Energy & Fuels 1995,9, 467-478
467
FLASHCHAIN Theory for Rapid Coal Devolatilization Kinetics. 6. Predicting the Evolution of Fuel Nitrogen from Various Coals Stephen Niksa Molecular Physics Laboratory, S R I International, 333 Ravenswood Avenue, Menlo Park, California 94025 Received August 2, 1995. Revised Manuscript Received January 17, 1995@ The release of fuel nitrogen during the primary devolatilization of any bituminous coal involves only two mechanisms, the shuttling of nitrogen as an element in tar molecules and the conversion of char-N into HCN. This modeling study characterizes these processes for heating rates from 5 to lo4 Ws, temperatures to 1550 K, and pressures from vacuum to atmospheric. Evaluations against a database compiled from the behavior of 40 coals of rank from lignite through lowvolatile bituminous demonstrates that predicted evolution histories of tar-N, HCN, and char-N and the nitrogen contents of tar are within experimental uncertainty throughout this domain. Under conditions of rapid heating, tar shuttling is the only mechanism for nitrogen release as long as tar is being expelled, but for slower heating conditions it is overlapped by HCN production from char-N. With FLASHCHAIN, no additional parameters or rate expressions are required t o quantitatively predict the contributions from tar shuttling from any bituminous coal at any operating conditions, including the nitrogen contents of tar. The data evaluations also show that tar shuttling imparts the familiar dependences of tar evolution on coal rank, heating rate, and pressure onto the evolution of the nitrogen species distributions. Conversion of char-N into HCN is analogous to the release of H2 during the graphitization of char a t high temperatures. Only one reaction rate expression is needed to predict the yields and evolution rates of HCN during the primary devolatilization of any coal, provided that it is based on an extremely broad distribution of activation energies. The nominal rates of HCN production must also be reduced for coals of progressively higher rank in proportion t o their O N ratios.
Introduction The nitrogen in coal is responsible for major environmental problems. Pulverized coal flames in utility boilers generate NO, primarily by converting nitrogen incorporated into coals’ organic components, because conversion of the nitrogen in air is effectively inhibited by regulating flame temperatures. Of course NO, is implicated in acid rain and photochemical smog formation. And polynuclear aromatic compounds that contain nitrogen are among the most mutagenic species present in the products of coal combustion or liquefaction. A detailed mechanism for fuel nitrogen evolution from coal remains to be established, although it is already known to be far more complicated than the direct production of HCN that initiates fuel nitrogen conversion with most liquid and gaseous fossil fuels. Whenever coal is rapidly heated to high temperatures, fuel nitrogen is expelled in two or more stages. It is first liberated during primary devolatilization as an element in heavy, aromatic compounds collectively called tar.l Among succeeding stages, this one is the most sensitive to specific properties of different coal types. In the second stage, additional fuel nitrogen is expelled from char as HCN (and occasionally NH3) on time scales that are considerably longer than those for tar evolution. At the same time, volatiles in hot gases undergo secondary pyrolysis that converts some of the nitrogen in tar into HCN.2,3The remainder is incorporated into soot along @Abstractpublished in Advance ACS Abstracts, April 1, 1995. (1) Freihaut, J. D.; Zabielski, M. F.; Seery, D. J. Symp. (Int.) Combust. [Proc.], 1983, 19, 1159.
0887-0624/95/2509-0467$09.00/0
with the aromatic components of tar molecule^,^,^ counteracting the devolatilization of fuel nitrogen in the primary devolatilization stage. A third stage occurs if oxygen contacts the char and soot. Combustion liberates additional nitrogen either by direct chemical conversion to NO or by thermal dissociations induced by the higher particle temperatures associated with char combustion. Ultimately, the fuel nitrogen is partitioned into NO and N2 by homogeneous chemistry in flame zones and postflame gases. Even though this phenomenology is reasonably well established by laboratory observations, it has not yet been implemented in modeling. In virtually all models for nitrogen evolution from coal in the literature, HCN is released in proportion t o the total rate of weight loss, and the impact of coal rank and the operating conditions is reproduced by adjusting rate constants. In the most recent model,6 four separate reaction channels for the simultaneous production of HCN and NH3 supplement the shuttling of fuel nitrogen in tar. A total of nine rate parameters were tuned to fit the NH3 and HCN yields for pyrolysis of eight coals. Two factors are responsible for such coarse treatments. First, current elementary reaction models for nitrogen chemistry in the gas phase are incompatible with the chemistry of polynuclear (2) Bruinsma, 0. S. L.; Geertsma, R. S.; Bank, P.; Moulijn, J. A. Fuel 1988, 67, 334. (3)Nelson, P. F.: Bucklev, A. N.; Kellev, M. D. Srmp. - - (Int.) Combust. [Pro~.], 24, 1992, 1259. (4) Chen, J. C.; Niksa, S. Symp. (Int.) Combust. [Proc.], 24, 1992, 1269. (5) Chen, J. C.; Castagnoli, C.; Niksa, S. Energy Fuels 1992,6,264. (6) Bassilakis, Y.; Zhao, Y.; Solomon, P. R.; Serio, M. A. Energy Fuels 1993, 7, 710.
1995 American Chemical Society
468 Energy & Fuels, Vol. 9, No. 3, 1995
aromatics that contain nitrogen; and, second, it was not possible to predict the yields and evolution rates of tar within useful tolerances until very recently. The analysis in this paper circumvents the second factor by using FLASHCHAIN to describe the crucial partitioning of coal-N between the volatile species that contain nitrogen and the residual nitrogen in char (but all secondary volatiles chemistry is ignored). As demonstrated recently,' FLASHCHAIN is an ideal foundation for models of nitrogen evolution because the ultimate analysis is the only sample-specificinformation it requires to simulate the evolution rates and yields of tar and noncondensible gases from any coal at any operating condition^.^,^ Indeed, results in this paper demonstrate that, for the most part, reliable predictions of fuel nitrogen evolution during devolatilization rest upon an accurate model for tar evolution, simply because tar is the main shuttle for fuel nitrogen during primary devolatilization and also because tar evolution depends on the coal properties more than any other aspect of devolatilization behavior. The theory must be expanded in two modest ways t o predict nitrogen evolution. Only a simple accounting scheme without any new adjustable parameters or rate expressions is needed t o account for the nitrogen shuttled away in tar molecules. And only one simple reaction mechanism must be added to represent the evolution of HCN from char-N. These extensions will be evaluated against a database that represents heating rates from 5 to lo4 Ws, pressures from vacuum t o atmospheric, and temperatures to 1500 K. The database also features nitrogen species distributions and tar nitrogen contents from 40 different coal samples representing ranks from lignite to low-volatile bituminous.
Extensions to the Theory As explained elsewhere in more detail,1° FLASHCHAIN represents coals' cross-linked macromolecular structure as a mixture of chain fragments ranging in size from a monomer to the nominally infinite chain. The diverse assortment of structural components in real coals is rendered coarsely with four structural components: aromatic nuclei, labile bridges, char links, and peripheral groups. Aromatic nuclei are refractory units having the characteristics of the hypothetical aromatic cluster based on 13CNMR analysis. They also contain all the nitrogen in the coal, consistent with X P S characterizations that indicate a predominance of pyrolic and pyridinic nitrogen in all coal types, with appreciable quaternary nitrogen only in low-rank ~ o a l s . ~ J lExcept - ~ ~ for HCN production from their nitrogen, nuclei are immutable. Nuclei are interconnected by two types of linkages, labile bridges or char links. Labile bridges represent groups of aliphatic, alicyclic, and heteroatomic functionalities, not distinct chemical bonds. They contain all the oxygen, sulfur, and aliphatic carbon, but no aromatic components. Being refractory, char links are completely aromatic with no heteroatoms. Peripheral groups are the remnants of broken bridges having the same composition. (7) Niksa, S. Symp. (Int.) Combust. [Proc.], 25, in press. ( 8 ) Niksa, S. Energy Fuels 1994,8, 659. (9) Niksa, S. Energy Fuels 1994,8, 671. (10)Niksa, S.; Kerstein, A. R. Energy Fuels 1991,5,647. i l l ) Burchill, P.; Welch, L. S. Fuel 1989,68, 100. 112)Wallace, S.; Bartle, K. D.; Perry, D. L. Fuel 1989,68, 1450. (13)Kambara, S.; Takarada, T.; Yamamoto, Y.; Kato, K. Energy Fuels 1993,7 , 1013.
Niksa
Connectedness among nuclei is an important aspect of coal rank independent of chemical constitution. The initial distribution of fragment sizes in this model is empirically related to extract yields in pyridine because, qualitatively, fragment distributions skewed toward smaller sizes correspond to coals with substantial amounts of readily extractable material. Throughout devolatilization, fragments in the condensed phase disintegrate as bridges break and reintegrate as char links form, so that the amount of tar precursors, called metaplast, passes through a maximum value that depends on coal rank. Fragment statistics incorporate the chemical kinetic rates of bridge scission and recombination to describe the changing fragment size distributions, including that for tar precursors. Tar release rates are related t o metaplast concentrations with the flash distillation in which a phase equilibrium relates the instantaneous mole fractions of like fragments in the tar vapor and condensed phase. No finiterate mass transport phenomena are involved because all volatiles are presumed to escape in a convective flow that is initiated by the chemical production of noncondensible gases. In the model's four-step reaction mechanism, labile bridges are the key reaction centers because bridge conversion governs both tar and gas formation. Conversion of a bridge initiates two distinct reaction pathways, either to generate smaller fragments including precursors to tar or to form a new refractory char link accompanied by the immediate release of noncondensible gases. The pathway to char links depletes the bridge population without inducing fragmentation, thereby suppressing the production of tar precursors. Oxygen plays a distinctive role in both processes, promoting the bridge conversion rate and shifting the selectivity between scission and spontaneous char condensation toward the production of char links8 Of course the oxygen concentration in bridges is a strong function of rank. As an analog t o cross-linking, additional char links and gases may also form by bimolecular recombination between the ends of smaller fragments. This framework for primary devolatilization is adapted to describe fuel nitrogen evolution by accounting for two additional processes, (1)the release of nitrogen as an element in the aromatic nuclei in tar molecules and (2) the production of HCN from the nitrogen in aromatic nuclei throughout the condensed phase, collectively called char-N. Both processes are monitored in terms of the average moles of nitrogen per mole of nuclei in the condensed phase at any instant, denoted by v. Assuming that the nitrogen in coal is randomly distributed among nuclei regardless of the size and constitution of their host macromolecules, we first relate the initial value of 17 to the nitrogen mass fraction in the whole coal, N%(O),according to
where 70is the moles N/moles nuclei in the whole coal, the moles nuclei per volume in the whole coal, MWN the molecular weight of nitrogen, and eo the initial bulk density of the whole coal. This expression can be rearranged to define 70 in terms of several variables introduced in the original
A0
(14) Niksa, S. AIChE J . 1988,34,790.
Energy & Fuels, Vol. 9, No. 3, 1995 469
Rapid Coal Devolatilization Kinetics
. ....
00
0
Q3-O0
v
v v 0
00.0
v
0
on
0
since it is relegated to the aromatic nuclei in FLASHCHAIN, it does not participate in the bridge conversion mechanisms that govern both tar formation and the production of the major noncondensible gases. Nitrogen chemistry need not be resolved to explain nitrogen evolution because the dominant mechanism for nitrogen release during devolatilization is the release of tar molecules that contain nitrogen within some of their aromatic nuclei, called tar shuttling. The rate of nitrogen evolution by tar shuttling is directly proportional to the evolution rate of tarj-mers, rj. This rate determines how much of the coal's nitrogen is present in the cumulative tar sample released up to time t. More formally, a balance is written in terms of the fraction of the nitrogen in the original coal that has been released into the cumulative tar sample, and the instantaneous shuttling rate of nitrogen in the tarj-mers according to
0
.
Cm,
5 25
8 fm Q
20
0
15
5
c
t
10
.
0
J*
B 5
v0(Gm/dt)= r [ z r j l
fNTAR(0)= 0
(3)
j=l
0
65
70
75
80
as
90
95
Carbon Content, daf wt.% Figure 1. Nitrogen levels (daf wt %) and 70values (top) and O/N ratios (bottom) for the coal samples used in the laboratory Chen and Niksa18(VI; Freihaut studies of Kambara et al.13(0); et al.1 ( 0 ) ;Cai et aLl9 (W); and Freihaut et aLZ0(0).
formulation of FLASHCHAIN in part 1:lo
where
G(p,pb,pC)= 1
+ pbMW, + (p - pb+MWC i- (1 - P) MWA MWA
p ( 0 ) is the fraction of intact links among all chains in the coal, pb(0)the fraction of labile bridges among all chains in the coal, and p e ( 0 )the fraction of chain ends having peripheral groups in the coal, equal to unity. Note that 70is evaluated from the sample's nitrogen content and previously defined probabilities for intact linkages, labile bridges, and peripheral groups. Values for the 40 coals in the data evaluations in this paper appear in Figure 1 almg with the weight fractions of nitrogen in the coals. It is impossible to discern any rank dependence in either the nitrogen weight fractions or the 70 values. Coals of any rank tend to have nitrogen contents between 1and 2 wt %, although a very weak maximum may be evident for samples with carbon contents of about 85 wt %. The 70 values range from 0.2 to 0.45, showing even less dependence on rank than nitrogen content. This range corresponds to one nitrogen atom for every 2-5 aromatic nuclei. Considering that most coals have 2 or 3 aromatic rings per nucleus, on average, there is only one nitrogen atom in every 5-15 aromatic rings. Since nitrogen is such a sparse element on a molar basis, its decomposition chemistry cannot play an important role in the dominant mechanisms of devolatilization, depolymerization, and cross-linking. And
By definition, the rj are normalized by the coal's molar concentration of nuclei, Ao. The summation therefore compiles the fraction of the original coal nuclei being released as tar molecules at any instant. Since 7 is the average value for all chains in the condensed phase, it is applied to all the nuclei in all the escaping tar molecules regardless of the fragment size. The factors that affect the conversion of char-N into HCN are less evident in the available database, so we propose the following phenomenological mechanism. The average number of rings per nucleus becomes progressively larger for coals of higher rank,15 and throughout the devolatilization of any particular coal. One can reasonably expect that nitrogen in nuclei that have more rings to be more resistant to thermal decomposition. Consequently, the rate of HCN release should diminish as nuclei grow during devolatilization, and for coals of progressively higher rank. This rather complex process can be modeled with a single distribution of activation energies, because the magnitude of a distributed-energy rate decreases as the precursors associated with the lowest available activation energy are converted. In the context of HCN release, the precursors associated with the lowest energies should be recognized as the nitrogen atoms in the smallest nuclei. In FLASHCHAIN, HCN is produced from multiple independent first-order reactions whose activation energies are represented as a Gaussian distribution, according t o m
dYHcN/dt = ~ H C N ~ [ G ~YHCN(O) C ~ I =0
(4)
j=1
where YHCNis the molar yield of HCN. The xj are the moles of fragments of any size in the condensed phase. The summation therefore compiles the fraction of the original aromatic nuclei that are present in the condensed phase at any instant. The rate constant, kHCN, is the equivalent first-order rate constant that gives precisely the same thermal response as the activation energy distribution. As formulated elsewhere,16 k H c N (15)Solum, M. S.; Pugmire, R.J.;Grant, D. M.Energy Fuels 1989, 3 . 187.
(16)Niksa, S.;Lau, C.-W. Combust. Flame 1993,94, 293.
Niksa
470 Energy & Fuels, Vol. 9, No. 3, 1995
is defined as
The fraction of coal nitrogen converted to HCN is given by
where
CCN
and the mean energy and standard deviation in the distribution of energies for HCN production are denoted and OHCN, respectively. The same frequency as EHCN factor, AHCN, is applied t o all channels in the manifold. However, this parameter is diminished for coals of progressively higher rank. Its value is correlated with the O/N ratio of the whole coal which, as seen in Figure 1,falls monotonically with increasing rank. Oxygen has already been implicated in the production of NH3 from low-rank coals, via attack of HCN by the OH radi~a1.I~ Even if such a role for OH is validated someday, the correlation ofAHcN and O/N cannot be corroborated this way because virtually all the oxygen from any coal will be expelled by the time HCN begins to be released from char-N under conditions of rapid heating. The O/N ratio is nothing more than a convenient monotonic scale for the slower release of HCN from coals of progressively higher rank. The factor r appears in eqs 3 and 4. Its instantaneous value is determined from a conservation law that relates the time rate of change of the total moles of nitrogen in the condensed phase to the production rates of HCN and tar-N, according to
= (14/27)wHCN/N%(0)
(84
The instantaneous mass percentage of nitrogen in char is given by
N%c,
i--I
=
(84
WCmMWAG(p(0),p'( 0)p " ( 0 ) ) The coal nitrogen fraction in char is
,$
= N%c,RWc,R/N%(0)
(8e)
Finally, the molar concentrations of all condensed species are converted to a mass basis with the average molecular weight of monomers, MWMON. Since the weight of the nucleus is affected, on average, by the release of HCN, the molecular weights of both nuclei and monomers must be adjusted continuously, according to
(MWJ =-?zMWA TO
+
(MWA- 27)
(80
J*
Model Parameters Once the derivative on the left-hand side is expanded we have m
[%I,.j=1
dr dt
+-1;1
dsx, j=1
dt
m
- -kHCNr>xj j=1
J*
- rzrj j=1
Since the time evolution of the moles of nuclei in the condensed phase equals the loss of nuclei in all tar molecules, the conservation law for 7 reduces to a firstorder process:
Once eqs 3, 4, and 7 are solved for instantaneous values of fim,YHCN, and r , the more familiar variables for nitrogen evolution are defined as follows: The mass percentage of nitrogen in the cumulative tar sample is given by
where W T is~the tar yield expressed as fractional coal mass. The yield of HCN expressed as fractional coal mass is given by (17) Aho, M. J.; Hamalainen, J. P.; Tommavouri, J. L. Combust. Flame 1993,95, 22.
FLASHCHAIN'S extended submodel for coal constitutions drastically reduces the input requirements by incorporating regressions among elemental compositions and all other input data. The ultimate analysis now specifies initial values for all the structural parameters. Whereas most of the reactivity parameters are the same for all coal types, the handful that change for every sample are correlated with the elemental compositions of labile bridges, which are also evaluated from the ultimate analysis for the whole coal. None of the relations for either the structural or reactivity parameters have been adjusted to simulate nitrogen evolution, so the tar yields, tar-N, and weight loss in this paper are predictions in the most literal sense. The only parameters assigned to predict nitrogen evolution are those in KHCN. This rate was assigned by matching the predictions to data reported by Kambara et al. on the partitioning between volatile-N and char-N from 20 different coals (shown in Figure 10, be10wI.l~ The final correlation of AHCNagainst the O/N ratio in the whole coal sample appears in Figure 2. It is implemented with the same mean value and standard deviation in the activation energy distribution for HCN production for all coal types, 209 and 66.9 kJ/mol, respectively. The rate parameters for HCN production are remarkable in two respects. First, the energy distribution for HCN production is much broader than any assigned for bridge conversion in even the lowest rank coals, where bridge compositions are extremely heterogeneous because of the abundance of oxygen.
Rapid Coal Devolatilization Kinetics
8-
-
7-
-
B
I
T
-
0
1.0
-
I
1
I
1
I
,
O
c
,
,
,
,
,
,
,
! -
-
D
0 T O
0
-
-
6 2 - 0 0
Energy & Fuels, Vol. 9, No. 3, 1995 471
0
i
-
- 1 . .
5-
-
4-
I
0
5
3
I
$
1
1
10
8
I
,
I
/
,
15
,
,
, I , , , I I , , , , l , 20 25 30
OIN
Figure 2. Regression of the logarithm of the A factor in the rate constant for HCN production from char-N.
HCN is expelled over a temperature range that is broader than the ranges for any other product formation channel, in all likelihood, because physical and morphological factors determine its evolution rate. The second remarkable aspect in the HCN rate parameters is the rank dependence of AHCN,which decreases by 4 orders of magnitude for ranks from lignite to anthracite (cf. O N vs. %C in Figure 1). The curvature in this correlation is probably an indication of a shift in mechanism near O M = 10.
General Guidelines for the Data Evaluations Considering the technological impetus to lower NO, emissions from pulverized coal combustors, it is surprising that no one dataset in the literature resolves the major nitrogen conversion channels during primary devolatilization for a diverse assortment of coal types. The strategy behind this data evaluation is to isolate the contribution from tar shuttling before proceeding through several cases that compound this situation with substantial contributions from the conversion of char-N into HCN. Three laboratory studies monitored the fractional evolution of fuel nitrogen among tar, HCN, and char for coals across the rank spectrum. This group contains both of the only experimental studies that assigned closures on the nitrogen balances for individual runs,lJ8 as well as a very recent laboratory study of nitrogen evolution during primary devolatilization that examined a wide range of heating rates.lg A fourth study that also reports the closure of the nitrogen balances for 20 different coal samples but does not resolve the nitrogen in tar provided the basis for the correlation of AHcN with the O M of the whole c0a1s.l~A fifth study provided the data on nitrogen mass fractions in tar.20 The coalification diagram in Figure 3 is based on all the coal samples in these five studies, showing nearly complete coverage of the rank spectrum. Chen and Niksa used their radiant coal flow reactor operated in its mode that eliminates secondarypyrolysis of the volatiles to monitor nitrogen species distributions throughout primary devolatilization from 8 coals, emphasizing low volatility samples.18 Only cases with (18)Chen, J. C.; Niksa, S. Energy Fuels 1992,6, 254. (19) Cai, H.-Y.; Guell, A. J.;Dugwell, D. R.; Kandiyoti, R. Fuel 1993, 72, 321. (20) Freihaut, J.D.; Proscia, W. M.; Seery, D. J. Energy Fuels 1989, 3, 692.
0.4
0
0.1
0.2
0.3
OIC Figure 3. Coalification diagram of the samples used in the data evaluations, according to the symbol key defined in Figure 1.
reaction times just slightly longer than the period of tar evolution are considered, so this data isolates the impact of the tar shuttling mechanism; contributions from conversion of char-N are minimal, as seen below. This entrained flow furnace heats coal suspensions with radiant fluxes as large as 60 W/cm2 at atmospheric pressure for which calculated particle heating rates exceed lo4 Ws. Simulations of these data are based on uniform heating a t 1.5 x lo4 Ws to 1550 K with no isothermal reaction period, and quenching at 13 300 Ws. Cai et al.19 resolved heating rate effects in recent wiregrid experiments for rates from 5 t o 5000 Ws. Their samples were dispersed in a layer only a few particles deep. Thermometry for this system has been subjected to detailed characterization and is thought to be among the most reliable in the literature. The reported timetemperature histories and pressures were used directly in the simulations. In their atmospheric pyrolysis experiments, they imposed uniform heating rates from 5 to 5000 Ws to 1225 K, followed by a 5 s isothermal reaction period. The much earlier studies of vacuum pyrolysis on a wire grid by Freihaut et a1.l imposed nonuniform heating rates, variable reaction periods at intermediate temperatures, and then slow approaches to extended soaking at higher ultimate temperatures. Only the total reaction time was uniform at about 10 s in all cases. Ultimate temperatures in calibration runs used to assign the power programs range from 800 to almost 2000 K, but the ultimate temperatures with coal are much cooler. Actually, the entire recorded thermal histories with coal on the mesh heater are substantially different than the records for the calibration trials. Consequently, the ultimate reaction temperatures in the calibration trials are not imposed in the actual experiments, although, following Freihaut et al., the nitrogen evolution data will be plotted against these values. Thermocouple records of several operating conditions were reported for the stainless steel mesh used for temperatures to 1300 K (but not for the tungsten screens used at hotter temperatures). As illustrated in Figure 4 for three representative cases, close fits €or the simulations were developed with two stages of uniform heating and one or two variable isothermal reaction periods t o intermediate temperatures, and a uniform cooling rate. The segments in the simulated thermal histories for all the experimental cases are collected in
Niksa
472 Energy & Fuels, Vol. 9, No. 3, 1995
and
0
800
e
600 0.
400
200
5
0
10
15
Heating Time, s
Figure 4. Reported (dotted curves) and simulated (line segments)time-temperature histories for the vacuum pyrolysis experiments of Freihaut et al.' The simulated programs are specified by an initial heating rate (q-1) and isothermal reaction period (itp-1)followed by a second heating ramp (q2) and isothermal reaction period (itp-21,followed by cooling at a uniform rate (q-c), Cases c', d , and e' are for calibration temperatures of 1070, 1225, and 1375 K, respectively. Table 1. Operating Conditions for Several Cats Simulations case
T,,I,"C
a-1
Ti, "C
itp-1
a-2
Tf, "C
it^-2
a-c
a' a'
550 600 675 795 950 1005 1100
60 80 100 170 335 410 460
290 340 435 575 590 625 675
0.0 0.0 0.0 1.8 1.4 1.2 0.8
20 25 35 45 110 130 180
445 505 590 710
0.0 0.8 1.5 2.6 4.8 5.6 6.0
210 210 210 210 210 210 210
b' Ct
d e' f
have been added to predict fuel-nitrogen
personal microcomputer,although the complex thermal histories in Figure 3 took up t o 10 min to simulate.
?i
k I-
UHCN)
release. A simulation of each thermal history usually L required less than 3 min on a 20-MHz, 386-type
'*0°
870 915 1000
Table 1. Given such close fits to the recorded thermal histories over nearly all stages, uncertainties associated with the thermal histories are not inordinate. The simulations are based on a pressure of 0.015 MPa. Only the data that closes the nitrogen balances to within 10% are included. Kambara et al.13 monitored the nitrogen remaining in the chars from 20 coals after heating at a nominal rate of 1000 W s to temperatures to 1490 K in a Pyroprobe. Only the data for an ultimate temperature of 1490 K are used here. The reported thermal histories with coal on the heated ribbon show heating rates that diminish exponentially and temperatures that nearly reach an asymptotic isothermal value after 2 or 3 s. These experiments are simulated with an initial stage of heating at 1045 W s t o 1345 K immediately followed by uniform heating a t 55 Ws. The pressure is atmospheric. Freihaut et a1.20used an entrained flow reactor to recover tar samples and determine their nitrogen contents. The simulations for these cases are based on uniform heating at rates from 2000 to 6000 W s , depending on the ultimate temperature, followed a 0.1 s isothermal reaction period. In the forthcoming FLASHCHAIN simulations, the operating conditions of temperature, heating rate, time, and/or pressure were varied to match those in the experiments. The ultimate analysis of each coal sample is the only sample-specific information input to the simulations. All other structural and reactivity parameters were evaluated from the regressions presented in parts 1l0and 4.8 Only the three new parameters for HCN evolution reported earlier in Figure 2 (AHCN, EHCN,
Results The comparison with the case that highlights tar shuttling for very rapid heating conditions appears in Figure 5. The fractional nitrogen evolution data are plotted against the extent of primary devolatilization, evaluated as observed weight loss for progressively longer reaction times expressed as a fraction of the observed ultimate yield for this heating rate and pressure. The respective ultimate yields in order of increasing rank from these four coals are 52.5, 55.9, 58.4, and 31.4 daf wt %. This abscissa circumvents the need to relate the measured residence times to thermal histories, which were not measured. Two pairs of predictions are shown for tar-N and HCN. Both predicted solid curves are for the case in which fuel nitrogen is shuttled by the larger tar molecules but not by oils. (Oils are single-ring aromatics including benzene, toluene, xylene, phenol, and cresol.) As explained elsewhere,8 predicted tar yields from FLASHCHAIN also include the oils yields, because both products come from the coal's aromatic nuclei. The predicted nitrogen fraction in the combined products was apportioned to tar-N and HCN in proportion to the yields of oils and tar reported by Chen and Niksa.l' The pair of predictions indicated by the broken curves are for the combined amount of tar and oils, assuming that the nitrogen contents of oils and tar are the same. The HCN for this latter case comes solely from conversion of char-N. For this case of transient devolatilization during heatup at a rate exceeding lo4 Ws, the predicted evolution of char-N is within experimental uncertainty throughout primary devolatilization for all four coals. Predicted tar-N levels are also within experimental uncertainty for all four coals, provided that tar is the sole nitrogen shuttle and the oils contain no nitrogen. Recall that no additional parameters are involved in the predicted char-N or tar-N levels. Predicted HCN yields are within experimental uncertainty for all coals except the subbituminous (14881, again for the case with no nitrogen in the oils. It is interesting t o note that the predicted contribution of HCN that is proportional to the oils evolution rate very closely follows the observed evolution of HCN. Note that the first appearance of HCN coincides with the onset of discrepancies in predicted tar-N values based on nitrogen shuttling by both tar and oils. This feature suggests that oils are expelled as a tar decomposition product, perhaps with the simultaneous release of HCN. The amounts of HCN that accompany the production of oils are larger than yields from the direct conversion of char-N as long as tar is being released. The discrepancies between the predicted and observed HCN levels for the subbituminous coal are in all likelihood due to the production of NH3. Ammonia was not monitored in the experiments, but it is the most likely reason that the nitrogen balance could not be closed by the observed levels of tar-N, charN, and HCN for this coal only. Of course, the predicted level of HCN remains a reliable indication of the total amount of fuel-nitrogen among all noncondensible products because the predicted levels of char-N and tar-N are so accurate.
Rapid Coal Devolatilization Kinetics
Energy & Fuels, Vol. 9, No. 3, 1995 473
1.o C Q) 0 ) 0.8 E
e
f
$
0.6
0 .e
0
0 .c
0.4
0
E
0.2 0
1 .o
c Q)
0.8 c
2
2
-
0.6
0 0
-
.-0
0.4
0
2
0.2 0 0
20
40
60
1000
80
20
40
60
80
100
Extent of Primary Devolatillzation, % Extent of Primary Devolatillzallon, o/o Figure 5. An evaluation of FLASHCHAIN predictions against the char-N (01,tar-N (v),and HCN (W) observed by Chen and NiksaI8 from a subbituminous (1488), Illinois No. 6 (1493),Pittsburgh No. 8 (1451), and mv bituminous (1516) during primary devolatilization during heatup at more than lo4 Ws. The abscissa indicates the percentage of the ultimate weight loss for each of the observed residence times. Predictions shown as solid curves assume that nitrogen is shuttled by tar but not oils. The broken curves denote the predicted tar-N and HCN levels when tar and oils are assigned the same nitrogen content.
"
65
70
75
80
85
90
95
Carbon Content, daf wt Yo
Figure 6. An evaluation of the ultimate distributions of char-N (01, tar-N (71, and HCN (W) observed by Chen and Niksals for primary devolatilization of several coals after heatup a t more than lo4 Ws. Predictions are indicated as the respective open symbols connected by curves. They are for the case where fuel-N is shuttled by t a r but not oils.
A broader range of coal rank is evident in the comparison in Figure 6. These are reported and predicted nitrogen distributions a t the end of primary devolatilization following heatup in excess of lo4 Ws. Each case is for primary devolatilization during heatup at rates in excess of lo4 Ws t o the point where asymptotic weight loss was achieved, so there were no isothermal reaction periods. As seen elsewhere,s the predicted tar and gas yields for these experiments are within experimental uncertainty, except for one of the low-volatility coals. The predicted nitrogen distributions have tar as the sole shuttle for fuel nitrogen and nitrogen-free oils. Predicted char-N levels correctly
exhibit the minimum nitrogen retention observed for hvA bituminous coals and are within experimental uncertainty for all coal types. Again, the predicted tar-N levels are within experimental uncertainty for all coals except the sample with 89%C, correctly exhibiting a maximum for hv bituminous coals. The lone discrepancy is because the tar yield for this coal is badly underpredicted, at 16 vs 27 wt %. The HCN yields are also reliable except for the subbituminous coal. But predicted HCN levels are insensitive to coal rank for the short contact times in these experiments because the contributions from char-N conversion are so small. It also seems very likely that the discrepancy for the subbituminous coal can be attributed to NH3 production, which is ignored at this point in the model. Another perspective on tar-N is seen in the comparison between predicted and observed nitrogen mass percentages in Table 2. The data are from the entrained flow, atmospheric pyrolysis experiments of Freihaut et a1.20 The predicted tar-N contents are highest for the hvA bituminous and lowest for the subbituminous coal and also increase as the cumulative tar yields grow for progressively hotter reactor temperatures, in accord with the data. Predicted values are fairly accurate for the hvA and medium bituminous coals, but too large by 25-50% for the subbituminous coal. Although, here too, the database is very limited, the tendency for tar nitrogen contents well below those in the parent coals was reported for two other low-rank coals by these same authors.21 Perhaps under these conditions the abundance of oxygen in low-rank coals promotes NH3 production on a time scale even shorter than that for tar (21) Freihaut, J. D.; Proscia, W. M.; Mackie, J. C. Combust. Sci. Technol. 1993,93, 323.
474 Energy & Fuels, Vol. 9, No. 3, 1995
Niksa
Table 2. Tar Nitrogen Contents (dafwt 70) %N
coal sample
Tn K
observed
predicted
F1520
980
0.68 0.80 0.92 0.98 1.67 1.72 1.69 1.73 1.74 1.73 1.39 1.29 1.45 1.48
1.14 1.17 1.21 1.26 1.51 1.54 1.58 1.61 1.68 1.70 1.27 1.33 1.38 1.42
1100 1335 1515 980
F1451
1100 1215 1335 1435 1515 980
F1516
1100 1215 1335
i
Total
.a
/=
/ *
-1
+/,-* **. Utah Bit.
20
i -
' ' ' ' I ' ' ' ' I ' ' ' ' I ' ' ' ' 1 800 1000 1200 1400 1600
OW'
0.8
5
0.5
P)
5
.--
0.4
Calibration Temperature, K
I L
J
c
4 4
Vol-N
Figure 8. Comparisons between predicted and observed weight loss ( 0 )and tar yields (0) from a (top) subbituminous coal and (bottom) Utah hv bituminous coal during vacuum pyrolysis reported by Freihaut et a1.l Thermal histories used in the simulations for the various calibration temperatures are described in Table 1. The abscissa indicates the ultimate temperatures in calibration trials and values with coal are substantially lower.
coal a t 5-5000 Ws to 1225 K with a 5 s isothermal reaction period. The predicted total and tar yields are c 0 0.2 within experimental uncertainty for all heating rates. 0 0 The predicted volatile-N levels are slightly low at all 0.1 but the fastest heating rate. The tar-N values are within experimental uncertainty a t all heating rates O F ''llOD1l ' ' ~ ~ J except the fastest. However, this is probably not a flaw 1 10 102 103 1o4 in the predictions because 5000 W s is probably faster Heating Rate, Ws than coal particles can actually be heated in wire mesh Figure 7. An evaluation against the (top) weight loss ( 0 )and heaters. Note that the equal observed values for 1000 tar yields (0) and (bottom) volatile-N ( 0 )and tar-N (0) from and 5000 Ws are consistent with this rationale. Both an Illinois No. 6 coal for various heating rates to 1225 K with of the nitrogen predictions correctly grow for faster a 5 s hold period a t 0.1 MPa reported by Cai et al.19 heating rates, especially for tar-N. Tar-N increases in evolution during primary devolatilization, depleting the proportion to the tar yields which increase for faster low-rank tars of their nitrogen before they are released heating rates. But the loss of fuel nitrogen in tar diminishes the concentration of precursors to HCN, so from the coal matrix. Alternatively, low-rank coal tars are relatively unstable in hot gases, so these tars may the volatile-N levels are less sensitive to heating rate be decomposing preferentially a t their nitrogen sites, than the tar-N values. along the same lines suggested by the correspondence The next evaluations in Figures 8 and 9 are for nitrogen evolution under vacuum on much longer time between oils yields and HCN levels in Figure 5. Direct determinations of the complete N-species distributions, scales from four bituminous coals. In these experiincluding NH3, from several low-rank coals during ments, nominal heating rates were varied from 60 to 550 Ws depending on the ultimate temperature, alprimary devolatilization are needed to clarify the discrepancy in the predicted tar-N contents of low-rank though total reaction time is fixed at about 10 s for all coals. cases. The indicated temperatures are the calibration The tar shuttling mechanism in FLASHCHAIN acvalues which are significantly higher than the actual curately predicts the amounts of fuel nitrogen in tar for ultimate temperature with coal on the heater. Predicdiverse coal types, accounting for nearly all the nitrogen tions appear only to 1300 K because the thermal release during primary devolatilization at rapid heating histories for the runs at higher temperatures were not rates. We next examine cases with slower heating rates characterized in the published reports. The representative cases in Figure 8 demonstrate that total yields are and longer reaction times that bring in substantial accurately predicted at all temperatures but that the contributions from the conversion of char-N into HCN. The evaluation in Figure 7 resolves the total and tar predicted tar yields match the observed values only to the point where a maximum is observed. This discrepyields and the total fractional nitrogen release along with the fractional tar-N from heating an Illinois No. 6 ancy is a clear indication that tar was permitted to 0
E
0.3
Tar-N
' " l t l l l l
' ' 1 1 1 1 1 1 1
Energy & Fuels, Vol. 9, No. 3, 1995 475
Rapid Coal Devolatilization Kinetics
800
1000
1200
1400
16O(BOO
1000
1200
1400
1600
Calibration Temperature, K Calibration Temperature, K Figure 9. Comparisons between predicted and observed char-N (0 and solid curves),tar-N (V and dotted curve), and HCN (B and dashed curve) from a (a) subbituminous coal; (b) Utah hv bituminous; (c) Colorado hv bituminous; and (d) Pittsburgh hv bituminous during vacuum pyrolysis reported by Freihaut et a1.l Thermal histories used in the simulations for the various calibration temperatures are described in Table 1. The abscissa indicates the ultimate temperatures in calibration trials and
values with coai are substantially lower. decompose in this experiment. For this reason, the reported yields of HCN for temperatures higher than for the maximum tar yield were reduced by the difference between the maximum tar-N level and the observed tar-N level a t the hotter temperature, assuming that all nitrogen expelled during tar decomposition was released as HCN. This must be appropriate because only datasets that close the nitrogen balance to within 10% are included here. As seen in Figure 9, the predicted evolution of char-N is within experimental uncertainty for all coals except for the underprediction by 8% for the Pittsburgh No. 8 hvA bituminous. Predicted tar-N levels are within experimental uncertainty throughout for the subbituminous coal, and just 3% too high for the Utah bituminous. Discrepancies are somewhat larger for the Colorado bituminous. The asymptotic tar-N level for the Pittsburgh No. 8 is within experimental uncertainty, but the predictions are much too low for temperatures below 1100 K. However, the data indicate that tar evolution is faster for this coal than for all the other coals of lower rank, which is contradicted by an enormous d a t a b a ~ e . The ~ , ~ predicted HCN yields are within experimental uncertainty throughout, except for the underpredictions for the Colorado bituminous. It also appears that the observed values for temperatures above 1300 K lie along the trajectories of the predictions. Note also that these cases convey the mistaken impression that HCN and tar-N are generated simultaneously; in fact, the extended soaking periods at each temperature in these experiments provide time for substantial HCN production after the tar-N has reached its asymptotic value at every temperature. A parity plot for the dataset of 20 coals used t o define the regression of AHCNin Figure 2 appears in Figure
0
e"
0
I
I
, : '
,e'
0.2 F;
,.6
,
1
I
I
I
f
1
1
1
I
I
I
I
I
t
]
.
Energy & Fuels, Vol. 9, No. 3, 1995
Niksa
10
OF 1
' " ' f l t l '
10
'
' ' 1 1 1 1 1 '
' ' 1 " 1 1 1 '
102
I o3
' 'imLyJ 1o4
Heating Rate, Ws
Figure 11. An evaluation against the (top) weight loss ( 0 ) and tar yields (0) and (bottom) volatile-N ( 0 )and tar-N (0) from a mv bituminous coal (Tilmanstone) for various heating rates to 1225 K with a 5 s hold period a t 0.1 MPa reported by Cai et al.I9 Dashed curves are based on the regression value ofAHCN for this coal from Figure 2. Solid curves are for a value that is 350 times greater.
This database clearly indicates that HCN expelled by the conversion of char-N becomes slower for coals of progressively higher rank, as evident in the regression for AHCNwith O/N in Figure 2. This tendency is also consistent with reported HCN yields from char-N from the four coals in the study behind Figures 8 and 9, as explained e l ~ e w h e r e .But ~ there is one notable exception to this trend shown in Figure 11. Cai et al. subjected a medium volatile bituminous coal to the heating rate tests described in connection with an Illinois No. 6 coal in Figure 7. Clearly the predicted tar and total yields and tar-N levels are within experimental uncertainty for all thermal histories examined at atmospheric pressure. But the volatile-N levels based on the value of AHCNin Figure 2 for this coal are less than half the observed values. This data can only be reproduced with a value of A H Cthat ~ is a factor of 350 larger than the regression value of 3.4 x lo4. Considering that the regression in Figure 2 is corroborated by the behavior of 39 coal samples, it is hard to regard this case as any more than a highly improbable aberration. Discussion Nitrogen is a sparse element in coal, being present in fewer than a third of all aromatic nuclei and in only about 1 in 10 ring structures, on average. So nitrogen decomposition chemistry does not play any discernible role in the chemistry of devolatilization, particularly the bridge conversion chemistry responsible for the formation of tar precursors. While tar molecules are expelled during the initial stage of primary devolatilization, fuel nitrogen is simply shuttled into the vapor phase as one of their elements. The nitrogen released by this mechanism is just a random sample of the total population of fuel nitrogen atoms among all the nuclei in condensed phase species. No additional chemical reactions o r
physical transformations are needed to describe the shuttling of fuel nitrogen by tar. Although simple in the sense that nitrogen chemistry need not be understood to explain nitrogen release, this description had never before been implemented in modeling because tar evolution was impossible to predict within useful tolerances. Now that it has been demonstrated, it is clear that the tar shuttling mechanism imparts all the attributes of tar evolution onto the evolution of the nitrogen species distribution during devolatilization. The rank dependence of the total nitrogen release follows that of the tar yields by exhibiting a broad maximum for hv bituminous coals before falling off for low-volatility coals. Extents of nitrogen release increase for faster heating rates, especially for high-volatile bituminous coals. But the enhancement in the total nitrogen release is far smaller than the enhancement in tar-nitrogen, analogous to the way that tar yields increase for faster heating rates by much more than the enhancement in total weight loss. The impact of pressure on nitrogen release, tar yields, and total weight loss is also similar, as discussed el~ewhere.~ Additional support for the tar shuttling mechanism in FLASHCHAIN is seen in the way that tar nitrogen contents change in time throughout devolatilization. Tar nitrogen levels increase throughout primary devolatilization because tar increments added later to the cumulative tar product have fewer labile bridges, and hence a higher proportion of aromatic nuclei and hence a higher mass fraction of nitrogen. But the quaternary nitrogen functionalities in low-rank coals may be labile enough to decompose before tar is expelled at even the fastest heating rates of interest. More data is needed to clarify this aspect. Under conditions of rapid heating, tar shuttling is the only mechanism for nitrogen release as long as tar is being expelled. Considering the evaluations with Chen and Niksa's data (in Figure 41, it even appears that the earliest detectable amounts of HCN do not come from the decomposition of char-N. Rather, the HCN observed during the latest stages of tar evolution is probably expelled during the decomposition of tar into oils, not from precursors in nuclei in condensed-phase species. Certainly, the oils do not contain nitrogen. But beyond that, this connection is very hard t o validate because, after tar shuttling has been exhausted, HCN is almost always released simultaneously and vigorously from both decomposing tars and from char-N. The relative rate of these two processes is affected by a multitude of operating conditions, so obtaining conclusive evidence about which one occurs first will require an extraordinary level of control in the laboratory. For all coals except the lowest rank samples, fuel nitrogen evolution consists of only two stages, the (1) rapid initial release via tar shuttling followed by (2) HCN release from the decomposition of the nitrogen in the nuclei in char on longer time scales. Once the contribution from tar shuttling is accounted for and the observed levels of HCN are adjusted for any contributions from tar decomposition during secondary pyrolysis, the release of char-N can be recognized as a relatively simple process. The second stage is generally analogous to the second stage of primary devolatilization, when additional amounts of C1 and Cz hydrocarbons, CO, and HzO are released after tar evolution has ended.18 However, insofar as HCN continues to be expelled from
Rapid Coal Devolatilization Kinetics char over an enormous temperature range, the release of H2 is really the best analogy for HCN production. Release mechanisms for both these products probably involve the graphitization of aromatic nuclei into larger domains. The kinetics of this process have not yet been elucidated, so the distributed-energy rate expression incorporated into FLASHCHAIN for HCN production is only an expedient modeling tool, albeit a very effective one. The most distinctive features of char-N conversion are revealed most clearly in the observed growth in volatile-N for faster heating rates. It is impossible to represent this trend with an ordinary rate constant in a single first-order reaction regardless of the parameter values. With any single first-order reaction, volatile-N is predicted to diminish slightly as heating rates to the same ultimate temperature are increased, simply because there are fewer precursors t o HCN in the char whenever tar evolution is enhanced, as it is for faster heating rates. Since the first-order rate of HCN production would be proportional to the initial concentration of precursors, relatively less HCN is released when there are fewer precursors. In contrast, for HCN production with a distribution of activation energies, asymptotic yields of HCN are achieved at any sustained ultimate temperature, irrespective of heating rate. The same relative conversion is achieved for any initial precursor concentration. So with this mechanism the volatile-N, being the sum of tar-N and HCN, increases for faster heating rates only because the tar-N does. Consequently, the data in Figure 7 conclusively demonstrate that HCN production from char exhibits an extremely broad thermal response that requires a distribution of activation energies. The heating rate dependence also reveals that a kinetic competition apportions the fuel nitrogen between the tar shuttling and HCN production channels. Instead of resolving this competition, all previous models for fuel nitrogen evolution incorporate hypothetical ultimate yield parameters for HCN and NH3 which must be adjusted for different thermal histories, pressures, etc. FLASHCHAIN does not, yet it accurately predicts the partitioning between tar-N and HCN for diverse operating conditions with any coal type. Taken literally, the breadth of the activation energy distribution assigned for ~ H C Nindicates a diverse multitude of precursors for HCN in char-N. But we already know that virtually all the nitrogen in chars is present in the pyrrolic and pyridinic The exceptionally broad thermal response of HCN evolution is probably due to morphological effects acting across large domains of polynuclear aromatics. As aromatic nuclei graphitze into large domains, the precursors to HCN become more resistant to thermal decomposition. Even though the nitrogen functionalities are similar among all nuclei, the broad size distribution of the nuclei imparts a broad range of decomposition energies to the HCN precursors. Moreover, the similar thermal responses for the release of HCN and Ha suggest that graphitization may be the rate-determining process for both products. This interpretation also explains the tendency for slower HCN evolution from coals of progressively higher rank because their much larger aromatic nuclei would make HCN precursors more stable and also promote graphitization. Simultaneous determinations of the release rates of HCN and Hz over a broad temperature range
Energy & Fuels, Vol. 9, No. 3, 1995 477 would characterize this process in more detail. Ideally, such experiments would cover the release of all the nitrogen and hydrogen from diverse coal samples and their chars prepared under assorted thermal treatments. Whereas tar shuttling and char-N conversion are the only mechanisms needed to model nitrogen release from bituminous coals, ammonia is also released from lignites and subbituminous coals. The assigned parameter values for ~ H C Nfor these ranks accurately represent their relatively fast rates of conversion of char-N, and the predicted levels of HCN equal the sum of the observed levels of HCN and NH3. So modeling NH3 production during primary devolatilization does not necessarily entail a separate rate mechanism. Kambara et al.13recently reported that the fractional yield of N H 3 during pyrolysis equals the relative amount of quaternary nitrogen in a coal’s X P S spectra. Whenever such information is available, the proportions of HCN and NH3 can also be predicted with the nitrogen release mechanism in FLASHCHAIN by apportioning all of the quaternary nitrogen into an NH3 production channel.
Conclusions 1. The release of fuel nitrogen from any bituminous coal involves only two mechanisms, tar shuttling and the conversion of char-N into HCN. Tar shuttling and HCN production occur consecutively during primary devolatilization at the fastest heating rates, and the first HCN appears to be released when tars decompose into oils, rather than from char-N conversion. These two release mechanisms occur simultaneously a t heating rates slower than a few hundred kelvins per second. 2. Tar shuttling imparts all the dependences of tar evolution on coal rank, heating rate, and pressure onto the evolution of the nitrogen species distributions. With FLASHCHAIN, no additional parameters or rate expressions are required to quantitatively predict the contributions from tar shuttling from any coal a t any operating conditions. 3. Conversion of char-N into HCN is analogous to the release of Hz during the graphitization of char at high temperatures. Only one reaction rate expression (but no hypothetical ultimate yield parameter) is required to predict the yields and evolution rates of HCN during the primary devolatilization of any coal, provided that it is based on an extremely broad distribution of activation energies. Nominal rates of HCN production become slower for coals of progressively higher rank.
Nomenclature Ao
AHCN EHCN
$2 J* ~ H C N
MWA
molar concentration of aromatic nuclei in original coal pseudofrequency factor in the distributed-energy rate of HCN production mean energy in the Gaussian energy distribution for HCN production, f l E ) coal-nitrogen fraction retained in char coal-nitrogen fraction released as HCN coal-nitrogen fraction released as tar maximum degree of plymerization of metaplast and tar equivalent first-order rate constant for the distributed energy rate of HCN release defined in eq 5 molecular weight of an aromatic nucleus in original coal
Niksa
478 Energy & Fuels, Vol. 9, No. 3, 1995 instantaneous molecular weight of an aromatic nucleus throughout devolatilization molecular weight of a labile bridge molecular weight of a char link instantaneous molecular weight of a monomer molecular weight of nitrogen daf wt % nitrogen in original coal daf wt % nitrogen in char daf wt % nitrogen in cumulative tar sample instantaneous fraction of intact links among all chains in the condensed phase instantaneous fraction of labile bridges among the chains in the condensed phase instantaneous fraction of ends having peripheral groups among all chains in the condensed phase
WCWHCN WTAR Xj
YHCN
char yield, daf weight % HCN yield, daf weight % tar yield, daf weight % moles of any condensed phase j-mer normalized by Ao molar HCN yield normalized by A,
Greek Symbols molar release rate of tarj-mers r.i average moles of nitrogen per mole of aromatic rl nuclei in the condensed phase bulk density of the original coal eo in BE)for HCN production std dev about EHCN UHCN
EF9401561
