Why do iron particles fail to burn? Numerical investigations into the interaction of turbulence and iron powder combustion
Speaker
About this event
Industrial iron powder combustors represent a highly promising carbon-free renewable energy technology, but their practical application is frequently hindered by combustion inefficiency, specifically the presence of unoxidized particles. To understand if such flames can be self-sustaining, an investigation into the observed inefficiency is necessary. This investigation hinges on two fundamental questions: 1. Can all particles be ignited in the first place? 2. If particles are successfully ignited, can they quench midway through their combustion? In real combustors, the flow is inherently turbulent, which drives the fluid dynamic phenomenon of preferential concentration. This phenomenon results in particle clustering, which could have a detrimental effect on the combustion process. In order to understand precisely how preferential concentration affects the overall combustion of the particles and to deduce whether turbulence interaction could be an answer to the posed questions, a dedicated numerical framework to simulate turbulent, particle-laden iron flames is developed. The framework is based on a two-way coupled Eulerian-Lagrangian approach, tracking the iron particles with the point-particle assumption and employing the “switch-type” kinetics from the oxide-layer model by Mi et al. to model particle reactions. This framework is implemented in the high-fidelity solver NTMIX-CHEMKIN to perform robust Direct-Numerical-Simulations (DNS) and in the commercially-available OpenFOAM to test large-scale capabilities with Large-Eddy-Simulations (LES). Using this framework, iron particle combustion in a Homogeneous Isotropic Turbulence (HIT) field is first analyzed using DNS [3]. The results demonstrated that preferential concentration causes massive localized oxygen depletion, which can elongate particle burn times by up to eight times compared to isolated particles. However, the most critical message from this study is that although the combustion time is significantly elongated by clustering, the flame does not quench; the particles smoothly complete their oxidation. This behavior is further confirmed by the DNS of a turbulent mixing layer, which also showed the extension of combustion time and underscored the critical importance of oxygen depletion in particle-dense regions. The mixing layer study additionally revealed that while small particles (under 20 μm) maintain the laminar structure of the flow, larger particles (over 28 μm) possess the momentum to actively perturb the mixing layer and induce turbulent-like behavior. Because particles do not quench once ignited, the primary focus for solving combustion inefficiency should be strictly on the ignition phase of the combustion process. Experimental results from pilot burners, such as those analyzed by Niek van Rooij and Jesse Hameete, highlight that iron particles can completely fail to ignite, possibly due to the gradual growth of the oxide layer and subsequent hindering effects on ignition. Hence, accurately predicting ignition becomes the most vital part of the numerical modeling of turbulent iron flames. The jet-in-hot-coflow burner developed by Jesse Hameete is numerically modeled as LES in OpenFOAM. Two distinct reaction models—the first-order Damköhler model and the oxide-layer model—are compared against experimental data. Ultimately, neither numerical model is currently able to perfectly capture the observed ignition behavior. While the first-order model overpredicts ignition percentages, the oxide-layer model exhibits the same trend in percentages as seen in the experimental results, albeit with a substantial error in critical temperature. However, this study confirmed that particles do not partially oxidize in such large-scale flames; rather, they fail to ignite. Can preferential concentration enhance ignition? A numerical study on ignition in three-dimensional spherical suspensions shows that a clustered particle distribution has a substantially lower ignition temperature than a random Poisson distribution in space. Hence, partial clustering before particle injection could lead to better ignition in large-scale combustors. Ultimately, this research establishes that the combustion inefficiency in industrial turbulent iron flames is primarily a failure of ignition rather than mid-combustion quenching. While preferential concentration significantly delays oxidation through localized oxygen depletion, it does not inherently extinguish ignited particles; conversely, clustering may actually lower the thermal threshold required for initial ignition. These findings shift the paradigm for optimizer strategies toward the pre-ignition phase. Future research must bridge the gap between current kinetic models and experimental reality, specifically investigating how multi-stage turbulent mixing dynamics can be leveraged to optimize combustion efficiency and improve the overall performance of large-scale iron powder combustors.
Host
