Robert H. Essenhigh

Ignition of coal particles: a review

Abstract This review of ignition concepts and mechanisms is set in the context of the historical development of the subject, to explain the reason for the slow development of validated ignition theory until quite recently, but leading to the present theoretical developments and supporting experiments for ignition of a single particle. The principal focus is on the heterogeneous ignition concepts as these are most fully developed at this time on the basis of Thermal Explosion Theory (TET), but the requirements of a homogeneous ignition analysis are also discussed, with particular attention given to the well known branching radical mechanism in a generalized formulation. Extension of the single particle results to batch clouds and flames is examined, with theoretical results for clouds now available but awaiting verification. The problem of a standing flame in a continuous flow is seen as a problem of a different order. Overall, however, work of the last two decades has developed quantitative and analytical understanding of particle ignition that has superseded the largely postulational views held previously. This is specifically the case with the long-standing problem of heterogeneous vis a vis homogeneous ignition. Nevertheless, extension of that largely single-particle work to large dispersions is now required for credibility of such research in relation to engineering practice.

Burning of forest materials under late Paleozoic high atmospheric oxygen levels

Theoretical models suggest that atmospheric oxygen reached concentrations as high as 35% O2 during the past 550 m.y. Previous burning experiments using strips of paper have challenged this idea, concluding that ancient wildfires would have decimated plant life if O2 significantly exceeded its present level of 21%. New thermochemistry and flame-spread experiments using natural fuels contradict these results and indicate that sustained burning of forest fuels at moisture contents common to living plants does not occur between 21% and 35% O2. Therefore, the fires under atmospheres with high oxygen concentrations would not have prevented the persistence of plant communities. Times of high O2 also agree with observations of concurrent fire-resistant plant morphology, large insects, and high concentrations of fossil charcoal.

Mapping and structure of inverse diffusion flames of methane

Six different flame types have been identified in a mapping of inverse methane diffusion flames. Only Types I to III were stable. Types II and III were similar and most “representative” of the IDFs; they also covered more than 50% of the total flame map. All but Type IV were underventilated but, yet, were essentially non-luminous with no carbon formation. Only the Types II and III showed any luminous region, as an orange-yellow cap on top of a blue cone. For a single Type III flame, radial and axial temperature and species concentration profiles were measured and found to agree well with prediction, except at the top of the flame where a pool of high CO and H2 was found and in the central air jet with 1/2% H2 that diffused in. The Burke-Schumann flame surface with instantaneous reaction was used as the physical basis for the model. The mathematical model is an extension of the Gosman et al13 analysis, covering continuity, momentum, energy, and 5 species (CH4, O2, N2, CO2, H2O). The solution methods developed by Gosman were used. The predicted flame profile was also in reasonable agreement with the experimental profile determined by the peak temperature profile and direct observation. All experimental data for species and temperature from all points in the flame were reduced to common curves by plotting against the local equivalence ratio. With corrections for analysis base, the curves for inverse diffusion data are almost identical quantitatively to the Mitchell data. This provides direct confirmation of the reasonable assumption that there is no distinction mathematically between the normal and inverse diffusion flames.

Determination of global kinetics of coal volatiles combustion

Values of global kinetic constants for the combustion of coal volatiles have been determined for 14 coals ranging from 40 to 80% VM content (bituminous to lignite). The combustion rates of the volatiles from the different coals differed somewhat, but not markedly, and without any evident dependence on coal rank. Activation energies (E) were mostly in the range 9 to 14 kcal/mol, but higher and lower values were evidently the result of statistical scatter in the determinations, as shown by some double and triple repeats. In addition, the values of log( k 0 ) and E in 22 pairs of measurements showed a high auto-correlation, supporting the conclusion that the variations between the different volatiles were not significant at the present level of measurement. The volatiles were pyrolysed from crushed coal pulled on trays through a muffle furnace, and were burned at the top of a vertical combustion tube in an intense back-mixed region treated theoretically as a stirred reactor using the usual equations. The total data base on which the kinetic determinations were based was in excess of 5000 measurements, yielding over 600 values of the velocity constants (k) from which the total of 22 paired values of activation energy and frequency factor were obtained for the 14 coals.

Influence of pressure on the combustion rate of carbon

The influence of pressure on the combustion rates of carbon (or coal) particles is shown, by comparison of prediction with experiment, to be zero to minor in the temperature range studied. This result is contrary to the empirical ( n th order) assumption widely adopted in much of the literature that predicts a substantial pressure dependence at all temperatures. Two models were used in the comparison, and the results were compared with three independent experimental sets of data. These experiments were measurements of burning times of single coal particles by Tidona [19] at 1, 1.5, and 2 atm; reaction rates of char particles by Monson et al. [15] at 1, 5, 10, and 15 atm; and (noncritical) ignition temperatures of coal particles in the pressure range 0.4–1.7 atm [20]. The first model was based on the fundamental Langmuir-Nusselt-Thiele suite of theoretical equations in the form of the extended resistance equation (ERE) [35]. The second model combined the Nusselt BLD analysis with the empirical n th order assumption that the reaction rate at all temperatures is proportional to the n th power of the partial pressure of the oxygen concentration ( p ox n ) [2,3]. The ERE model was able to predict the structural form of the experimental results with adequate prediction of numerical values, particularly of the reaction rates measured by Monson et al. In particular, the ERE predictions and experiments jointly showed small to no dependency of the rates on pressure, contrary to the predictions of the empirical model. We conclude that the empirical model has no experimental support for the assumptions made and that fundamentally based equations can be developed or already exist that can be used to predict carbon combustion reaction rates at elevated or reduced pressure with acceptable confidence.

Influence of initial particle density on the reaction mode of porous carbon particles

In combustion of porous carbon particles, the relation between particle density and diameter: σσ0 = (dd0)α; was obtained in an earlier analysis [1]. This present work extends that earlier analysis to show with experimental support that (1) the power index, α, varies with the initial char density, σ0, according to α = α0(1 − σ0σs), where σs is the He density, and (2) that the limit value of α at zero density, α0, is inversely proportional to the square root of the velocity constant of the reaction. Available experimental data values for α0 yielded an effective value of activation energy, E, of 23 kcal/mol, and 46 kcal/mol after correction for Zone II behavior in agreement with earlier analysis of the same data by a different method [4]. For values of α, supporting the linear decline of α with σ0, the experimental data base was a set of 37 values obtained from 15 literature sources. The set showed a surprisingly small range in α values, from 0 to 7, and with most in the range 1–3, in spite of the different methods of experiment, char types, origins, particle sizes, and reaction temperatures. This suggests that, while Zone II reaction may be common, the depth of penetration of the reaction in Zone II combustion may be quite shallow. It also supports the conclusion that intrinsic reactivity from different char sources may be very similar and that differences in combustion rate may be determined less by chemical reactivity differences and more by physical access to the surface and interior of the particle, with access to the interior largely determined by initial char density, σ0, at the start of char combustion.

An integration path for the carbon-oxygen reaction with internal reaction

An analytical extension of the Thiele analysis for reaction in a porous particle relates the variation of density and diameter during reaction by the power index equation: σ/σ o =( d/d o ) α . This provides specification of a path that permits integration of the inexact DE for combustion of a particle when there is also internal reaction. Re-formulation of the particle rate equation along the σ- d path for a given α leads to a new “Extended Resistance” Equation (ERE) that combines the processes of boundary layer diffusion, adsorption, desorption, and internal diffusion. This is proposed as an alternative to the commonly used but empirical “ n -th” order kinetic equation. The integration also introduces a second factor: e=1+α/3. This factor is shown to be a divisor for the experimental reaction rates to calculate “intrinsic” rates. Values of e can be very sensitive to temperature. At flame temperatures the values are typically 2 or 3 and the diameter varies substantially with burn-off. At low temperatures, however, the values can be as high as 10 4 or 10 5 and the diameter is essentially constant through the whole of burn-out.

A comparison of prediction and experiment in the gasification of anthracite in air and oxygen-enriched steam mixtures

Gas composition and temperature profiles for anthracite gasified in a combustion pot have been measured and shown to be in good agreement with prediction. The pot was refractory of 6.5″×15.5″ and bed depths were up to 4′ deep. Gasification was in air, and in steam up to 38.5% with oxygen enrichment up to 31%. Measurements confirmed that the initial temperature rise at the bottom of the bed was very steep, at about 1000°C/cm. Comparison with model predictions showed that: the gas-solid reaction in the combustion region was mainly endothermic reduction of CO 2 in a double film, with the main heat release in the ambient gas by reaction of CO to CO 2 ; and that the bed temperature in this region was maintained by radiation as the principal heat transfer mechanism in the bed. This double film mechanism replaces the single film assumption that is common to essentially all past theories of solid bed combustion. The model is otherwise similar to a prior construct (2) but with an improved solution procedure in which computation for the two parabolic differential equations starts at the boundary between the combustion and gasification regions. In other experimental results with computational agreement it is shown: that with sufficiently high or low velocity, the reaction zone exhibits classical blow-off and flashback; that bed depth has no significant effect on the reaction zone width; that the main effect of a deep bed is to act as a heat exchanger; and that the model presented is able to predict with good accuracy the temperature and composition profiles for a wide range of air rates and reactive gas compositions of air with steam and oxygen enrichment.

Prediction and measurement of ignition temperatures of coal particles

Ignition temperatures of one bituminous coal, one anthracite, and three chars have been measured at 5 particle sizes in the range of 60 to 230 microns by injecting small quantities into an electric tube furnace. The technique determined the lowest air temperature (Tg,i) required for ignition, and the results were used to verify a theoretical preiction for the variation of Tg,i with diameter based on a modification of the Semenov thermal explosion analysis. As predicted, Tg,i rose with decreasing diameter, and the quantitative verification permitted determination of the activation energies, frequency factors, and specific reaction rates for the different fuels. With these coefficients determined, the particle ignition temperatures, Ti, were calculated. The activation energies ranged from 19 to 29 kcal/mole; these are interpreted as implying heterogeneous ignition in carbon-reaction Zone II (partial penetration of the reaction zone). Heterogeneous ignition had to be true of the chars. It was notable that there was no evident differentiation of ignition temperature between the coals and chars, with more variation due to diameter than due to fuel type. There was a similar small variation in the specific reaction rates, with all being in the region of 10−4 g/cm2 sec, which is in agreement with a value extrapolated from data given by Smith (22). This supports the accuracy of the measured ignition temperatures which, being in the range of 700±50° K, are considerably lower than the few previously-published values of any validity; they are however, in line with commercial practice for safe operation of pulverized coal mills.

Pyrolysis and combustion of corncobs in a fluidized bed: Measurement and analysis of behavior

Abstract A study of corncob combustion in a fluid bed, for investigation of kinetic behavior, provided reproducible data on the variation of mass, volume, and density with time that was interpreted quantitatively using boundary layer diffusion (BLD) theory, and qualitatively using heat transfer and pyrolysis kinetics. The core of a corncob is an inner cylinder of pith surrounded by two annular cylinders: an inner cob, of about 80% of the total weight, and an outer hull. A total of 120 corncob segments, roughly of length equal to their diameter, and each about 13 g were used. They were immersed in a fluid sand bed at 650°C for periods up to 160 s, using a cage to withdraw the particles after immersion. The cobs were initially cylindrical but they became spherical during reaction so that combustion theory for spheres could be used in the analysis. In the first 10 s, about 10% of the corncob mass burned off, as loss of a fraction of the outer (lower density) hull, and with an increase in mean density by about 20%. In the next 80 s, as heat penetrated into the cob, there was, simultaneously: pyrolysis of the cob, governed jointly by the rate of heat penetration and the pyrolysis kinetics; and reduction of volume by combustion of the hull, quantitatively predicted by BLD theory. The char residue after pyrolysis was about 20% of the original mass, and was spherical, of low density. Its rate of combustion was also quantitatively described by BLD combustion theory. These results showed that the behavior of corncobs in combustion was reproducible, predictable, and amenable to mathematical analysis that could be applied to model descriptions of fluid bed combustion of these materials.