Phd Degree / Doktora

Permanent URI for this collectionhttps://hdl.handle.net/11147/2869

Browse

Filters

2012 - 201922020 - 20242

Settings

Has files: YesSubject: search.filters.subject.Finite element method

Search Results

Now showing 1 - 4 of 4
  • Doctoral Thesis
    Reliability Assessment Based on Structural Health Monitoring Data and Bayesian Updating of Structural Models
    (01. Izmir Institute of Technology, 2024) Uzun, Ertuğrul Türker; Aktaş, Engin; Hızal, Çağlayan
    Finite element (FE) models are commonly used in numerical modeling of structures, but their assumptions can lead to inaccuracies and uncertainties. To address this, FE model update methods have been developed, calibrating the model based on structural health monitoring (SHM) data. However, a general framework for realistic life cycle performance assessment of structures using monitored data has not yet been presented. Bayesian modeling can characterize uncertain structural parameters as random variables, but it is complex and time-consuming. Metamodeling techniques, which are effective stochastic predictors, can be used to decrease the computational burden of model updating. Adapting a Polynomial-Chaos-Kriging (PCK) metamodeling technique to Bayesian model updating in order to reduce uncertainty and circumvent computational challenges using SHM data in order to assess the reliability of structures more precisely is the objective of this research. Therefore, the effectiveness of the proposed method has been tried and demonstrated through experimental and numerical studies. An experimental study of a bridge column is used to evaluate the reliability of structures subjected to various corrosion effects. As a result, the proposed solution method reduces computational costs and enables an updated FE model to be closer to real structure measurements. The updated models are found to be more reliable in reliability evaluations, providing more accurate predictions on issues like structure safety, service life, and maintenance cost compared to non-updated models.
  • Doctoral Thesis
    Designing Composite-Based Cylindrical Structures and Manufacturing Composite Prototypes by Filament Winding Method
    (01. Izmir Institute of Technology, 2024) Martin, Seçkin; Tanoğlu, Metin
    This study reports the design, finite element modeling, optimization, fabrication and testing of relatively thick (radius/thickness ~ 7) and long carbon fiber reinforced polymers produced by filament winding against buckling damage under axial loading. The optimum winding angle and stacking sequence against Linear (Eigenvalue) buckling were determined in accordance with the predetermined design requirements utilizing genetic algorithm (GA) optimization via MATLAB. During the optimization process, the critical buckling load factor (λcr) was assigned as objective function, design constraints were natural frequency (fn) and angle of twist (Φ), and ply angles were considered to be variable and restricted with 20 to 87-degree continuous fiber angles in the laminate sequences. As a consequence of the test results, λcr of the proposed optimum model was found to be 3.2 times better than the reference model and both the analytical and finite element model satisfactorily predicted the critical buckling load for all CFRP rods consistent with the test results. The critical buckling loads calculated by applying a KDF of 0.95 for the finite element model and a KDF of 0.9 for the analytical solution were found to be reasonably appropriate for use in the preliminary design input. Additionally, results showed that a higher axial to the circumferential ratio of axial and bending stiffness (A11/A22, D11/D22) promises better buckling performance than other possible candidates. Finally, the microstructures of the produced rods were examined and the fiber volume ratios were calculated by means of chemical characterization.
  • Doctoral Thesis
    Enriched Finite Elements Method for Convevtion-Diffusion Problems
    (Izmir Institute of Technology, 2012) Şendur, Ali; Pashaev, Oktay
    In this thesis, we consider stabilization techniques for linear convection-diffusionreaction (CDR) problems. The survey begins with two stabilization techniques: streamline upwind Petrov-Galerkin method (SUPG) and Residual-free bubbles method (RFB). We briefly recall the general ideas behind them, trying to underline their potentials and limitations. Next, we propose a stabilization technique for one-dimensional CDR problems based on the RFB method and particularly designed to treat the most interesting case of small diffusion. We replace the RFB functions by their cheap, yet efficient approximations which retain the same qualitative behavior. The approximate bubbles are computed on a suitable sub-grid, the choice of whose nodes are critical and determined by minimizing the residual of a local problem. The resulting numerical method has similar stability features with the RFB method for the whole range of problem parameters. We also note that the location of the sub-grid nodes suggested by the strategy herein coincides with the one described by Brezzi and his coworkers. Next, the approach in one-dimensional case is extended to two-dimensional CDR problems. Based on the numerical experiences gained with this work, the pseudo RFBs retain the stability features of RFBs for the whole range of problem parameters. Finally, a numerical scheme for one-dimensional time-dependent CDR problem is studied. A numerical approximation with the Crank-Nicolson operator for time and a recent method suggested by Neslitürk and his coworkers for the space discretization is constructed. Numerical results confirm the good performance of the method.
  • Doctoral Thesis
    Stabilized finite element methods for time dependent convection-diffusion equations
    (Izmir Institute of Technology, 2012) Baysal, Onur; Tanoğlu, Gamze
    In this thesis, enriched finite element methods are presented for both steady and unsteady convection diffusion equations. For the unsteady case, we follow the method of lines approach that consists of first discretizing in space and then use some time integrator to solve the resulting system of ordinary differential equation. Discretization in time is performed by the generalized Euler finite difference scheme, while for the space discretization the streamline upwind Petrov-Galerkin (SUPG), the Residual free bubble (RFB), the more recent multiscale (MS) and specific combination of RFB with MS (MIX) methods are considered. To apply the RFB and the MS methods, the steady local problem, which is as complicated as the original steady equation, should be solved in each element. That requirement makes these methods quite expensive especially for two dimensional problems. In order to overcome that drawback the pseudo approximation techniques, which employ only a few nodes in each element, are used. Next, for the unsteady problem a proper adaptation recipe, including these approximations combined with the generalized Euler time discretization, is described. For piecewise linear finite element discretization on triangular grid, the SUPG method is used. Then we derive an efficient stability parameter by examining the relation of the RFB and the SUPG methods. Stability and convergence analysis of the SUPG method applied to the unsteady problem is obtained by extending the Burman’s analysis techniques for the pure convection problem. We also suggest a novel operator splitting strategy for the transport equations with nonlinear reaction term. As a result two subproblems are obtained. One of which we may apply using the SUPG stabilization while the other equation can be solved analytically. Lastly, numerical experiments are presented to illustrate the good performance of the method.