J. Quanr. Specwosc. Radial Printed in Great Britain
Transfer Vol. 33, No. 3, PP. 281-289,
SPECTRAL
1985
0022~4073/85 $3.00 + .oo 0 1985 Pergamon Press Ltd.
EMISSION FROM MHD COMBUSTION MIXED WITH PULVERIZED COAL
GAS
KAZUO ONDA, KIYONAMITAKANO and KEN KATO Energy Systems Division, Electrotechnical Laboratory, Umezono l-l-4, Sakura-mura, Niihari-gun, Ibaraki 305, Japan (Received 4 November 1983)
Abstract-The spectral emission from MHD combustion gas mixed with pulverized coal was measured between 0.38 and 2.7 pm. Twenty lines of K, eight lines of Li, Na, Rb, and Ca, and two molecular spectra of calcium compounds were observed. The experimental line shape for the 4P-4S transition of K agreed well with the Lorentzian shape combined with the measured farwing absorption cross sections. The population temperature of the K electronic state in the optically thin limit agreed well with the wing-reversal temperature. Continuous emission from coal-ash particles was not observed.
1. INTRODUCTION
The spectroscopic features of MHD combustion gas are required to estimate radiation heat transfer in a commercial MHD generation channel. Due to its high temperature and large optical thickness, the radiant heat flux will be comparable to or greater than the convective flux. Optical diagnostic techniques are frequently adopted for the measurement of important parameters in high-speed MHD combustion gas plasma. However, since no detailed spectroscopic study on MHD plasma has been reported, we measured and analyzed the spectroscopic features of MHD combustion gas mixed with pulverized coal. Because non-luminous emissions are well known, this study deals only with the specific spectra of coal MHD plasma. We first identified the spectra of the K seed and the components of coal ash in the wavelength region between 0.38 and 7.5 pm by placing these constituents individually into a diffusion flame of city gas and oxygen. Many strong emissions were observed in the wavelength region shorter than 1.6 pm. Based on this preliminary test, MHD combustion gas mixed with pulverized coal was measured between 0.38 and 2.7 pm in a test duct set at the combustor outlet. The gas temperature varies along the optical axis, which must be allowed for in the analysis. The temperature profile was estimated from data measured at low temperature and velocity. Accurate main-flow temperatures are also necessary in the analysis. Wingreversal temperatures’ were measured and compared with TE temperatures. 2. ANALYSIS
2.1. The radiative transfer equation and absorption coejicients The monochromatic emission intensity I at wavelength X is dI - = -a(X)(I - Zb), dy where a(h) is the absorption coefficient, I,, the blackbody emission intensity, and y the distance along the optical axis. The value of a(A) is obtained for LTE in atmospheric combustion gas.’ For a uniform temperature profile and small optical thickness, the integrated emission intensity J over the whole line is3
(2) where h is Plan&s constant, k Boltzmann’s constant, v = c/h, e and m, the electric charge and mass of the electron, respectively, c the velocity of light, to the perr&ttivity of 281
K. ONDA ef al.
282
Fig.
I. Energy levels and their observed spectral transitions
for neutral
K electronic
states4
free space, f the oscillator strength, 2y,., the emission length, g the statistical weight, E, the upper state energy of the transition, and T the temperature. Subscripts i, 1, and 0 identify the line center, the lower and the ground state of the transition, respectively. For emissions from the same emitter, Eq. (2) can be rearranged to
= _ -E,
(3)
kT’
>
The population temperatures are obtained by plotting In (J/Yi3fgo) vs E,/k. Many K lines were observed in the MHD combustion gas. Figure 1 shows the energy levels of neutral K;4 the observed transitions are shown by inclined solid lines. Figure 2 shows the energy levels of other atoms4 and their observed transitions. 2.2. Spectral line shape In terms of the absorption
cross section
S and the lower state density
nl, we define
a(X) = Srz,. For Lorentz
(4)
broadening S = eff12atom,cAv,/([2(v
where the collision
- vi)/Avt,]’ + 1 1,
(5)
+ (llm~)]}“‘.
(6)
half width is Avr_ = nFai’{(8kT/~)[(llmE)
‘s 1’
‘P:” i2
2PP +?
Rb
Fig. 2. Energy levels and observed
spectral
transitions
of electronic
state?
for neutral
Li, Na, Rb, and Ca.
Spectral emission from MHD combustion gas
283
In Eq. (6) nr is the number density of the combustion gas, at the optical collision cross section, m the mass of an atom or a molecule, and subscripts E and F stand for emitter and combustion gas, respectively. For high emitter densities, collision broadening between emitters cannot be neglected. The half width for self broadening’ Aur = nEaE2(16kTlrmE) 112
(7)
should be added to Eq. (6) to obtain the complete line shape of K. The far-wing spectrum of the 4P-4S transition of K cannot be explained by Lorentz broadening. 6,7Goodwin and Mitchner’ showed experimentally that S in the far wing may be approximated by d d S = jw - wilP= awp ’ where d and p are determined by gas composition, temperature, wavelength region, etc., w is the wave number, and Aw the wave number displacement from the line center. Lorentz broadening is valid only near the line center and the quasi-static theory’ can explain the far-wing spectrum. We also measured far-wing absorption cross sections. Taking S as a combination of Lorentz and measured far-wing cross sections, a(X) becomes a function of temperature and wavelength. For a given temperature profile, the line shape is calculated by integrating Eq. (1) and can then be compared with the experimental shape. 3. EXPERIMENTAL
RESULTS
AND
DISCUSSION
3.1. Test to identify the spectrum Spectral emissions from K seed and coal-ash components were first identified by placing these constituents individually into about a 2 cm diameter diffusion flame of city gas and oxygen. The optical arrangement shown in Fig. 3 was used for measurement. The photodetector was a photomultiplier, dry-ice cooled PbS or liquid-nitrogen cooled Au-Ge, depending on the wavelength region between 0.38 and 7.5 pm. Abundant elements in coal ash are Si, Al, Fe, Ca, Mg, Na, and S. Water solutions of nitrates (Al, Fe, Ca, Mg), water glass (Na20 - nSi02 - mH20), sulfuric acid and K seed were individually atomized in the flame. The spectra of ash elements and K seed were distinguishable from the background flame emission. Many strong emissions were observed from alkali-metal atoms, alkaline-earth atoms and their molecules in the wavelength region shorter than 1.6 pm.
cl 44
Pressurs
Fig. 3. Schematic diagram of the experimental arrangement.
284
K. ONDA et al. Table
I. Proximate
and ultimate
analyses
Proximate analysis
Moisture (mass %)
4.09
Ash (mass %)
15.65
Volatile (mass %)
47.14
Fixed carbon (mass %)
33.12
High beating 26.02 value (M.J/kg)
of Taiheiyo
coal (as received)
Ultimate analysis (mass X) Carbon
5.73
Nitrogen
0.83
Oxygen
19.04
Phosphorus
0.05
Sulfur(bum)
0.09
Sulfur (unbum)
0.12
Ash
3.2. Experimental
58.76
Hydrogen
15.38
conditions
Spectral emissions were measured in the test facility shown in Fig. 3, where LPG and pulverized coal were burned with oxygen-enriched air, K seed and SOZ. In a coal MHD power-generation system, about 80% of the coal ash is rejected in a slag-tap combustor, with the remaining 20% draining into a generation channel. The coal flow rate in our tests was adjusted to yield a similar level of remaining ash content and no attempt was made to reject the slag in the combustor. Proximate and ultimate analyses of the coal are shown in Table 1, together with the measured heating value. Table 2 shows mineral analyses of the coal ash and of the slag deposited in the test duct. The test duct was a water-cooled peg-wall channel, with a constant cross section of 30 X 34 mm2. Spectral emissions were measured from the LPG combustion gas, from the seedstomized combustion gas and from the coal-mixed combustion gas to discriminate the specific spectra from seed and coal ash. Thermal input Qin and total mass flow G were kept nearly constant during the tests, as is shown in Table 3, where 4, and $k are, respectively, the mass fractions of ash and K, C#J rlo the stoichiometric ratio of fuel to 02, &,,O the mass fraction of supplied N2 to 02, C$ s,k the mole fraction of supplied S to IS, Qi,,, the heat loss at the combustor and nozzle, and T, the optical pyrometer reading on the surface with seed or slag deposits. Because a temperature profile along the optical axis is required for calculating line shape, we measured velocity and temperature profiles of low-temperature combustion gas in another duct. The hot junction of a 0.5 mm wire Pt(20/40)-Rh(80/60) thermocouple was set along the isothermal plane and orthogonal to the flow. A 0.4 mm tap in a watercooled total-pressure tube of 2 mm diameter was set orthogonal to the flow. Measured profiles of the velocity u/u, and the temperature (T - T,)/(T, - TJ are shown in Fig.
Table 2. Mineral
analyses
of Taiheiyo Coal
Composition
ash (mass %)
coal ash and of the deposited Duct deposiL ash (mass %)
SiOz
54.45
31.0
A1103
25.65
21.6
Fez03
4.59
cao
8.10
4.83
MgO
2.02
2.15
Na>O
0.88
KzO
0.78
2.84
0.36 26.6
SO6
2.73
7.06
MIIO
0.08
0.05
P205
0.73
0.09
unburnt L
1.42
slag
Spectral emission from MHD combustion gas
285
Table 3. Combustion conditions and comparisons of experimental and analytical gas temperatures and K number densities Run #
CMrY16 LPG
Fuel
CM&7 LPG
+coa1
CMd18
LPG +coa1
LPG
LPG +coa1
LPG
G (g/s)
38.11
37.48
38.25
37.68
38.12
37.50
aa
(%)
0.492
0
0.494
0
0.496
0
$
(%)
1.01
1.00
1.00
0.99
1.02
1.02
1.05
1.05
1.05
1.05
1.06
1.05
0.201
0.200
0.200
0.199
0.200
0.200
0.503
0.503
0.506
0.508
0.496
0.493
'F/O @N/O @S/K
Qi,
(kw)
353.9
351.8
355.1
354.5
355.8
351.2
Qloss(k")
124.9
127.0
122.3
123.7
123.1
124.0
T, (K)
1750
1390
1750
1390
1750
1390
R ex hm (nm)
1.73xX?
1.73x10"
1.74x10'
1.76~10"
1.73x10'
1.73x10s
763.2
763.2
1179.0
1179.0
763.2
763.2
T
2572.5
2574.0
2557.0
2594.0
2574.0
2566.5
T;:I :::
2518.8
2574.7
2533.5
2598.4
2537.7
2586.3
AT/T~ (%)
2.13
-0.027
-0.77
7.42~10~
-0.17 _
1.43
8.16x]gY
0.93 -
a19xioY
8.6OXlou
nKem
(cc-l)
nKLcW1)
a7ixxY
8.57Xlf
-
An/nc (X)
-6.3
-13.4
-
_
8.72xlff
8.51x1@'
-
-6.1
1.1
4, together with calculated laminar and turbulent boundary layers assumed to develop from the duct inlet. The measured velocity and temperature profiles nearly coincide and are approximated by the turbulent boundary layer for the l/7 power law. Referring to the spectrum-identifying test, the spectra of MHD combustion gas were measured between 0.38 and 2.7 pm. The relative sensitivity of the detector system and the transmittance of the light source system were calibrated with a standard lamp. Measured flame emissions were converted to emissivities eA relative to the blackbody emission of the main-flow temperature Tgk by utilizing the reference lamp intensity. The spectral resolution of the monochromator was 0.22-0.16 nm for wavelengths between 0.38 and 0.9 pm; it was 2-1.4 nm between 0.9 and 2.7 pm. 3.3. The potassium 4P-4S spectrum Table 4 shows typical spectra for a test. Because of the low spectral resolution, some lines are not separated. These emissivities are shown at the middle of the two possible transitions. The strong 4P-4s spectrum of K is extensively broadened due to the high K densities. Taking account of the small optical thickness and cold boundary layers, S for the far wing is approximated by S = ~J2y,n,.
Distance
Fig. 4. Nondimensional
from
center
(9)
of
flow
profiles of gas velocities and temperatures.
286
K.
ONDA
et al.
Table 4. Emissivities, oscillator strengths, and optical collision cross sections for the observed spectra
The circles in Fig. 5 show cross sections for wings of the 4P-4S transition. Our data lie between those of Vasil’eva et ~1.~ and Goodwin and Mitchner’: none of the data can be explained by Lorentz broadening. Assuming d in Eq. (8) is proportional to the oscillator strength and p is the same in the violet wings and in the red wings for the doublet, d and p are determined as in Table 5. When S is represented by Lorentz broadening combined with the measured far-wing cross section, numerical integration of Eq. (1) yields the solid line in Fig. 6. The measured intensities are shown by circles and agree well with the numerical results. The Lorentzian line shape is shown by the dotted line in Fig. 6, and does not correspond to the measured intensity, except in the vicinity of the line center. The 5P-5S line of Rb and the 6S-4P line of K lie on the broad wings of the 4P-4S line of K (see Fig. 6).
Spectral emission from MHD combustion gas
Wave
number
dltferenca
w-w,
hi’)
Wow
number
dithrmce
287
w2-w
IcnT’I
Fig. 5. Far-wing absorption cross section for the 4P-4S transition of K.
The main-flow temperature r,, was measured by the wing-reversal method’ and was used to calculate the line shape. The method was modified to take into account the farwing cross section, with great improvement in the accuracy of the main-flow K density nKS. The measured temperature T_, and K density nk~,,, are shown in Table 3; T,,, is the calculated temperature based on the combustion-gas table and the main-flow enthalpy, and nK&c the TE density corresponding to TgLc. These TE values agreed well with the experimental data. In Table 3, Rex is the Reynolds number at a measurement position x from the duct inlet and X,,, the wavelength, while AT/T, and An/n, are ratios of measurement errors to calculated values. In one test the wing of the 3D5,2-4P3,2 line of K was used to measure the gas temperature. We could not correct this wing-reversal temperature because of the unknown optical collision cross section. The error caused by self-resonance absorption was estimated to be small because the lower state density for this transition is much smaller than the most populated ground-state density. The measured temperature without any correction, which is shown in Table 3, agrees well with the TE temperature. 3.4. Other spectra Published data of oscillator strengths8-I0 and optical collision cross sections” are shown in Table 4 for the observed spectra. With inadequate spectral resolution, the measured line shape is not accurate. Nevertheless, the integrated absorption coefficient can still be measured fairly accurately. ‘* We therefore obtained a population temperature for the K electronic states by plotting the logarithms of the integrated intensities in the optically thin limit against the upper energy levels of the transitions [cf. Eq. (3)]. The population temperature obtained from Fig. 7 agreed well with the wing-reversal temperature.
Table 5. Constants in Eq. (8) for the far-wing absorption cross section (cm*) Transition
Red or violet wing
4P w/z-4s,,=
Violet wing
Red win'g
4PL,2 -4s I,2
Violet wing
Red wing
Wave number region(cni')
d
P
Aw 5 500
1.103~10-"
1.202
Aw > 500
l.070~lw'z
2.303
Aw 5 500
2.142xlC"'
1.250
Aw > 500
2.217x10-=
2.355
Aw < 500
5.483~ IO-'
1.202
Aw > 500
5.319x10-'s 2.303
Aw 5 500
1.065~10"~
Aw > 500
1.102~10-'* 2.355
1.250
K.
288
ONDAet al.
lLPG*coall Tw
I o-1
=1750K
3
,
.-
w-w,=500cm-1
4 _~,_L_p-
740
760
of experimental
and analytical
~~_
780
Wavelength
Fig. 6. Comparisons
_
resolution
w-w,=lOOcm-1
4
720
Spectral
(nm
800
820
I
emission
intensities
CMC#l8
1
(LPGtcool)
-
for the 4P-4S
transition
of K.
IO"2.5x IO4 Upper
Fig. 7. Population
3.0x IO4
energy
temperatures
level
3 5x104 Eu (cm-‘)
for K electronic
states
Ten lines belong to emitters other than K. A Li line was observed from the combustion gas mixed with pulverized coal. Composition analyses showed that the coal ash contained 0.003 mass % of Li. Both Na and Rb were contained in the seed and the coal ash. The aqueous solution of KOH used as the seed was found to contain 0.20 mass % of Na and 0.00147 mass % of Rb, which are 1.83 and 2.54 times the amounts in the coal ash. One line of Ca atom and two band spectra of Ca compounds were observed from the combustion gas mixed with pulverized coal. The two band spectra in Table 4, where Xi is the wavelength for the peak intensity and w is the width at half maximum, probably belong to CaOH and Caz02.‘3 We did not observe continuous spectra from ash particles. Ariessohn et al. I4 measured Sauter diameters and volume densities of ash particles under similar conditions. Assuming that their results can be applied to our condition, the emissivities of ash particles at 325 nm and 3.39 pm are estimated to be 0.008 and 0.003, respectively. These small emissivities were hardly distinguished from the zero drift in our measurement system. Acknowledgments-This for MHD power suggestions.
work was supported generation. The authors
by MIT1 under the National thank M. Kubota, National
Research Institute
and Development Program of Chemistry. for helpful
REFERENCES 1. K. Onda, Y. Kaga and K. Kato, JQSRT 26, I47 (1981). 2. J. R. Greig, Br. J. Appl. Pbys. 16, 957 (1965). 3. S. S. Penner, Quantitative Molecular Spectroscopy and Gas Emissivities. p. 475. Addison-Wesley. Massachusetts (1959).
Reading,
Spectral emission from MHD combustion gas
289
4. E. M. Charlotte, “Atomic Energy Levels,” Vols. I, II, NSRDS-NBS 35, National Bureau of Standards, Washington, D.C. 20402 (1971). 5. T. Hiramoto, J. Phys. Sot. Jpn 26, 785 (1969). 6. I. A. Vasil’eva, L. V. Deputatova and A. P. Nefedov, Proc. 2nd U.S.S.R.-U.S.A. Colloquium on MHD Electrical Power Generation,p. 231. NTIS, U.S. Department of Commerce, Springlield, VA (1975). 7. D.G. Goodwin and M. Mitchner, Proc. 20th Symp. on Engng Aspects of MHD. p. 8.4. Redondo Beach, California (1982). 8. H. R. Griem, Plasma Spectroscopy,p. 45. McGraw-Hill, New York (1964). 9. 0. S. Heavens, J. Opf. Sot. Am. 51, 1058 (1961). IO. R. C. Weast, CRC Handbook of Chemistry and Physics, p. E-334. CRC Press, Boca Raton, Florida (1982). 11. F. W. Hofmann and H. Kohn, J. Opt. Sot. Am. 51,512 (1961). 12. H. J. Kostkowski and A. M. Bass, J. Opt. Sot. Am. 46, 1060 (1956). 13. R. W. B. Pearse and A. G. Gaydon, The Identificationsof Molecular Spectra, p. 130. Chapman & Hall, London (1976). 14. P. C. Ariessohn, R. H. Eustic and S. A. Self, Proc. 7th Int. Co@ on MHD Electrical Power Generation, p. 807. Symposia on the Energy Aspects of MHD, Cambridge, Massachusetts (1980).
QSRT
33:3-G