range, the product will immediately revert
INV1RC>NMEIMTAIJ8TB,IIMC.
-------
56
to reactants. Thus, the reaction is observable only at
lower temperatures.
Indeed, the plume from the cement kiln displayed a
temperature dependence which was associated with the time of
day. The plume was most visible "during early morning
hours, but diminished rapidly as the day progressed" and as
the ambient temperature consequently rose. Thus, it appears
that the reaction is dependent on the thermodynamic
stability of the product ammonium sulfite in the effluent.
The authors of the report on the cement kiln plume
suggested yet another reaction pathway to account for the
anomalous plume. In this alternative pathway, sulfur
dioxide dissolves in water droplets and is subsequently
oxidized by dissolved oxygen. Dissolved ammonia promotes
the reaction by increasing the solubility of the sulfur
dioxide in the water, and also reacts with the dissolved,
oxidized sulfur to form ammonium bisulfate (NH.HSCK). If
this scheme is correct, then the daily variation may reflect
the fact that the necessary water droplets will form more
readily when the plume contacts the cooler morning air.
Here, the thermodynamic stability of liquid water droplets
determines the outcome of the condensation reaction.
Another example of how thermodynamics affects the
outcome of condensation reactions involves hydrates. As was
discussed earlier, some substances can associate with water
INTROPY
•••••
I NVIRONMEIMTAUBTa,INC.
-------
57
molecules to varying degrees. (For the purposes of the
following argument, it is assumed that temperature is the
controlling factor in determining the degree of hydration.) >
Water molecules involved in hydration are not
necessarily weakly held. Heating a hydrated compound to
o
100 C does not necessarily boil off the water molecules. In
addition, the extent to which increased temperature causes
water to leave a hydrated substance is dependent on the
chemical identity of the substance. The compound magnesium
sulfate can be used as an example. Heating solid magnesium
o
sulfate heptahydrate to 150 C results in loss of six of the
waters of hydration.
StpSOff*) -^ AljSQf '
Increasing the temperature to 200 C results in the loss
of the remaining water molecule and produces anhydrous
magnesium sulfate.
o
(Heating the anhydrous compound to 1124 C decomposes the
substance into magnesium oxide (MgO) and sulfur trioxide
(S03) , "base" and "acid," respectively.)
Hydrates formed in the effluent, on the filter, or
during the laboratory phase of the method may display
similar patterns of stability. In this regard, hygroscopic
-------
58
particulate matter, i.e., particulate which will take up
moisture, collected at the filtration temperature, may
acquire additional water when it is exposed to cooler
ambient air.
One must recognize the impact of stabilities of
condensible substances in the effluent, if one is to predict
condensation reactions. This recognition presupposes that
the identity of the condensible material is known.
Stability arguments, however, do not always dictate the
observed particulate mass variability. Kinetic factors, the
subject of the following section, must be considered also.
Rates of Reactions - Chemical Kinetics
The presence of a reactive system and the conditions
necessary for stable products do not necessarily lead to a
measurable reaction within the effluent. It is the time
dependence of a reaction which determines the extent of the
reaction. Reactions may occur quickly or slowly; thus, a
reaction that is slow relative to the time scale of the
particulate measurement may not be observed at all, because
insufficient products will exist at the time of measurement.
In addition, reactions which can produce condensed
particulate may not make significant contributions to the
measured particulate because of time constraints imposed by
reaction rates.
iisrmopY
IIMVIRONMEIMTAUBTB,INC.
-------
59
The rates of chemical reactions are determined by: (1)
the chemical identities of the reacting substances, (2)
temperature, and (3) the concentrations of these substances.
It is important to understand that reaction rates are
measured by experimentation. As a result, it is not
possible to predict the rates of chemical reactions £
priori. Instead, the identity of the reactive system must
first be determined; then, the reaction rate may be
determined experimentally. Sometimes, rate predictions can
be made by referring to earlier experimental data, if such
data exist. Obviously, prediction will be compromised by
insufficient data. Unfortunately, chemical kinetic data for
effluent systems are either lacking or are difficult to
apply, because of the enormous complexity of such systems.
Chemical Identity
Identifying all the components of even a simple
reactive system can be a difficult (and often an impossible)
task. The complexity of most effluent streams explains why
chemical kinetic data for such systems are few. The
combustion of coal can serve as an example again. Not only
is the elemental composition of coal a variable, but its
chemical composition is poorly defined and is also the
subject of current study. As a starting point for obtaining
IIMTROPV
•••••I
INVIRONMENTAU8T8,IIMC.
-------
60
kinetic data, the identities of the reactants are
determined. The reactants are identified first, because the
problem gets horrendously complex when the reaction starts,
and most of the kinetic data are obtained during the
reactions. With coal, identification is a complex problem
to start with.
The question of reactant identity is made more
difficult when catalytic processes are operative, and the
chemical complexity of effluent streams in general argues
strongly for the presence of catalytic reactions. As a
result, extra scrutiny is often necessary to establish the
roles of catalysts in reactions. Moreover, catalysts may
exert their effects in relatively low concentrations, with
the result that the task of identification becomes more
difficult, because the catalysts are difficult to detect.
Identifying all the reactants that determine the rates
of reactions occurring in effluent streams is presently not
possible, and represents a formidable task. Most of the
major species, however, can be identified, so that it may be
possible to make predictions regarding some potential
condensation reactions.
INTROPV
m^mmm
IIMVIRONMEI\rrAIJBTB,INC.
-------
61
Temperature
Most chemical reactions display increasing rates with
increasing temperature. This generality predicts a decrease
in reaction rates as an effluent cools. Thus, with cooling,
chemical reactions, leading to condensation products should
contribute less to the particulate loading, because of
slower rates. This argument, however, is difficult to
aPPly» because of the interplay of numerous other processes.
The effect of effluent temperature on chemical reaction
rates must be viewed within the constraints of
thermodynamics and with regard to physical states. For
example, reactions generally occur faster in solution than
in the gas phase (when all the reactants are gases). Lower
temperatures predict slower reaction rates, but in this
case, lower temperatures favor the formation of condensed
phases, which in turn provide reaction conditions which
favor accelerated reaction rates. The reaction of oxygen
with sulfur dioxide to form (ultimately) condensed sulfate
occurs slowly at stack temperatures, when the reactants are
all gases. If the reactants are dissolved in water, however,
the reaction proceeds much faster. Condensed water, which
would be present only at relatively low temperatures,
provides an alternate reaction pathway, with a faster rate
of reaction.
IMTROPY
•••••i
INVIRONMEIMTAUBTa,INC.
-------
62
In general, the role of temperature in influencing
rates of chemical reactions is overshadowed by the role of
temperature with regard to providing condensed phases that
promote reactions with faster rates.
Concentration
The preceding paragraphs touched upon the effect of
concentration on reaction rates. Generally, reaction rate
is proportional to concentration, i.e., if the
concentrations of reactants are increased, the rate of
reaction is increased. Condensed phases promote greater
reactant concentrations relative to the gas phase; thus,
reactions occurring in a liquid medium ordinarily proceed
faster than similar reactions in which the reactants are all
gases.. The impact of this fact is that the presence of
condensed material may accelerate the formation of
additional condensed material - condensible particulate. A
good example of the potential effect can be found with the
impingers (back-half) of the EPA Reference Method 5 train.
A condensed phase, the water, dissolves (concentrates)
sulfur dioxide, which may oxidize to ultimately form solid
sulfate. This reaction occurs much faster in the impingers
than in an effluent stream.
IIMTROPY
•••••
INVIRONMENTAUBTS.INC.
-------
63
The filter of the Reference Method 5 train provides
another example. Particulate matter in the effluent is
concentrated on the filter during sampling. This, of
course, is a necessary step in measuring the particulate.
Nevertheless, this concentrating will result in increased
rates of reactions if reactants exist within the filter
cake. If such condensation reactions make a measurable
impact, the biases will be positive. An obvious dilemma
exists: the method of measuring particulate may potentially
result in a measurement which has an unknown relation to the
particulate which actually exists within the effluent.
IIMTROPV
•••••
IIWIRONMEI\rrAU8TS,INC.
-------
64
Summary
Understanding the origin of condensible participate
both within effluent streams and within the Reference
Method 5 train requires knowledge of the chemical identity
of the effluent at generation and all the physicochemical
reactions that can occur after that point. For most
effluents such an understanding is presently not possible,
because of their chemical complexity and because of the
inherent difficulty of analyzing chemical and physical
changes which operate concurrently. Nevertheless, the
observation of condensible particulate can be rationalized.
The temperature is the single most important parameter
affecting the condensible particulate which is ultimately
measured. The formation of condensible particulate can be
extremely sensitive to temperature, and consequently, the
relation of particulate formation to the temperature profile
of the effluent should be known, if accurate interpretations
of the particulate catch are expected. Finally, because the
filter ultimately provides the measure of particulate, its
temperature is crucial when an effluent with a high
condensible particulate loading is sampled.
IMVIRaNMEISrrAUBT8.INC.
-------
67
Reference Method 5 is applicable to those effluent
streams where physicochemical changes are relatively small.
In effect, the method assumes that the effluent is
physicochemically static once it has been generated or
controlled, (i.e., the mass loading, size distribution, and
chemical identity of the particulate matter are constant).
When physicochemical changes producing condensible
particulate occur within the filtration temperature range,
and when significant mass is involved in such changes,
Method 5 gives a biased measure of the performance of
particulate control devices.
In situations where the evaluation of control equipment
performance is not the important issue, a physicochemically
reactive system may require extra attention to the sampling
procedure in order to ensure reproducible data. For extreme
conditions, the sampling location temperature and the probe
and filter temperature control may be very important factors
affecting particulate measurement results.
Because the occurrence of condensible particulate is
sensitive to temperature changes, precise filtration
temperature control is a prerequisite for obtaining precise
particulate measurements when condensibles are present. The
filtration temperature control of Reference Method 5 is
ordinarily sufficient for measuring particulate in effluents
where the particulate loading is independent of temperature.
IIMTROPY
mf^^m
IIMVIRaiMMENTAUBTB,INC.
-------
68
1 2
However, the results of studies ' comparing the Reference
Method 5 filtration temperatures to the temperatures of the
thermostated box (the temperature actually monitored)
indicate that significant disparities may exist between the
two. Thus, the assumption that the temperature control of
the Reference Mehod 5 train is sufficient may not be valid
for sources with high loadings of condensible particulate.
No data exist which show the relation between source
operation - control equipment performance and the resulting
loading of condensible particulate in the effluent stream.
Nevertheless, it can be anticipated that the relative
contribution of condensible particulate to the measured
particulate will increase as the efficiency of particulate
control devices increase. Again, if condensible particulate
makes up a significant fraction of the total particulate
loading, precise measurements will demand precise
temperature control. Thus, the extension of Reference Method
5 sampling to sources with low particulate loadings may be
limited by the precision of the method's filtration
temperature control.
Reference Method 5 is often used to measure particulate
at sources not subject to NSPS. These sources represent a
more varied and extensive source population, and, therefore,
the potential occurrence of effluents with high condensible
loadings is correspondingly greater. As a result, the
reference method is often unintentionally applied to sources
IIMTROPY
mmmmm
IIMVIRONMEIMTAUSTa,INC.
-------
69
with high loadings of condensible particulate.
Consequently, either the quality of the particulate
emissions data is poor, or the condensible loading is so
great that sampling is aborted.
Reference Method 5 was developed in conjunction with
NSPS, and, as such, it was never intended to be used as a
general method for measuring particulate. The general
applicability of Reference Method 5 may not be assumed. A
flexible approach is required when particulate tests are
conducted at source categories which cannot be compared to
those covered by NSPS. Any particulate sampling methodology
should be chosen so that it not only is compatible with the
source operating conditions, but, of greater importance, it
should also provide data with a known relation to the system
being evaluated.
As a criterion pollutant for NSPS sources,
"particulate" differs from the other criterion pollutants —
sulfur dioxide, nitrogen oxides, hydrogen sulfide, etc. —
in that particulate is defined in terms of physical state,
rather than in terms of chemical identity. Sampling and
quantification are greatly simplified when the chemical
identity of the substance of interest is known, because if
the chemical identity is known, the physical properties of
the substance can be easily obtained. Thus, when the
chemical identity of a substance is known, sampling and
INTROPY
•••••
IIMVIPONMEIMTAUaTB.IIMC.
-------
70
quantification take advantage of both the substance's
chemical and physical properties. As a consequence of
particulate being defined on a physical basis, the
methodology for its sampling and quantification lacks
specificity and precision relative to the methods used for
determining emissions of other NSPS criterion pollutants.
In addition, the physical state of particulate is defined in
terms of only one physical parameter — the temperature of
filtration; the pressure drop across the filter is not
considered. "Particulate" can therefore be interpreted as
"any substance which condenses above the filtration
temperature and which has a vapor pressure which is
negligible relative to the pressure drop across the filter."
The definition (as interpreted by the procedures of
Reference Method 5) makes only one distinction of chemical
identity: the filtration temperature is held above the
boiling point of water to prevent it from being measured as
particulate. A similar situation exists for fossil fuel
fired steam generators where high loadings of sulfuric acid
mist exist. A higher filtration temperature is permitted to
prevent the measurement of sulfuric acid mist as
particulate. Both choices of filtration temperature are
based on the same idea: the temperature is selected to
exclude a known compound which is not functionally related
to the system being measured. This idea points to the
fundamental problem with the definition of particulate; this
IIMTROPY
mmmmm
INVIRONMEIMTAUBTB.INC.
-------
71
problem is manifested by the issue of condensible
particulate.
The presence of condensible particulate during
particulate measurements for compliance determinations
compromises the results of such determinations, because
condensible particulate has an undefined relation to control
system performance. The relation is undefined as a
consequence of the fact that the chemical identity of the
condensible particulate is generally unknown. Without
knowledge of the chemical identity of a substance, it is
difficult to understand the physicochemical history of the
\
substance within an effluent stream.
One recurring theme throughout this paper has been that
condensible particulate does not reflect control system
performance if it forms in the effluent stream after the
control system. If the chemical identity of the condensible
particulate were known, it would be possible to resolve the
problem of the relation between condensible particulate and
control system performance, because it would be possible to
determine at what point the condensible particulate formed
in the effluent stream.
The fact that the chemical identity of particulate
matter is undefined (and its physicochemical history within
the effluent stream is unknown) leads to yet another problem
with interpreting particulate emissions data: at what point
IIMTOOPV
-------
72
in an effluent stream is particulate classified as
condensible? Answering this question is impossible, because
as particulate is currently defined, a distinction between
"true" particulate and condensible particulate cannot be
made without some knowledge of the chemical identity of the
entire effluent.
The issues discussed above all point to the same basic
limitation of Reference Method 5. The limitation has its
origin in the current definition of particulate matter. If
an effluent stream is physicochemically reactive and
condensible particulate is forming, it becomes difficult to
determine what is being measured, because the parameter used
in the measurement has a poorly defined relation to the
effluent stream.
FOOTNOTES
R. F. Vollaro, "An Evaluation of the Current EPA
Method 5 Filtration Temperature Control Procedure," in
"Stack Sampling Technical Information, A Collection of
Monographs and Papers," Vol. IV, EPA-450/2-78-042d, October
1978.
2
E. T. Peters and J. W. Adams, "Evaluation of
Stationary Source Particulate Measurement Methods, Volume
III. Gas Temperature Control During Method 5 Sampling,"
EPA-600/2-79-115, June 1979.
IMTROPY
•••••
IIWIRONMEIMTAUSTB.INC.
-------
73
RECOMMENDATIONS
1. Because the formation of condensible particulate is
extremely sensitive to temperature, precise control of
the Reference Method 5 filtration temperature is
imperative, if precise particulate measurements are
desired for effluents with high condensible particulate
loadings. Preliminary investigations should be
conducted to assess the feasibility of directly
monitoring the filtration temperature during Reference
Method 5 sampling.
2. The chemical identities of condensible particulates
should be determined, and their contributions to
particulate measurements should be quantified. In
addition, investigations should be made of the
functional dependence of condensible particulate on
source performance, control equipment performance, etc.
With knowledge of the chemical identity, the mass
loading, and the functional dependence on performance,
it may then be possible either to adjust particulate
measurements to account for the presence of condensible
particulate, or to modify the sampling methodology to
remove the condensible particulate contribution.
IIMTROPY
mmm^m
IIWIPONMENTAUBT8,IIMC.
-------
74
3. Future investigations should focus on identifying
chemical indicators of system performance. The
specificity of an indicator of known chemical identity
would remove the ambiguity associated with the term
"particulate" and would, in all probability, result in
correspondingly specific (and precise) methods of
sampling and quantification.
As a hypothetical example, iron may be a specific
indicator of the performance of an electrostatic
precipitator. Thus, sampling for iron would be the
basis for evaluating the performance of the
precipitator.
INTROPY
•••^H
INVIRONMENTAUSTB,IIMC.
------- |