The response of lean (ϕ 6 0.4) premixed hydrogen flames to maintained homogeneous isotropic turbulence is investigated using detailed numerical simulation in an idealised three-dimensional configuration over a range of Karlovitz numbers from 10 to 1562. In particular, a focus is placed on turbulence sufficiently intense that the flames can no longer be considered to be in the thin reaction burning regime. This transition to the so-called distributed burning regime is characterised through a number of diagnostics, and the relative roles of molecular and turbulent mixing processes are examined. The phenomenology and statistics of these flames are contrasted with a distributed thermonuclear flame from a related astrophysical study.

We extend our previous three-dimensional, full-star simulations of the final hours of convection preceding ignition in Type Ia supernovae to higher resolution using the adaptive mesh refinement capability of our low Mach number code, MAESTRO. We report the statistics of the ignition of the first flame at an effective 4.34 km resolution, and general flow field properties at an effective 2.17 km resolution. We find that off-center ignition is likely, with radius of 50 km most favored and a likely range of 40 to 75 km. This is consistent with our previous coarser (8.68 km resolution) simulations, implying that we have achieved sufficient resolution in our determination of likely ignition radii. The dynamics of the last few hot spots preceding ignition suggest that a multiple ignition scenario is not likely. With improved resolution, we can more clearly see the general flow pattern in the convective region, characterized by a strong outward plume with a lower speed recirculation. We show that the convective core is turbulent with a Kolmogorov spectrum and has a lower turbulent intensity and larger integral length scale than previously thought (on the order of 16 km s −1 and 200 km, respectively), and we discuss the potential consequences for the first flames.

We investigate turbulence generated by cosmological structure formation by means of large eddy simulations using adaptive mesh refinement. In contrast to the widely used implicit large eddy simulations, which resolve a limited range of length scales and treat the effect of turbulent velocity fluctuations below the grid scale solely by numerical dissipation, we apply a subgrid-scale model for the numerically unresolved fraction of the turbulence energy. For simulations with adaptive mesh refinement, we utilize a new methodology that allows us to adjust the scale-dependent energy variables in such a way that the sum of resolved and unresolved energies is globally conserved. We test our approach in simulations of randomly forced turbulence, a gravitationally bound cloud in a wind, and the Santa Barbara cluster. To treat inhomogeneous turbulence, we introduce an adaptive Kalman filtering technique that separates turbulent velocity fluctuations on resolved length scales from the non-turbulent bulk flow. From the magnitude of the fluctuating component and the subgrid-scale turbulence energy, a total turbulent velocity dispersion of several 100 km/s is obtained for the Santa Barbara cluster, while the low-density gas outside the accretion shocks is nearly devoid of turbulence. The energy flux through the turbulent cascade and the dissipation rate predicted by the subgrid-scale model correspond to dynamical time scales around 5 Gyr, independent of numerical resolution.

The large range of time and length scales involved in Type Ia supernovae (SNe Ia) requires the use of flame models. As a prelude to exploring various options for flame models, we consider in this paper high-resolution, three-dimensional simulations of the small-scale dynamics of nuclear flames in the supernova environment in which the details of the flame structure are fully resolved. The range of densities examined, (1Y8) ; 10 7 g cm À3 , spans the transition from the laminar flamelet regime to the distributed burning regime where small-scale turbulence disrupts the flame. The use of a low Mach number algorithm facilitates the accurate resolution of the thermal structure of the flame and the inviscid turbulent kinetic energy cascade, while implicitly incorporating kinetic energy dissipation at the grid-scale cutoff. For an assumed background of isotropic Kolmogorov turbulence with an energy characteristic of SNe Ia, we find a transition density between 1 and 3 ; 10 7 g cm À3 , where the nature of the burning changes qualitatively. By 1 ; 10 7 g cm À3 , energy diffusion by conduction and radiation is exceeded, on the flame scale, by turbulent advection. As a result, the effective Lewis number approaches unity. That is, the flame resembles a laminar flame but is turbulently broadened with an effective diffusion coefficient, D T $ u 0 l, where u 0 is the turbulent intensity and l is the integral scale. For the larger integral scales characteristic of a real supernova, the flame structure is predicted to become complex and unsteady. Implications for a possible transition to detonation are discussed.

The nature of carbon burning flames in Type Ia supernovae is explored as they interact with Kolmogorov turbulence. One-dimensional calculations using the Linear Eddy Model of Kerstein (1991) elucidate three regimes of turbulent burning. In the simplest case, large scale turbulence folds and deforms thin laminar flamelets to produce a flame brush with a total burning rate given approximately by the speed of turbulent fluctuations on the integral scale, U L . This is the regime where the supernova explosion begins and where most of its pre-detonation burning occurs. As the density declines, turbulence starts to tear the individual flamelets, making broader structures that move faster. For a brief time, these turbulent flamelets are still narrow compared to their spacing and the concept of a flame brush moving with an overall speed of U L remains valid. However, the typical width of the individual flamelets, which is given by the condition that their turnover time equals their burning time, continues to increase as the density declines. Eventually, mixed regions almost as large as the integral scale itself are transiently formed. At that point, a transition to detonation can occur. The conditions for such a transition are explored numerically and it is estimated that the transition will occur for densities near 1 × 10 7 g cm −3 , provided the turbulent speed on the integral scale exceeds about 15% sonic. An example calculation shows the details of a detonation actually developing.

This paper presents three-dimensional direct numerical simulations of lean premixed hydrogen flames at an equivalence ratio of ϕ = 0.4 over a range of turbulence levels from Ka = 1 to 36. The simulations form part of a larger effort to construct a DNS database that can be used by the community for model construction and validation. We have focussed on producing well-resolved simulations at conditions representative of atmospheric laboratory-scale flames. After an overview of phenomenological trends with increasing Karlovitz number, we examine the factors that lead to an observed decorrelation between fuel consumption and heat release in the flame at Ka = 36. We show that in this flame the fuel consumption is greatly enhanced in regions of positive curvature. We also show that the radical pool is enriched throughout the entire flame as Ka is increased. In particular, we identify three reactions that, driven by high molar concentrations of radicals at low temperatures, are responsible for high levels of heat release away from regions of fuel consumption, thereby accounting for the observed decorrelation between fuel consumption and heat release.

Implicit LES methods are numerical methods that capture the energy-containing and inertial ranges of turbulent flows, while relying on their own intrinsic dissipation to act as a subgrid model. We present a scheme-dependent Kolmogorov scaling analysis of the solutions produced by such methods. From this analysis we can define an effective Reynolds number for implicit LES simulations of inviscid flow. The approach can also be used to define an effective Reynolds number for under-resolved viscous simulations. Simulations of maintained homogeneous isotropic turbulence and the Taylor-Green vortex are presented to support this proposal and highlight similarities and differences with real-world viscous fluids. Direct comparison with data from high resolution DNS calculations provides validation of the effective viscosity and effective Kolmogorov length scale.

No abstract