A new formulation of Delayed Detached-Eddy Simulation (DDES) based upon elliptic relaxation is derived and implemented within a finite-volume framework. This new formulation is based upon the φ − f RANS model which has previously demonstrated both improved modelling of the near-wall physics and numerical robustness for industrial applications. The φ − f DDES model is calibrated and validated using Decaying Isotropic Turbulence (DIT) to establish the validity of the derivation and to calibrate the model constants. In light of the numeri- cal scheme requirements for DDES, a hybrid numerical scheme is proposed and implemented, which is shown to perform in the intended manner.Initially, three DDES formulations (SA-DDES, SST-DDES and φ − f DDES) are compared on the 2D periodic hills test case at Re = 10590 and Re = 37000. This test case primarily serves as a validation case to evaluate whether the im- plementation and calibration were correct. The flow over a NACA0021 airfoil post-stall at 60o incidence is then evaluated; a test case that DDES was origi- nally devised for (i.e massive separation from an airfoil). The three formulations are then evaluated on a 2D wall-mounted hump which exhibits largely geometry induced separation, but is still sensitive to the modelling of the initial separated shear layer and upstream turbulence levels. The final case is the Ahmed car body which combines both geometry and pressure-induced separation from a 3D surface. This complex flow is challenging for any turbulence modelling approach and is sensitive to the underlying RANS model.A general sensitivity to the underlying RANS model is demonstrated for the majority of the test cases investigated. The φ − f DDES model is shown to have encouraging performance on these wide range of test cases compared to the established SST-DDES and SA-DDES models. Whilst the φ − f DDES model is not a fix for the shortcomings of DDES, it is shown to be a practical and robust alternative to the established SST-DDES and SA-DDES variants that have become the de facto choice for many DDES users.

The purpose of this thesis is to study the shock/wave turbulent boundary layer interaction by using delayed-detached-eddy simulation (DDES) model with a low diffusion E-CUSP (LDE) scheme with fifth-order WENO scheme. The results show that DDES simulation provides improved results for the shock wave/turbulent boundary layer interaction compared to those of its predecessor the detached-eddy simulation (DES). The computation of mesh refinement indicates that the grid density has significant effects on the results of DES, while being resolved by applying DDES simulation. Spalart in 1997 developed the Detached-Eddy Simulation (DES) model, which is a hybrid RANS and LES method, to overcome the intensive CPU requirement from LES models. Near the solid surface within a wall boundary layer, the unsteady RANS model is realized. Away from the wall surface, the model automatically converts to LES. The Delayed-Detached-Eddy Simulation (DDES) was suggested by Spalart in 2006 to improve the DES model previously developed. The transition from the RANS model to LES in DES is not grid spacing independent, therefore a blending function is introduced to the recently developed DDES model to make the transition from RANS to LES grid spacing independent. The DDES is validated by computing a 3D subsonic flat plate turbulent boundary layer. The first case studied using DDES is a 3D transonic channel with shock/turbulent boundary layer interaction. It consists of two straight side walls, a straight top wall, and a varying shape in span-wise direction for a bottom wall. The second case studied consists of a 3D transonic inlet-diffuser. Both results are compared with experimental data. The computed results of the transonic channel agree well with experimental data.

The displacement of the flow by a passing freight train can often result in dangerous conditions for railway equipment and people standing in the vicinity of the train. In this work, Computational Fluid Dynamics (CFD) simulations are performed to study the flow development around a moving freight train comprised of a Class 66 locomotive and four container wagons. The results will give a better insight into the effects that each flow structure can have in the flow within the train's slipstream. Both two- and three-dimensional simulations are carried out around the freight train using three different RANS turbulence models: the Spalart-Allmaras, the SST k-ω and the W&J EARSM. Two cases of 10o and 30o crosswinds are also considered and compared to the no-crosswind case, as side-winds characterize the majority of real-life situations and are known to amplify the slipstream effects. The results are validated against available experimental and numerical data and they are thoroughly presented and discussed. The 30o crosswind case is also computed using a DDES simulation. A meshing strategy which involves the assembly of different mesh blocks with a non-matching interface boundary condition to create the complete domain is used and assessed, as an alternative meshing approach that can simplify and accelerate the set-up of different case-studies. Additionally, the two-dimensional study is used to assess the influence of different parameters on the solution, such as the grid resolution and the moving-ground boundary condition.

La pollution atmosphérique et ses conséquences sur la santé et l'environnement constituent un domaine d'étude complexe à cause du nombre de phénomènes physiques mis en jeu. L'objectif de ce travail est d'étudier les principales caractéristiques de l'écoulement et de la dispersion d'un scalaire passif au sein de la canopée urbaine. Pour ce faire, un code numérique a été développé. Il résout les équations de Navier-Stokes dans le cadre d'un écoulement incompressible pour une atmosphère neutre en faisant usage de la méthode de la compressibilité artificielle selon la méthode des volumes finis. Le modèle de Spalart-Allmaras a été utilisé pour la modélisation de la turbulence. La canopée urbaine est explicitement prise en compte et est modélisée par un groupe d'obstacles de forme cubique. Le code fut d'abord testé pour des configurations bidimensionnelles avec un seul et 4 obstacles en configuration alignée selon deux approches : une simulation stationnaire RANS et instationnaire URANS qui reproduit le décrochement tourbillonnaire. La prise en compte du décrochement tourbillonnaire se traduit par une diffusion dans le sillage turbulent du groupe d'obstacles. Les résultats ont été comparés à des mesures expérimentales et d'autres résultats numériques de référence dans la bibliographie et montrent l'amélioration du champ de vitesse moyen par l'approche code fut ensuite testé sur un cas tridimensionnel avec un groupe d'obstacles organisés selon 2 configurations géométriques: alignée et en quinconce. Afin d'éliminer les effets des conditions aux limites, l'écoulement fut calculé sur un volume élémentaire de calcul en utilisant des conditions aux limites périodiques. Deux types de simulations furent réalisés: l'approche RANS classique et la version DDES du modèle de Spalart-Allmaras. L'écoulement obtenu par la DDES améliore de façon significative les résultats par rapport au RANS en comparaison de mesures expérimentales de simulation directe et montrent la bonne potentialité du modèle. La dispersion d'un scalaire passif émis au sein de la canopée fut obtenue sur un domaine plus important comprenant 16 volumes élémentaires par le biais des conditions aux limites périodiques utilisées. Une analyse du champ de concentration a ensuite été réalisée et des comparaisons effectuées en fonction du type de calcul et de la configuration géométrique. / Atmospheric pollution and its impact on health and the environment depend on many physical phenomena, and this makes it a difficult subject to study. The main objective of this work is to investigate the main characteristics of the flow and dispersion of a passive scalar in the urban canopy. Specifically, the urban canopy is simulated by a group of cubical obstacles in a neutrally-buoyant atmospheric boundary layer. A numerical code bas been developed as a tool to aid in this study; flow is computed by solving the Navier-Stokes equations for an incompressible flow, using a finite volume approach, and the method of artificial compressibility. The turbulence is modeled using the method proposed by Spalart and Allmaras. The code was tested first in a 2-D configuration, for flow over a single obstacle, and over a group of 4 obstacles; in both cases two types of simulation were studied: a stationary RANS simulation, and an unsteady RANS (URANS), which reproduced vortex shedding from the obstacles. The explicit inclusion of vortex shedding in the URANS simulation leads to diffusion in the obstacle wakes, and the results compare better with experimental measurements and other published numerical simulations than do those for the RANS simulations. The code was then tested for some 3-D cases consisting of a group of obstacles arrangcd either in aligned or staggered configurations. In order to avoid the influence of boundary conditions, the flow field was simulated using periodic boundary conditions and a small sub-unit from the group of obstacles. Two types of simulation were performed: a classical RANS type calculation and the DDES proposed by Spalart and Allmaras. The results obtained using the ODES agree much more closely with experimental measurements and the results of other numerical simulations than do those obtained using RANS, and indicate the potential of this approach. The dispersion of a passive scalar in the urban canopy was simulated on a much larger domain consisting of 16 of the sub-units used to compute the flow field. The concentration fields were analyzed to show the influence of the geometrical configuration and the type of model.

Enhancing the ability to predict airflow around the Vertical Tail Plane (VTP) of an aircraft is vital in the aviation industry. The size of the VTP is driven by a particular flight condition - loss of an engine during take-off and low speed climb. Nowadays, Computational Fluid Dynamics (CFD) is the main tool used by engineers to assess VTP flows. However, due to uncertainties in the prediction of VTP effectiveness, aircraft designers keep to a conservative approach, which risks oversizing of the tail plane, adding more drag. Uncertainties emerge from difficulties in predicting the massive separation that occurs on the swept tail when it is approached by a flow at high incidence. Furthermore, the deployment of the control surface (the rudder) over the tail plane and the skewed flow along the span increase the CFD challenges. Improved predictive capabilities of the flow around VTPs would enable a more optimal design approach with potential drag saving. The correct prediction of flow separation is the essence of this study. Currently, the industry uses steady Reynolds-Averaged Navier-Stokes (RANS) simulations to analyse VTPs flow. In order to assess RANS performance, the study of airflow detaching from a backward rounded ramp is performed and the results are compared to Large-Eddy Simulations (LES). The analysis shows that, even though RANS may predict the onset of flow separation correctly, they completely miss the location of flow reattachment over the ramp, and this affects the whole flow solution. Moreover, the flow features a strong anisotropy at the onset of separation, difficult to be captured by RANS. The analysis shows that RANS cannot predict production of turbulent kinetic energy in the detached flow region correctly, discouraging flow mixing, and delaying flow reattachment. A hybrid RANS/LES carried out on the same test case shows the benefits of using eddy-resolving simulations for detached flows. The prediction of the locations of the separation and reattachment points differs by only 1% from the highly-resolved simulation. The VTP investigation carried out in this thesis uses a wind tunnel model tested at Airbus. The study starts with steady RANS approaches for different turbulence models. RANS simulations produce acceptable results for the flow at low incidence levels. On the contrary, at high incidence, when flow separation occurs, RANS methods fail. The second step of the research consists of using unsteady RANS (URANS) simulations for VTP flows at high sideslip angles. The introduction of time-accuracy brings important benefits. Nevertheless, the results still show some inaccuracies (around 20% error). Finally, restarting from the flow solutions obtained by URANS simulations, higher fidelity hybrid RANS/LES techniques in the form of Delayed Detached-Eddy Simulations (DDES) are used to assess the characteristics of the separated flow around the tail plane. Results show a remarkable improvement of the flow solution. The pressure distribution matches experimental results favourably, and this translates into an improved prediction of the aerodynamic loads over the VTP. This leads towards a new strategy for the assessment of the flow over aircraft VTPs, amounting to an important contribution to the design of future aircraft.

Crashback operation of a propeller is a common emergency slowing maneuver for ships and submarines. The reversing of the propeller while the vessel is moving forward results in large loads on the propeller blades and highly detached flow, which presents both practical concerns and fundamental fluid physics inquiries. This thesis contains a comprehensive numerical analysis of two propellers in crashback operation. Available numerical and experimental data for David Taylor Model Basin (DTMB) 4381 propeller are used for validation of the computational fluid dynamics solver used, REX. A second propeller, Maritime Research Institute Netherlands (MARIN) 7371R is used to classify the common crashback flow behavior into regimes. Four regimes were identified, each existing for a range of operating conditions. The most prominent and deciding feature of the flow regimes is the presence of a ring vortex, resulting from the opposing action of the free-stream flow and the propeller induced flow. The position, shape and strength changes between regimes, dominating the dynamics of the flow by altering the induced flow into the propeller disk. Flow conditions resulting from regime transitions are described. Changes in the ring vortex structure lead to two stable flow conditions of interest. One condition produces a reduction of thrust despite the increase in flow speed into the propeller and negligible side-forces. The other condition creates large side-forces capable of rotating a vessel, resulting from an asymmetry forming in the ring vortex. Additionally, massive flow separation occurs at high free-stream speeds that cause extreme blade loading. An extensive description of each flow regime is provided, with further investigation and discussion of the flow regimes that present more practical concerns and novel characteristics of the crashback flow.

Distributed Discrete Event Simulation (DDES) has received much attention in recent years, owing to the fact that uniprocessor based serial simulations may require excessive amount of simulation time and computational resources. It is therefore natural to attempt to use multiple processors to exploit the inherent parallelism in discrete event simulations in order to speed up the simulation process. In this dissertation we study the performance of distributed simulation of queueing networks, by analysing queueing models of message flows in distributed discrete event simulators. Most of the prior work in distributed discrete event simulation can be catego­rized as either empirical studies or analytic (or formal) models. In the empirical studies, specific experiments are run on both conservative and optimistic simulators to see which strategy results in a faster simulation. There has also been increasing activity in analytic models either to better understand a single strategy or to compare two strategies. Little attention seems to have been paid to the behaviour of the interprocessor message queues in distributed discrete event simulators. To begin with, we study how to model distributed simulators of queueing networks. We view each logical process in a distributed simulation as comprising a message sequencer with associated message queues, followed by an event processor. A major contribution in this dissertation is the introduction of the maximum lookahead sequencing protocol. In maximum lookahead sequencing, the sequencer knows the time-stamp of the next message to arrive in the empty queue. Maximum lookahead is an unachievable algorithm, but is expected to yield the best throughput compared to any realisable sequencing technique. The analysis of maximum lookahead, therefore, should lead to fundamental limits on the performance of any sequencing algorithm We show that, for feed forward type simulators, with standard stochastic assump-tions for message arrival and time-stamp processes, the message queues are unstable for conservative sequencing, and for conservative sequencing with maximum lookahead and hence for optimistic resequencing, and for any resequencing algorithm that does not employ interprocessor "flow control". It follows that the resequencing problem is fundamentally unstable and some form of interprocessor flow control is necessary in order to make the message queues stable (without message loss). We obtain some generalizations of the insta­bility results to time-stamped message arrival processes with certain ergodicity properties. For feedforward type distributed simulators, we study the throughput of the event sequencer without any interprocessor flow control. We then incorporate flow control and study the throughput of the event sequencer. We analyse various flow control mechanisms. For example, we can bound the buffers of the message queues, or various logical processes can be prevented from getting too far apart in virtual time by means of a mechanism like Moving Time Windows or Bounded Lag. While such mechanisms will serve to stabilize buffers, our approach, of modelling and analysing the message flow processes in the simulator, points towards certain fundamental limits of efficiency of distributed simulation, imposed by the synchronization mechanism. Next we turn to the distributed simulation of more general queueing networks. We find an upper bound to the throughput of distributed simulators of open and closed queueing networks. The upper bound is derived by using flow balance relations in the queueing network and in the simulator, processing speed constraints, and synchronization constraints in the simulator. The upper bound is in terms of parameters of the queueing network, the simulator processor speeds, and the way the queueing network is partitioned or mapped over the simulator processors. We consider the problem of choosing a mapping that maximizes the upper bound. We then study good solutions o! this problem as possible heuristics for the problem of partitioning the queueing network over the simulator processors. We also derive a lower bound to the throughput of the distributed simulator for a simple queueing network with feedback. We then study various properties of the maximum lookahead algorithm. We show that the maximum lookahead algorithm does not deadlock. Further, since there are no syn­chronization overheads, maximum lookahead is a simple algorithm to study. We prove that maximum lookahead sequencing (though unrealisable) yields the best throughput compared to any realisable sequencing technique. These properties make maximum lookahead a very useful algorithm in the study of distributed simulators of queueing networks. To investigate the efficacy of the partitioning heuristic, we perform a study of queue­ing network simulators. Since it is important to study the benefits of distributed simula­tion, we characterise the speedup in distributed simulation, and find an upper bound to the speedup for a given mapping of the queues to the simulator processors. We simulate distributed simulation with maximum lookahead sequencing, with various mappings of the queues to the processors. We also present throughput results foT the same mappings but using a distributed simulation with the optimistic sequencing algorithm. We present a num­ber of sufficiently complex examples of queueing networks, and compare the throughputs obtained from simulations with the upper bounds obtained analytically. Finally, we study message flow processes in distributed simulators of open queueing networks with feedback. We develop and study queueing models for distributed simulators with maximum lookahead sequencing. We characterize the "external" arrival process, and the message feedback process in the simulator of a simple queueing network with feedback. We show that a certain "natural" modelling construct for the arrival process is exactly correct, whereas an "obvious" model for the feedback process is wrong; we then show how to develop the correct model. Our analysis throws light on the stability of distributed simulators of queueing networks with feedback. We show how the stability of such simulators depends on the parameters of the queueing network.

No description available.