John Ziegler, California Institute of Technology
Detailed understanding of detonation ignition and propagation in gases is vital for assessing the threat from accidental explosions, for example, in piping systems of petrochemical or nuclear fuel processing plants and hydrogen energy source delivery systems.
Detonations propagate at supersonic velocities between 1000-3000 m/s and exhibit unsteady spatial structures over a large range of scales, from microns to centimeters in typical fuel-air mixtures. Experimental observations are just beginning to provide flow field data within the reaction zone; and due to the high speeds and large range of scales, experimental characterization remains challenging. Today’s terascale supercomputers provide an alternative to laboratory experiments by simulating the governing thermo- and hydrodynamic equations at the required accuracy. While typical simulation codes for detonation waves are based on the inviscid Euler equations, the aim of this ongoing research is to employ the extended model of the reactive Navier-Stokes equations instead. The ultimate goal will be the Direct Numerical Simulation (DNS) of the Mach reflection phenomenon in gaseous detonations. This triple point structure of detonation fronts and shock waves is an essential feature of the cellular structure observed in propagating detonations and is fundamental for gaining knowledge of the underlying physical processes. Due to a lack of computational power, resolution, and efficient algorithms, previous research has mostly modeled detonations using either 2D or 3D reactive Euler equations with simplified chemistry. These models ignore viscous and diffusive processes and reduce the number of species and reactions modeled, in order to make the problem more tractable. However, new algorithms for partial differential equations, such as higher order shock-capturing methods and solution-dependent adaptive mesh refinement, implemented in codes on supercomputers, have made the simulation of the full reactive Navier-Stokes equations possible.
Abstract Author(s): Jack Ziegler