Browsing by Author "Ecder, Ali."
Now showing 1 - 18 of 18
Results Per Page
Sort Options
Item A framework for the analysis of coupled-physics models using adaptive multi-level techniques(Thesis (Ph.D.)-Bogazici University. Institute for Graduate Studies in Science and Engineering, 2010., 2010.) Turan, Erhan.; Ecder, Ali.This object of this study is to develop a computational framework to analyze coupled-physics problems within the context of multi-level methods. Adaptive solution strategies in conjunction with Newton-Krylov and Domain Decomposition Methods are used to investigate different problems. Two model coupled-physics problems are selected for simulation: a fluid-structure interaction problem and a multiphase flow problem. First problem is on the deformation of a bimetallic strip exposed to natural convection. Two non-conforming and overlapping domains are created to handle the changes on the boundaries so that the deflection of the solid is applied only some portion of the fluid region. Displacements on the strip are calculated using decoupled thermoelasticity with plane strain assumption. In the second problem, collapse of a water column into the air is modeled. The interface is tracked using the Volume of Fluid method and the results are compared against experimental studies. To let the physics interact with each other and to unify different numerical solution methods, a solver called DEMONA (Decomposition Enhanced Mechanics Optimized Numerical Analysis) is developed which is verified on numerous benchmark problems. A new technique, based on an idea to reduce the solution sets is implemented into the solver, as well. With this methodology, the unknowns are filtered using various reduction criteria which are either applied in run-time or decided prior to the computations so that a specific solution approach is employed. Consequently, an adaptive strusture is attained and different solution techniques are allowed to be tested with a single model definition.Item Aerodynamic analysis of a paraglider wing using domain decomposition techniques(Thesis (M.S.)-Bogazici University. Institute for Graduate Studies in Science and Engineering, 2007., 2007.) Başol, Altuğ Melik.; Ecder, Ali.The flow around a curvature tube is investigated using it as a simple 2D model of a paraglider wing based on the similarity between their top views. The solutions are obtained for five different angle of attacks varying from 00 up to 420, for two different radius of curvatures and for two different Reynolds number namely for Re=80 and for Re=160. Two non-matching overlapping domains are used. The flow at these domains is solved separately and the solutions are transferred into each other using the alternating multiplicative Schwarz technique. With the aid of this technique a solver is developed which is capable of solving the flow around various forms of the tube and at different angle of attacks without making any modifications in the outer boundaries and without using grid generation. Newton's methods combined with three different Krylov sub-space solvers are applied. Implementing also Jacobi, symmetric Gauss-Seidel (SGS) and incomplete LU decomposition (ILU(0)) preconditioners into the solvers their effects on the convergence behavior are investigated. It is observed that the ILU(0) has a superior effect on the convergence behavior than the others have. However, its unavailability for the matrix free algorithms makes SGS preconditioned inexact Newton's method a better option as a solver because of its low storage load and low code development period.|Keywords: Newton's method, Krylov solvers, overlapping domain, preconditionersItem Computational analysis of cyclonic separation(Thesis (M.S.) - Bogazici University. Institute for Graduate Studies in Science and Engineering, 2014., 2014.) Sasan, Salour.; Ecder, Ali.Cyclonic separation is a method to remove particles from a uid ow using a vortex for separation. This method of separation can be used to separate ne droplets of liquid or particles from a gas ow. This device is a kind of stationary mechanical device that uses centrifugal force to separate any kind of particles from a uid stream. There are two fundamental types of air cyclones according to the direction in which the cleaned gas leaves the cyclone. The reverse ow cyclone is one of them and it is actually the most frequently used type in industry. While the clean gas leaves from upper lid, separated particles can exit through a bottom apex. Cyclones have simple manufacturing, low energy requirements, capability to work at high pressures and temperatures. In spite of the simplicity in manufacturing and operation, the mathematical formulations used for anticipating the separation e ciency are very complicated. In contrast to most of the previous studies which use a single input, in this study two slot type cyclone separators are considered and compared together, for both laminar and turbulent ows. In this research, mathematical modeling of cyclonic separation is based on Muschelknautz Method formulation (MM) and the computational uid dynamics (CFD) analysis is performed using the COMSOL Multi-Physics program. Some of the well-known models such as Muschelknautz D type, Swift High E ciency, Stairmand High E ciency and Lapple General Purpose with single and double inputs are analyzed and compared. A model with di erent input sizes with single and double inputs is introduced and parameterized in terms of pressure drop pro le, velocity magnitude, number of particles that each cyclone can separate, separation e ciency and stream lines in laminar and turbulent streams. The particle tracing module of COMSOL is used for that purpose.Item Convergence acceleration procedures for the computation of 2-D transonic flows(Thesis (Ph.D.)-Bogazici University. Institute for Graduate Studies in Science and Engineering, 2014., 2014.) Türk, Uğur; Ecder, Ali.This study addresses a novel adaptive time stepping procedure, which leads to selection of larger time steps allowed by the physics of the problem. Information about the gradients of the flow variables can be regarded as an indicator for determining proper amount of time step, in which the system evolved. The signals from the pressure sensors, which act according to the pressure gradients, are chosen as a measure to determine the magnitude of the local CFL number. Thus, the aimed methodology for the selection of the local time step with the use of Pressure Sensor introduces optimal time steps to the implicit solution method by accounting for the pressure gradient in the solution domain, such that sharp pressure gradients encourages small time steps and vice versa. To illustrate the effect of proposed procedure, Newton Krylov (NK), with implicit pseudo time stepping method, has been employed to solve the compressible Euler equations for steady transonic case by turning on the pressure switch. Numerical experiments show that the introduced adaptive time stepping procedure decreases the computation time and the number of iterations, effectively. Additionally, a comparison study on the performances of Newton Krylov (NK) and nonlinear multigrid (FMGFAS) methods are presented. The longer computation time required by NK can be a result of the requirement of Newtons method for a better initial guess. When the free stream values are used as initial guess, a more sophisticated method for time step selection is needed for a better NK performance especially at the start up phase.Item Coupled analysis of turbomachinery blades(Thesis (M.S.)-Bogazici University. Institute for Graduate Studies in Science and Engineering, 2008., 2008.) Düz, Bülent.; Ecder, Ali.; Sönmez, Fazıl Önder.Three-dimensional analysis of uid ow over turbomachinery blades and the deformation of the blades under aerodynamic loads are modeled and simulated using computational techniques. Decomposition into computational sub-domains and using multi-level partitioning hierarchy characterize the procedure of solving the non-linear set of governing partial di erential equations. Grid generation techniques and various formulations of Navier-Stokes equations are studied. These equations are linearized using Newton's method, and the resulting system of equations are solved using matrix- free implementations of the preconditioned Krylov techniques. Pressure values determined as a result of uid analysis are implemented as bound- ary conditions into ANSYS environment. Displacements are found after solid analysis performed by ANSYS and transferred into uid analysis to update the geometry. The computations were carried out for di erent Reynolds numbers and the results of the numerical simulations are discussed and compared.Item Domain decomposition analysis of aerodynamic coupling(Thesis (M.S.)-Bogazici University. Institute for Graduate Studies in Science and Engineering, 2008., 2008.) Aydın, Özkan.; Ecder, Ali.The main objective of this research is the computational modeling and analysis of aerodynamic-coupling effects of multiple bodies in motion in a fluid using domain decomposition techniques. Simulation of such a flow field requires the solving of the nonlinear equations of the fluid motion. The computational model and techniques are first tested in the analysis of the internal flow problem. However, the objective of this thesis work is to analyze the aerodynamic coupling observed in the external flow problem. Computational domains is separated into simple cartesian subdomains. The flow at these subdomains is solved separately using stream function, vorticity formulation and these solutions are transferred into each other using the alternating multiplicative Schwarz technique. Newton’s method combined with two different Krylov sub-space solvers is applied to solve the flow problem in rectangular region. In order to have global convergence, backtracking algorithm is used within Newton method. In order to decrease computational work, local mesh refinement strategy is used in critical areas of domains. In the internal flow problem, the effects of Reynolds number on the flow are investigated for three different Reynolds numbers, namely, Re=50, Re=100, Re=150. In the external flow problem, influences of the angle-of-attack, Reynolds number, and longitudinal distance between the objects on the flow properties are investigated.|Keywords: Newton’s method, backtracking, Krylov solvers, domain decomposition, internal flow, external flowItem Level set analysis of two-fluid interfacial flows(Thesis (M.S.) - Bogazici University. Institute for Graduate Studies in Science and Engineering, 2012., 2012.) Sheikhi, Aidin.; Ecder, Ali.In this study incompressible immiscible two-phase ows have been analyzed in two dimensions using Level Set method. Firstly, a brief explanation about other numerical methods existing in the field of modelling interfacial two-phase flows is presented. Then, Level Set methodology and its coupling with Navier-Stokes equations are introduced in detail. Discretization method used in this study is Finite Difference method and derivations of Level Set function and convective terms are done using fth order Weighted Essentially Non-Oscillatory (WENO) scheme. In order to solve the pressure field implicitly, a projection method is applied. Afterwards, the reliability of method is tested by making comparison with Smoothed Particle Hydrodynamic (SPH) method for di erent test cases such as Rayleigh-Taylor instability, rising bubble, droplet fall, bursting bubble at free surface and droplet fall onto free surface. Effects of non-dimensional parameters like Reynolds (Re) and Eotvos (Eo) numbers on the evolution of interface in two-phase flows are investigated. Lastly, dam breaking problem is considered in order to analyze the performance of Level Set method in simulating free surface flows.Item Mixed convection about a rotating sphere(Thesis (M.S.)- Bogazici University. Institute for Graduate Studies in Science and Engineering, 1986., 1986.) Ecder, Ali.; Aksan, Emre.This report presents a theoretical analysis of flow and heat transfer characteristics of the effects of rotational speed, buoyancy force and the Prandtl number on laminar boundary layer over a rotating sphere in forced flow. Applying the finite difference method, numerical computations are carried out for various values of the above parameters. Both assisting and opposing flows are considered. Although the heating condition of uniform wall tempeuture is used in the analysis, the case of uniform surface heat flux is also studied in the formulation. After an introduction to the subject and an examination of the previous works, the theoretical background chapter supplies a general formulation. In the seotion which follows, the problem is specified. Then the results of the numerical solution are displayed in graphical form. Finally, the results are discussed and conclusions are arrived at. The computer program is also supplied.Item Multiple-domain analysis of 3-dimensional flow over an ellipsoidal body of aeronautical interest(Thesis (M.S.)-Bogazici University. Institute for Graduate Studies in Science and Engineering, 2006., 2006.) Sevinç, Erkan.; Ecder, Ali.Numerical solutions to partial differential equations form the backbone of mathematical models that simulate gas flow, which in the present case is assumed to be airflow around an arbitrary atmospheric body. Significance of simulating airflow in three dimensions reveals itself with the developing aviation industry and increasing needs for obtaining accurate results during the design process of atmospheric vehicles. A three dimensional approach is deemed necessary for accuracy reasons and the opportunity to compare results with that of 2-D flow assumptions. As the number of nodes in the computational domain increases by a factor of the number of nodes in one dimension, excessive computational work is expected. As such, performance of the numerical solution algorithm gains importance more than anything in the whole burden. Adapting a multi-grid algorithm is hence expected to be a wise step towards solution. The effects of leveling in the multi grid domain are of primary interest. The primary objective of this study is to develop computational tools that facilitate and speed up the solution of three-dimensional external flows around non-symmetric bodies. Adapting multiple domains in the computational space is a sort of domain decomposition technique. Overlapping domains where each domain of coarser mesh encloses that of the finer meshes have been employed. This provided a multi-grid/multilevel formulation which resulted in faster convergence by way of reducing smooth errors. This formulation comprises constant-size sub-problems which might also exhibit good applicability in parallel computation. The effects and the advantages of multi-domains applied for finite difference formulations in three spatial coordinates were numerically experimented. Faster convergence of numerical algorithm using a multi level approach in three dimensions was achieved.Item Nonlinear preconditioning applied to internal flows(Thesis (M.S.) - Bogazici University. Institute for Graduate Studies in Science and Engineering, 2018., 2018.) Yıldız, Ersin.; Ecder, Ali.Solution of large and sparse nonlinear system is required in many computational science and engineering areas. Well-known inexact Newton method converges fast with a good initial guess, however such an initial guess is hard to obtain for the problems with unbalanced nonlinearity, such as fluid flow problems with a high Reynolds number. Nonlinear preconditioning has been proposed to deal with this issue by constructing a preconditioned nonlinear system and solving it by inexact Newton. This approach is regarded as left nonlinear preconditioner since it changes the nonlinear system. On the other hand, right nonlinear preconditioner eliminates some previously chosen compo nents of the unknowns to form a better approximate solution to participate in inexact Newton iterations. We consider three preconditioning methods: additive Schwarz pre conditioned inexact Newton (ASPIN) which is based on domain decomposition meth ods, field-split preconditioned inexact Newton (FSPIN) which splits the components and solves them in an additive Schwarz manner, and nonlinear elimination precondi tioned inexact Newton (NEPIN). In this thesis, various studies have been conducted to observe the behavior of the algorithms with respect to two different implementa tions of finite difference method on boundaries, some tolerance parameters and some algorithmic parameters for steady-state lid-driven cavity problem at several values of Reynolds number. Our numerical tests obtained on parallel show that NEPIN is more robust than inexact Newton and other nonlinear preconditioning methods.Item Numerical analysis of multipilicty and transition phenomenon in natural convection(Thesis (M.S.) - Bogazici University. Institute for Graduate Studies in Science and Engineering, 2016., 2015.) Kafil, Hadi.; Ecder, Ali.This research is devoted to the numerical investigation of the transition and multiplicity phenomena in natural convection. The numerical simulation is performed for a 2-D closed domain using FLUENT, a commercial computational uid dynamics package. In this study, the e ects of di erent parameters such as Rayleigh number (Ra), aspect ratio (AR), and Prandtl number (Pr) are numerically analyzed. The complex phenomenon of onset of instability is studied to nd the critical Ra numbers and bifurcation points. In a speci c range of critical Ra numbers, transition and multiple results are observed using Rayleigh, boundary condition and pseudo-transient continuation methods. This method uses higher order temperature boundary conditions as an initial value on vertical walls. Multiple steady state results are observed at di erent Ra numbers. In a range of Ra numbers, the transition and multiple steady state results can completely change the convective ow pattern within the domain.Item Numerical analysis of surface-driven non-isothermal viscoelastic flow(Thesis (Ph.D.)-Bogazici University. Institute for Graduate Studies in Science and Engineering, 2010., 2010.) Kaptan, Yalın.; Ecder, Ali.; Atalık, Salim Kunt.The numerical investigations of the moving edge non-isothermal viscoelastic flows are simulated by using two example problems (lid driven cavity (LDC) and rotating disc in a cylindrical enclosure (RDCE) flows) in this study. The viscoelastic behavior of the fluids is modeled by adopting three differential constitutive relations namely Upper Convected Maxwell (UCM), Oldroyd B and Giesekus models. The comparisons reveal that the Giesekus model is the most realistic one and the maximum Weissenberg number limit is higher compared to the others. Two separate solvers are used in the simulations; PETSc and IN-GMRES solvers. PETSc code is used as a solver for the Newtonian flows and a benchmark tool for the Krylov subspace methods and preconditioners. PETSc analyses reveal that BiCGStab with ILU(5) preconditioning is the most effective solver in the simulations of the Newtonian flows. IN-GMRES solver is used to simulate the non-isothermal viscoelastic flows and it is based on the matrix free preconditioned inexact Newton-Krylov methods. To obtain higher Weissenberg number limits in the simulations, the numerical tools such as the continuation, the upwind differencing scheme, the higher order discretization schemes, the slanted stencils and similar others are implemented in the IN-GMRES algorithm. In the non-isothermal part of the study, besides the advection and diffusion, the viscous dissipation is also included and it is understood that the viscous dissipation is very important in simulations of non-Newtonian flows. The viscosity is modeled as temperature dependent by adopting the approximate Arrhenius formulation and it is realized that the viscosity changes can alter the flow field. The effects of the Reynolds number, the Weissenberg number, the Prandtl number, the Brinkman number, aspect ratio and some of the material parameters are documented within this study.Item Numerical investigation of convective-diffusive transport using the smoothed particle hydrodynamics technique(Thesis (M.S.) - Bogazici University. Institute for Graduate Studies in Science and Engineering, 2012., 2012.) Danış, Mustafa Engin.; Ecder, Ali.; Orhan, Mehmet.In this thesis, Incompressible Smoothed Particle Hydrodynamics (ISPH) technique is implemented to analyze various convective-diffusive transport problems numerically. First of all, a brief summary of applications and developments in SPH is given. Then, SPH methodology is presented. A detailed discussion on frequently used high order SPH approximation schemes is made and an expansion of existing projection methods developed for isothermal flow to non-isothermal and double-diffusive flows is introduced. The code developed during this thesis study is validated for isothermal problems such as lid-driven cavity and vortex spin-down. Furthermore, in order to test the upper limit of accuracy of SPH computations, a grid-based ISPH approach is proposed. Grid-based ISPH is applied to natural convection in a square cavity problem and the results are compared to the data available in the literature. Similar to gridbased ISPH code, meshless ISPH code is used to solve natural convection problems. Rayleigh-B́enard convection is studied and multiple states of the solutions are obtained. Moreover, natural convection is investigated at the onset of instability in which multicellular and oscillatory flow patterns are observed. Apart from natural convection, ISPH code also introduced to double-diffusive transport problems. In terms of doublediffusion, ISPH performance in aiding and opposing flows are provided. Furthermore, ISPH code is also implemented to two-phase flows. In the context of two-phase flows, topological changes of the interfaces and Rayleigh-Taylor instability problems are simulated. ISPH results for two-phase flow are compared to results obtained by Level Set Method.Item Numerical investigation of high Knudsen number flow in rectangular enclosures(Thesis (Ph.D.)-Bogazici University. Institute for Graduate Studies in Science and Engineering, 2006., 2006.) Orhan, Mehmet.; Ecder, Ali.; Tezel, Akın.Nowadays, enlightening unknown aspects of rarefied gas flow is one of the critical issues of fluid dynamic research to ensure correct and proper operations of manyMicro- Electro-Mechanical-Systems (MEMS). Thermally driven motion of rarefied gases is gaining in importance to develop Knudsen compressors having better performance or to improve single crystal growth processes. Therefore, accurate prediction of the physics lying behind the thermal creep in the transition regime as well as slip flow regime is one of the main motivations of this study. The other emphasis is possible flow instability of the rarefied gases in enclosures. For this purpose, an asymptotic approximation has been performed in the first part of the study to find analytical solutions. In the second one, linear disturbance theory of hydrodynamic stability has been applied to the problem to determine bounds of instabilities. Analytical solutions of two-dimensional stability analysis have been introduced. Critical states have been identified for different models and for varying Knudsen numbers. More generally, eigen-spectrum of the perturbation equations has been identified in three-dimensions. At the last part, by applications of an artificial viscosity scheme, a computer program has been constructed to solve Burnett and also Navier-Stokes equations. Mechanisms of the thermal creep flow have also been verified by inspecting stress tensors of Burnett equations. Most importantly, the insufficiency and the failure of Navier-Stokes equations for the creeping flows have been proved. Moreover, it has been shown that Burnett equations can correctly model such creeping flows.Item Numerical simulation of two phase flow using the homogeneous flow model(Thesis (M.S.) - Bogazici University. Institute for Graduate Studies in Science and Engineering, 1999., 1999.) Dal, Murat.; Ecder, Ali.The objective of this study is to understand physical and numerical modelling aspects of two phase flow occurring in heat exchangers used in domestic refrigerators, and construct a computer model to simulate one dimensional two phase flow using the homogeneous flow model.This study is mainly composed in two parts. First part focuses on the two phase flow characteristics of refrigerants in tube: Flow regimes, pressure drop, void fraction and heat transfer correlations are investigated. Empirical correlations are evaluated and compared for a typical evaporator condition. Second part, introduces governing equations for 1-D two phase flow and describes the development of a finite difference type numerical scheme for a simulation of the transient behaviour of two phase flow in heat exchangers. A second order accurate numerical method, namely MacCormack, is applied for solving the equations. Simulation results are compared with the literature and similar flow characteristics are found. Although the presented model is second order accurate and easy to apply, it is too slow for use in an overall refrigeration system simulation. Also, it requires small CFL number and fine grid for some applications which means more execution time.Homogeneous flow model was the simplest model. An improved model, at least including slip effects should be preferred. For a complete system simulation, a simplified system of equation should be taken with an advanced implicit method.Item Numerical study on airfoils in steady and oscillating streams(Thesis (M.S.) - Bogazici University. Institute for Graduate Studies in Science and Engineering, 1999., 1999.) Özalp, Mehmet Rifat.; Ecder, Ali.In this study, a numerical method has been developed in order to simulate steady and unsteady flow fields around airfoils. The unsteady flow field around the airfoil has been obtained by imposing periodically oscillating far field boundary conditions. In the interior domain, the Navier-Stokes equations have been solved by using Krylov subspace methods. The non-linear part of the problem has been linearized by Newtons method and GMRES is used as the linear solver with ILU preconditioner for efficient calculations. In order to reduce storage requirements, matrix free method is employed. The method of body-fitted curvilinear coordinate generation has been used to construct a finite-difference solution of the full, incompressible Navier-Stokes equations for the laminar viscous flow about arbitrary two-dimensional airfoils or any other two dimensional bodies. The governing equations in generelized curvilinear coordinate system were solved by using the stream function-vorticity formulation, which eliminates the pressure term. The velocity components have been calculated from the dependent variables, namely stream function and vorticity.Item Parallel and unstructured computation of 2-D and 3-D vortex flows(Thesis (M.S)-Bogazici University.Institute for Graduate Studies in Science and Engineering, 2005., 2005.) Dayı, Oğuz.; Ecder, Ali.; Özturan, Can.Vortex flows are main flows which lead to turbulence phenomenon. Turbulence isa complicated and many-faceted problem of fluid dynamics. These flows have striking and varied spatial organizations. Nearly every flow that can be observed in naturehas turbulent character. Consequently, engineering systems in various applicationsare affected by turbulence. Significant examples of such cases are severe efficiencydecreases of pumps and turbines which are designed for a specific design load. There are many cases in combustion or mixing applications where the high mixing and heattransferrates of the turbulent boundary layer are desirable. So, better understandingof turbulent flow regimes requires further investigation of vortex flows. Consistent withthis purpose, several groups of study were performed to cover different types of vortex flows within the context of this study.Firstly, a basic vortex flow is analyzed with various numerical techniques in thisstudy. This test problem is lid driven cavity flow. An implicit finite volume discretizationtechnique is used in terms of primitive variables for the unstructured code. Unstructured grids are handled with staggered grid arrangement for irregular domainsin 2D. Fractional step method is used as a velocity-pressure coupling.Finally, in addition to these analysis Taylor vortex flows in two different 3Dgeometries and 3D Centrifugal Pump flow analysis are conducted with the aid of variousturbulence models. LES (Large Eddy Simulation), k-epsilon, k-omega turbulencemodels are used. Analysis that are mentioned above is conducted with parallel processingtechnique.Item Wavelet analysis in computational fluid dynamics(Thesis (M.S.) - Bogazici University. Institute for Graduate Studies in Science and Engineering, 2018., 2018.) Kahraman, Ali Berk.; Ecder, Ali.Wavelets are compact functions in space and time which enable easy data com pression through multi-resolution analysis. The compression is done by scattering dif ferent resolution levels of wavelets onto the domain, and then discarding the wavelets with small energies. Using these properties, wavelets can be an efficient and easily applied tool to construct an adaptive grid for the solution of a Partial Differential Equation (PDE) with local structures. In this work, wavelets are used in this manner of compressing the interpolated data, being combined with finite difference discretiza tion to solve the PDE. To calculate the spatial derivatives of the compressed interpo lation, algebraic polynomial fits and cubic splines are used on the irregular adaptive grid. These two approaches are compared with each other and finite differences on regular grids. Various problems in 1-D and 2-D are solved. As model problems, Pois son’s Equation and Helmholtz Equation are solved with artificially created Gaussian Pulse as the solution. The results seemed to be in agreement in terms of the order of error with the known exact solutions of the model problems. A new application for wavelet optimized finite differences is also suggested. To that purpose, split-step (projection-correction) time scheme is implemented for the Navier-Stokes Equations governing infamous lid-driven cavity problem. The qualitative results seemed to be in agreement with other results in literature, however the method is observed to be not of any advantage for this problem as this problem does not have strong localized structures. PETSc framework provided the high-level tools, such as matrix and vector operations and linear solvers, and the work is conducted through this perspective.