Browsing by Author "Patel, Sanjay"

Now showing 1 - 3 of 3
  • Thumbnail Image
    ItemOpen Access
    Computational modelling of environment flows featuring gas dispersion
    (Cranfield University, 2010-07) Gerousi, Loukia; Patel, Sanjay; Drikakis, Dimitris
    This study particularly aims at understanding flow and pollutant dispersion when flat terrain, single hill and hill with obstacles are present. The emissions of ethylene from a point source are located in eight different positions. For the hill cases, the sources are located downwind from the top of the hill, and the data are collected at various locations. The commercial software packages Gambit 2.4.6, Fluent 6.3.26 and Tecplot 360 are used for the two-dimensional mesh generation, for the flow simulation and for the validation respectively. The numerical results are compared with experimental and numerical data for the single hill case and for the point source using the Spalart- Allmaras model, k-ε Standard, k-ε RNG and k-ε Realizable models. The comparison of the results shows that the k-ᵋ Standard model is in good agreement with the experimental and numerical data. Results also show that the mass fraction of ethylene is highest for the flat terrain case. The next highest mass fraction of ethylene is found for the case with the hill and obstacles, and the single-hill case has the lowest. Moreover, upwind of the first obstacle the average mass fraction is larger than inside the first and the second canyons, and the minimum pollutant is downwind of the last obstacle. The average mass fraction of ethylene is measured at the corners of the canyons, and the results show that generally the bottom left corners have a higher mass fraction than the middle and bottom right side.
  • Thumbnail Image
    ItemOpen Access
    Computational modelling of instability and transition using high-resolution methods
    (Cranfield University, 2007) Patel, Sanjay; Drikakis, Dimitris
    This thesis concerns the numerical investigation of suddenly expanded flows featuring separation, instabilities and transition, in the context of Implicit Large Eddy Simulation (ILES). The study of separated flows through suddenly expanded geometries is a classic yet complex area of research. These types of flows feature instabilities which may lead to bifurcation. Non-linear bifurcation is of great importance when considering hydrodynamic stability and the mechanism of laminar to turbulent flow transition. A detailed numerical investigation of various high-resolution methods and their ability to correctly predict the flow through a suddenly expanded and contracted geometry demonstrates that the choice of the particular numerical method employed can lead to an incorrect solution of the flow. The key di erence between the various highresolution methods employed is in the calculation of the nonlinear wave-speed dependent term. It is shown that the nonlinearity of this term provides an asymmetric dissipation to the flow which triggers symmetry-breaking bifurcation in a fully symmetric computational set-up. High-resolution simulations of three-dimensional flow through a plane suddenly expanded channel at low Reynolds numbers show that this type of flow is characterised by a symmetric separation of the fluid which is nominally two-dimensional in the spanwise direction. Increasing the Reynolds number reveals a symmetry-breaking bifurcation of the fluid flow which becomes three-dimensional as Reynolds number is further increased. Simulations confirm that it is this threedimensional disturbance which leads to the onset of time-dependent flow characterised by the periodic shedding of vortices from the upstream recirculation zones. Preconditioning techniques which aim to alleviate sti ness in the calculation of the advective fluxes for low Reynolds number flows are shown to be unsuitable for flows featuring instabilities. The added dissipation to the flow causes the prediction of an incorrect stable solution or to an improper estimation of the size of the separation bubbles. Simulations of a synthetic jet issuing into quiescent air using various slope limiters manage to capture the flow physics relatively well. Limiters are used to avoid a scheme from being oscillatory and provide non-linear dissipation in the region of excessively large gradients. The various limiters di er with regards to the amount of dissipation they provide to the flow, hence the solution obtained is dependent on the limiter used.
  • Thumbnail Image
    ItemOpen Access
    High resolution methods for incompressible, compressible and variable density flows.
    (2004-01-01T00:00:00Z) Drikakis, Dimitris; Hahn, Marco; Patel, Sanjay; Shapiro, Evgeniy
    The uid dynamics community has dealt with a number of numerical challenges since the 1950's. These include the development of numerical methods for hyperbolic conservation laws with particular interest in capturing shock wave propagation and related phenomena, solution algorithms for the solution of the incompressible Navier-Stokes equations - a numerical challenge arises here due to the absence of the pressure term from the continuity equation - methods/ techniques for the acceleration of the numerical convergence, modelling of turbulence and grid generation techniques. Within each of those areas di erent numerical approaches have been pursued by various researchers aiming to achieve higher accuracy and ef ciency of the numerical solution. Continuous interest exists in relation to the development of accurate and e cient numerical methods for the computation of instabilities, transition and turbulence. It has been observed for more than a decade that high-resolution methods can be used in (under-resolved) turbulent ow computations without the need to resort to a turbulence model, but this approach has only recently gained some theoretical support and structural explanation for the observed results [3, 15]. Because of this there is a necessary overlap between the classical modelling of turbulence and its computation through high resolution methods [5]. These methods are currently used to simulate a broad variety of complex ows, e.g., ows that are dominated by vorticity leading to turbulence, ows featuring shock waves and turbulence, and the mixing of materials [23]. Such ows are extremely dif cult to practically obtain stably and accurately in under-resolved conditions (with respect to grid resolution) using classical linear (both second and higherorder accurate) schemes. Further, new applications at micro-scale, e.g. micro uidics, microreactors and lab-on-a-chip, have raised a number of challenges for computational science methods. In this paper we provide a brief overview of highresolution methods in connection with some of the above problems. An extensive description of these methods for incompressible and low-speed Flows can be found in [5].