Search

A large amount of low-momentum boundary-layer flow ingesting into a flush-mounted inlet can cause significant total pressure loss and distortion to the extent beyond operability of a fan/compressor. To improve the quality of incoming flow into the engine, shape optimization of the surface geometry at the inlet entrance has been carried out using the discrete adjoint method; the inlet-floor shape is parameterized by the use of control points on B-spline surface patches. To resolve the complicated geometry flexibly and wail-bounded turbulent flow accurately, an overset mesh system is well-suited for integrating the flow analysis code, sensitivity analysis code, and grid modification tools. To enhance the convergence characteristics of the sensitivity analysis code, additional numerical dissipation for the discrete adjoint formulation is introduced. After using this optimization procedure, the new inlet yields a significant improvement in performance: a more than 50% reduction in flow distortion and a 3% increase in total pressure recovery. High performance at off-design conditions is also realized with only slight degradation, confirming the capability of the adjoint method for a practical design problem. Finally, the physical meaning and implication of the performance improvement are elaborated upon in relation to the flow characteristics resulting from the new design. DOI: 10.2514/1.J050222

For accurate and efficient computations of compressible gas--liquid two-phase mixture flows, the AUSMPW + and RoeM schemes (for which the accuracy, efficiency, and robustness have been successfully demonstrated in gas dynamics) are extended to two-phase flows at all speeds. From the mixture equations of state, a new shock-discontinuity-sensing term suitably scaled for two-phase flows is derived and its performance is validated. In addition, several numerical difficulties appearing in the development of the two-phase AUSMPW+ and RoeM schemes are analyzed and successfully cured. The two-phase AUSMPW+ and RoeM schemes are then efficiently preconditioned for the simulation of all Mach number flows by employing the existing AUSM or Hartenu-Lax--van Leer with contact restoration types of preconditioning strategies. Various gas--liquid two-phase flows, from highly compressible to nearly incompressible flow conditions, are tested. The numerical results show the accurate and robust behavior of the proposed schemes for all speeds of two-phase flows.

An adjoint-based design approach for the delicate treatment of complex geometry is presented by using an overset mesh technique. Overset blocks, such as collar and tipcap grids, which are commonly used in accurate drag prediction, are employed to evaluate the applicability of the proposed design approach to practical problems. Various pre-and postprocessing techniques for overset flow and sensitivity analyses are implemented to develop a robust gradient-based optimization method on an overset mesh topology. In preprocessing, overlap optimization, which can provide an accurate overset solution and enhance convergence characteristics, is adopted to automatically construct the block connectivity. A new postprocessing method, the spline-boundary intersecting grid scheme, is introduced by reflecting the ratio of the surface cell area for accurate prediction of aerodynamic coefficients and a convenient evaluation of sensitivities under a parallel computing environment. For the sensitivity analysis, the adjoint formulation for the overset boundary condition is implemented in the fully hand-differentiated sensitivity analysis code. A three-dimensional discrete adjoint solver on the overset mesh system is developed by exploiting the overset flow analysis techniques. Good convergence characteristics of the adjoint solver can be achieved by using the automatic construction process of block connectivity. The derivatives of aerodynamic coefficients can be obtained by an efficient and accurate postprocessing technique. The present overset adjoint formulation and flow analysis techniques are validated by comparing the flow and sensitivity analyses, as well as the design results, with those of a single-block case for a transonic wing. Finally, careful designs are carried out by minimizing the drag of a three- dimensional wing--body configuration. The design results successfully demonstrate the capability of the present design approach.

Detailed numerical simulations are conducted to investigate aerodynamic characteristics of unsteady-force generation by a two-dimensional flapping motion under a forward-flight condition. A realistic wing trajectory called the figure-eight motion is extracted from a blowfly's tethered flight under freestream. Computed results show complex vortical flowfields that exhibit very interesting and distinctive unsteady characteristics. Lift is mainly generated during downstroke motion by a high effective angle of attack due to translational and lagging motion. On the other hand, a large amount of thrust is abruptly generated at the end of upstroke motion. Vortical structure in the wake and the pressure field shows that vortex-pairing and vortex-staying mechanisms can be presented as strong evidence for the abrupt large thrust generation, which is fundamentally different from the inverse Karman vortex that is used to explain thrust generation by sinusoidal oscillating airfoil. Additional numerical simulations are conducted to examine the effects of motion components of figure-eight motion. From numerical results and comparisons, it is observed that wing rotational motion at the end of upstroke is crucial in generating the pairing and staying of vortices, which eventually leads to the abrupt thrust generation.

Unsteady, viscous, incompressible flows over an airfoil under flapping motion are numerically investigated. Depending on key parameters such as Reynolds number, reduced frequency, and flapping amplitudes, a flapping airfoil could experience complex flow fields. The trailing-edge vortex plays an important role to induce inverse Karman-vortex street which is a jetlike flow on the downstream and then generates thrust. And, the leading-edge separation vortex is closely related on the propulsive efficiency. Through careful computations of several pitching, plunging, and plunging combined with pitching modes in terms of flow and/or geometry parameters, the key physical flow phenomenon dictating the aerodynamic characteristics of flapping airfoil is identified. Based on the analysis of thrust coefficient and propulsive efficiency a new airfoil shape for optimal aerodynamic performance is proposed. The improved performance of the new flapping airfoil is validated in terms of thrust coefficient and propulsive efficiency in various low-Reynolds number flow regimes.

The present paper investigates methods to control dynamic stall using an optimal approach. Exploiting an un-steady aerodynamic sensitivity analysis based on a direct differentiation method from a two-dimensional unsteady compressible Navier-Stokes solver including a two-equation turbulence model, dynamic-stall control is conducted by minimizing an objective function defined at an instant instead of integrating for a period of time. Unsteady sensitivity derivatives of the objective function are calculated by the sensitivity code, and then optimization is carried out using a linear line search method at every physical time step. Adopting several control parameters such as nose radius, and maximum thickness of airfoil and suction, their effects on the delay or suppression of dynamic stall are examined in detail. Numerous results and comparisons indicate that the present approach yields predictive and more efficient results compared to conventional methods, and suction near the leading edge is essentially more effective than the geometric changes of airfoil in controlling dynamic stall.

A feasibility study is carried out by investigating the effects of a usual assumption of constant turbulent eddy viscosity on the aerodynamic design using an adjoint variable method, one of the most efficient gradient-based optimization techniques. Accurate unsteady and steady flow analyses are followed by the aerodynamic sensitivity analysis for the Navier-Stokes equations coupled with two-equation turbulence models. A challengeable approach for high-lift design optimization at higher angles of attack is also proposed, which is based on unsteady ensitivity analysis using a dual time-stepping method and the chimera overset grid scheme. Through the comparison of the sensitivity gradients with respect to all of the design variables including angle of attack, it is observed that the constant turbulent eddy-viscosity assumption might provide inaccurate gradients in sensitivity analyses such as transonic airfoil with a strong shock and high-lift airfoil at a high angle of attack close to stall angle. Simultaneously, however, the final design results indicate that both approaches are acceptable in engineering applications. Both the single- and multi-element airfoil design optimizations using the constant eddy-viscosity assumption are carefully assessed in terms of design accuracy, computer memory overheads, and total design time in various design examples.

The inverse design method is applied to a design problem of transonic wing planform and section shape design.

An inverse design method for a new dual-point design procedure for transonic wings is presented. Weighted averages of the residuals were minimized and target pressure distributions were optimized using a genetic algorithm. The characteristics of single-point design wingers were transferred to dual-point design wings in proportion to weighting factors.

Summary: Edge computing focuses on processing near the source of the data. Edge computing devices using the Tegra SoC architecture provide a physically distinct GPU memory architecture. In order to take advantage of this architecture, different modes of memory allocation need to be considered. Different GPU memory allocation techniques yield different results in memory usage and execution times of identical applications on Tegra devices. In this article, we implement several GPU application benchmarks, including our custom CFD code with unified, pinned, and normal host/device memory allocation modes. We evaluate and compare the memory usage and execution time of such workloads on edge computing Tegra system‐on‐chips (SoC) equipped with integrated GPUs using a shared memory architecture, and non‐SoC machines with discrete GPUs equipped with distinct VRAM. We discover that utilizing normal memory allocation methods on SoCs actually use double the required memory because of unnecessary device memory copies, despite being physically shared with host memory. We show that GPU application memory usage can be reduced up to 50%, and that even performance improvements can occur just by replacing normal memory allocation and memory copy methods with managed unified memory or pinned memory allocation. [ABSTRACT FROM AUTHOR]

An insect-type flapping wing micro aerial vehicle offers high aerodynamic efficiency and maneuverability in confined spaces. The complicated aerodynamic/structural behavior of flapping wing micro aerial vehicle, however, causes difficulties regarding the dynamic control and parametric design. This paper develops a moderately accurate numerical framework taking into account the passive motion of the main wings. Finite-element-based multibody dynamics and two-dimensional unsteady aerodynamics are combined to simulate the hover of a flapping wing micro aerial vehicle. In addition, flexible and rigid wings are compared through numerical simulation considering the flexibility. In terms of the average thrust, numerical simulation by fluid–structure interaction shows good agreements against the experimental results within 5% discrepancy. [ABSTRACT FROM AUTHOR]

Aerodynamic sensitivity analysis is performed for the Navier-Stokes equations, coupled with two-equation turbulence models using a discrete adjoint method and a direct differentiation method, respectively. Like the mean flow equations, the turbulence model equations are also hand differentiated to calculate accurately the sensitivity derivatives of flow quantities with respect to design variables in turbulent viscous flows. Both the direct differentiation code and the adjoint variable code adopt the same time integration scheme with the flow solver to solve the differentiated equations efficiently. The sensitivity codes are then compared with the flow solver in terms of solution accuracy, computing time, and computer memory requirements. The sensitivity derivatives obtained from the sensitivity codes with different turbulence models are compared with each other. Using two-equation turbulence models, it is observed that a usual assumption of constant turbulent eddy viscosity in adjoint methods may lead to inaccurate results in a case of turbulent flows involving strong shocks. The capability of the present sensitivity codes to treat complex geometry is successfully demonstrated by analyzing the flows over multielement airfoils on chimera overlaid grid systems.

Viscous turbulent flows over high-lift airfoils are investigated by using unsteady, incompressible, and compressible Reynolds-averaged Navier--Stokes equations under a parallel computing environment. Compressibility effects can be significant in the leading-edge region of high-lift airfoils with a highly loaded element. Thus, both the incompressible and compressible computations are performed to study the compressibility effects. The compressible code involves an upwind-differenced scheme for the convective terms and a lower--upper symmetric Gauss--Seidel scheme for temporal integration. The incompressible code with a pseudocompressibility method also adopts the same schemes as the compressible code. Both codes are parallel processed by using message passing interface programming method and show good parallel speedups. Three two-equation turbulence models (the standard k--[Epsilon], the k--[Omega], and the k--[Omega] shear stress transport model) are carefully evaluated by computing the flows over single-element and multielement airfoils. The compressible and incompressible codes are validated by predicting the flow around the RAE 2822 transonic airfoil and NACA 4412 airfoil, respectively. In addition, both the incompressible and compressible codes using the Chimera overlapping grid scheme are used to compute the flow over the NLR 7301 airfoil with flap and the NASA GAW-1 high-lift airfoil. Compressibility effects on surface pressure coefficients, velocity profiles, and skin friction coefficients are numerically simulated.

Notwithstanding the effectiveness of ventilated cavitation as a safe way to accelerate and achieve natural supercavitation, detailed physics of ventilated cavitating flows and hydrodynamic effects on underwater vehicles have been rarely explored. The present work deals with computational investigation on ventilated supercavitating flows around a high-speed underwater vehicle and its hydrodynamic characteristics. The homogeneous mixture model is adopted for describing the cavitating flow field composed of water, water vapor, and a ventilated gas. After extensively validating the flow solver with experiments of ventilated cavitating flows, we carry out a series of computations of supercavitating flows around a high-speed underwater vehicle with four control fins, and investigate the flow physics and the vehicle's hydrodynamic characteristics. The computations are performed under different freestream velocities, ventilation rates, and angles of attack. Furthermore, the unsteady effects of turbulence are examined by comparing the computed results based on unsteady Reynolds-Averaged Navier–Stokes (URANS) simulation and Detached Eddy Simulation (DES). The computational results show that the cavitating flows cause substantial nonlinear characteristics in cavity shapes and hydrodynamic forces such as lift, drag, and pitching moment, due to the presence of natural and ventilated cavitations and their interaction. This can be beneficially exploited in understanding and predicting ventilated supercavitating flows around a high-speed underwater vehicle, and establishing an accurate dynamic model of the vehicle's motion. • Computations of ventilated cavitating flows around a vehicle are presented. • All the results exhibit non-linear hydrodynamic characteristics of the vehicle. • As a cavity passes through the fins, noticeable drag and lift changes are observed. • As the AOA increases, the pitching moment changes from nose-down to nose-up. • URANS and DES computations are compared for a specific flow condition. [ABSTRACT FROM AUTHOR]

• A new cavitation model is proposed by adopting physical bubble growth rates. • Empiricism inherent in the bubble number density is substantially reduced. • Smooth temperature recovery in cryogenic cavitation is well captured. • Existing characteristics under isothermal conditions are kept. Cavitation in thermosensitive fluids is accompanied by a noticeable temperature drop inside a cavity. Motivated by the fact that most popular cavitation models have been developed without considering thermodynamic effects, the present study proposes physics-based corrections to existing cavitation models. Two primary corrections are adopting the physical bubble growth rate and separating bubble number density in the vaporization and condensation terms. The bubble growth rate in the cavitation model is chosen as the limiting factor between the inertia-controlled rate and heat diffusion-controlled rate from the bubble growth theories. To implement the theoretical growth rate of the thermal regime into the computational framework based on the homogeneous mixture model, the bubble time and liquid temperature are newly modeled. Furthermore, the bubble number density, which is often set uniformly throughout the whole computational domain, is separated for the vaporization and condensation processes. The vaporization bubble number density is further modeled by experimental data to reduce the empiricism inherent in the user-defined coefficients. The relation between the remaining model coefficients and characteristic temperature drop (Δ T *) is developed and it is verified to be effective in various types of fluids and working conditions. The newly proposed model is validated by cryogenic cavitating tests and applied to the computations of the 3-D cryogenic cavitating flows around a turbopump inducer. Unlike the sudden temperature recovery by the previous inertial rate-based models, the proposed model predicts a gradual temperature recovery inside a non-isothermal cavity, which is consistent with the experimental observations. It is also confirmed that the new model chooses the physical inertial rate under isothermal conditions, and thus, the model can be used regardless of the flow conditions. [ABSTRACT FROM AUTHOR]