Dataset for modeling NOx formation in pulverized biomass flames under air and oxyfuel conditions

Abstract This dataset is published to support the investigation and development of predictive models for NOx formation during biomass combustion under both air and oxyfuel atmospheres. It is generated using simulations of a drop tube furnace (DTF) configuration, in which biomass particles are injected from the top into a hot, vertically oriented combustion chamber. During thermochemical conversion, the particles release nitrogen-containing volatiles, which oxidize in the gas phase and contribute to NOx formation. The computational domain is designed to capture the full volatile flame length and the complete devolatilization process, which governs the primary mass loss of the solid particles. Simulations were run until a statistically steady state was reached, from where the flame structure and averaged NOx concentrations at the outlet remained constant. The dataset was used in the manuscript "Numerical investigation and modeling of NOx formation in pulverized biomass flames under air and oxyfuel conditions", submitted to Combustion and Flame, and was also presented at the 19th International Conference on Numerical Combustion in Kyoto, Japan. The dataset includes detailed gas-phase and particle-phase fields for Miscanthus combustion in both air and oxyfuel environments. In addition, it provides simulation results used to evaluate reduced-order models based on the fixed volatile composition (FVC) assumption, which is a common simplification in solid fuel kinetics and flamelet tabulated chemistry models. To assess the impact of this assumption on NOx prediction, results from the classical FVC model are compared against reference simulations with dynamically evolving volatile composition. A newly proposed FVC-N model, developed to improve NOx predictions, is also included in the dataset. The dataset consists of statistically steady-state gas-phase and particle data, along with key routines for processing and visualization. It is a valuable resource for researchers and model developers working on pollutant formation in biomass flames. Please note that this is a reduced version of the complete dataset due to size limitations of public repositories. The full dataset and additional technical documentation are available upon request. For more information, contact: [e-mail] Technical Details The simulations were performed in a down-scaled version of the inherited drop tube furnace (DTF) geometry, with domain dimensions of 45 mm × 18 mm × 18 mm. These dimensions were selected to ensure that the domain is sufficiently large to capture the complete flame structure and the full volatile release process during biomass combustion. The computational domain was discretized using a three-dimensional Cartesian mesh with a central resolution of approximately Δx ≈ Dₚ, where Dₚ is the particle diameter. This resolution was chosen to resolve finite-size particle effects and the local flame structure surrounding individual particles. Each simulation consisted of approximately 2.5 million computational cells. To assess NOₓ emissions accurately, the simulations were run for more than two particle-flow-through times (tₛₛₛ = 0.33 s), allowing the flame and NOₓ concentrations at the domain outlet to reach a statistically steady state. Both single-particle and particle-group injection scenarios were studied using torrefied Miscanthus (tMIS) as the fuel. The elemental and proximate properties of the fuel, which govern devolatilization and combustion behavior, are listed in the following table: Proximate analysis (daf-wt.%) Ultimate analysis (dry-wt.%) C H O N S Fixed Carbon Volatile Ash 52.8 5.7 38.16 0.18 0.46 22.3 75 2.7 For the solid-phase kinetics, the CRECK-S model was used. This model requires the fuel to be characterized based on a triangulation approach using specific reference species included in the CRECK-S scheme. The necessary routines for fuel characterization, along with the CRECK-S kinetic mechanism, are provided as part of this dataset. For the gas phase, a biomass-specific detailed chemical mechanism was employed and solved using the finite-rate chemistry approach. This mechanism is also included in the dataset for transparency and reproducibility. Particles with an average diameter of 90 μm and an initial temperature of 300 K were injected into the hot furnace. A uniform wall temperature, equal to the inlet gas temperature, was imposed to replicate the thermal boundary conditions used in the experimental setup. The oxidizing co-flow was introduced from the inlet at a velocity of 0.175 m/s. In the particle-group configuration, a mass flow rate of 0.2 g/min was used to match the experimental conditions. To investigate the influence of oxidizer composition on NOₓ formation, simulations were conducted under both air and oxy-fuel atmospheres, with varying oxygen concentrations. The specific oxidizer configurations used in the simulations are summarized in the following table: Air20 Oxy20 Oxy30 X_O2 0.21 0.21 0.31 X_N2 0.79 0 0 X_CO2 0 0.79 0.69 Field data: The gas phase data has an HDF5 format, which contains selected scalars relevant to model visualization and model development. A time instance, in which the simulations reach the statistically steady state, is used to provide the data. Each HDF5 dataset has the following structure: Scalars (/sd_box/data/scalars/SC): Temperature(T), OH, ZMIX, CO, CO2, O2, H2O, H2, NH3, HCN, C5H5N, N2, N, NH, HNO ,NO, NO2, N2O Source terms (/sd_box/data/scalars/chemSRC) : ST, SOH, ,SCO, SCO2, SO2, SH2O, SH2, SHCN, SNH3, SC5H5N, SN, SNH, SHNO, SNO, SNO2, SN2O, SN2 Other fields (/sd_box/data/cv_data_real): Gas: RHO, WMIX, Cp Particle: particle location (ND_prt), particle temperature (T_prt), normalized mass of the particle (Mnorm_prt), devolatilization rate (DVDT_prt), char oxidation rate (DCHDT_prt) Averaged data: In order to perform the flux analysis and overall accumulated NOx produced, the time and volume-averaged NOx in the entire domain and the plane-averaged NOx at the exit plane at each time until the statistically steady state are also provided. These data are in columned "txt" format that can be read for visualization. Also, the Nitrogen element Fluxes are provided, which can be used as an input for the flux analysis. Scripts: Different scripts are provided for a reader to enable the first-time use of the data, such as visualizing the gas phase quantities, calculating the conditional mean and statistical analysis, and the flux analysis using the averaged data. Here is a brief information regarding the scripts: wrapp_HDF5.m: This script can be used for loading the HDF5 file and visualising the field, joint PDF, and joint correlation of different quantities with respect to different parameters. ElementBasedReactionFluxAnalysis: This script package performs the reaction flux analysis based on the nitrogen element. For this script, the integrated averaged fluxes can also be generated using the volumetric averaged data and the gas-phase mechanism, which are also provided. A comprehensive readme file is also provided to explain how to use the flux analysis package.