Douglas A. Schwer

Towards Efficient, Unsteady, Three-Dimensional Rotating Detonation Engine Simulations

Rotating detonation engines (RDEs) represent an alternative to the extensively studied pulse detonation engines (PDEs) for obtaining propulsion from the high efficiency detonation cycle. Numerical simulations play an important role in understanding the basic physics of the RDE and will be important in optimizing the geometry and flow-field conditions for an RDE, but are necessarily unsteady and three-dimensional. The current paper describes efforts to develop a new code, Propel, for simulating complex engine designs. Propel has support for structured and unstructured meshes, several different numerical algorithms and limiters, and can be run on both CPUs and GPUs in HPC to laptop environments. This paper compares two and three dimensional solutions using Propel for a detonation tube and baseline RDE with our current RDE simulation tool. As an example of the capabilities of the new Propel code, we examine some preliminary calculations of the expansion flow region of a Rotating Detonation Engine as the expansion geometry of the combustion chamber is modified.

Physics of Heat-Release in Rotating Detonation Engines

Rotating detonation engines (RDEs), along with the well-studied Pulsed Detonation Engine (PDEs), represent an alternative for obtaining propulsion and power generation using the high efficiency detonation cycle. In the last half decade, the RDE has been studied aggressively internationally due to its advantages over conventional PDEs. Many previous studies have focused on the main characteristics and performance of RDEs in various configurations and conditions. The performance of any combustor is tied closely to the local heat-release conditions in the combustor. Unlike more conventional combustors or even the closely related Pulsed Detonation Engines, heat-release occurs under a wide range of conditions within RDEs from detonation to deflagration to even non-premixed burning, and so describing it solely as a “detonation cycle engine” is not completely accurate. In this paper we use a deflagration model in simulations with different configurations to try to more accurately pinpoint where heat is being released, and how this affects the performance of the resulting engine. In particular, we examine how changing the pressure, the injector configuration, and the mixedness of the reactants may change the different heat-release regimes and ultimately the performance of the RDE. Results show that for cases using the ideal injection model, low pressure reactions are insignificant and have only a small effect on performance. When computing finite-sized injectors, a separate low pressure reaction model has to be used due to overprediction of deflagration and recovery zone reactions that eventually cause the detonation wave to fail. In these cases, less than half of the reactants are burned within the detonation wave, although the detonation wave is still the dominant feature of the flow-field. Detonation wave speeds varied from 1637 m/s to 1844 m/s, and specific impulses varied from 4650 s to 4980 s for 10 atm feed stagnation pressure.

Area Effects on Rotating Detonation Engine Performance

Prior work shows that the thermodynamic cycle of the rotating detonation engine can be analyzed as a ZND planar detonation that is transformed into a rotating frame of reference. The resulting cycle model is independent of geometry and consistent with the Euler turbomachinery equation. A two-dimensional inviscid numerical model confirms the essential features of the analytical cycle. A subsequent examination confirms the operation of the Euler equation in a three-dimensional numerical simulation and is consistent with the modified ZND model. The geometry of the RDE is varied over nine different geometries of area and radius values. Specific impulse is shown to be proportional to area ratio. The range of exit swirl angles are shown to be proportional to the ratio of exit to inlet radii. The correlation of performance or swirl with geometry suggests other factors are equally important.

Implementation of Thermochemistry and Chemical Kinetics in a GPU-based CFD Code

The implementation of multi-species thermochemistry and chemical kinetics in a GPUbased CFD code is described, focusing on issues which are specific to GPUs. The physical model and its numerical formulation are described in detail. Performance results are presented for two multidimensional test cases: non-reacting supersonic flow over a forward-facing step and the reacting flow of a cellular detonation in a low-pressure H2-O2Ar mixture. The performance results are analyzed to determine the performance when simulations are run on GPUs versus CPUs, the scalability of the solver on both types of compute devices, as well as the cost of the thermochemistry model relative to simpler thermodynamic models.

Blast Loading on the Head Under a Military Helmet: Effect of Face Shield and Mandible Protection

In order to assess the affect of helmet system geometry on blast loading on the head, a study was performed computing the under-helmet pressures due to a blast event for various combinations of the helmet shell and liner, mandible protection, and face shield for the Conformal Integrated Protective Headgear System (CIPHER) prototype geometry. As in previous studies, pressure waves penetrate the gap between the head and perimeter of the helmet shell. These waves interacted with the head, the suspension geometry and other waves to generate a complex pressure field under the helmet. Significant variations in peak pressures under the helmet were predicted for various combinations of components and blast orientation. In some cases, waves trapped by the geometry produced an increased pressure when reduced pressure was expected. The presence of protective equipment often reduces the pressures in some areas on the head (typically the surfaces facing the blast source) while amplifying pressures elsewhere. Finally, wave reflections from the torso are shown to be critically important in assessing wave infiltration and peak pressures under the helmet.

Assessing Blast Loading Within Obstacle Arrays

The current study characterizes how pad size, shape, and spacing affect the pressure infiltration under a military helmet during a blast event. The pads are represented as an array of static obstacles in two-dimensions, and computational fluid dynamics is used to predict the interaction of a blast wave with the array. Unlike obstacle arrangements for defending the perimeter of an object such as a building, the pressures generated among the obstacles (and not just transmitted beyond the obstacle array) drive the performance assessments. Three baseline obstacle arrangements consisting of 7, 19, and 37 circular pads blocking the same total area have been defined. Testing indicates that the array of more, smaller pads experienced lower peak pressures at the “core” area at the center of the pad array as compared to the array of fewer, larger pads. Additionally, altering the shape of the perimeter pads demonstrated that both reducing the gap between the pads and presenting a flat face to the incoming wave reduce pressure infiltration into the array.

Efficient Utilization of a CPU-GPU Cluster

This paper will investigate the performance of a mixture of central processing unit (CPU) and graphical processing unit (GPU) codes on a multi-CPU, multi-GPU cluster. This cluster attempts to balance IO, GPU, and CPU performance to accommodate a wide variety of codes. When designing this cluster, the design goal of a balanced system was one of many options that could have been taken. The GPU, is essentially a video graphics card, found in every desktop or laptop computer. High-end graphics cards such as those used by a computer gamer are capable of extremely high floating point performance. The GPU utilizes the CPU to initialize the GPU, to transfer data from memory/storage to and from the GPU, and to launch the computation kernels that run on the GPU. The Jet Engine Noise Reduction (JENRE) code implements a compressible flow solver which is under development for the simulation of supersonic jet flow and its acoustic properties. A major emphasis of this codes development is on ensuring that the code is capable of fully exploiting emerging massively parallel, high-performance computing architectures for either GPUs or multi-core CPUs. The JENRE codes performance using GPUs is currently 2.1 times that with CPUs, and thus is run typically on the GPUs in the cluster. The cluster is also used for a variety of MPI-based jobs as well as single node OpenMP shared-memory jobs. These jobs utilize the CPU only, and the GPUs are left idle. Typically, a user requests that an entire node (or set of nodes) is allocated to a single job (CPU or GPU) so that there is no contention for resources with other jobs. Since jobs are either CPU or GPU-based, this leads to significant under-utilization of the computational resources. This paper will examine the overall utilization of the cluster and performance of a mix of CPU codes with the GPU-based JENRE code running simultaneously on the same nodes of the cluster. Results indicate that careful and cooperative scheduling of jobs can result in a tripling of the computational capability of the cluster.

Optimizing the Arrangement and Shape of Grooves in Microfluidic Components

The Toolbox approach to the automated design of microfluidic components chooses combinations of features based on a precomputed library of shapes in order to optimize tasks such as mixing and surface delivery in channel-based microfluidic components. This approach is extended to include parameterization of feature shapes, meaning that the design algorithm can now blend discreet shapes from the library to produce intermediate groove shapes. Based on the user-supplied performance metric (and not a user-supplied prototype), the Toolbox identifies promising prototypes and then refines them into optimized designs. Components previously optimized for mixing in pressure-driven flow using a limited set of groove shapes are compared to new designs that exploit the additional degrees of freedom in the geometries.