Multi-fidelity computational modeling of non-equilibrium plasma assisted combustion: interaction of chemical kinetics, plasma dynamics and ignition kernel evolution

Abstract

Plasma-assisted combustion (PAC) is an emerging technology that leverages non-equilibrium plasma to enhance ignition, flame stability, and combustion efficiency, and abate perilous emissions in a variety of propulsion and power generation applications. The motivation for PAC stems from the challenges associated with lean, low-temperature combustion, ammonia combustion, and high-speed combustion. These challenges include long ignition delays, high ignition energy requirements, flame extinction challenges, and excessive emissions of nitrogen oxides (NOx ). By generating reactive species at low temperatures, introducing ultra-fast gas heating, and the interplay of vibrational - translational relaxation, non-equilibrium plasma offers a promising pathway to overcoming these obstacles, thereby enabling cleaner and more efficient combustion. Despite its potential to improve ignition, flame stabilization and controlling emissions, modeling PAC remains a significant challenge due to the complex interaction between chemical kinetics, electrostatics and plasma dynamics, thermodynamics and turbulent gas dynamics. Plasma introduces vastly disparate spatial and temporal scales, requiring multi-fidelity computational approaches to accurately capture ignition and pollutant emission kinetics, regime transitions of the discharge based on the kinetics, dynamics, and thermodynamics, and the evolution of the ignition kernel to eventually form a sustained flame. Moreover, the plasma is bi-directionally tightly coupled with the reacting flow, i.e. not only does the non-equilibrium plasma enhance ignition, flame stabilization and abate emissions, but the evolving thermochemistry and turbulent gas dynamics also affect the formation and regime of the plasma discharge. The multiphysics and multi-scale nature of PAC necessitates the development of varied modeling frameworks used to probe into different aspects in order to balance physical accuracy and computational feasibility for system level simulations.This thesis describes three such modeling approaches, and uses them to investigate the underlying physics and chemistry of plasma assisted ignition. Suggestions on leveraging one model to improve the fidelity of another have also been proposed. As a first step, a zero-dimensional (0D) solver for modeling coupled plasma and combustion kinetics has been developed, verified and validated. The governing equations, numerical methods, and verification/validation studies are presented. The PAC chemical kinetics of ammonia-air mixtures are explored under varying conditions, highlighting the effects of the reduced electric field, equivalence ratio, pulse repetition frequency, pulse energy density and pressure on ignition delay and NOx emissions. The production of key radicals due to the dissociative quenching of electronic excited states of N2 at some of these conditions, fast gas heating due to exothermic quenching and electron ion recombination reactions, and the vibrational-translational relaxation induced slow gas heating effectively help to enhance the reactivity of the mixture. Furthermore, the role of intermediates such as NH2 and HNO in the production of fuel-bound as well as air-bound NOx has been elucidated. Given the computational cost of fully resolved multi-dimensional plasma simulations, a phenomenological model that captures the essential effects of plasma—ultra-fast gas heating, ultra-fast radical generation, and vibrational energy transfer—without explicitly solving for charged species kinetics and the coupled electric field dynamics, has been implemented. Large eddy simulations (LES) are performed to investigate plasma assisted ignition due to nanosecond pulsed high frequency discharges, under turbulent conditions. Interesting constructive and destructive coupling mechanisms between kernels produced by successive discharges are found that helped to explain some of the observations in lab experiments. The model was verified against numerical benchmarks and validated against experimental data, and further improvements to the model to account for the spatiotemporal evolution of the plasma power density are proposed for enhanced accuracy. To better understand the fundamental physics of plasma discharges, two solvers - Multigrid plasma solver 1D (mps1d) and an adaptive mesh refinement (AMR) enabled, CPU-GPU compatible solver, Vidyut3D, are used to assess the dependence of the plasma formation on the local thermochemical state. Both these solvers are verified against several popular benchmarks in the literature. mps1D is first used to simulate avalanche to streamer transition and streamer propagation in a 1D domain at four different states of a freely propagating laminar ammonia-air flame - unburnt reactants, pre-heat zone mixture, reaction zone mixture and burnt products. The dependence on the applied electric field, total gas number density and the composition of the mixture, on the streamer formation and connection times were analyzed. Next, 2D axisymmetricmodeling using the drift-diffusion-reaction equations along with a photoionization model is performed to simulate cathode-directed streamer propagation and transition to nanosecond glow and nanosecond spark regimes at atmospheric pressure. The kinetics and thermodynamics which determine the fast gas heating and fast radical production in the spark phases are described in detail for discharges in same four thermochemical states of the ammonia-air flame. The timescales associated with the radical generation pathways were found to become comparable with some reactivity inhibiting, chain termination pathways, after the voltage was turned off, that resulted in negligible gas heating and radical generation in the after-pulse phase of a discharge in the unburnt reactants mixture. Moreover, a secondary ionization wave induced electron density hotspot resulted in extreme fast gas heating and production of O, OH, H and NH2 near the anode, in the reaction zone mixture. Commonly known pathways to produce O radicals and fast gas heating in dry air are contrasted with the pathways found for these ammonia-air flame states. Finally, a way to extend the phenomenological model to account for such complex chemical kinetics and spatiotemporally varying radical production and gas heating effects is proposed. The computational challenges in these simulations have been highlighted as well. Finally, a brief summary of this dissertation along with several recommendations for future research in this area are provided.