Stephen Niksa

Coal conversion submodels for design applications at elevated pressures. Part I. Devolatilization and char oxidation

Numerous process concepts are under development worldwide that convert coal at elevated pressure. These developments rely heavily on CFD and other advanced calculation schemes that require submodels for several stages of coal chemistry, including devolatilization, volatiles combustion and reforming, char oxidation and char gasification. This paper surveys the databases of laboratory testing on devolatilization and char oxidation at elevated pressure, first, to identify the tendencies that are essential to rational design of coal utilization technology and, second, to validate two well-known reaction mechanisms for quantitative design calculations. Devolatilization at elevated pressure generates less volatile matter, especially tar. Low-rank coals are no less sensitive to pressure variations than bituminous coals; in fact, coal quality is just as important at elevated pressure as it is at atmospheric pressure. Faster heating rates do not enhance volatiles yields at the highest operating pressures. The FLASHCHAIN® predictions for the devolatilization database depict the distinctive devolatilization behavior of individual samples, even among samples with the same nominal rank. The only sample-specific input requirements are the proximate and ultimate analyses of the coal. There were no systematic discrepancies in the predicted total and tar yields across the entire pressure range. Char oxidation rates increase for progressively higher O2 partial pressures and gas temperatures, but are insensitive to total pressure at constant O2 mole fraction. Char burning rates become faster with coals of progressively lower rank, although the reactivity is somewhat less sensitive to coal quality at elevated pressure than at atmospheric pressure. An expanded version of the carbon burnout kinetics model was able to represent all datasets except one within useful quantitative tolerances, provided that the initial intrinsic pre-exponential factor was adjusted for each coal sample.

Optically determined temperatures, sizes, and velocities of individual carbon particles under typical combustion conditions☆

This study utilizes an in situ optical method to determine the temperature, size, and velocity of individual particles of burning pulverized fuels. Temperature-size and velocity-size correlations were determined for both nonreacting and ignited suspensions of a spherical, nonvolatile, molecular sieve carbon flowing in one dimension. The measurements at three axial positions when no oxygen was present illustrate the strong sensitivity to particle size in the transient temperatures. A comparison with predictions from a heat transfer model with no adjustable parameters assesses the experimental errors. Sizes were determined to within 10 μm (120 < dp, μm < 240), temperatures to within 50K (1150 < Tp, K < 1950), and velocities to within 5% (150 < νp, cm/s < 300). Four temperature-size correlations for suspensions burning in 24% excess oxygen show that heterogeneous reaction steepens the correlations beyond their levels for conduction heating alone. For a 75 μm size range, the observed ranges of temperature during heatup and ignition span several hundred degrees, which raises serious doubts about the utility of using average temperatures and sizes to assign kinetic parameters throughout the first 100–150 ms of this process.

Fragmentation during carbon conversion: Predictions and measurements

The first quantitative predictions of fragmentation phenomena during carbon conversion are presented. Predictions of the porosity at which fragmentation occurs based on deterministic and statistical models are compared. Analysis of the statistical models is based on percolation theory, which is the geometrical theory of the connectedness of irregular objects. Percolation theory predicts fragmentation at a porosity of about 0.7 for homogeneous samples. Corrections for some aspects of heterogeneity are provided by the theory. In addition, percolation theory predicts that the mass distribution of fragments will exhibit a power-law range f(m)≈m −2.15 . The distribution of fragment radii is predicted to exhibit a power-law range g(r)≈r −3.86 . The fragments are predicted to be highly irregular, with surface roughness characteristics different than those of the unreacted sample. These predictions are qualitatively consistent with previous observations. New measurements of the porosity at fragmentation for six carbon composite materials are presented. These measurements indicate that the porosity at fragmentation is a reproducible quantity for a given material, and that it depends in a complex way on material properties. These theoretical and experimental results indicate the likelihood that fragmentation contributes significantly to weight loss during conversion. In addition, fragmentation may influence the size distribution of particulate effluents produced during conversion.

Heat and Mass Transfer in the Vicinity of a Devolatilizing Coal Particle

Abstract A computer model has been developed to describe heat and mass transfer in the vicinity of a single reacting coal particle entrained in a laminar gas stream. The model describes the various processes that occur during devolatilization and combustion in the boundary layer between the particle surface and the bulk gas stream. Global reaction rates are used in the gas phase, with light and heavy volatiles represented by methane and benzene. Although the model was developed for a two-dimensional treatment. one-dimensional simulations are shown to yield meaningful results at low particle Reynolds numbers. Simulations indicate that the region of homogeneous volatiles combustion may extend for several particle diameters, implying that flame sheet approximations are not applicable during devolatilization. Parametric studies indicate that particle heating is dominated by homogeneous volatiles combustion for large particles and by heterogeneous reaction for small particles.

The Distributed-Energy Chain Model for Rapid Coal Devolatilization Kinetics. Part I: Formulation

Abstract The distributed-energy chain model (DISCHAIN) interprets coal devolatilization in terms of independent influences from chemical reaction rates and from macromolecular configuration. Coal is represented by three components: (1) aromatic units that are attached pairwise by (2) labile bridges to form nominally infinite linear chains, with (3) peripheral groups branching from the aromatic units. These components are the building blocks for unreacted coal, free monomers (mobile aromatic units), gas, tar, and char. Four chemical reactions represent bridge dissociation, peripheral group elimination, and tar and char formation. Analytic probability expressions and competitive reactions describe the conversion of bound aromatic units into free monomers, and enter into the formation of all products except gas. There are no hypothetical ultimate yields. The model is introduced in two parts. Here in Part I, the coal model, chemical reactions, and chain statistics are derived and formulated into rate equations. Mechanisms leading to major products are identified, including a novel mechanism for yield enhancement by faster heating. Whenever bridge dissociation and char formation occur concurrently, as for slow heating, the subsequent generation of monomers is inhibited. Aromatic units are thereby excluded from the competition between tar and char formation. Conversely, bridge dissociation and char formation occur consecutively for rapid heating, and a greater proportion of the original bound aromatic units become monomers and, ultimately, tar.

The distributed-energy chain model for rapid coal devolatilization kinetics. Part II: Transient weight loss correlations

Abstract The Distributed-Energy Chain Model (DISCHAIN) is formulated in a companion paper [1]. In this paper, numerical predictions are compared to the transient weight loss from a bituminous coal in vacuum during heatup and throughout isothermal pyrolysis for two heating rates (10 2 , 10 3 K/s) at temperatures between 700 and 1300K. Based on seven independent reaction rate parameters, the model predicts transient yields for unreacted coal, gas, tar, and char. Except during heatup at the lower heating rate, the model correlations are in good quantitative agreement with the measured weight loss, and agree with the observed yield enhancement for greater heating rates within the experimental error. This study also includes data correlations from the Distributed Activation Energy Model, to identify the most important model parameters, and model predictions from DISCHAIN for typical pulverized coal combustion conditions, to further illustrate the extent that transient devolatilization on short time scales is determined by transient thermal histories.

On the role of heating rate in rapid coal devolatilization

We have measured time-resolved weight loss from two bituminous coals during heatup at heating rates varied from 10 2 to 10 4 K/s independently of reaction temperatures (to 1300 K). The results indicate that both ultimate yields and evolution rates increase with faster heating, particularly for condensed liquid products. Increasing the pressure from vacuum to 0.19 MPa diminished the sensitivity to heating rate and suppressed the reactive volatiles. A non-isothermal kintic analysis of a single reaction model describes yield suppression during heatup and predicts the faster generation rates with faster heating. When coupled into a competitive rate scheme, this model could account for the enhanced yield.

A simple numerical model to estimate the effect of coal selection on pulverized fuel burnout

The amount of unburned carbon in ash is an important performance characteristic in commercial boilers fired with pulverized coal. Unburned carbon levels are known to be sensitive to fuel selection, and there is great interest in methods of estimating the burnout propensity of coals based on proximate and ultimate analysis--the only fuel properties readily available to utility practitioners. A simple numerical model is described that is specifically designed to estimate the effects of coal selection on burnout in a way that is useful for commercial coal screening. The model is based on a highly idealized description of the combustion chamber but employs detailed descriptions of the fundamental fuel transformations. The model is validated against data from laboratory and pilot-scale combustors burning a range of international coals, and then against data obtained from full-scale units during periods of coal switching. The validated model form is then used in a series of sensitivity studies to explore the role of various individual fuel properties that influence burnout.

Time-resolved weight loss kinetics for the rapid devolatilization of a bituminous coal

The weight loss kinetics for the rapid devolatilization of a bituminous coal were measured over broad ranges of temperature (to 1000 C) and heating rate (10 2 to 10 4 K/s) in vacuum. The reaction period was resolved in 0.1 s time increments with rapid quenching. In addition, the effects of pressure (13.3 Pa–10 MPa) and particle size (50, 85, and 125 μm) were studied. We found that the rate of generation of volatiles increases slowly with temperature andis relatively insensitive to pressure changes. However, the heating rate does affect the devolatilization rate and the total yield. Below 600 C, two stages of product evolution were observed, one with and the other without coliberation of tarry liquids. At higher temperatures these two stages coalesced. Although the concept of simultaneous, parallel reactions can account for the low apparentactivation energy of the first stage of the devolatilization process, a multiple reaction model with a Gaussian activation energy distribution does not adequately describe the total time-temperature history. In addition, extant models with competitive rate processes to describe the redeposition of volatiles and their escape from the particle seem incompatible with the effects of particle size and heating rate observed in vacuum.

A global NOx submodel for pulverized coal flames at elevated pressures

ABSTRACT This study formulates a global NOX submodel for deployment in CFD simulations from a database on flames of three diverse coals at pressures to 3.0 MPa for broad ranges of stoichiometric ratio (S.R.). A new reaction scheme was formulated from a sensitivity analysis of simulations based on detailed reaction mechanisms for all tests. It shares many elements in common with commercial submodels, yet it correctly predicts that (1) less coal-N is converted into NO; and (2) HCN persists to higher S.R. for progressively higher pressures. Explicit dependences on O2 concentrations are responsible for the first feature, because the variations in O2 concentrations mimic the ways that the oxyhydroxyl radical pool shrinks at progressively higher pressures, which shifts HCN conversion toward N2 production. The second feature was depicted by resolving the intermediate products of HCN decomposition in the global scheme. Discrepancies surfaced when the new submodel was applied to different coals without re-adjusting rate parameters, which probably reflects a generic limitation of global NOX production submodels for coal combustion.