Biomass Particle Ignition in Mill Equipment

Abstract

In the last decades, interest in replacing fossil fuels with renewable energy sources has increased, largely motivated by environmental concerns. One alternative to coal and gas is biomass, especially for combined heat and power production. Biomass can play an important role in an all-renewable energy scenario: unlike solar or wind power, it can easily be stored in large quantities and converted when demand for heat and electricity arise. Power plants designed for pulverized coal firing have successfully been converted to biomass (especially wood) dust firing. Use of biomass dusts to replace pulverized coal in combined heat and power plants has however led to several incidents in which fires broke out in the plants’ mills. Mill fires are likely caused by self-heating and spontaneous ignition of settled dust accumulations. Transport air passing through the mills is pre-heated, which serves to both dry the fuel and also as a method of internal heat recovery in the power plant. High combustion air temperatures are therefore desirable from the point of view of overall plant efficiency, but poor predictability of critical conditions for mill fires has led operators to adopt cautiously low inlet air temperatures (≤423 K). In this work, results from a study on single pellets (0.25 g) suggested that heterogeneous oxidation of biomass is the main mechanism leading to mill fires. Small wood pellets spontaneously ignited when exposed to temperatures around 500 K and above. Ignition led to large temperature increase of the sample, emission of significant amounts of CO and CO2 and near-complete conversion of the organic matter. Ignition occurring at low temperatures was distinct from flaming modes of combustion. Flaming required higher ambient temperatures (>700 K). From a carbon balance, it was concluded that a pyrolytic decomposition occurs in parallel to the oxidation. Both processes likely occur at rates in similar order of magnitude. A simple reaction mechanism was developed to describe parallel oxidation and pyrolysis of biomass at low temperatures. The mechanism takes the composition of biomass into account, distinguishing between volatilzable fractions of extractives, hemicellulose, lignin and cellulose, as well as char. Analytical data was available for six biomasses: beech and pine wood, wheat straw, sunflower husk pellets, and two types of commercial wood pellets. Kinetic parameters were fitted to data measured in stepwise-isothermal thermogravimetric analysis at 423–523 K. The models were in good agreement with the experimental data, and could to some degree also be extrapolated to higher heating rates (5 K/ min) and temperatures (ca. 550–650 K). It was found that low temperature pyrolysis required a four-species model (extractives, hemicellulose, lignin, cellulose), while the volatilizable fraction of biomass could also be treated as a lumped component for the heterogeneous oxidation reaction. For each component considered, the same activation energy could be used, regardless of which biomass it appeared in. Within the range considered, the oxidation reaction could be described with an activation energy of 퐸a = 130 kJ/mol and reaction orders in oxygen between 0.4–0.5 (dependent on the biomass). Differences between the biomasses were modelled by the pre-exponential factor 푘0 and the reaction order in the solid conversion. It was further found that the reactivity of the different biomasses could to some degree be related to its composition: biomass rich in extractives had increased mass loss rates at low temperatures (<470 K), while biomasses rich in potassium (in the inorganic fraction) had high reaction rates above 500 K. Lab scale experiments on loosely packed dust beds (10–40 g) were carried out to gain a further understanding of the ignition process. Onset of oxidation reactions was detectable by low concentrations of CO and CO2 from around 373 K, which gradually increased with temperature. Larger sample bulk densities and higher ambient oxygen concentrations favored thermal runaway of the samples, which could otherwise (i.e., under ’subcritical’ conditions) also stabilize and slowly oxidize. Critical conditions for thermal runaway mainly depended on the material tested. Qualitatively, the ignition behavior of the biomasses tested in this part of the study agreed with the thermogravimetric data: Sunflower and pine showed an early reaction onset, while the reactivity of beech and wheat increased towards higher temperatures (>500 K). Both experimental results as well as estimate calculations suggest that ignition was mainly kinetically controlled in the experiments. Acceleration of reactions on thermal runaway may be linked to a rapid increase in oxidation of the cellulose component. A one-dimensional numerical model was developed to describe self-heating and ignition in dust beds, including reactions, mass and heat transfer. The reaction mechanism derived from thermogravimetric analysis was used in the model, while values for material property parameters could be found in the literature. Central aim of the model is to predict critical temperatures for ignition under different conditions. The lab scale experiments could be simulated with good accuracy, where ignition temperatures where predicted 2–5 % lower than measured. No parameters were fitted to fixed bed experiments. Sensitivity to various parameters used in the model was investigated. Onset of thermal runaway appeared to be controlled to a great extent by the kinetics of the oxidation reaction. Material properties, in comparison, had a very low influence. Further simulations were carried out to investigate effects of sample size and oxygen availability. Increasing the characteristic length scale or the bulk density of the sample drastically lowered the predicted ignition temperatures (as low as 395–460 K for large/dense samples). Thus, the scaling study demonstrated that self-ignition of settled biomass dust beds could plausibly explain mill fires. In summary, self-heating and self-ignition were studied both experimentally and numerically in this work. Results show that critical conditions for self-ignition can be predicted based on the reaction kinetics of heterogeneous oxidation of the biomass. The kinetics of the oxidation reaction can to some degree be tied to the structural composition of biomass, and it is suggested that further research be carried out to investigate this link.