ABSTRACT. Rotorcraft icing on Earth and planetary landings in outer space are characterized by the two-phase flow of compressible air-droplet and gas-particle, respectively. Computational modeling of these flows is challenging due to large variations in temperature, particle concentration, including the near-zero limit, and flow velocity, as well as the complexities and nonlinearities of the flow involved in rotorcraft with rotor blades and planetary landers with rocket motors.

    The present lecture first considers a mathematical model of the non-strictly hyperbolic, pressureless gas dynamics type that governs dispersed phase transport within a continuous fluid phase. The main features of this model are occurrences of delta shock waves and vacuum states, which bring non-trivial numerical challenges. Next, it introduces a new hybrid aerodynamic solver based on the nonlinear vortex lattice method, vortex particle method, and Navier-Stokes-Fourier (NSF) CFD method to efficiently predict the flow field around the rotorcraft fuselage at any advance ratio, considering rotor wake effects. The hybrid solver is then unified with water droplet impingement and ice accretion solvers, all of which were developed in a finite volume framework.

In a planetary landing, compressible gas-particle flow is formed when the rocket plume of the lander impinges on a dusty surface and causes erosion and dispersal of solid particles into the flow field. The present study proposes a full continuum framework based on a finite volume approach for simulating plume surface interaction when a plume interacts with a bed of particles in the near-vacuum condition. It utilizes an Eulerian–Eulerian approach for modeling two-phase flow physics, which is computationally more efficient than the Lagrangian counterparts. It also effectively describes the high non-equilibrium effects in the rarefied condition of the planetary atmosphere based on the second-order nonlinear coupled constitutive relation (NCCR) beyond the conventional first-order NSF constitutive relation, which was systematically derived from the Boltzmann kinetic equations based on the closinglast balanced closure in a thermodynamically compatible manner.

ABSTRACT. Artificial Intelligence has played a key role from predicting, minimizing and stalling Pandemic outbreak such as Coronavirus to making autonomous vehicles a reality. The world of computing is going through an incredible change. With deep learning and AI, computers are learning to write their own software. This session will explore areas of building and improving deep learning models for applications related to scientific computing and physical systems defined by differential equations.

ABSTRACT. Low Reynolds number flows past UAVs, MAVs and drones have gained significant interest because of their surveillance, communication, weather monitoring, etc. We first study the transition of flow past an end-to-end wing [1]. The flow becomes unsteady beyond Re=600 approximately via the primary instability of the wake leading to vortex shedding. Three-dimensionality sets in at Re~1280.9 via mode C instability and hairpin vortices. The spatio-temporal structure of the flow is examined via spanwise variation of lift coefficient. Interaction of the shear layer vortices with the separated boundary layer leads to formation of a Laminar Separation Bubble (LSB) at Re~20,000, causing a delay in flow separation. The airfoil experiences a very significant increase in lift and decrease in drag.

            Wing-tip vortices form on a finite wing. We study their effect on the flow at low Reynolds numbers. Numerical simulations for flow past a finite rectangular wing with NACA 0012 section at Re=1000 for various semi-aspect ratios (0.25≤sAR≤7.5) over a range of angle of attack (0°≤α≤14°) reveal streamwise vortices, which increase in strength and number to occupy an increasing spanwise extent with increase in α [2]. Unlike the prediction from the Lifting Line Theory (LLT) by Prandtl [3], they result in non-monotonic spanwise variation of local force coefficients for α>8°. Viscous and pressure drag dominate for low- and high-sAR respectively. The time-averaged drag coefficient first decreases and then increases with increase in sAR. This is in contrast to the prediction from LLT and will be discussed in detail in the presentation. Vortex shedding, for α=14°, is single cell and parallel for sAR<3. Shedding is in two cells with oblique angle that varies with time, leading to large spanwise variation in rms of local force coefficients for higher sAR. Various type of dislocations, reported earlier in wakes of bluff bodies [4], are seen for different α and sAR. Dislocations, for α=14°, appear at same spanwise location for sAR=3 and at different spanwise locations for sAR≥4. Vortex shedding for α=14° and sAR=5 exhibits one cell structure in the near- and two-cell in the far-wake due to splitting and reconnection of vortices near midspan in the moderate wake. Linkages form between counter rotating spanwise vortices for sAR≥1. Additional linkages between shed- and wing-tip vortices are observed for lower sAR. The strength of wing-tip vortex increases with α. At each α, the strength and radius of its core, estimated using a forced-free vortex model, increases up to a certain sAR beyond which it is approximately constant. The incompressible flow equations are solved via a stabilized finite element method [5].

            The seam of a cricket ball affects the transition of boundary layer on a cricket ball and results in its swing and reverse swing. We show that the pressure difference on the seam and non-seam side results in wing-tip vortices. Similar flow structures are found on a frisbee as well.

ABSTRACT. Oberbeck-Boussinesq (OB) approximation for buoyancy-driven flows has been in use for many years. It essentially limits the involvement of density changes to buoyancy terms in the governing equations. The experimental data suggests this is a valid approximation for small density changes. However, attempts to mathematically reason OB approximation are not successful, leaving one to think of the effect of density changes in other terms, namely the mass conservation equation. The incompressibility condition is often assumed while analyzing OB-approximated thermally stratified flows. This talk looks at possible deviations from the divergence-free velocity field for non-Oberbeck-Boussinesq(NOB) Rayleigh-Benard flow. A pressure evolution equation for NOB flows is derived, which accommodates density changes due to temperature, while the density changes due to pressure are minimised. We will also show that the entropically damped artificial compressibility (EDAC) model results in a parabolic pressure evolution equation, which eliminates the need for the Poisson equation for other 'truly' incompressible flows. Along with Navier-Stokes and energy equations, the pressure equation, instead of the continuity equation,   forms a closed system to solve the thermal buoyancy-driven flow. We validate the method by simulating the NOB effects for Rayleigh Benard convection in air and water.

ABSTRACT. An approach leveraging a Machine Learning framework to enhance the CFD based prediction of intake dynamic distortion is presented in this study. A Neural Network model has been trained to predict the RMS turbulence pressures at Aerodynamic Interface Plane (AIP) for a fixed set of flow variables obtained using steady CFD simulations. The predicted RMS turbulence pressures are then employed to reconstruct unsteady pressure variation across the AIP, enabling the estimation of dynamic distortion using the reconstructed unsteady pressure signal. The proposed approach has been successfully validated against wind tunnel data, demonstrating a favorable agreement with experimental observations.

ABSTRACT. In this work, we demonstrate how physics-informed neural networks (PINNs) can solve various fluid dynamics problems at high-speed. The goal is to solve fundamental fluid dynamics problems by solving non-linear partial differential equations (PDEs) at high speed. We will deal with one-dimensional and two-dimensional Euler equations to solve Sod shock-tube problems, oblique shock, and expansion wave problems with vanilla PINNs and modified PINNs like weighted PINNs, and domain extension. The proposed method can learn the discontinuity present in the domain with dispersion error, but modified PINNs can minimize the error. This paper also demonstrates how the positivity condition can be enforced to solve the entropy equation across the shock. The proposed method solves the flow field for the problems mentioned above.

ABSTRACT. In Computational Fluid Dynamics (CFD), a field highly reliant on numerically solving partial differential equations (PDE) governing fluid motion, a new technique called Physics Informed Neural Networks (PINNs) has been developed (introduced by Raissi et al. in 2019), where the governing PDEs are solved by manipulating the loss function in a feed-forward neural network. In this paper, we perform PINNs simulations for steady flow past a circular cylinder and the lid-driven cavity test cases and compare the results with that from an in-house finite volume solver–PRAVAHA. The simulations of the lid-driven cavity at Reynolds numbers 100 and 400 and that of laminar flow past a cylinder at a Reynolds number of 20 are presented. We compare the velocity contours of the FVM solution with the PINNs solution. It is observed that the PINNs solution matches closely with the FVM solution and experimental data in the case of low Reynolds numbers.

ABSTRACT. Deep learning has gained significant interest in the modeling and analysis of fluid turbulence. One application is using super-resolution (SR) algorithms to reconstruct small-scale structures from their larger-scale counterparts in turbulent flows. However, existing SR algorithms either require supervised training or unpaired high-resolution reference data, limiting their practical use in fluid flow scenarios. Therefore, it is crucial to develop physics-guided models that leverage the multi-scale nature of turbulence. In this study, we propose a self-supervised workflow based on deep neural networks for reconstructing small-scale structures in homogeneous isotropic turbulence. We evaluate the quality of the reconstructions using various statistical metrics and find a good agreement with the ground truth data, despite not being included in the training. This research paves the way for reconstructing small-scale structures from large-eddy simulation data.

ABSTRACT. In this work, we explore the application of shallow decoder (SD) Erichson et al. [1] in reconstructing the full flow field using surface measurements. The Re = 100 flow past a circular cylinder is used for the exploration. Using SD, the mapping between vorticity - vorticity, vorticity - pressure, pressure - pressure, vorticity - vorticity & pressure is learnt. The results show that SD can reconstruct the flow variables to a good accuracy.

ABSTRACT. Blast waves are pressure waves expanding outward resulting from an explosion. An example of such explosion is the Beirut explosion which is a bear witness to the destructive power of the blast waves to life and property. Sudden raise and drop in pressure are the characteristics of blast waves which leads to long term damage to internal organs. Creating a blast wave using shock tube is the simplest and safest means to study its effects on the materials and structures. The present study explores the effectiveness of the shock tube to generate blast waves in a safe, consistent and inexpensive manner, which is detrimental for studying blast wave phenomena and developing blast mitigation strategies.The idea of modifying compressed gas driven shock tube to generate a blast signal is the main focus of the study. The blast tube relations yield the driven section length, peak over pressure and decay time which are then used along with Kingery-Bulmash equations to obtain the impulse, wave form factor, TNT equivalent and the stand-off distance. Friedlander wave profile is hence plotted analytically which is then compared with CFD data for both viscous and inviscid simulations.

ABSTRACT. The performance of unpowered scramjet integrated vehicle depends upon aerodynamic parameters and associated fidelity. Internal flow through the scramjet duct plays a significant role in overall aerodynamic coefficients. The data obtained from CFD simulations and experiments is compared. Simulations are carried out using commercial as well as in-house packages. Cumulative distribution and viscous contributions are discussed to understand the differences between the data obtained from various computational sources.

ABSTRACT. Mach 6.47 flow over a hollow cylinder has been studied with OpenFOAM's conjugate heat transfer (CHT) module (chtMultiRegionFoam) to estimate the heat flux and temperature transients. The CHT solver was validated by comparing the results with experiments and a density-based solver (rhoCentralFoam). The computed wall pressure, temperature, and heat flux results agree well with the experimental data.

ABSTRACT. The spreading and entrainment of jet strongly depends on the variations in ambient conditions. Further, the strong acoustics waves influence the formation of Kelvin-Helmholtz vortices at the free-shear layer of the jets. Here, the modifications in transient jet emanating from an open ended shock tube due to the interaction of blast wave reflected from different walls of an enclosure is numerically simulated using the inviscid and the one equation Spalart-Allmaras turbulence models for a diaphragm pressure ratio of 40. These results are compared with the transient jet in an open atmosphere. Simulations were carried out using the commercial software ANSYS Fluent with a structured mesh containing 175000 elements and 176541 nodes. The shock tube has a driver and driven section length of 200 mm and 1165 mm. It is found that the mean flow characteristics were similar during the flow development in both models. However, a huge reduction in strength of vortical structures is noticed with Spalart-Allmaras model during later stage of evolution. The unsteady pressure load during impingement of jet is also reduced drastically in Spalart-Allmaras model due to strong dissipation of vorticity.

ABSTRACT. Development of air breathing propulsion is one of the niche areas in the field of aerospace engineering. Few space organizations around the globe were able to demonstrate the technology and one among them is ISRO, India. A two staged rocket with booster and fins, scramjet engines mounted on the sustainer stage was used for the demonstration. CFD studies were carried out to estimate the overall aerodynamic characteristics of the vehicle using PARAS3D software for the flight scale model at an angle of attack of 2deg and post processing using PARAVIEW software. In this paper, actual test flight conditions (i.e. altitude data corresponding to flight trajectory) were used in the CFD simulations. Cartesian grid was generated to study the configuration. Grid independence, Solution convergence is achieved. Flow features at select Mach numbers are discussed. Also, variation of the drag coefficient with respect to Mach numbers and the cumulative drag force distribution over the scramjet engine and the test vehicle is clearly brought out in this paper.

ABSTRACT. This paper presents the extension of the in-house solver, ACTFlow, to simulation the high-Mach number non-equilibrium multi-species flows that occur during the re-entry of spacecraft into planetary atmospheres. The two-temperature model is utilized to account for thermoal non-equilibrium, and finite-rate chemistry model is used to capture the chemical reactions. The developed solver has been validated by solving various benchmark problems, demonstrating its ability to predict hypersonic flows accurately.

ABSTRACT. This study delves into the exploration of a novel automated process to analyze and optimize the design of a high-speed intake duct for a hypothetical TBCC propulsion system. Our approach incorporates the versatile CAESES as an integration platform, seamlessly blending its parametric geometric building functionalities with GridPro for the automatic creation of a structured mesh, and harnessing the computational power of MISTRAL CFD solver for accurate simulations. A diverse set of 20 distinctive variations were meticulously generated and subjected to comprehensive analysis during this investigation. The findings unequivocally demonstrate a remarkable advancement in the pressure recovery coefficient of the duct, showcasing an impressive 6 percent boost at the throat and a notable 4 percent enhancement at the outlet.

ABSTRACT. FSI optimization of a wing-winglet configuration integrating aerodynamics and structures, is presented. Aerodynamic shape optimization by maximizing the L/D is combined with a structural problem of minimizing the bending stress at the wing root in a multidiscplinary optimization framework (MDO). The optimized geometry which is the L-solution of the Pareto front shows a 1.2% improvement in L/D and around 39% reduction in the root bending stress

ABSTRACT. Adjoint based sensitivities provide a quick way for identifying the design hot-spots which require modification. The critical components based on a drag based cost functional have been identified for a typical trainer aircraft configuration based on the solution of a discrete adjoints of the RANS solution. Modification of the geometry has been proposed based on drag sensitivities. Critical drag hot spots have been identified and the geometry modified. The modified geometry shows reduced drag sensitivities at the critical spots. The final geometry shows around $72$ counts reduction in drag.

ABSTRACT. Designing and Development of Aerodynamically efficient scramjet inlets is very essential as the inlet is the part which determines the pressure at combustion chamber. Also, keeping in account about the structural and other aerodynamic penalties the inlet has to be designed. As the combustion chamber pressure is important factor which determines the efficiency of the combustion, the inlet compresses the air which is then passed on to the combustion chamber through Isolator. Hence, developing an inlet which has a better total pressure recovery at the Isolator end and compresses air significantly for proper combustion is required. This study aims to get an optimized geometry for scramjet inlets as a preliminary design procedure. The pressure validation is done with the experimental data to verify the computational results. For this, meshing is done using commercially available software and for the computation Ansys FLUENT is used. Theoretical formulation and calculation for both constrained and unconstrained geometrical optimisation is done using MATLAB software and its inbuilt Optimisation Toolbox. To optimize the geometry Evolutionary Algorithm i.e., Genetic Algorithm is used to get maximum total pressure recovery. The optimized geometry is then taken to the computational study to check the flow physics. For this unconstrained Inviscid simulation is done and also shown that there is increase of around 60% of the total pressure recovery better than the baseline geometry. Further, the study deals with comparison of different turbulence models in predicting the flow physics of the inlet. Then the constrained optimisation is done for the baseline model considering ramp angles as the design variables. Here, physical aspects like compression ratio, minimum temperature required for the combustion, minimum mass flow rate, pressure losses are taken into consider for the study.

ABSTRACT. Design and development of air intake is one of the most crucial requirements of any air breathing propulsion system. The performance of the intake ultimately decides the performance of the propulsion system and the aircraft as a whole. Intake design affects both internal and external aerodynamics of the aircraft and hence the design problem involves a large number of variables and multiple design requirements. The current study looks into the design and optimization of S-Duct intake. Sensitivity study is carried out using full factorial sampling and optimization for two design variables for multiple Length-to-Offset ratios and entry face aspect ratios are considered in the design. Computational Fluid Dynamics (CFD) analysis is conducted for all the configurations at multiple angles of attack and side slip angles and the corresponding data is evaluated to arrive at an optimized design. Total pressure recovery and Total pressure distortion coefficients are evaluated for all the configuration and the results are discussed.

ABSTRACT. Breakthroughs in aerodynamic optimization have made it possible to develop efficient modes of transport with lesser exploitation of valuable resources. This makes it crucial for technical professionals such as engineers and scientists to understand the methodologies behind carrying out such optimizations. A common approach towards improving the aerodynamic properties of a vehicle is to alter its physical shape, which has concurrently been a very strenuous process given the time consumed to remodel the vehicle for each simulation process. This research aims to tackle this problem by using intelligent techniques to automate the step-by-step process of remodeling the car and arriving at a final optimized solution with a significantly lower drag coefficient, a quantity used to measure the amount of drag force acting on a vehicle. This is achieved by assigning particular parameters to ensure guided improvement of the airfoil in a process known as parametrization, followed by implementing a response surface methodology primarily to circumvent the strenuous task of performing a large number of CFD simulations by employing surrogate models to generate a response surface between selected independent variables. Further, evolutionary algorithms such as Genetic Algorithms have gained momentum in the optimization studies carried out during product design by selecting the optimum parameters from the available design spaces on the basis of natural evolution. The proposed method of optimization has been implemented on a prototype vehicle with the aim to reduce drag and increase efficiency.

ABSTRACT. Use of asymmetric deflection of airbrake for generating yawing moment in transonic and supersonic Mach number regimes for a generic fighter aircraft, is explored using computational data and experimental results from wind tunnel tests. When the aft mounted airbrakes on a fighter aircraft is deflected on one side, it generates drag and affects the pressure distribution on the vertical fin leading to the generation of finite side force. The drag generated by airbrake and the side force generated by the resulting pressure distribution on the fin contributes to the generation of finite yawing moment. Computational studies were carried out to simulate the asymmetrically deflected airbrake geometry, to understand the effects of both drag and unbalanced side force on the overall aerodynamic characteristics of the fighter aircraft. The results from the CFD computations were validated against that from the wind tunnel experimental results obtained for the same configuration. Appropriate utilization of such unconventional means of generating yawing moment can be used to design a rudder with smaller surface area and lower capacity rudder actuator.

ABSTRACT. The RANS CFD simulations are performed to investigate the effect of cross wind on aerodynamics of aircraft at wide range of sideslip angles. The effect of free stream velocity is also investigated. The computed results obtained are compared with available wind tunnel experimental data. It is found that computed results match quite well with wind tunnel data.

ABSTRACT. Numerical estimation of airspeed and altitude error on transport aircraft due to external store installed near Static port.

Two semi-cylindrical shaped external stores (one on each LH and RH side) of Pollution Surveillance System (PSS) is to be installed on fuselage sidewall of a transport aircraft.

To measure the airspeed and altitude accurately, the static ports must be away from changes in airfield due to external store installation. Therefore, to study the flow interactions and static pressure on fuselage wall, CFD simulations were carried out by iterating locations of external store on the aircraft wall near static port zone and results of each configuration were compared with clean configuration.

A semi-cylindrical domain, an unstructured tetrahedral mesh with prism layer over half body model in ICEM CFD was prepared. The computational time was reduced significantly using half body model. High level of clustering at external store was used to capture the flow interactions accurately over the operational flight envelope of AoA (α) and AoS (β). The mesh was solved in Ansys CFX solver and results were processed in CFX Post.

ABSTRACT. This study aims to evaluate the performance of the Entropic Lattice Boltzmann Method (ELBM) in predicting the aerodynamic forces on a NACA0012 airfoil at high angles of attack. To validate the ac- curacy of the ELBM results, we compare the lift coefficient obtained from simulations with wind tunnel experiments conducted by Sandia National Laboratories [1]. As the angle of attack increases, the lift coefficient of the airfoil becomes nonlinear. At high angles of attack, the lift coefficient experiences a sudden drop, known as the stall. Our focus is on simulating the turbulent flow near the stall region of the airfoil using the higher-order ELBM solver developed in-house by SankhyaSutra Labs (SSL). We choose two angles of attack, 8.2 and 13.2 degrees, to simulate the linear and nonlinear regions of the lift coefficient, respectively. The simulation was performed on the Rudra HPC facility by SankhyaSutra Labs. By investigating the flow characteristics in the near stall region, we aim to provide insights into the stall behavior of the NACA0012 airfoil and improve the accuracy of aerodynamic modeling using ELBM.

ABSTRACT. The work presented here has been carried out to investigate the effect of cavitation on the performance of hydrofoil which are majorly employed cross-section for control surfaces such as rudder, active fin stabilizers etc. in marine applications. Cavitation occurs in a flow when local static pressure falls below the vapour pressure of water. Cavitation causes the reduction in lift force & severely deteriorate the lift generating capability of hydrofoil. In this paper, steady state two-dimensional multi-phase CFD study has been carried out on NACA 0018 profile considering various free stream velocities to vary cavitation number. Commercial CFD tool ‘ANSYS Fluent v19.1’ has been used to analyse flow & to calculate hydrodynamic forces & moment. K-w SST turbulence model has been used to solve turbulence numerics. CFD Post has been used for post-processing activities. Initial few sections are dedicated for introduction & validation of the results obtained from CFD with experimental results for Göttingen profile 460. In subsequent sections, CFD domain & meshing has been discussed and results have been presented in form of coefficient of lift, drag & moment. The effect of cavitation on the performance has also been reported.

ABSTRACT. In this work, analytical Jacobian for the implicit operator of AUSM family schemes, namely AUSM, AUSM+, AUSM+ -up and AUSM+ -up2 are implemented in the unstructured grid framework of open-source CFD code SU2. These implementations are assessed for standard test cases, ranging from very low-speed flows to hypersonic flow conditions. Test cases include inviscid flows as well as Laminar and RANS computations. Detailed convergence assessment is carried out against inconsistent Jacobian due to Rusanov and Roe flux. Brief assessment shows the advantage of consistent Jacobian over inconsistent Jacobian; various observations are also highlighted in detail.

ABSTRACT. A simple modification to HLL-CPS scheme for computing low speed viscous flows is presented here. The modification is based on scaling the velocity jump across the cell interface by a local Mach number function. The modified scheme shows significant improvement in accuracy at low Mach number. The scheme retains the robustness of the original HLL-CPS scheme at high speed. A set of numerical test cases are solved to demonstrate the efficacy of the present scheme.

ABSTRACT. A finite volume CFD solver comprises of numerical schemes for gradient calculation, reconstruction, inviscid flux and time integration. Each of these introduce numerical dissipation. This affects the solution accuracy. Different numerical schemes and implementations vary the amount of dissipation introduced. In this study, effect of changing the stencil used for Venkatakrishnan limiter in reconstruction as well as gradient scheme on overall coefficient of a winged body, a launch vehicle and heatflux on a blunt wedge are discussed.

ABSTRACT. Artificial Compressibility Methods (ACM) rely on an artificial equation that links the pressure and velocity fields to model incompressible flows. These hyperbolic/parabolic equations can rapidly converge to a ‘nearly’ divergence-free flow field in contrast to the methods based on the elliptic pressure Poisson equation. We compare the computational efficacy of different ACMs, namely, the Bulk Viscosity ACM (BVACM) and Entropically Damped Artificial Compressibility (EDAC) recently proposed in the literature. These are implemented in the in-house high-order finite difference solver, COMPSQUARE, and are validated for the test case of a 2D double periodic shear layer (DPSL). Although both ACMs give competitive predictions, the divergence-free velocity field from BVACM is superior to that obtained using EDAC. We further extend EDAC formulation to handle distorted and dynamically deforming grids and verified its performance on the test cases of a 3D Taylor Green Vortex (TGV), and a 3D pitching airfoil (PA). The results of EDAC on dynamically deforming grids are observed to be in encouraging agreement with the reference solutions on static grids. A comprehensive comparison of different ACMs across 3D test cases will be included in the final version of the manuscript.

ABSTRACT. The robustness and accuracy of the meshfree least squares kinetic upwind method (LSKUM) depends on the condition numbers of the weighted least-squares matrices associated with the approximation of spatial derivatives in the upwind scheme. In computational domains with a highly stretched or anisotropic distribution of points, the least-squares matrix experiences large condition numbers, which results in either loss of accuracy or code divergence. This paper presents the development of optimally weighted LSKUM with minimal conditioning. The optimal weights that yield minimal condition numbers are found using discrete adjoint based optimisation. Numerical results have shown that the LSKUM with optimal weights resulted in a more accurate solution than the current strategies for choosing weights.

ABSTRACT. We apply the recently developed mode multigrid method to accelerate the convergence of a linear convection-diffusion equation to the steady state. Mode multigrid is a data-driven technique that constructs reduced order representation of the flow field using the solution vectors available at different time instants. We show that the additional time spent building the reduced order representation and other related operations is negligible compared to the total simulation time. A speedup factor of more than 4.5 is achieved for the simulation on the considered fine mesh.

ABSTRACT. Flow energy harvest is one of important renewable energy source.  A typical way to harvest flow energy is to use a wind turbine.   In the past decade, several novel ways to convert flow energy to electricity have been proposed.   Flow-induced vibration is one of novel ways.  A blunt body is often used in this new approach.   Due to flow-induced vibration, the blunt body can be vibrated and then drive a power take-off device [1].   In order to obtain maximum vibration, the so-called lock-in phenomenon is expected to move the blunt body.   The lock-in phenomenon only happens when the flow-blunt body interaction is within a specific range.

In practice, not only a blunt body is used in a flow energy harvest device, but two or more are included.   To let those blunt bodies vibrate in the lock-in condition, it will need some special approaches to reach the purpose.  In this study, two novel approaches are proposed to enhance vibration of bunt bodies in the lock-in condition [2].  They are passive ways, so no extra energy is needed.   To validate the passive ways in the enhancement of flow-induced vibration, the in-house computing code, Tiger-C, was built.  It is based on the finite volume method and parallel computation algorithm is incorporated in the code.  Open-MP is used to parallelize the code.  Since the flow-blunt body interaction problem is 3D and turbulent, numerical simulations were undertaken in TAIWANIA, the high performance computer cluster in Taiwan.  In conclusion, the proposed passive ways can provide the flow energy converter based on flow-induced vibration to harvest more energy from turbulent flow.

REFERENCES

[1] M. M. Bernitsas, K. Raghavan, Y. Ben-Simon, and E. M. Garcia, Journal of Offshore Mechanics and Arctic Engineering, 130, pp.1–15 (2008).

[2] Y.H. Irawa, Y.H. Chiu, S.A. Raza and M.J. Chern, Physics of Fluids, 35(8), 085124 (2023).

ABSTRACT. The invent of programmable capability of Graphical Processing Units (GPUs) has paved new opportunities in the domain of CFD. It is preferred by the high performance computing community to off-load the computationally demanding tasks onto these many core accelerator based processors. In the present paper, we have adopted OpenACC programming model to port a portion of our FORTRAN code developed for Staggered Update Procedure (SUP) i.e. a higher order accurate finite volume solver. The identification of the data parallel region and the porting of the compute intensive task i.e. the gradient calculation routine onto GPU accelerators are presented in detail. We could establish superior algorithmic scalability with CPU+GPU heterogeneous system for 2D scalar convection-diffusion equation. The results show that a speedup of 8.5 is obtained with respect to explicit residual calculation on one NVIDIA Tesla V100 GPU.

ABSTRACT. Wall-modeled large-eddy simulations of the NASA 67 transonic fan stage are performed using the GPU-accelerated moving-mesh compressible flow solver Fidelity CharLES. The computational mesh is generated by computing Voronoi diagrams which provides both computational efficiency and flexibility. The solver employs a low-dissipation, nonlinearly stable numerical scheme. A formally second order accurate extended gradient operator for all interior cells combined with an efficient one-sided compact gradient operator to treat the stationary-moving part interfaces provides accuracy and flexibility in building part interfaces and is particularly suitable for turbomachinery simulations. The performance curve of the NASA stage 67 at 100% rotation speed will be calculated and compared against experimental measurements. The computational cost of the GPU-accelerated solver and its comparison with the CPU-based solver will also be reported in the full paper.

ABSTRACT. Graphical Processing Units (GPUs) are now commonly used to accelerate fluid flow simulations. In this work, we present a case study of the performance results of an in-house CFD solver - PRAVAAH code, on Nvidia A100 and V100 cards using OpenACC directives. PRAVAAH is a modern high fidelity, compressible flow, high-order finite-difference, LES solver. OpenACC helps to maintain the original form of code and simplifies porting and optimizations with minimum intrusion through compiler directives. Special attention is given to code restructuring for increasing parallelism and data movement optimization. Through the code optimization, we were able to achieve more than 50× speedup compared to a single CPU.

ABSTRACT. This study concerns a computational investigation of a supersonic flow over a compression-expansion corner at freestream Mach number 2.85. The flow is turned at both compression and expansion corners by an angle of 20 degrees. Steady and unsteady RANS simulations with a well-validated SST k-ω turbulence model are used for the current investigation. The combined effects of successive distortions on the boundary layer are analyzed along with static pressure distribution and skin-friction coefficient distribution along the ramp. Simulations suggest strong changes in the boundary layer due to distortions and it is relaminarized after the expansion corner.

ABSTRACT. High order methods have garnered significant interest recently for simulating compressible flows due to their low dissipation nature. In this regard, the use of discontinuous Galerkin method has been actively pursued due to the advantages ensuing from its compact interpolation. In this work, we present a performance comparison of a set of open-source compressible flow solvers: an emerging discontinuous Galerkin solver FLUXO, and the popular finite volume solvers OpenFOAM (specifically rhoCentralFoam) and SU2, in the context of viscous shock tube simulations. To this end, we perform an error vs cost analysis using these three solvers on different sets of meshes. We find that while OpenFOAM and SU2 perform similarly as regards the accuracy/cost metric, FLUXO (using cubic basis polynomial degree) outperforms both these two solvers, consuming an order of magnitude less cost to achieve the same accuracy. These findings encourage the use of high order disconsinuous Galerkin discretisation for transient viscous compressible flow simulations.

ABSTRACT. The significance of shock wave boundary layer interactions (SWBLI) for supersonic transport vehicles' aerodynamic performance and fatigue life cannot be overstated. In this study, we investigate the impact of an oblique shock wave impinging on a turbulent boundary layer across a range of Mach numbers and deflection angles using Direct Numerical Simulation (DNS) with an open-source solver. Our chosen flow conditions produce a mild separation of the incoming boundary layer. Our numerical results for wall pressure and pressure rms are validated against experimental findings. We determine the interaction length for each case by analyzing Schlieren images and time-averaged - spanwise averaged velocity field contours and find that it increases as the deflection angle increases and decreases as the Mach number increases. Our study further investigates the physical phenomenon that is predominantly responsible for a long-studied low-frequency unsteadiness. Our visualizations of the interactions reveal the vigorous flapping motion of both the reflected and incident shock and the discharge of coherent structures at the separation point, indicating this instability.

ABSTRACT. The Aerospike rocket engine is conceptualized in the quest for innovations to reach orbit using a single-stage rocket. The Aerospike nozzle is known for its altitude compensation ability. This study develops a numerical approach to design an axisymmetric aerospike nozzle using a method of characteristics in conjunction with stream functions. The nozzle used in this study is designed for exit Mach number 3 and 10 bar chamber pressure. This algorithm gives efficient contours for user-defined exit Mach number and its accuracy is verified from the results given by Fluent. A numerical simulation is performed to analyze the flow behavior using computational fluid dynamics CFD code ANSYS FLUENT. A 30% truncated aerospike nozzle with two pressure tappings at spike contour is 3D printed with a resin material for experimentation purposes and Schlieren flow visualization is set up to visualize the flow downstream of the nozzle. The pressure data acquired from the experiment is then validated with numerical data.

ABSTRACT. Transition control is crucial for hypersonic vehicles to protect the surface from the severe thermal and mechanical stress that will be induced due to transition or to enhance the mixing in combustion. In this work, the effect of suction, which is an active technique, on boundary layer transition is investigated using the recently published modified γ-model. The computational analysis is done using a well-validated in-house Navier-Stokes solver implemented with various transition and turbulence models. The effect of the suction slot’s location, width, numbers, and suction pressure is analysed. The results show that suction slot location plays a major role in delaying the transition compared to other parameters.

ABSTRACT. Shock Wave Boundary Layer Interactions (SWBLI) occur due to the convergence of an inviscid shockwaves and a viscous boundary layer. The resulting separation bubble is a major setback for the performance of high-speed air intakes. This paper presents the numerical investigation of effectiveness of a Pressure Feedback Technique (PFT) to mitigate the shock induced flow separation in a Scramjet engine intake at Mach 4. PFT is a novel self-sufficient simultaneous suction and injection flow control technique. The two main output parameters used to support the hypothesis of the present study are the size of the separation bubble as well as the total pressure recovery at the isolator outlet. The most prominent observation of this study is that with the installation of the PFT channel, the static pressure at the exit is noted to rise by almost 30% without having significant drop in total pressure recovery. The effectiveness of the pressure feedback technique in reducing the intensity of the separation bubbles is found to be depending on the diameter of the PFT, pitch to diameter ratio and PFT wall temperature.

ABSTRACT. Aircraft drag is calculated on a book-keeping method so that the drag and thrust are accounted for correctly and only once. The drag is split in two components, throttle independent and throttle dependent drag. The throttle independent drag is accounted in aero data set and the throttle dependent drag consists of spillage drag and aft-body drag which are accounted in propulsion data set. Estimation of spillage drag is well established for axis-symmetric air intakes through empirical methods. However, it remains a challenge to predict spillage drag for complex air intake geometries found in today’s fighter aircraft. In this paper we describe a method to easily compute spillage drag from CFD simulations. The method is validated by comparing the drag values computed through CFD and experiments on an axis-symmetric intake. This method can be used to compute spillage drag for any kind of air intake geometry and is not restricted to axis-symmetric intakes.

ABSTRACT. This paper presents a methodology for aerodynamic optimization of a long endurance UAV wing employing a genetic algorithm. It is well known that genetic algorithm-based optimization procedure requires a large number of function evaluations to arrive at a globally optimal configuration. In the present study, the function under consideration is the endurance parameter which depends on the lift and drag coefficients of the wing. Use of Computational Fluid Dynamics (CFD) tools for function evaluation is prohibitively expensive. To address this, a low-fidelity aerodynamic tool based on numerical lifting line theory (NLLT) working in conjunction with a high fidelity sectional aerodynamic data set generated using RANS solver is developed. NLLT significantly reduces the cost of function evaluation without compromising accuracy. The endurance parameter of the optimal wing obtained through a GA-based optimization procedure employing the NLLT tool is compared with CFD, and the results show a very close match between them. This clearly brings out the efficacy of the present procedure to obtain an aerodynamically optimal wing for long endurance UAV configuration. In addition, the study also establishes the need for considering a large population size for a reliable GA-based optimization process for achieving optimal solution.

ABSTRACT. The ground effect studies have been a talk of subject in recent years due to its improved aerodynamic efficiency during take-off and landing approaches for an unmanned air vehicle designed with a high aspect ratio wing. Towards this, an attempt has been made to understand the ground effect by using computational approach based on the articles published in an open domain. For this purpose, a high aspect ratio wing aircraft is designed on original FX 63-137 airfoil. A commercial CFD software Fluent along with γ-Reθ transition model is used to characterize the flow phenomena around the aircraft. A ground effect is simulated by using a moving wall boundary condition. It is observed that the ground effect significantly influences the longitudinal aerodynamic coefficients like lift, drag and pitching moment when the aircraft is in close proximity to the ground. The vortex deformation in the downstream direction of fuselage becomes strong in the ground effect than in freestream. The volume streamlines over a wing surface at different heights are also shown in ground effects and freestream. Further, aerodynamic coefficients over a vehicle with and without ground effects are compared.

ABSTRACT. The present paper describes the development of a mesh-free Computational Fluid Dynamic technique capable of simulating store separation problems on large parallel computing platforms. The developed technique is capable of simulating simultaneous release of multiple store from an aircraft (Salvo release scenarios). A mesh-free method based CFD solver, which uses a collection of points distributed in the flow domain (point-cloud) as its discretization strategy has been coupled with a pre-processor which generates the required point-cloud and a 6-DoF solver. The mesh-free solver uses Least Square Kinetic Upwind Method (LSKUM) along with MCIR split fluxes. The MCIR split fluxes enables achieving second order accuracy using a compact stencil, this aids in achieving better parallel efficiency and scalability of the mesh-free solver. The preprocessor is capable of handling multiple moving stores by translating, overlapping and merging multiple point clouds into a single large cloud on which the mesh-free CFD code operates. A quasi-steady approach has been used to couple mesh-free CDF solver with 6DoF equations of motion solver. Quasi-steady approach neglects the aerodynamic damping due to angular motion of store. Transpiration boundary condition has been imposed on the store boundaries to capture the damping effects. To enable handling large problems and to exploit large scale parallel cluster machines, distributed memory parallelization of the mesh-free solver has been carried out using Message Passing Interface (MPI) technique.

ABSTRACT. This paper presents the development of a Regent based parallel meshfree solver capable of running either on GPUs or CPUs. The meshfree solver is based on the Least Squares Kinetic Upwind Method (LSKUM) for 2D Euler equations. Benchmark simulations are performed on coarse to very fine point distributions. The computational efficiency of the Regent solver on an A100 GPU is compared with an equivalent LSKUM solver written in CUDA-C. The codes are then profiled to investigate the differences in their performance. Later, scalability results on CPUs are shown to assess the performance of the Regent solver with an explicitly parallel Fortran solver.

ABSTRACT. This paper proposes the implementation of wall function in meshless framework for Spalart-Allmaras bases RANS solver. The wall function approach is based on the assumption of universality of the wall law for turbulent boundary layers. Numerical results for turbulent flow past NACA0012 airfoil are presented where the effect of usage of wall function on skin friction and drag coefficient is demonstrated.

ABSTRACT. The numerical simulation of gas-solid multiphase flows characterized by suspended solid particles has been an emerging topic for the last few decades. This has been a fundamental topic for various applications as well. Compared to their low-speed counterparts, particle-laden compressible flows introduce additional physics that further complicate modeling efforts. In the Eulerian-Lagrangian framework, the carrier gas is treated as a continuum that is modeled using the Navier-Stokes equation, and the dispersed granular particles are tracked in a Lagrangian framework using the discrete element method. The interaction between the two phases is modeled in a four-way coupled manner, which involves the effect of fluid on particles, the effect of particles on the fluid, particle disturbance of fluid locally affecting other particles, and particle collisions affecting the motion of other particles.

ABSTRACT. Shock wave boundary layer interaction is the most extensively studied topic in the field of gas dynamics since, it has tremendous effects on the performance, efficiency and structural integrity of the vehicle. Particularly, the shock wave boundary layer interaction (SWBLI) due to curved shock is studied prominently, as most military aircrafts, missile designs undergo strong SWTBLI due to its curved or blunt shape. These interactions effect the aerodynamic stability of the vehicles at supersonic flow regime. The flow phenomena associated with the boundary layer (BL) remains the same as SWBLI using an external wedge or compression ramp. The main difference being the shock strength which decreases downstream of the bow shock and bow shock topology as the deflection angle of the body increases. Thus, in the present study the interaction between shock wave and turbulent boundary layer (SWTBLI) is considered for a cylindrical shock generator. The investigations are carried out for varying diameter of the cylinder, D and for ratio ∆y/D i.e., the ratio of vertical distance between the flat plate and cylinder. The numerical investigation is carried out using RANS based turbulence models. The numerical results show good agreement with the experimental results for both surface pressure and heat flux.

ABSTRACT. In this paper implementation and testing of Delayed Detached Eddy Simulation (DDES) proposed by Spalart et. al. [1] into the existing multiblock RANS code MB-EURANIUM is discussed. This model is based on the Spalart & Allmaras model [4]. The DDES model will work as a RANS model closer to the wall and Large Eddy Simulation (LES) subgrid like model when the flow is separated and away from wall region.

ABSTRACT. Micro drones with flapping wings have advantages like camouflage, useful for gathering military intelligence in stealth. This necessitates the study of flapping flight of wing sections at low speeds. This paper focuses on the motion of a high-lift airfoil, SD7003, at a Reynolds number of 10,000 and a high reduced frequency of 3.93, in pure plunging mode. Computational results obtained using ANSYS Fluent have been validated against experiments and numerical predictions from literature, using both laminar and turbulent flow simulations. Results indicate that little variation is observed in contours of velocity and vorticity for laminar and turbulent simulations while the lift and drag coefficients predicted are virtually identical. In the final paper, modification of the kinematics will be performed by shifting the peak of a baseline sinusoidal plunge equation. In the final paper, the effects of these modifications on leading edge vortex (LEV) formation and shedding will be presented, along with its impact on lift generation and drag penalties

ABSTRACT. The Aerodynamic Behavior of DrivAer car models was analyzed using open-source computational fluid dynamics (CFD) program, openFOAM. The abbreviation of openFOAM is Open-source Field Operation and Manipulation and it is a C++ toolbox for the development of customized numerical solvers, and pre-/post-processing utilities for the solution of continuum mechanics problems, most prominently including computational fluid dynamics (CFD). In this research we have analyzed the drag coefficient of DrivAer car models at three different velocities of air 20m/s, 30m/s, 40m/s. The main objective of this research is to reduce the aerodynamic drag force on the DrivAer car models using Bio-inspired Sharkskin denticle geometry along with three rib-lets. We have focused on the two types of DrivAer car models fastback and Estate, since most of the passenger car models falls under this category. The effect of sharkskin denticle geometry was analyzed by placing it on the roof of the car model. The drag coefficient is reducing with increasing the velocity of the air and relation between them is parabolic. The drag reduction of nearly 2-2.5% was achieved with this technique. The results are showing that drag force reduction with the help of sharkskin denticle geometry as an inexpensive and better method for aerodynamic drag force reduction.

ABSTRACT. The present study numerically evaluates and compares the performance of 3D single-layered and doublelayered graphite and silicon microchannels for electronic chip cooling. The heat source is an electronic chip with dimensions of 1 cm × 1cm. Since graphite is anisotropic, high in thermal conductivity and less dense than other materials, we have tried to highlight the advantages of using it in light-weight and high heat flux applications. Conjugate heat transfer in rectangular microchannels with a hydraulic diameter of 0.24 mm is simulated for this study. Additionally, superhydrophobic walls were considered. The superhydrophobic walls produce better results regarding the power required and maximum temperature. The performance of single-layered and double-layered microchannels is evaluated with respect to the power required to operate those microchannels. A superhydrophobic graphite double-layered microchannel is found to have a better cooling capacity for a given operating power than other microchannels

ABSTRACT. Various investigations have previously been conducted on viscous flow for drag reduction and lift enhancement of an immersed body. This paper focuses on using moving boundaries on the surface of an airfoil as an active flow control method to achieve the same. In addition to the drag reduction and lift generation, the moving boundary also suppresses the vortices and eliminates the flow unsteadiness.

ABSTRACT. This paper presents the Computation Fluid Dynamics (CFD) studies carried out on a generic fighter aircraft configuration to understand the free floating characteristics of a close coupled canard with and without the wing. During the design stage, it is essential to understand the free floating characteristics of the canard as it is one of the main inputs for canard actuator design. Additionally, the estimation of free floating characteristics is essential for the design of flight control laws, support structural loads, actuator control system etc. The free floating characteristics of the canard is estimated by analysing the canard hinge moment (moment about pivot axis) characteristics with respect a set of pivot (hinge) locations along with the presence and absence of the wing. This paper explains the various observations from CFD that help in arriving at a good estimate of the optimum pivot (hinge) location in the transonic and supersonic regimes.

ABSTRACT. Release of stores from an internal weapons bay cavity involves opening the cavity door, and ejecting a store into the outside atmosphere, followed by the closing of the doors. To design the door and the mechanism controlling its movement, one needs to estimate the forces and moments experienced by the door during the whole operation. While the experimental determination of these forces and moments is possible, it is a time-consuming and expensive procedure. Here, transient flow over the opening doors of the cavity has been simulated in ANSYS Fluent using the SST-SAS turbulence model. Smoothing and re-meshing methods available in ANSYS FLUENT have been used to update the volume mesh in the deforming regions subject to the constrained motions defined at the boundaries. Door motion is calculated separately and incorporated into ANSYS Fluent as a user-defined function (UDF). With the relative motion between parts A and B of the door, the high-speed stream (M 1.3) exerts an oppositely directed Y-force on these components. This indicates that with relative motion between door components, the Y-force-generated moments on Parts A and B tend to balance each other.

ABSTRACT. The current study aims to capture the drag characteristics of the Ahmed Body with a rear slant angle of 25◦ at a Reynolds number, Re = 7.68 × 10^5, based on the vertical height of the Ahmed model. High fidelity simulations have been conducted with SankhyaSutra Taral™, where the Entropic lattice-Boltzmann Method (ELBM) is employed in tandem with the 67 lattice model for discretization of the velocity space. SankhyaSutra Taral™ makes use of a structured multi-resolution grid arrangement, and the interpolations are carried out using gradient and Hessian from lattice operators over a Body Centered Cubic (BCC) lattice. Spatial discretization has been done with a BCC arrangement of grid points, and no explicit turbulence models have been used in this study. The total drag and total lift coefficients of the Ahmed Body have been reported and validated against the results of Ahmed et al. and Bayraktar et al., respectively. The wake flow has been studied, and averaged velocity profiles at the symmetry plane have been compared with the experimental findings of Lienhart et al. The calculated drag coefficient (C_D), and lift coefficient (C_L) exhibit excellent conformity with experimental results. The velocity profiles in the wake of the Ahmed model also display an excellent match with experiments.

ABSTRACT. We present the implementation, verification and validation of an immersed boundary method (IBM) along with a wall model (WM) to simulate atmospheric boundary-layer flows over complex terrain. The framework presented here has two novel aspects over standard IBM implementations. First, the underlying schemes are global in nature and require specification of values throughout the solid region. Second, to enable high-Reynolds number simulations, a wall model is coupled to the IBM. The proposed numerical framework is shown to have second-order accuracy. The framework is validated by simulating flow over a smooth cosine-squared hill and comparing to previously published experimental results. The mean velocity and turbulence intensity are reproduced accurately by our LES.

ABSTRACT. This work demonstrates the need of accounting fluid structure interaction (FSI) in the internal flows by modeling flows through the channels. For comparison, the channel walls have modelled as both rigid and flexible using Immersed Boundary Method (IBM). When the walls are elastic, because of the wall deformation recirculation zones are present which impact the flow behavior and lead to complex flow patterns. The analysis of the flow pattern revealed that the rigid wall assumption overestimated wall shear stress data. This study emphasizes the significance of accounting FSI in channel flows and demonstrates the effectiveness of using IBM to model the interaction between fluid and structures.