-------
TABLES
Number Page
1 Acronym Key 9
2 Laser Operating Points 2A
3 Calibration Cell Measurement Results 37
v i i i
-------
SECTION 1
INTRODUCTION
Acid rain with its harmful effects on plant and animal life has focused
considerable interest in the sources of sulfur and nitrogen oxides. Atmos-
pheric emissions of these substances come from the burning of fossil fuels,
the smelting of sulfide ores, volcanoes and other natural sources. Human ac-
tivities have made significant changes on a continental scale and these ef-
fects are expected to rise further with increased dependence upon coal and
lower qua 1i ty oil.
In the combustion of sulfur containing fuels the primary product is sul-
fur dioxide which can be oxidized and hydrolized in the atmosphere to form
sulfuric acid. However, stationary sources such as electric power generation
plants emit sulfuric acid directly from their stacks. As much as ten percent
of the sulfur may be converted to sulfur trioxide either by atomic oxygen in
the combustion zone or by efficient catalysis on heavy metal contaminents in
the fuel. At moisture levels above four percent and temperatures in the
range of 150-200°C, theoretical models indicate that sulfuric acid vapor
should predominate over sulfur trioxide.
To better understand the factors controlling acid emissions from stacks
an in situ, real time monitoring systen for sulfuric acid is needed. Current
measurements are confined to wet chemical methods which require long sampling
times, are sensitive to operator techniques, and may not discriminate suffi-
ciently between sulfuric acid and sulfur trioxide.
This report describes the development and testing of a spectroscopic
instrument designed for continuous monitoring of sulfuric acid vapor concen-
trations. Although spectroscopic measurements of species such as S02 and CO
are routine, sulfuric acid presents a far nore difficult problem. This mole-
cule displays no sharp absorption features in the IR and is not amenable to
detection systems based upon gas filter correlation. In general, fixed fre-
quency laser techniques are limited by interference from the absorption lines
of water, carbon dioxide and other species. Laser Analytics' approach to the
measurement problem is a dual wavelength tunable diode laser (TDL) system ca-
pable of selectively avoiding interference lines and discriminating against
scattering from particulates and dirty optics.
Monitoring sulfuric acid by spectroscopic techniques requires that tne
infrared absorption spectrum be well characterized. Laboratory measurements
are difficult to obtain because of the very low vapor pressure at room temp-
erature and the dissociation, equilibrium and chemical attack problems in a
heated cell. Quantitative low resolution (-1 cm--) spectra were obtained by
Burch et al at Aeronautroni c FordO and by Majkowski^) at the General
1
-------
Motors Research Laboratory. The two absorption bards of interest occur at
1220 cm"' and 880 cm-*. Peak absorption coefficients for each of these bands
were of the order of 2-5 en-1 atm"1. While these absorption strengths appear
adequate for infrared monitoring of sulfuric acid in stack, gas the task would
have been simplified by the location of sharp spectral features. Eng
et al.^' ' at Laser Analytics, Inc. performed high resolution (.0001 cm"1)
measurements on both the 1220 and 880 cm"1 bands. Both low pressure and at-
mospheric pressure observations were made. No resolved structure was ob-
served in either case. Measured peak absorption coefficients for the 1220
cm-1 and 880 cm-1 bands were found to 6.5 and 6-9 cm-1 atm-1, respectively.
Analysis of the data in these experiments included the effects of dissocia-
tion of the sample when the observation cell was maintained at a temperature
higher than the liquid reservoir. Reanalysis of iiajkowski's data on this
basis gave a value in very good agreement with Eng's measurements.
Majkowski^ has reported a new value for the peak absorption coeffi-
cient for the 1220 band which is four times greater than his first published
va 1 ue (2) . This sharp increase upon remeasurement was attributed to inproved
experimental technique, particularly the elimination of hot spots, and to a
re-evaluation of the equ i 1 i br iurr, controlled dissociation of the 'n t'ie
absorpt ion ce!1.
2
-------
SECTION 2
CONCLUSIONS
1. A tunable diode laser differential absorption instrument has been
demonstrated to be a promising approach to the search for a reliable sulfuric
acid vapor monitor. The first continuous, real time measurement of sulfuric
acid concentration as a function of combustion conditions has been realized.
2. Stability studies which included over 600 hours of laboratory mea-
surements demonstrated that drifts in indicated response could be kept within
0.01 absorbance units over a three hour period. The limitations were found
to be primarily related to detector motion in the focal plane, depending upon
LN2 levels in the detector dewars. Spatial variations in the element respon-
sivity, V/W, combined with the relative notion of the detector elements pro-
duce unequal responses for the two laser images. This is seen as system
drift.
3. The system zero level is established through relatively straight-
forward but precise alignment procedures. However, the introduction of win-
dows in the analytical path was found to increase the uncertainty in the
zeroing. This uncertainty would be completely eliminated if the measurements
were made through a purgeable pipe.
4. Measurements of the peak absorption coefficient of the 880 cm"' band
of H^SO^, made with the laboratory calibration cell, yielded a value of
13.3 x 10"^ ppm-' n-'. Discussions with previous investigators and re-
examination of some of their data leads us to conclude that the earlier
studies were adversely affected by hot spots and large thermal gradients in
the absorption cell.
5. Field testing of the monitor confirmed the feasibility of the design
concept. A one week test program on the stack of a 650 MW electrical
generation plant yielded the first continuous record of sulfuric acid emis-
sions. The unit tested was burning high sulfur oil of Venezuelan origin.
Minimum vapor concentrations measured on the first two test days ranged be-
tween 3 and 8 ppm. On the third and final test day, the lowest level re-
corded was 25 ppm. On two test days, combustion conditions were varied by
increasing the excess air input from 0.5 to 1.5 percent. Measured increases
in acid vapor concentrations were 32 ppm and ^6 ppm.
6. The feasibility of dual laser monitoring was confirmed. Simulta-
neous operation of two tunable diode lasers (TDL) in one temperature control-
led refrigerator (TCR) was demonstrated for the first time in a practical
measurement system.
3
-------
7. Optical feedback could be completely eliminated from the system.
The significant steps taken to eliminate, feedback were rotation of the de-
tector off the optical axes, offset return of the laser beam and spatial
f i1ter i ng.
8. Linelocking of the lasers assured reliable, unattended operation.
This technique was found to be stable and rugged under field conditions.
k
-------
SECTION 3
RECOMMENDATIONS
1. Optical deternination of the zero level for the monitor was achieved
in the laboratory through reasonably straightforward alignment procedures.
Problems with this technique appeared when the stack windows were included in
the system. These windows exhibited an intrinsic differential attenuation of
0.36 absorbance units between the infrared frequencies used for monitoring.
However, a slight wedging in the windows led to apparent values of differen-
tial absorption which were critically dependent upon window orientation.
Thus, the major contribution to the uncertainty in the system zero in the
field was a window effect which could not independently be checked once the
instrument was mounted on the stack.
We, therefore, recommend the use of a purgeable pipe for future mea-
surements. The high precision of the dual wavelength monitoring instrument
can be realized with a purgeable path. The system is capable of making cor-
rections for.apparent differential absorption when used with a purged pipe.
In the purged pipe configuration the system zero could be routinely confirmed
to within the system noise level of approximately 0.005 absorbance units.
2. An experimental program should be carried out on the stack simulator
at the EPA Environmental Sciences Research Laboratory. This would serve to
establish the limitations of zeroing procedures. In addition, the introduc-
tion of CO2, H2O, NO2, and other interfering species at various concentra-
tions and temperatures would demonstrate the immunity of the monitor to
interference from these abundant stack gas components.
3. Essential to the successful design of the monitor was the prototype
dual laser temperature controlled refrigerator (TCR). Two improvements in
its design are needed to make the monitor more versatile. First, the maximum
attainable temperature separation between the lasers should be increased.
This would relax the tight constraints on laser selection and lessen the
chance for delays in laser replacement. Second, the laser mounts could be
designed to permit the installation of up to four lasers in the TCR. Only
two lasers would operate simultaneously but the additional lasers used sepa-
rately or as a pair would permit rapid switching to measurements of other
gases. Installation of additional lasers would also reduce the per channel
cost of the system.
k. Although the system has been designed to monitor h^SO^ which has ab-
sorption bands with no find structure it could easily be equipped with lasers
suitable for monitoring species such as CO, SO2, N0X> an^ HNO^. Only minor
5
-------
software changes would be required for measurements based on sharp spectral
1i ne features.
5. A modest research effort to evaluate the feasibility of mounting de-
tector elements inside the TCR would be desirable. This would eliminate the
requirement for supplying cryogens to the detectors- It is also anticipated
that such an arrangement would reduce the magnitude of detector element dis-
placements and thus lead to improved system stability.
6. A smaller and more rugged optical pallet should be developed. The
weight would be reduced through the use of relieved panels instead of solid
aluminum plate. Experience with the present system has shown that the pallet
area can be reduced by 25 percent. A bO percent reduction would be possible
if the detectors were mounted inside the TCR.
7. The volume and weight of the electronics and data systems could
be reduced at least 40 percent by eliminating redundant functions and un-
necessary features. Digital control would provide more reliable unattended
operation and permit automatic switching of experimental conditions.
8. Because of the large increase in the measured value of the h^SOii ab-
sorption coefficients further refinements in intensity measurements are war-
ranted. It has been shown that serious errors can result from large tempera-
ture variations over the reservoir, connecting line and absorption cell. In
the present work, differences as small as two degrees Celsius were accommo-
dated without difficulty. Future measurements should be made with total
temperature variations not greater than two degrees.
6
-------
SECTION 4
DESIGN CONCEPTS
DUAL WAVELENGTH DIFFERENTIAL ABSORPTION
The principle of dual wavelength differential absorption is best illus-
trated by referring to Figure 1 and the acronym key given in Table 1. The
combined outputs of the two TDL's at infrared frequencies v. and vR are
divided by beamsplitter BS2. The fraction which is directed to detector
via mirror M2 establishes the local sample of the laser power. The trans-
mitted fraction passes across the stack to a retrorefIector and returns to
BS2 where it is directed to the return detector D^ via mirror M3. Laser sig-
nal intensities are related to measured intensities by the relations:
AL = k!
1 AL
AR = k2
1 AR
BL = k3]
1 BL
BR = V
1 BR
(1)
where the k's include beamsplitter and other optical efficiencies. Absorp-
tions at the two IR frequencies can be expressed as
°al + "a = tn ['ai/'ar] '2'
°bl + "b = tn I'bl/'br] (3)
where a^ and are the H^SO^ vapor absorption coefficients at the two respec-
tive IR frequencies; L is the absorption path length and ca include ab-
sorption due to interfering molecular species, particulates ana optics.
Subtracting Equation (3) and (2) yields:
^A " V" + (CA " CB^ = £n ^,ALIBR/IAR,BL^ ^
CH2S0i{ = ^SA ~ V L jZn ^'alIBR/!ARIBL^ " ^°A " °B^j
where k = (k j k^)/(k2k3), Ch2SCl is the su,furic acId concentration and SA and
Sg are the absorption coefficients per unit H^SO^ vapor concentrations at the
analytical and background IR frequencies and vg respectively. in princi-
ple, the normalization constant k should be unity if the beamsplitter and
7
-------
CO
'AL
BS,
TCR
'BH
CTS
CTS
PSD
PSD
PSD
PSD
PSD
PSD
LCM,
LCM
D,
Figure 1. Schematic of HjSO^ stack monitor optics and electronics.
-------
TABLE 1. ACRONYM KEY
A -subscript-Analytical laser frequency
Aj, A^ Aperture
ADC Analog to Digital Converter
B -subscript-Background laser frequency
BSj, BS^ Beamsplitter
CTS Cryogenic Temperature Stabilizer
D , D Detector
L K
f^, etc laser modulation frequency
l^L> etc laser intensity
He Ne Helium Neon laser
L -subscript- local
L 1 ens
LCM Laser Control Module
M|, etc Mi rror
MP Mirror-Pellicle mount
PH Pinhole
PSD Phase Sensitive Detector (Lock-in amplifier)
R -subscript- Return
Rp etc Retroreflector
TCR Temperature Controlled Refrigerator (cold head)
TDL Tunable Diode Laser
-------
optical efficiencies are independent of infrared frequency. In practice it
is found to be within a few percent of unity for a properly aligned system.
Calibration of the measurement depends upon the difference in the ab-
sorption coefficients at the two IR frequencies. Discussion of the determi-
nation of this value is given in Section 7-0. The result, (S^ - Sg) =
12.2 x 10"^ ppm-' m~' implies that to achieve a detection sensitivity of
5 ppm-rn the instrument must register a change in differential absorbance of
0.006.
Effects of interfering species, (a^ - o^), are minimized by a careful
selection of monitoring frequencies. Figures 2-6 show calculated spectra for
H2O, CO2 and NO2 in the regions of the analytical 882.029 cm-', and back-
ground, 962,160 cm"', frequencies. These plots were generated for the condi-
tion of 15% CO2, 12% H2O and 25 ppm NO2 over a path of 5 meters at 150°C, the
temperature of the stack gases encountered in the field test. From the fig-
ures the total interfering absorbances are found to be .007 and .006 at the
analytical and background frequencies. The net differential absorbance,
(c^ - ag), is then 0.001. This illustrates the effectiveness of the high
resolution, tunable laser system in reducing the effects of interfering mole-
cular species. Particulate matter will also absorb and scatter infrared
radiation. However, over the wavelength separation of the two laser probes
there is no significant differential absorption.
Extended studies of interference absorption lines in the analytical and
background regions are given in Appendix A. An inspection of these figures
clearly shows that the number of interfering spectral features is non-
negligible. Low resolution techniques cannot adequately exclude these ef-
fects, because they effectively integrate over a large number of lines.
WAVELENGTH LOCK AND MODULATION
The key design concept which has made dual wavelength monitoring witl^
TDL's feasible is Laser Analytics' modulation and line locking technique.
This eliminates the need for a mode separating monochromator and permits re-
liable unattended operation of the lasers.
The TDL's are independently tuned by varying the current from the Laser
Control Modules (LCM). If the applied current is given by:
J = Jj s in (2ttf t) (6)
where is the DC component and the second term represents a dither current,
then the laser output frequency will have the approximate form:
v = Vj + Avj sin(2irf t) (7)
Where Av^ is the dither frequency magnitude. When this frequency modu-
lated radiation is passed through a cell containing a low pressure gas which
has a strongly absorbing spectral feature centered at only a laser mode as
Vj will be amplitude modulated. The laser frequency (wavelength) lock loop
is completed by a phase sensitive detector (PSD) tuned to the modulation fre~
10
-------
CM
m
cm
Figure 2. Calculated H2O interference in the region of the analytical
frequency.
00
pj
QD
CSJ
CD
cm
Figure 3- Calculated CO2 interference in the region of the analytical
f requency.
-------
uL
0. m
cm
Figure Calculated interference in the region of the background
f requency.
aL
0, 00
cm
Figure 5. Calculated CO2 interference in the region of the background
f requency.
12
-------
aL
0. 00
(M
CD
Cm
Figure 6. Calculated NC>2 interference in the region of the analytical
freouency.
13
-------
quency, f. Output cr the PSD, whicn corresponds to a first derivative of the
absorption profile, provides tne correction signal to the LCM.
Analysis of the laser output power shows that the dither across the lock
cell absorption line produces an anplitude modulation at the frequency 2f.
Another PSD tuned to this second harmonic will then monitor the relative re-
ceived laser power.
Referring to Figure I, it is seen that the coirplete system involves two
infrared detectors, D. and D^.and six PSD's. The local detector, DL, sup-
plies input to both tne lock loop (If) and signal level (2f) PSD's. Be-
cause the lasers are modulated at two distinct frequencies the PSD's serve to
separate the multiplexed information. The two PSD's connected to the return
detector, D^, function in the same manner.
The line lock and modulation cell can be filled with any gases which
provide suitable absorption features at the desired wavelengths. Optimum
modulation is achieved with a Doppler limited absorption line and a dither
excursion which is approximately 1.2 times the Doppler halfwidth.
Outputs of the PSD's, 1^, I , and I ^ are digitized and supplied
to a calculator which averages the results and calculates the H2SOI4 concen-
tration from Equation (5). Results are printed at operator definable
i nterva1s.
Selection of the lock lines to be employed is a compromise among the
stable gases which have strong, sharp, assignable absorptions falling in the
available interference free wavelength regions (Appendix A). Initially lines
of NO2 and NH3 at 881.A10 and 962.25^+ cm-' respectively were selected to con-
trol the analytical and background infrared frequencies. Early tests in-
dicated that improved operation could be obtained by switching to a mixture
of NH^ and OCS. The final selections, employed both in laboratory and field
testing, were the R(62) transition of the v. band of OCS at 882,029 cm-' and
SQ (9,9) transition of the band of NH, at: 9&2.160 cm"', These choices do
not represent a unique compromise but rather proved to be effective and con-
venient. The selection can also be influenced by operating temperature con-
straints of the dual laser Temperature Controlled Refrigerator (TCR).
I b
-------
SECTION 5
OPTICAL SYSTEM
DUAL LASER MOUNTS
Simultaneous operation of two TDL's was achieved through the design of a
dual laser Temperature Controlled Refrigerator (TCR). This modification of a
standard Laser Analytics' product was developed under a NSF program for long
path monitoring.(7) Two independently regulated laser mounts were attached
to the heat sink of a closed cycle cryogenic refrigerator.
Attainable temperature separation of the two lasers as a function of
the set temperature of one of them is illustrated in Figure 7. This plot was
obtained by adjusting the Cryogenic Temperature Stabilizer (CTS) for one
laser position and following the unregulated (lower) temperature of the
other position. Maximum temperature separation with both lasers operating
will be reduced and will depend upon the specific laser characteristics. At
the time of the field test, the lasers were operated with a separation of
M°K. Figure 8 is a plot of the power provided to each of the temperature
regulating heaters as a function of the separation of the two temperatures.
For this case, T^ was held fixed at 32°K. It is seen that the operation is
not quite symmetrical because of an unequal thermal resistance between each
laser position and the cox-non heat sink.
OPTICAL PALLET
The original optical configuration conformed to the schematic concept
illustrated in Figure 1. The two laser beams were combined by a mirror-
beamsplitter assembly directly in front of the lasers. The collimated radia-
tion passed through the lock cell and was divided by a beamsplitter, BS^,
which served both to provide detector D with local reference intensities and
to direct the returning radiation to detector D^. This direct retroreflected
approach proved vulnerable to standing wave or etalon effects which responded
sensitively to mechanical and thermal displacements. The net effect was in-
stability in the zero baseline of the instrument. In extreme cases, feed-
back of the laser radiation occured causing large instabilities in laser
output.
Three examples of etalon effects are shown in Figure 9. These fringes
were identified with the following paths: a) calibration cell to TCR window,
b) local detector to return detector, c) local detector to local retro-
reflector to return detector.
15
-------
10-
6
AT
(°K)
6+
4 -
2
16 20 24 28
Figure 7•
Maximum temperature separation
32 36 40 44 46 50
T2 (°K)
n dual laser cold head (TCR) versus operating temperature.
-------
Position
Position 2
T
1
-6-4-2 0 2 4 6
AT (° K)
Figure 8. Thermal regulation power as a function of temperature separation
in dual laser cold head with no laser load.
17
-------
Reproduced from
best available copy.
Fiqure 3 • Eta ion e~rec ts observed in optical systeT. a) ca1i brat i cn cell to
TCP, window, b) local detector to return detector, c) local detector to
local retroref 1 ectctc return detector.
-------
To completely eliminate feedDack anc other etalon effects, the direct
retrorefIected return bean was changed to a staggered configuration. The
beam was directed off center onto the corner cube retroref1ector so that it
returned on a displaced, parallel path. An added benefit was that all of the
returned radiation could be brought to the return detector. In the original
configuration beamsplitter BS2 directed only 35 percent of the returning
radiation to the detector. The detectors were also repositioned so that the
element normals were skewed about 10° relative to the optic axis.
Considerable effort was devoted to defining alignment procedures which
would lead to reliable determination of the zero level for the instrument.
Referring to Figure 1 and Equation 5 this is seen to mean that in the absence
of absorption, the value of the normalization constant k given by
k = 'al'BR/'AR'BL ^
should be the same for a beam returned from a remote retroreflector Rj and
for the case where another retroref1ector, R_ , short circuits the path. In
other words, it is necessary to insure that the value of k obtained from R_,
which is on the optical pallet, represents the correct normalization for trie
distant reflector, Rj.
The energy emitted from a TDL is generally contained within an f:l cone.
However, the distribution of energy within this cone is not homogeneous; the
radiation tends to be confined to separate narrow filaments within the larger
cone. These filaments can be emitted from separate regions of the laser
crystal which may represent an effective object with typical dimensions of
the order of a few hundred microns. These characteristics have two important
consequences. First, the finite effective object size determines the residu-
al divergence of a collimated bean. Second, the filamentary composition of
the emissions yields inhomogeneous beams after collimation.
The residual divergence is of the order of 1 milliradian for the optical
system. Over a path of several meters, the expanded beam becomes large
enough to vignette on the optical apertures. If the radiation pattern were
homogeneous, this would not necessarily be a serious problem. The actual
power distribution, however, can lead to unequal losses in the received power
from each laser and hence, a change in the effective normalization. Addi-
tionally, with spatially inhomogeneous beams it is not possible to find an
optinum signal level in scanning an image across the detector, because the
image plane has hot spots. The severity of this problem was not fully ap-
preciated prior to the initiation of this program.
A further problem arises when the characteristics of the infrared de-
tector elements are considered. These one millimeter square elements are
larger than the images produced by the focusing mirrors and M , However,
the response over the surface of the elements is not uniform. Recent studies
at Laser Analytics, Inc. performed under a separate contract1, ' have provided
mappings of detector elements. Results for the detectors employed in the
H^SO^ monitor are shown in Figure 10.
19
-------
S/N 3776
Figure 10. Detectivity maps of typical detector elements
-------
These contours reveal differences of a factor of two or greater. When
one considers the superirposed images of the two lasers on the return de-
tector, it is easy to see that small displacements which come with the
switching from the near to the far retro mirror can result in sizeable
changes in the ratioed intensities.
Consideration of the homogeneity problems and unsatisfactory results in
our efforts to obtain reliable normalization procedures suggested that the
laser outputs should be spatially filtered. This was accomplished through
the incorporation of a pinhole in the system. The final optical layout is
illustrated in Figure 11. The lenses, L, image the laser outputs on the pin-
hole, PH. The mirror, , and beamsplitter, BS^ , serve to combine the two
beams. Radiation emerging from the pinholes is directed by mirror M to an
off-axis parabaloid (OAP), M , which serves to collimate the beam. After
traversing the lock/modulation cell, LC, a sample of the beam is directed to
which is an OAP that images on the detector element of D^.
The probe bearr. continues from BS^ to M where it is directed through the
stack and returns to M,. The OAP, Mg, images onto detector to complete
the optical circuit. A helium-neon laser, HeNe, can be coupled into the
optical path through a removable mirroi—pellicle module, MP. The visible
bean is used to align both the components on the pallet and the retroreflec-
tor. Figure 12 is a photograph of the complete optical pallet. All compo-
nents can be identified by reference to Figure 11.
LASER TESTS
TDL's can be tailored to emit in desired wavelength regions by control of
the energy gap through adjustment of the chemical composition of ternary semi-
conductor materials. Once fabricated a diode's precise emission wavelength
within the operating region is a function of temperature. Regulation of both
the laser crystal temperature and the injection current are required to main-
tain emission at a given wavelength. Each laser has its own peculiar
wavelength-temperature-current characteristics which may vary with age and
cycling between cryogenic and room temperatures.
Selection of appropriate lasers begins by applying the criterion that a
given pair operate within the temperature separation constraints of the dual
laser TCR. A selected pair must next be characterized with a monochromator.
The laser beam is deflected from the pallet after it has passed through the
lock cell and sent to the monochromator. The appropriate lock lines are
located and the current and set point temperature of each LCM and CTS are re-
corded. The lasers should be temperature cycled and the procedure repeated
to insure reproducibility.
Several lasers were tested in the course of this effort. Final selec-
tion was determined primarily on the basis of cycling stability. Table 2
summarizes the operating condition of the selected lasers over a ten week
period ending with the field test. Temperatures are recorded in terms of CTS
sensor voltage readings which are reproducible to greater accuracy than the
absolute temperature calibration. Different polarities of the two diodes
merely reflects differences in the details of the fabrication process. Both
21
-------
¥ A
mr
N3
ro
M,
M
HeNe
M,
D„
BS,
MP
Figure I]. H2S0^ monitor optical system schematic
TCR
11
TDL^j
LC
Md
-------
i cure 12. HoSC^ monitor optical syster, photog-aoh.
23
-------
TABLE 2. LASER OPERATING POINTS
Date
S/N 9054-3
881.410 cm'
T
•1
S/N 9178-31
962.254 cm ^
19 July
5 Aug
1.0766
1.C78S
1.0783
-0.573
-0.633
-0.810
cvcled
1.0664
1.0660
i.0677
1.730
1.390
1. 715
8 Aug
1.079C
-0.842
:jo chakge
12 Aug
13 Aug
20 Sept
1.0791
1.0791
1.0783
1.0803
-0.840 1.0677
cycled
-0.851 1.0677
-0.809(improved operation)
cycled
-0.924
1.707
1.715
25 Sept
2 Oct
1.0803
1.0800
-0.930 1.0686 1.675
cycled
-0.926 1.0684 1.668
3 Oct
1.0803
•0.945
1.0686
1.667
4 Oct
1.0G0]
-0.926
1.0686
1.668
2k
-------
diodes were operated with injection currents near threshold in order to re-
duce the nunber of nodes emitted.
STABILITY TESTS
Continuous monitoring sensitivity is United by long term internal drift
in the instrument. For a continuous monitor, frequent adjustment and cali-
bration should not be necessary. An intensive effort was, therefore, made to
identify and remove sources of instability. More than 600 hours of stability
tests were conducted following assembly of the instrument. These tests led
to improvements in pallet rigidization, removal of sources of optical feed-
back, strengthening and replacement of some optical mounts, spatial filtering
as well as development of reliable alignment procedures.
Laser amplitude variations are effectively removed from the system re-
sponse by the ratioing scheme. No drifts in the PSD receivers or other
electronics were of consequence. System limiting stability in the current
optical configuration is associated with the cooling of the infrared de-
tectors. The detector elements are mounted on the tips of liquid nitrogen
dewars. Changes in the coolant level cause displacements of the elements.
The laser images consequently move across the elements. Both vignetting and
the uneven response of the elements cause these motions to be recorded as
output variations. These effects are most pronounced when coolant is added
and ray display recovery times of several minutes. Careful focusing and
precise alignment are essential to the reduction of these effects.
Figure 13 shows the level of performance achieved prior to the incorpo-
ration of the pinhole. Each strip covers 3-3 hours and shows variations
within 0.01 absorbance units (1 au = 820 ppm-m). These tests were run in the
laboratory with the retrorefIector at a distance of 3 meters from the optical
pallet. Stability results obtained following the spatial filtering modifica-
tion are shown in Figure 1^. The sharp deviation in the first hour resulted
from replenishing the coolant in the dewar.
In general, our results with the pinhole in place were more predictable
than those without a pinhole, and stability was typically better than 0.01
absorbance units on time scales of 1/2 hour.
STACK COUPLING FLANGES
The flange and window assemblies were specifically tailored to the re-
quirements of the field test site. Access to the stack was through a ^-inch
diameter slotted pipe. The pipe was terminated with gate valves at each end.
The valves had standard schedule 150, ten inch diameter flanges.
To avoid deposits on the windows, a short section of pipe containing an
aluminum honeycomb material was designed to attach to the gate valve. This
material would inhibit diffusion of soot and droplets toward the windows and
also act as a field of view limit. Cell size was 0.125 inches and cell depth
was 3 inches. Alignment was made through two threaded shafts in the outer
flange. The complete assembly is illustrated in Figure 15.
25
-------
MINUTES
Figure 13. Results of laboratory stability tests of optical system before pinhole filter install-
ation.
-------
0. 15 ..
ID
e. 05
0. 15
Figure 1^. Laboratory stability test results for optical system after pin-
hole i nsta11 at i on.
27
-------
o
ANGES FOR H' N3V. NA
BEVELED SURFACE
BaF, WINDOWS
NOMINAL Fir E .
Figure 15- Stack coupling flange and windows.
28
-------
Laboratory tests of this system showed that it introduced greater atten-
uation of the laser beam than expected. It was difficult to adjust and
interfered with visual alignment. Prior to the field test, the honeycomb was
removed. The only window contanination encountered in the field test was
liquid condensation on one side of the stack. This was eliminated by warn-
ing the flange with electrical heater tape.
29
-------
SECTION 6
ELECTRONICS AND DATA ACQUISITION
The electrical schematic for the complete monitor system is shown in
Figure 16. A photograph of the portable electronics cases is given in
Figure 17. In the photograph, the case to the left contains the laser con-
trol modules, LCM, and the lock loop amplifiers, PSD. Also contained in this
case is an oscilloscope used to view the infrared detector outputs during
alignment and line locking of the lasers. Above the case are the two cryo-
genic temperature stabilizers, CTS. In field operation, these units must be
located in close proximity to the optical pallet whereas the two cases may be
remotely positioned.
The case to the right in Figure 17 contains the data acquisition system.
This consists of four lock-in amplifiers, PSD, which are coupled to the cal-
culator through the analog to digital converter, ADC. The calculator is out
of view inside the drawer at the top of the case. Specific details of the
interconnections are given in the Systems Operations Manual.
Data acquisition is controlled by a Hewlett Packard 9815S calculator. A
description of the software and a program listing are given in the Systems
Operations Manual.
30
-------
fB REF
f. REF
PREAMP
PREAMP
CTS
CTS
PSD
PSD
PSD
PSD
PSD
PSD
ADC
LCM
LCM
CALCULATOR
TCR
Figure 16. Laser control and data acquisition systen- block diagram.
31
-------
Figure 17, rnotoc-aph c£ electronics aid cis :a acquisition systeri.
3?.
-------
SECTION 7
h2so^ CALIBRATION CELL
DESIGN CONSIDERATIONS
The apparatus described in this section is designed for laboratory cali-
bration of the monitor instrument. It attaches directly to the optical pal-
let of the monitor. Measurements are conducted in the same manner as cross
stack measurements.
The absorption cell and its copper heat shield are shown in Figure 18.
The cell is constructed of one inch diameter pyrex tubing with an overall
length of 1inches. Two silicon windows are fused into the pyrex tubing
with an 8 inch absorption path between then. The fused windows provide a
rugged vacuum seal which can withstand thermal cycling and high temperature
H^SO^ vapor without deterioration. To maintain uniform temperature over the
absorption path, the 12 inch long heating jacket extends beyond the silicon
windows. The cell body extends three inches beyond each silicon window, and
the ends of the tube are closed with BaF2 windows which serve to prevent cool
air circulation to the silicon windows to the cell. To prevent etalon prob-
lems all windows are installed about three degrees off from norTal incidence
to the optical path. A 50 ml reservoir bulb contains the liquid sample. The
cell, reservoir and connecting neck are separately heated. To maintain uni-
formity over the absorption path, the cell is surrounded by a copper jacket 12
inches long. A cylindrical heating mantle is fitted over the copper jacket.
Reservoir temperature is controlled by a standard spherical heating mantle.
To insure that no cold spots occurred in the connecting neck, this region was
wrapped with a heating tape. Both of the heating mantles are regulated by
current proportional temperature controllers which have an accuracy of ±0.1C.
A variable autotransformer controls the necK heater. Ten thermocouple sen-
sors are located on the cell and copoer jacket. Their positions are indicated
in Figure 18. Temperature readout is provided by a digital thermometer with
a resolution of 0.2°C.
The optical pallet containing the calibration cell is shown in Figure 19.
Figure 20 is a Dhotograph of the cell attached to the monitor. Also shown is
the control cabinet containing the proportional controls and the digital ther-
mometer. The cell is supported by the cylindrical heating mantle which is
cradled in adjustable supports that permit optical alignment. The reservoir
bulb is suspended freely below the pallet. This arrangement is very stable
and shows no noticeable displacement of the cell through thermal cycling.
33
-------
SPLIT COPPER TACKET
TEMP. CONTROLLER
SENSOR ND. 2.
BdFz WINDOW
2 PLACES
SILICON WINDOW
2 PLACES
^EAL OFF NECK
¦TEMP. CONTROLLER
SENSOR ND. 1
Figure 18. H2S0^ calibration cell diagram showing liquid azeotrope reser
voir, absorption path and copper heating jacket.
-------
Figure 19- H2S0^ calibration cell pallet optical diagram.
-------
-------
Preparation of the cell began by filling the reservoir 2/3 fuli with a
37 normal solution. The cell was evacuated and heated to 180°C. An
azeotropic mixture was obtained by distilling away about i/3 of the liquid
before the cell was cooled. The first cell prepared was backfilled with dry
nitrogen before it was sealed. This atmospheric pressure version proved to
be unworkable because the partial pressure of could not equilibrate be-
tween the cell and the reservoir. An evacuated cell was adopteti when it was
demonstrated that an equilibrium could be achieved in this manner. The sealed
cell was tested with a tesla coil to verify that a good vacuum had been pre-
served .
CALIBRATION MEASUREMENTS
The calibration cell pallet was first aligned to the monitor with the
cell removed. The cell was then installed and positioned to give maximum
transmitted signal. Insertion loss for the cool cell approached 90 percent.
This loss comes primarily from the silicon windows.
Optical normalization was determined for the cell at room temperature.
The cell mantle was switched on and the cell was allowed to stabilize at a
set temperature. The change in differential window absorption was recorded.
The reservoir was then heated and allowed to stabilize at a temperature below
that of the cell or neck. Measurements were recorded only after the differ-
ential absorbance was observed to vary by less than 0.01 units over a period
greater than ten minutes. Results of six measurement runs are given in Table
3-
TABLE 3- CALIBRATION CELL MEASUREMENT RESULTS
Run
No.
^b u 1 b
°C
T , ,
ce 1 1
Absorbance
PH2S01(!Torr>
SA " ^B
(ppm-m~1x10L)
1
200
206
1.15
^.17
10.3
2
188
206
0.^9
2.12
8.7
3
190
192
0.98
2.90
12.6
k
185
192
0.81
2.20
13.8
5
169
192
0.3^
0.83
15.1
6
180
192
0.56
1 .6^4
12.8
The absorbance column in Table 3 gives the net differential absorbance
due to the r^SOj, vapor. Partial pressures given for h^SO^ are taken from
Lutschinski (9?. They have been corrected to account for dissociation in ;
going from the reservoir bulb to the hotter cell. The results given in the
last column are differential absorption coefficients per unit i^SO^ vapor con-
centration. From the work of Burch et al.(') the absorption coefficient in
the 962 cm"1 background region (Sg) can be estimated to be .09 of the peak
absorption coefficient (S/^) in the 880 cm-1 band. Taking an average value of
37
-------
j, - Og - 12.2 ± 2.3 x 10~4 Dpm-1n-1 the peak absorption coefficient, S^, for
the band is 13.3 x I0_i4 ppn"1^"1. This is a suostantial inc^r^ase over the
value of 7.0 x 1C~'H ppn_1,Ti"- reported by Eng and co-workers
icre;
,(3).
A closed forn analytical model incoroorating temperature dependent pa-
rameters has been formulated in order to better understand tne liquid-
vapor system. Partial pressures above the I^SO^ azeotrope at temperature T
are given by (fron Lutschinski) (9).
m _ , [9.603-'ili23/T] (I)
h o" "
pso it) (2)
PH .. (T) - 10l9-»59-3976/T] (j)
2
PTotal(T) = PH,S0,(T» * Vo(T> + PS0 ,lT) <*>
2 k 2 3
In our case the azeotrope is maintained at a temperature Tr and the ab-
sorption path is maintained at temperature T(- where To >_ Tr.
For the sulfuric acid azeotrope the temperature dependent equilibrium
dissociation constant, K, (T), is given by
V(T» pso
Km fff- • (5)
2^it
So the equilibrium dissociation constants for the two temperatures T^ and Td
can be calculated in a straightforward manner from relations (1) through (4).
If we consider the volume immediately above the azeotrope solution
to be in equilibrium, then
ph o'tr' pso (V
K(T ) = —^ rpl (6)
R H_S0, R
2 H
In the absorption cell, at temperature T^-, we approximate the equilibrium
dissociation constant by equation (7) where AP is a small change in pressure.
Vo(V + sp)
38
-------
We can calculate AP *rom equation (7), anc we calculate a final I^SO^ pres-
sure in the cell body from the relation
Pfinal(H2SV " (PH,S0 'V " iP) F-fsP (8)
2 h t
where Pt/(Pt + AP) is a small pressure correction and Pr. . is the sulfuric
acid pressure that contributes to the Beer's law absorptfcift.
Figures 21 and 22 present the results of Table 3 along with absorbances
calculated from the above relations for various assumed values of the absorp-
tion coefficient. Consistent results are seen in Figure 21 for a constant
cell body temperature of 192°C. Lower values for the absorption coefficient,
shown in Figure 22, were obtained when the cell was maintained at 206°C.
This tends to suggest dissociation problems associated with hot spots. How-
ever, there are not sufficient grounds for rejecting this data. Therefore,
our final value of the absorption coefficient is the average of the six meas-
urements given above.
In reviewing the earlier work the following distinctions should be con-
s i de red:
1. The absorption cell used in the present measurements was well con-
trolled thermally. Temperatures of both the cell and the reservoir were reg-
ulated by current proportional controllers. Thermocouple sensors on the cell
gave temperature profiles during each experimental run. The maximum varia-
tion noted was 2.7°C. In the earlier work only the reservoir was heated with
a feedback controller. The cell body was warmed by electrical heating tape
connected to a variable transformer. Temperature uniformity along the cell
was not adequately confirmed. Hot or cold spots with variations 10°C or
greater cannot be excluded.
2. Present measurements include smaller temperature separations between
cell and reservoir than previously attempted. In the limit of very small
differences the effects of further dissociation shoulc disappear. The pres-
ent study shows very good agreement between measurements at small and large
temperature separations. This confirms that extra dissociation effects have
been adequately treated.
3. Concern for aerosol generation led to the inclusion of a glass wool
plug in the previous work. The present system has an unobstructed neck con-
necting the reservoir to the cell body.
k. A narrow neck side arm contained the liquid sample in the earlier
measurement. In the present work the sample is contained in a bulb reser-
voir which provides a large area liquid-vapor interface. This affords great-
er confidence that the liquid and vapor are in equilibrium.
5. The H2S0^ monitor instrument used to make the present absorption co-
efficient measurements offers the considerable advantage of real time, con-
39
-------
Xr
O
1.0
1. 4 ..
1.2.
1.0..
01
u
rt
n
A
g B.0
wi
.n
<
0.0
0.4..
B.2..
0.0
3
t>M - 192 °C
3 3
Reservoir Temperature (°K)
IS
Figure 21. Calculated and measured absorbances of calibration cell as a function of reservoir temp-
erature. Cell temperature 192°C.
-------
1-6
ppm
L'o i ]
-1
0.0
0.6 ..
0.4
0.2 ..
in
in
¦¦in'-.nj-vrji i Tomjieraturp (°K)
Figure ?2. Calculated and measured absorbances of calibration cell as a function of reservoir temp-
erature. Cell temperature 206°C.
-------
tinuous xeasuremert of the absorption cell. This permits accurate deterrrina-
tion of stable conditions in the heated cell. Calibration measureTents were
accepted only after the vapor absorption remained stable for a period of from
10 to 1«0 mi nutes .
6. Partial pressures for the vapor above the I^SOr azeotrope used in
the present measurements were taken from Lutschenski (9). Eng and co-workers
(3) obtained independent results which are in good agreement with these val-
ues .
7. The present measurement is based upon a new modulation technique
which is sensitive to the second harmonic of the laser modulation frequency.
The earlier work was a conventional approach employing a beam chopping modu-
lator and a single detector. For the second harmonic (second derivative)
technique to yield a different result it is necessary to show that it offers
a nonlinear response. Intensity measurements on carbon monoxide made at
Laser Analytics, inc. have shown that the method is indeed linear over the
range of absorbance encountered in the H^SO^ measurements. We therefore are
confident that the modulation and detection methods employed in this work do
not introduce systematic errors into our results.
A reanalysis of the low pressure data taken by Eng and co-workers^3) has
suggested that average cell temperatures were much higher than originally as-
sumed. In this earlier work cell temperatures were taken to be close to the
values measured at the windows. However a calculated dissociation constant
Kp(T) derived from comparisons of measurements with different reservoir temp-
eratures but constant cell conditions implies a cell temperature of 228°C as
opposed to the value of 200cC stated in the report. Such a temperature dif-
ference would increase the reported value of 7.0 x 10"^ ppm m .
In reviewing the earlier atmospheric pressure measurement (3) with the
authors a possible cold spot was noted in the neck connecting the reservoir
to the cell. This decrease in the effective reservoir temperature could have
been as large as 6cC. In this case the measured absorption coefficient would
be raised to 13 x 10"4 Dpm~-rr_1.
Majkowski and co-workers^ have reported a peak absorption coefficient
of 17 x 10"4 pprrT-m 1 for I^SO^ band at 1220 cm"1. From the work of Burch
and co-workers) it appears that this band and the 880 cm"1 band have nearly
equal peak strengths. It must be noted that the present results on the 880
cm"1 band are low pressure measurements whereas results for the 1220 cm"1
band were obtained in a flowing stream of nitrogen at atmospheric pressure.
Nevertheless, for the broad h^SO^ bands considered the pressure broadening
effects at one atmosphere can not introduce large changes in peak absorption.
Majkowski^ ' previously reported the much smaller value of ^.16 x 10_l+ ppm"1
m-1 for the 1220 cm_1band. The jump in the value of the remeasurenent was
attributed to the elimination of hot spots and inclusion of excess dissocia-
tion corrections.
k2
-------
SECTION 8
FIELD TEST
SITE SELECTION
A requirement of this program was that a field test be conducted at an
electrical power generation plant burning high sulfur residual oil of
Venezuelan origin. Such fuels, which are characterized by a high vanadium
content, are used extensively in utilities in the eastern United States. The
presence of heavy metals in the fuel is thought to increase sulfate genera-
tion through catalytic oxidation of sulfur dioxide.
The test site selected is one of the larger electrical generation plants
in the northeastern United States. At the time of a site inspection in June,
the test unit was undergoing periodic maintenance which permitted the in-
stallation of a slotted pipe in the duct near the entry to the stack. The
650 MW unit is operated as a base load unit running continuously at or near
full output capacity. Fuels from 2-3 percent sulfur content are used, except
for periods when local air quality levels prohibit the use of high sulfur con-
tent oil. Experimental areas consisted of ^ x 8 foot iron platforms on either
side of the duct. These platforms were about 80 feet above ground and were
accessible via ladders from a catwalk which crossed the top of the duct.
Arrangements were made to shelter the electronics in a "covered bridge" sec-
tion of the overhead catwalk and to have riggers raise all equipment directly
from the ground to the platforn and catwalk.
To ensure adequate time for installation, alignment, and measurement,
the following test plan was prepared.
TEST PLAN
September 28
September 29
September 30
Installation and protection of equipment.
Prea1i gnment.
Contingency repair.
Completion of installation and alignment.
Measurements at normal operating conditions.
Measurements during excess air increase program.
Repeat of October 3's program.
Removal of equipment.
October 1
October 2
October 3
October ^
October 5
^3
-------
With a test area which was not protected fronn the weather, it was neces-
sary to allow for time lost due to inclement conditions.
The original test date was set for the week August \l~2k with contingency
plans for the following week. In the second week of August, the test unit
suffered a boiler tube failure. Uncertainty in the repair schedule required
postponement of the test until the first week of October.
TEST CONDITIONS
All of the equipment was installed by riggers and LAI personnel on
Friday, September 28. Plans to work over the weekend were aborted by a
steady rain. Clear weather on Monday permitted completion of the installa-
tion and preliminary alignment. During visual alignment, condensation on the
stack windows near the retrorefIector was discovered. This was apparently
the result of the combination of excess water leaking into the duct work and
breezes cooling the flange. The flange near the optical pallet suffered no
condensation problems probably because it was in a sheltered location and
therefore remained warmer. Tuesday morning the condensation problem was
eliminated by applying heating tape to the flange. Upon start up of the sys-
tem a component failure was discovered in one of the LCM units. This was
remedied by early afternoon. Sufficient time remained to conduct a startup
test of the instrument and to make a short duration measurement across the
stack.
The tests on October 3 proceeded on schedule until they were terminated
by a severe thunderstorm with heavy rain at 15:40 hours. The unit operated
at its normal level of 0.5 percent excess air until 10:30 when this was
raised to 1.0 percent. At 14:20 an increase to 1.5 percent was made and
maintained until local lightening activity forced a cessation of activity.
Altogether, over 5 hours of observations were made.
On the morning of October k, the test unit was found to be operating
above 0.7 percent excess air. This situation could not be corrected until
after 11:00 a.m. Two and one-half hours were required to reduce the excess
air and allow the unit to stabilize at the normal level of 0.5 percent. The
time lost ruled out the planned two-step air increase program. At 14:00 a
single step from 0.5 to 1.5 percent was executed. Over a 12 hour period the
monitor followed the H^SO^ production response to the increase and subsequent
decrease in excess air.
Equipment removal was completed on October 5 just before a second rain
began. In spite of wet weather conditions no part of the system suffered
discernable damage in the course of the test.
Throughout all of the tests an electrostatic precipitator was in opera-
tion. A fuel additive containing MgO was supplied continuously and was not
altered during excess air increase programs.
-------
RESULTS
Data collection on October 2 was limited because of the malfunction in
the LCM. The first tests were stability checks. Maximum drift was .017 ab-
sorbance units over a 36 minute period. These tests were performed with a
retroref1ector located before the stack. Several alignment iterations were
made to confirm the repeatability of the procedure. The first stack test
commenced at 17:22 and lasted about kO minutes. Results are shown in Figure
23. Although the software for the calculator permitted direct readout in
H2SOvapor concentration, testing was conducted with data output in absor-
bance units. This permitted better understanding of the magnitude of system
fluctuations and allowed evaluation of system response free from questions
concerning the exact value of the absorption coefficient. Interpretation of
the results also requires a measurement of the differential absorbance of the
stack windows. Prior to the field test this was measured in the laboratory
to be 0.36 + .0*t absorbance units. Following the field test the windows were
re-evaluated in the laboratory. No change attributable to surface deposits
could be determined. The uncertainty in the window differential absorption
is primarily attributable to the fact that the surfaces of each window are
not precisely parallel. This wedge cases small but detectable differences
between visible (HeNe laser) and infrared alignments.
Section 7 > the value of 12.2 x iu pp"1 m i=> lof.ch iui lug hcl u¦¦¦g¦en ~
tial absorption, (S^-S-), of I^SO^. Given the stack absorption pathlength of
5.5 m our results can Be interpreted using a scale of 1.5 Ppn per 0.01 ab-
sorbance units. The data in Figure 23 thus yields concentration ranging be-
tween 3 and 8 ppm.
October 3 measurements are presented in Figure 2b. At 10:30 a.m., de-
noted as point a, the excess air was raised from 0.5 to 1.0 percent. This
induced a rapid response followed by a steady rise until the excess air was
stepped to 1.5 percent at 1^:20; point b, on the plot. Less than 90 minutes
of data were obtained at the peak air level before the rain terminated the
run. All gaps in the plot represent periods during which the alignment and
normalization were being checked by putting the shorting retroref1ector in
place. The large downward spike at c was caused by personnel stepping onto
the platform to secure the instrument against the weather. Sulfuric acid
concentration results range from a starting level of 8 ppm to a final, maxi-
mum concentration of bO ppm.
Figure 25 is a plot of the excess air program recorded in the control
room for Unit No. 3. The trace showing the lesser values applies to the
section of the boiler which exhausted into the duct on which the test was
conducted.
Figure 26 gives the results of 12 hours of monitoring on October 4. The
corresponding excess air program is shown in Figure 27. The plot begins at
9:^ a.m. and initially reflects a reduction in excess air to bring the unit
to a base level of 0.5 percent. Particular features and events indicated
by the letters on the plot are identified as follows:
From the measurements on
^5
-------
;-t -. L H
S1IIVITY TIME HH.HM
-4.03 -01
1
0 . 0 0 - Q 3 5»2
-4.01 -o i
-3.97 -01
SI T I V1 T V
-4.02 -01
5.0y -03 PRINT IH7ERV
iL
-3.99 -01
( S E L 0 N L 0 ¦
-3.9S -01
fi-L SrfJ
SIT I VI TV 30
- 3. 9 6 - 0 1
i
fi, Pi fi - Z -4.12
-01
-3.96 -01
-4.11
"01
-3.97 -01
r; - } '¦ J.J
SIT I VI TV ~ 4 1 >"
-01
-3.95 -01
I
fi, fi M - ft 3 -4 , 1 1
-01
-3.97 -G1
-4,14
-61
-3.97 -01
-4. 12
"01
-3.99 -01
ft — hi U r h
SET -4.11
-01
-3.99 -SI
fi, ?i fi Pi -4.14
• -01
-4.09 -0i
-4. 1 1
-01
-3.9S -01
B-R OFF
p T — 4 . 1 d
-01
-3.95 -01
Rfi -4.3 1
-0 1
-2.97 -Q1
-4.0?
-01
- 3. 9 b - O l
Fi-L OFF
set "^-9?
"01
-3.92 -01
8 = O O 0 0 ~ 4. O 2
-01
-3.92 -01
- 4 « U j
-01
-3.8? -01
E-L OFF
SET ~4.02
-01
-3.91 -CI
0 . 0 0 0 O ~ 4 . U b
-0 1
-3.87 -01
-4, 0?
"0 1
-3.85 -0 1
-4.04
-01
-3.85 -01
-4.0 5
"01
-3.8 4 -01
MGRr- = i
.032 08 -4.01
-01
-3.85 -01
-4. 0b
-0 1
-3.82 -01
-4.64
O 1
-3.80 -01
R -- R S E11
SI 1 IViTV -4,06
~ t- i
-3.79 -01
£ , 5 fi - n-4.G7
-01
—3. / b - y 1.
-4.03
-01
¦ { O — 0 1
B - K E N
SIT!VI TV -4.0o
_ n i
-3.77 -Gi
2 . 5 0 - 0 3 - 4 . u 5
- 0 1
-3.7? —01
-'4. 0 4
-0 1
-3,7 5 -01
fi-L SEN
SIT I VI TV ~5*?§
- 0 i
1
X
0 .00 - 0 3 ~ ' -1 4'
-0 1
-3.87 -01
-4.85
-0 1
— -J a O j — 0 1
B-i SEN
SITIVITV -4.01
¦*
— o J
*
10.09-03 -4 . U 1
-0 1
-4.04
-0 1
- 3. 9 9
-01
-3.82 -0i
r. —, r. . .
-4. 04
-01
~-i.Bc -01
-4.00
-0 1
-3.86 -01
Figure 23-
Calculator printout of measurements on
Tuesday, October 2. In-
strunent sensitivity setting are given in the left
colurr.n and test results
are in the middle and right colunr.ns. Results given
in absorbance units.
ppnll-^SO^) =
150 (a.u. - 0.36)
^6
-------
0. 70 _
k2 -
0. 60 ..
30 -• aL
ppm -•
18 -¦
6 1
0. 50 ..
0. 40
Hours
Figure 2k. Measurement results for October 3. Arrows indicate
excess air increases. Blank areas are periods when the monitor
was in an internal test configuration. Time scale begins at
10 : 1 *4 hou rs .
kl
-------
Figure 25- Control room record of excess air levels, October
Units are excess air in percent. Left ^^trace applies to duct monitored
-------
0. 90
78 t
66
0. 80
5>i -¦
ppm
0. 70
aL
^2
0. 60
\/*V
30 -
0. 50 ..
T
18 "
6 1 0. 40
-I 1 h-
-H.ro m
Figure 26, Measurement results for October
Time scale begins at 09:44 hours.
-+-
—H
O)
CD
Q
in cd r-
Hours
rrows indicate excess air increase and decrease.
-------
Figure 27• Control room record of excess air levels, October
Left trace applies to duct nonitored. Units are excess air in percent.
50
-------
a. Untility officials inspected the optical pallet platform.
b. Detectors replenished with coolant.
c. Excess air increased from 0.5 to 1.5 percent at 14:00.
d. Detectors replenished.
e. Excess air reduced to nominal level.
f. Excess air over corrected.
g. Visitors on platform.
h. Detectors replenished.
The lowest concentration recorded before the excess air was stepped up-
ward was 25 ppm. Point d corresponding to 71 ppm was taken to be the best
representation of the upper level. The resulting differential change,
71 - 25 = 46 PDm should be compared to the corresponding change of 32 ppm
recorded the previous day.
The striking upward shift in the minimum H^SO^ level on October 4 (com-
pared to that on October 3) is attributed to the greater than normal level
of excess air encountered at the start of the day. Control room records re-
veal that the excess air had remained between 0.65 and 0.75 percent for over
eight hours prior to the October 4 test program. The integrated effect of
such a condition is to develop increased wall deposits on heat exchanger sur-
faces where catalytic oxidation of SO2 occurs.
Careful checks of the optical alignment and system normalization showed
that the monitor was functioning properly throughout the measurement period
on October 4. There was no indication of any instrumental condition which
would have given rise to a change in the minimum level.
SUMMARY OF RESULTS
Measurements made on October 2 and 3 indicated base values of sulfuric
and vapor concentration in the stack from 3 to 8 ppm. The minimum level of
25 ppm found on October 4 must be viewed in the perspective of the condition-
ing of the unit prior to the measurements. Measured changes of 32 and 46 ppm
in acid vapor concentrations recorded October 3 and 4, respectively, repre-
sent consistent results for similar changes in combustion conditions. These
measurements record the first continuous study of sulfuric acid concentra-
tions in a stack. They demonstrated the feasibility of the dual wavelength
differential absorption method to provide in situ, real time observations.
51
-------
REFERENCES
1. Burch, D. E., F. J. Gates, and N. Potter. Infrared Absorption by
Sulfuric Acid Vapor. EPA-200/2-76-191, U.S. Environmental Protection
Agency, Research Triangle Park, North Carolina, 1976. 27 pp.
2. Majkowski, R. F. Infrared Absorption Coefficient of H_S0. Vapor From
1190 to 1260 cm-'. J. Opt. Soc. Am., 67 (5): 62^-62/, T977
3. Eng, R. S., K. W. Nil!, and J. F. Butler, Spectral Measurements of
Gaseous Sulfuric Acid Using Tunable Diode Lasers. EPA-600/2-78-019,
U.S. Environmental Protection Agency, Research Triangle Park, North
Carolina, 1978. 6*4 pp.
-A -1
b. Eng, R. S., G. Petagna and K. W. Nill. Ultrahigh (10 cm ) Resolution
Study of the 8.2-um and 11.3-ym Bands of H^SO^: Accurate Determination
of Absorbance and Dissociation Constants. Applied Optics, 17 (11):
1723-1726, 1978
5. Majkowski, R. F., R. J. Blint, and J. C. Hill. Infrared Absorption
Coefficients of Gaseous H.SOi and S0^. Applied Optics, 17 (7):
975-977, 1978. 3
6. Mantz, A. W. , Feasibility Measurements of Low Pressure Modulation
Techniques for Frequency Selection. Laser Analytics, Inc.,
Internal Report, 1978.
7. Eng, R. S. A. W. Mantz, Investigation into Improved Sensitivity of
Laser Absorption Techniques for Pollution Monitoring. National
Science Foundation Grant ENV77-021, Semiannual Report 2/1/79-8/1/79
8. Todd, T. R., Mantz, A. W., Final Report: Model LS-2X Laser Source
Detector Test System, U.S. Army, CERCOM, Contract No. DAAB07-78-C-2^65,
1979.
9. Lutschinski, G. P., Zh.Fig. Khim. 30:1207, 1956
52
-------
APPENDIX
INTERFERENCE ABSORPTION SPECTRA
Figures A-l through A~9 are plots of calculated interference absorptions
from 1^0, CO2 and NO2 in the 880 and 960 cm"' regions. These plots are ex-
tensions of the calculated spectra given in Section From an inspection of
these figures it is clear that the number of spectral features resulting from
these three stack gas components is non negligible. This illustrates the
effectiveness of selective, high resolution minitoring techniques in minimiz-
ing sensitivity to stack gas interference as compared to techniques which
integrate over several wavenumbers.
53
-------
0. 06
0. 05 ..
0. 04
0. 02
0.01
in (o r- gdo>s—"(vaj't-in
Nr-r-t^r^mcpmrnaim
GDiDCD
-------
0.05 ..
0. 04 ..
m
(O
00
00
GD
-1
cm
Figure A-2. Calculated C0? background absorption spectra for 5 meter path at 523 K and ^73 K with a
pressure of 11'f Torr in the 880 cm"1 region.
-------
0. 06
0.05 ..
0. 04 ..
Figure A-3. Calculated ^0 background absorption
9&0 cm"' region.
spectra for 5 meter path at ^73"K, 91 Torr in the
-------
VJ1
•~vl
0.06
0. 05
0.04
0. 03
0. 02
0. 01
0. 00
in
in
O)
UPPER CURVE 523K
LOWER CCIRVt; 47 3K
(O
in
CD
in
O)
s
0)
a
CO
—f-
a
(O
en
cm
(D
O)
S
O)
-------
0. 07 _
0,06..
0. 04 ..
0. 03 ..
0. 02 ..
0. 01 ..
-I
cm
Figure A-5. Calculated N02 background absorption spectra for 5 meter path at A230K, 25 ppm concen
tration in the 880 cm ^ region.
-------
0. 07
0. 06 ..
a. 05 ..
B. 04
0. 23 ..
0. 02
0. 01
0. 00
w
0)
{N
0)
01
0)
in
0!
cn
in
0)
w
in
01
cm
Figure A-6. Calculated COj background absorption spectra for 5 meter path al 11k Torr i n the
960 cm"' region.
-------
0. 07 _
0. 06 ..
0. 05 ..
0. 24 ..
a
0. 03 ..
0. 02 ..
0. 01
0. 00
0)
in
0)
in
in
in
rH
in
-l
cm
Figure A~7. Calculated H2O background absorption spectra for 5 meter path at ^73°K, 91 Torr in the
950 cm"' region.
-------
0.07t
0. 05
i
a
0. 03
0. 02
0. 01
Figure A-8. Calculated H2O background absorption spectra for 5 meter path at ^73°K, 91 Torr in the
970 cm" region.
-------
0. 07 __
0. 05 ..
0, 04 ..
r.
0. 03 ..
0. 02 ..
0. 01 ..
ai
to
to
h>
•HI
fv
-1
cm
Figure A-g. Calculated CO2 background absorption spectra for 5 meter path at ^73°K, 1 I^ Torr in the
970 cm"' region.
-------
TECHNICAL REPORT DATA
(Please read /nurucrions on the reverse before completingj
1 . REPORT NO. 2.
EPA-600/2-80-1 74
3. RECIPIENT'S ACCESS! Ol* NO.
PB n
4. TJTLE AND SUSTITLE
A TUNABLE DIODE LASER STACK MONITOR FOR SULFURIC
ACID VAPOR
s"*^t5895f>Airf80 ISSUING DATE.
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
Edwin F. Pearson
Arlan W. Mant7
8. PERFORMING ORGANIZATION REPORT NO
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Laser Analytics, Inc.
Bedford, Massachusetts 01730
10. PROGRAM ELEMENT NO.
r07Ain/n(;-fmn (pv.nm
11. contrA:i763ant NO * '
68-02-2990
12. SPONSORING AGENCY NAME AND ADDRESS
Environmental Sciences Research Laboratory-"?- RTP, NC
Office of Research and Development
U.S. Environmental Protection Agency
Reserach Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
Final 9/78-3/80
14. SPONSORING AGENCY CODE
EPA/600/09
15. SUPPLEMENTARY NOTES
i®'ABST^ACfrield prototype instrument for continuous in-situ monitoring of sulfuric acid
vapor in industrial smoke stacks has been developed. The method of detection is dual
wavelength differential absorption in the infrared. Two tunable diode lasers are
locked to two specific frequencies that provide analytical and background absorption
information. The difference between the analytical and background absorption yields
the net absorption due to sulfuric acid.
Calibration of the monitor depends upon a precise determination of the differen-
tial absorption coefficient for the analytical and background frequencies. Measure-
ments made as a part of this program yield a value of 12.2x10"^ ppm"^ nH and tend to
resolve the discrepencies between previously published values. Temperature gradients
in the absorption cells severely influenced previous measurements.
The monitoring system was field tested at a 650 MW electrical generation plant
burning high sulfur oil of Venezuelan origin. The excess air supplied to the boiler
combustion chamber was varied from 0.5 to 1.5 percent. Measured baseline concentra-
tion of sulfuric acid ranged from 3 to 25 ppm. As the excess air was stepped up from
0.5 to 1.5 percent, the average measured increase in concentration was 39 ppm.
The system gave consistent results during the brief test period.
17. KEY WORDS AND DOCUMENT ANALYSIS
a DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
c. COSATI Kield/Gioup
Air pollution
* Sulfuric acid
* Vapors
* Flue gases
* Monitors
Infrared lasers
Field tests
Tunable diode lasers
13B
07B
07D
21B
20E
14B
18. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
19. SECURITY CLASS (ThisReport)
UNCLASSIFIED
21. NO. OF PAGES
71
20. SECURITY CLASS (This page)
UNCLASSIFIED
22. PRICE
-------