• combustion
• flashback
• hydrogen
• perforated burner

The decarbonization of energy systems, particularly within the heating sector, necessitates the deployment of low-emission technologies capable of integrating with existing infrastructure. Hydrogen-enriched combustion offers a promising route to this end; however, the propensity of hydrogen for flashback, defined as the upstream propagation of the flame into premixing regions, poses significant safety challenges in practical applications such as condensing boilers. This thesis presents a comprehensive numerical investigation of flashback and autoignition phenomena in premixed hydrogen-methane-air and hydrogen-air flames stabilized by perforated burner plates, with a specific focus on multi-slit geometries.

From a physical standpoint, the study analyzes how flame dynamics, heat transfer, and species transport influence flashback behavior across different burner configurations. Direct Numerical Simulations (DNS), incorporating conjugate heat transfer, indicate that flashback is strongly affected by thermal feedback mechanisms, preferential diffusion, and the Soret effect. Two distinct flashback regimes are identified, with the transition between them depending on hydrogen content and burner temperature. A strong correlation emerges between these mechanisms and the geometric characteristics of the burner, highlighting how local physical processes are coupled with design features. Three-dimensional simulations are found to be necessary to capture key phenomena, especially the localized interactions at slit ends (absent in traditional two-dimensional models) that enhance flashback velocities through intensified preheating and local fuel enrichment. The analysis also addresses autoignition-driven flashback at elevated wall temperatures, examining critical Damköhler-number regimes and geometric effects on ignition thresholds.

To support practical burner design, the research employs a suite of computational tools for high-dimensional uncertainty quantification and design space exploration. Surrogate models based on generalized Polynomial Chaos (gPC) and sparse grids are developed to approximate DNS-derived quantities such as flashback velocity and critical burner temperature. These models enable rapid parametric sweeps and global sensitivity analyses, quantifying the relative importance of geometrical and operational parameters in shaping flashback behavior. A surrogate-informed sparse grid framework is also proposed to explore the impact of slit geometry with reduced computational cost. These tools are intended to support design decisions under uncertainty for hydrogen-compatible combustion systems.

By combining detailed physical modeling with surrogate-based analysis, the thesis aims to contribute to the understanding and predictive modeling of flashback phenomena, with potential implications for the development of safer and more efficient hydrogen-fueled heating technologies.