This work applies a new combination of techniques for the fully resolved simulation of compressible, gas- particle multiphase flows. The adaptive wavelet collocation method is used to dynamically, and efficiently, adapt the computational grid to localized flow features and the particles. A characteristic-based volume penalization method that imposes arbitrary Dirichlet, Neumann, or Robin-type immersed boundary conditions, is used to enforce the no-slip condition at particle surfaces. A hard-sphere collision model is applied to capture the particle-particle collisions. Proof of concept test cases are presented, showcasing the dynamic grid adaptation and fully resolved two-way coupling between the phases that is possible with this approach. Results for a shock-driven single cylinder under viscous and inviscid conditions are presented along with a demonstration of a shock interacting with a cloud of randomly distributed cylinders and spheres.

In this work, we measure continuous thermal radiance from evolving clouds of liquid metal fragments ejected into vacuum, nonreactive, and reactive gas. We implement a model for the thermalization of the ejecta and gas and use this to constrain the absolute temperature of the ejecta cloud. This model enables further analyses of ejecta thermal behavior under a variety of conditions.

Molten metal atomization in close-coupled gas atomization dies can operate between two limiting conditions, jetting and filming, together with several complex mechanisms: liquid-gas drafting, downward/upward shearing, melt bouncing, etc. Liquid jet deformation depends on flow and geometric parameters, such as liquid Reynolds, liquid Weber, and gas Mach numbers, as well as gas jet apex angle and melt tube tip extension and aspect ratio, among others. Understanding their effect is of importance for the metal powder making industry. Numerical gas atomization studies can provide approximated flow information and consider a wide range of conditions, beyond experimental reach. Here, 3D high-resolution simulations employing a 5-equation compressible flow model coupled with the volume-of-fluid method are compared with experiments, for liquid Weber number in the range of 1–30 and liquid Reynolds number below 10,000. This validation explores the predicting capabilities of the numerical model.

An important aspect of scramjet development is characterizing liquid jet atomization in supersonic crossflow under startup conditions. Numerical simulations may be used to better understand the flow behavior. To this end, an interface reconstruction approach extended from a Tangent of Hyperbola for INterface Capturing (THINC) reconstruction scheme for use with the five-equation model has been developed. Coupled with a Harten-Lax-van Leer-Contact (HLLC) Riemann solver–modified to include the effects of surface tension–this method maintains the thickness of the gas-liquid interface throughout the simulation without impacting the underlying conservation of the scheme. The atomization of a liquid jet in subsonic crossflow is simulated and validated against experimental results to ensure the relevant flow physics are captured accurately. A qualitative analysis of the overall flow structure and quantitative comparisons to measured values of liquid jet atomization are performed, including: surface wavelengths, penetration height, and jet trajectory. After examining the subsonic case, a liquid jet in supersonic crossflow is simulated to demonstrate the strength of the method for compressible atomization applications. ∗Corresponding Author: jregele@iastate.edu Introduction Challenges currently exist during scramjet startup conditions when liquid fuel is injected into a supersonic crossflow. As experimental tests are often limited in their ability to observe all of the relevant flow physics, numerical simulation can provide useful insights to the understanding of such flows. Abundant literature exists on the simulation of low Mach number primary atomization, typical to that of a majority of combustion applications. However, little work on the simulation of primary atomization in supersonic conditions has been completed with the effects of liquid surface tension. The development and validation of a numerical model suitable for primary atomization in supersonic flows is vital to further the understanding of the processes involved in supersonic atomization. The method presented utilizes an interface capturing approach which parallels shock capturing schemes already commonly used in compressible flow algorithms [1, 2, 3, 4]. Such an approach is straightforward to implement in compressible solvers and easily extends to multiple dimensions without the need for complex geometric calculations. However, a major drawback of such an approach is the smearing of material interfaces over time similar to contact discontinuities in a compressible solver. A number of techniques have been presented to prevent or reduce interface smearing. One approach is to use WENO [4] or other high order methods with less numerical diffusion. However, such an approach still suffers from interface diffusion over long simulation times. Alternatively, interface compression methods, such as the one presented by Shukla [5], provide another approach. These approaches are often not discretely conservative or must be tailored specifically for the numerical scheme and discretization employed. Another method is the Tangent of Hyperbola for INterface Capturing (THINC) reconstruction method [6]. This approach reconstructs the material interface assuming a locally hyperbolic profile within cells that contain the interface. This method was extended to ρ–THINC in previous work where a similar approach is followed in the reconstruction for the phasic densities [7]. The approach results in a method that is discretely conservative and can maintain the interface thickness to a mesh representable profile through time. The goal of this work is to validate the solver against experimental results for a liquid jet in crossflow. Previous work has shown good agreement of the solver for a variety of test cases as presented in Garrick et al.[7]. Although supersonic atomization is the focus of the solver, regions of subsonic flow are expected behind the bow shock in front of the jet in supersonic conditions. Furthermore a sizable body of literature has been generated on the simulation of liquid jets in subsonic crossflow providing a baseline for comparison [8, 9, 10]. A number of subsonic and supersonic liquid jet experimental results are available in the literature [11, 12, 13]. Among this body of work a wide variety of conditions have been studied: non-turbulent [11, 12], turbulent [14], pulsed [15], and injection direction [16], among others. The large number of parameters can result in a number of uncertainties when attempting to replicate a given experiment in a numerical simulation. Furthermore, significant differences in both simulations [8, 9] and experiments [17] resulting from a slight changes in inflow parameters can be seen, and often such parameters are not completely quantified in experimental results. These can require additional inflow domain simulation or imposing boundaries based on an a priori simulation. A non-turbulent round jet in crossflow was selected to reduce the uncertainty in replicating experimental results for validation. For this work comparisons are made to the work of Wu et al. [11] and Sallam et al. [12]. Wu et al. looked at a number of correlations for non-turbulent round jets and investigated parameters including jet trajectory and break-up lengths. Sallam et al. expanded on this and carefully worked to reduce experimental uncertainty by minimizing the crossflow boundary layer as well as providing additional correlations to experimental data including surface instability wavelengths. The remainder of this paper is organized as follows. First, a discussion of the governing equations and numerical approach is presented. This is followed by the methodology of the recently extended ρ–THINC used to counter the effects of numerical diffusion on the material interface. A validation of the solver is then performed on a liquid jet in subsonic crossflow comparing against experimental data of non-turbulent round jets. Finally the capabilities of the solver for compressible atomization are demonstrated in the simulation of a liquid jet in supersonic crossflow. Governing Equations For the problem considered, the compressible multicomponent Navier-Stokes equations with capillary force terms govern the fluid flow. The five equation model of Allaire [3] is employed to model the given two-fluid problem:

A fluid dynamics video is presented that demonstrates an indirect detonation initiation process. In this process, a transient power deposition adds heat to a spatially resolved volume of fluid in an amount of time that is similar to the acoustic timescale of the fluid volume. A highly resolved two-dimensional simulation shows the events that unfold after the heat is added.

Acoustic timescale Deflagration-to-Detonation Transition (DDT) has been shown to occur through the generation of compression waves emitted by a hot spot or reaction centre where the pressure and temperature increase with little diminution of density. In order to compensate for the multi-scale nature of the physico-chemical processes, previous numerical simulations in this area have been limited to relatively small activation energies. In this work, a computational study investigates the effect of increased activation energy on the time required to form a detonation wave and the change in behaviour of each hot spot as the activation energy is increased. The simulations use a localised spatially distributed thermal power deposition of limited duration into a finite volume of reactive gas to facilitate DDT. The Adaptive Wavelet-Collocation Method is used to solve efficiently the 1-D reactive Euler equations with one-step Arrhenius kinetics. The DDT process as described in previous work is characterised by the formation of hot spots during an initial transient period, explosion of the hot spots and creation of an accelerating reaction front that reaches the lead shock and forms an overdriven detonation wave. Current results indicate that as the activation energy is raised the chemical heat release becomes more temporally distributed. Hot spots that produce an accelerating reaction front with low activation energies change behaviour with increased activation energy so that no accelerating reaction front is created. An acoustic timescale ratio is defined that characterises the change in behaviour of each hot spot.

A simple and robust method for solving hyperbolic conservation equations based on the adaptive wavelet-collocation method, which uses a dynamically adaptive grid, are presented. The method utilises natural ability of wavelet analysis to sense localised structures and is based on analysis of wavelet coefficients on the finest level of resolution to create a discontinuity locator function phi. Using this function, an artificial viscous term is explicitly added in the needed regions using a localised numerical viscosity that ensures the positivity and TVD non-linear stability conditions. Once the wavelet coefficients on the finest level of resolution are below the error threshold parameter epsilon, the artificial viscosity is shut off and any remaining physical waves are free to propagate undamped. Multiple examples in one and two dimensions are presented to demonstrate the method's robustness, simplicity and ease of extending to more complex problems.