D. Cecere

LES of the HyShot scramjet combustor

With the overall goal to clarify the physics of com pressible (supersonic) combustion, a 3D LES of the HyShot supersonic combustor has been performed and is reported in this paper. HyShot is an (originally) Australian p rogram to assess feasibility of supersonic combustion by means of a ballistic test flight. The HyShot combustion chamber is shaped as a box 75x9.8 mm in cross section and 300 mm long. Hydrogen is injected at 90 degrees with respect to the superson ic airstream 40 mm downstream from the combustor inlet by means of four 2 mm diameter choked orifices. Air enters the channel at a Mach number that, in the actual te st, depended on the flight trajectory; in this simulation, the trajectory poi nt is that at height = 28 km, where the Mach number was 2.79, P=82.11 kPa and T = 1229 K. A structured grid of about 14x10 6 nodes discretizes the actual combustor shape, wher e hydrogen-air combustion is treated by means of a detailed chemical kinetics model including 9 species and 37 reactions. Numerical results indicate that hydrogen penetrates in the air stream generating 3D bow shock structures upstream of the injection orifices as seen in experiments. In these regions recirculation zones u pstream and downstream of the fuel injection orifices are observed as expected; the OH predicted by LES indicates that a flame starts already in the upstream recircu lation zone. Interactions among the essentially 1D airstream entering the combustor, th e heat released and the 3D jets produce large vorticity rates and therefore enhance and accelerate turbulent mixing. Combustion is predicted very fast and efficient: on ly 0.5% of hydrogen is found unburned at the combustor exit.

The role of the baroclinic term in supersonic fuel/air mixing enhancement

Understanding of physics of supersonic combustion is mandatory for future hypersonic air-breathing propulsion systems. In fact, in order to overcome Mach 5 flight speeds, a supersonic combustion ramjet (SCRJ), where combustion takes place in supersonic conditions is required. In a scramjet, the air entering the combustor is supersonic: this means that in a residence time of about a millisecond air and fuel must mix and react. Hence, mixing in supersonic combustion plays a critical role on the the combustion efficiency and its understanding is critical to properly address a working engine. In this paper a rigorous analysis of the vorticity generation and transport in supersonic flows has been done in order to understand the key parameters to improve mixing and combustion. This 3D LES of the HyShot supersonic combustor performed by means of a fairly dense grid of 50Mnodes showed that the baroclinic term is the primary responsible of vorticity generation. In fact, interactions among the airstream entering the combustor and the H2 crossflow jet, the heat released and the shock waves produce a vorticity rate of order of 10 Hz. This vorticity generation is mainly due to the baroclinic term that creates spanwise vortices just upstream the H2 injection. These vortexes are afterwards tilted and stretched by the vortex stretching in the streamwise direction. LES predicts a very fast and efficient combustion: only 0.2% of hydrogen is found unburned at the combustor exit. Comparison of pressures distribution along the wall centerline at 1.32 ms shows a good agreement, mostly in the first part of combustor, where the grid is much more refined.

Shock/Boundary Layer/Heat Release Interaction in the HyShot II Scramjet Combustor

Previous computational work by these authors showed a very complex structure arising within the HyShot II combustor due to the interaction between the sonic crossflow injection and the airstream flowing at M = 2.78. In that work, a 3D Large Eddy Simulation (LES) of the HyShot II combustor was performed with a 14 ¤ 10 6 nodes structured grid. In the present work, in order to analyze in higher details the structures occurring within the HyShot II combustor, a fairly refined grid of 52 ¤ 10 6 nodes has been adopted. The LES simulations have been performed by means of a in-house code (S-HeaRT, Supersonic Heat Release and Turbulence): here, the LES model is based on a high order with lowdiffusion numerical schemes to accurately reproduce complex shock interactions, contact surfaces without artificially damping resolvable scales of turbulence. A hybrid method has been implemented to properly capture shocks while at the same time solving the transport equations away from discontinuities via a low dissipation, central scheme with fourth order accuracy. Simulations performed by means of the fairly accurated grid predicts very complex 3D flow structures arising within the flowfield due to the blockage induced by the H2 transverse injection. Ahead of each H2 injector a bow shock forms: the interaction of the bow shock and the boundary layer leads to a boundary layer separation zone where recirculation of H2 is allowed. In this recirculation zone, the presence of OH radical is predicted by LES indicating that a flame starts already in the injectors upstream the recirculation zone, downstream the flow separation. LES predicts the formation of barrel shock due to the H2 expansion: here the H2 jet expands untill it is definitively recompressed through the Mach disc. These flow structures, such us the bow and the barrel shocks, the Mach disk, the jet vortices and the horseshow vortices are in very good agreement with experimental results. Interactions among the airstream entering the combustor, the heat released and the shock waves produce a large vorticity rate that enhances and accelerates turbulent mixing. The vortex shedding, merging and tilting has also been analyzed, pointing out the contribution of the baroclinic term to the vortex generation and intensification. LES predicts a very fast and efficient combustion: only0.5% of hydrogen is found unburned at the combustor exit. Comparison of pressures distribution along the wall centerline at 1.2 ms shows a good agreement, mostly in the first part of combustor, where the grid is much more refined.

Unsteady simulation of a CO/H2/N2/air turbulent non-premixed flame

The Sandia/ETH-Zurich CO/H2/N2 non-premixed unconfined turbulent jet flame (named ‘Flame A’) is numerically simulated by solving the unsteady compressible reactive Navier–Stokes equations in a three-dimensional axisymmetric formulation, hence, in a formally two-dimensional domain. The turbulent combustion closure model adopted is the Fractal Model, FM, developed as a subgrid scale model for Large Eddy Simulation. The fuel is injected from a straight circular tube and the corresponding Reynolds number is 16 700, while the air coflows. Since the thickness of the nozzle is 0.88 mm, and the injection velocity high, ∼ 104 m s−1, capturing the stabilization mechanism of the actual flame requires high spatial resolution close to the injector. Results are first obtained on a coarse grid assuming a fast-chemistry approach for hydrogen oxidation and a single step mechanism for carbon monoxide oxidation. With this approach the flame is inevitably anchored. Then, to understand the actual flame stabilization a more complex chemical mechanism, including main radical species, is adopted. Since using this chemistry and the coarse grid of previous simulation the flame blows off numerically, attention is focused on understanding the actual flame stabilization mechanism by simulating a small spatial region close to the injection with a very fine grid. Then, analysing these results, an artificial anchoring mechanism is developed to be used in simulations of the whole flame with a coarse grid. Unsteady characteristics are shown and some averaged radial profiles for temperature and species are compared with experimental data.

Non premixed Supersonic flames: Combustion models

The aim of this work is to investigate the validity of combustion models that were developed for low-speed combustion and then traditionally extended to high-speed combustion. In fact, the assumption of fast chemist ry, as well as the flamelet chemistry model, must be validated in supersonic flows, where compressibility may affect the flame structure. LES of the HyShot test case, showed that the interactions between the airstream entering the combustor and the H2 sonic jet produce an average vorticity of order 10 5 Hz. The interaction between the hydrogen transverse jets and the supersonic air flow leads to bow shock formation and, accordingly, to boundary layer separation. This separation allows H2 to be convected upstream through the spanwise recirculation vortices created by the baroclinic effect. Once created, the vortices are t ilted, stretched, compressed and expanded according to the vorticity transport equation. Thes e vortices are the key structures responsible for the observed fast fuel air mixing. In this context, an analysis of the flame structure is of theoretical and numerical interest. In fact, depending on this structure, a appropriate kinetic and chemical/turbulence model can be chosen to correctly predict experimental results. The flame structure has been analyzed by means of the Burke and Schumann theory.

Mixing and turbulent kinetic Energy scaling in compressible reacting flows

Previous work by these authors analyzed in depth vorticity generation and transport in supersonic flows in order to understand the physics of supersonic combustion and to improve air-hydrogen mixing. In fact, the short combustor residence time (1010 s) minimizes the chance to completely mix and burn the fuel. Thus it becomes imperative to create there very energetic vortex structures, and a solution is to inject hydrogen in crossflow. In this paper, the 3D LES of the supersonic combustor flight tested in the HyShot project showed that the interactions between the airstream entering the combustor and the H2 sonic jet produce average vorticity of order 10 5 Hz, with much higher localized peaks. The interaction between the hydrogen jets and the supersonic airflow leads to a bow shock formation in front of each jet and boundary layer separation. This separation allows H2 to be convected upstream through spanwise vortices created by the baroclinic effect. Once created, vortices are tilted, stretched, compressed and expanded as predicted by the fully compressible vorticity transport equation. This paper is meant to expand and complement earlier works showing how vortices affect combustion and analyzing the main species distribution along the combustor. The spectral analysis of turbulent kinetic energy obtained by LES results demonstrates that where compressibility is not negligible, the turbulent kinetic scaling differs from Kolmogorov.

Theoretical and Numerical analysis of the Turbulence scaling in Supersonic

Supersonic mixing and combustion is critical to adv anced airbreathing propulsion systems able to push vehicles well beyond M=4. Research in this field is of interest around the world. In fact, vehicles capable of such speed are being tested now in the US (HyTech, HyV), Russia and the UK-Australia (HyShot), in Japan, India, China and Korea. EU is funding the project LAPCAT to study the feasibility of a long range hypersonic commercial transport. In a SCRJ, the air stream flow captured by the inlet is decelerated but still maintaining superson ic conditions. Since the residence time is very short (~1ms), the study of a efficient mixing and combustion is a key issue in the ongoing research in compressible flows. Due to experimental difficulties in measuring complex high-speed unsteady flowfields, the most convenient way to understand unsteady features of supersonic mixing a nd combustion is the use of computational fluid dynamics. The complexity of physics involved makes the proble m of considerable interest also from a numerical point of view. Therefore, resoluti on of a turbulent compressible reacting flow imply a threefold requirement: 1. a h ighly accurate non dissipative numerical scheme to properly simulate the strong gradients in the vicinity of the shock waves and the turbulent structures away from these discontinuities; 2. a proper modelling of the small subgrid scales for supersoni c combustion, including the effect of compressibility on mixing and combustion; 3. a highly detailed kinetic scheme accounting for the radicals formation and recombination to properly predict the flame anchoring. A hybrid method capable of capturing shocks and, at the same time, of resolving with low dissipation turbulent structures away from discontinuities has been implemented in the present paper. A new subgrid scale model accounting for the nature of the turbulent in compressible regime is p roposed. The introduction of detailed chemistry (the scheme of Warnatz, including 9 species and 38 reactions to account for radicals formation) results in more exp ensive computer run times and storage requirements. High velocity and density gr adients, and high hydrogen diffusivity also poses some numerical critical issu es. This work, based on this subgrid physical model and using LES shows that, in supersonic flows, the baroclinic and dilatational effects pump vorticity in the flow inf luencing the turbulent KE decay and the dissipative turbulence scale.