Bogdan Cranganu-Cretu
Politehnica University of Bucharest
Network
Latest external collaboration on country level. Dive into details by clicking on the dots.
Publication
Featured researches published by Bogdan Cranganu-Cretu.
IEEE Transactions on Magnetics | 2006
Joerg Marc Ostrowski; Zoran Andjelic; Mario Bebendorf; Bogdan Cranganu-Cretu; Jasmin Smajic
The main problems for applying boundary element methods (BEM) in computational electromagnetism are related to the large memory requirements of the matrices and the convergence of the iterative solver. In this paper, we solve a Laplace problem with mixed boundary conditions by making use of a variational symmetric direct boundary integral equation. The Galerkin discretization results in densely populated matrices that are here compressed by adaptive cross approximation. This leads to an approximation of the underlying BEM-operator by means of so-called hierarchical matrices (H-Matrices). These matrices are then used to construct an effective preconditioner for the iterative solver. Numerical experiments demonstrate the application of the method
IEEE Transactions on Magnetics | 2002
Gabriel Preda; Bogdan Cranganu-Cretu; Florea I. Hantila; Ovidiu Mihalache; Zhenmao Chen; Kenzo Miya
Pulse eddy currents are proposed as a nondestructive testing (NDT) technique to detect flaws in conductive structures with large thickness. The harmonic component of a pulse is rich, so that the pick-up signal containing the amount of information corresponds to a multifrequency analysis. Due to the short time length of the pulse, the amplitude of the excitation increases up to 100 times of the amplitude for an AC signal. Both direct simulation of pulse eddy-currents phenomena using an A-/spl phi/ FEM-BEM code and neural network-based inversion techniques are performed. Numerical results for the inversion of signals due to outer defects are shown.
international conference on electrical machines | 2010
Thorsten Steinmetz; Bogdan Cranganu-Cretu; Jasmin Smajic
Amorphous metal core dry-type distribution transformers have a significant place in today power market since they exhibit 60–70% lower no-load losses when compared to the standard steel based transformers. At typical loading regimes these losses are the most significant ones for a distribution transformer. In order to improve the design and to shorten significantly the time to market of such products numerical analysis of the no-load as well as of the load tests are of paramount importance. The present manuscript describes the FEM based numerical analysis performed on a 1 MVA unit, in an industrial environment, with the emphasis on the importance of non-linear and anisotropic magnetic modeling as opposed to isotropic for the accurate computation of no-load losses. The choices for the analysis regime and the specific analysis tools are described in details - and results are given to prove the salient points of this particular design of a transformer.
IEEE Transactions on Magnetics | 2010
Christian Magele; Alice Köstinger; Michael Jaindl; Werner Renhart; Bogdan Cranganu-Cretu; Jasmin Smajic
Many real world optimization problems turn out to be multi-objective optimization problems revealing a remarkable number of locally optimal solutions corresponding to the chosen objective function. Therefore, it seems desirable to detect as many of those solutions with as few objective function calls as possible. A Niching Higher Order Evolution Strategy (NES) can successfully be applied to locate a large number of these local solutions during a single optimization run. Additionally, it turns out that all of these solutions can be found next to the front of non-dominated solutions. Therefore, evaluating more than one objective function (in parallel or in series) yields a good approximation of the Pareto-optimal front. The proposed method will be tested against several test functions and then applied to the solution of a magnetic shunting problem.
IEEE Transactions on Industrial Electronics | 2010
Bogdan Cranganu-Cretu; Audrey Kertesz; Jasmin Smajic
This paper presents a Visual Basic software solution, aimed at driving a complex finite-element-method-based electromagnetic-thermal solver to allow for a range of virtual experiments for accurate simulations of numerically demanding stray-magnetic-flux problems in power devices. The development allows for both engineering experiments (by a nonexpert numerical user) and for numerical experiments to determine the best setup of a simulation attempt.
IEEE Transactions on Magnetics | 2008
Bogdan Cranganu-Cretu; Jasmin Smajic; Jörg Ostrowski; Werner Renhart; Christian Magele
There is a strong requirement within industrial environment for software integrated solutions for the design based on optimization that present a design engineer with little or no experience in numerical modeling with the possibility of obtaining meaningful engineering results. The core idea is to provide the user with an interface where only geometrical, material, excitation, or global field quantities are present, hiding from him the details of both the finite-element method (FEM) solution part as well as the optimization itself. Thus, a design engineer will be able to work within an environment with which he is highly familiar, and concentrate on the engineering task instead of the numerical intricacies. To exemplify, we have chosen a simple device, developed a driving application for the FEM field solver, identified meaningful parameters and then searched for suitable optimization procedures that we then coupled with the driving application.
international conference on electrical machines | 2008
Bogdan Cranganu-Cretu; Audrey Kertesz; Jasmin Smajic
This paper presents an ad-hoc application, developed in VB to drive a complex FEM based electromagnetic-thermal solver to allow for a range of virtual experiments. The development allows for both engineering experiments (by a non-expert numerical user) as well as for numerical experiments - to determine the best set-up of a simulation attempt.
IEEE Transactions on Magnetics | 2002
Bogdan Cranganu-Cretu; Florea I. Hantila; Gabriel Preda; Zhenmao Chen; Kenzo Miya
In order to accurately compute the magnetic field variation due to changes in material properties (e.g. material aging), we propose a technique that computes the difference magnetic field directly. Nonlinear materials are analyzed by means of the polarization technique. The linear field problem is solved using a fast Green function approach. A two-dimensional formulation is validated upon comparison with measurement data. Then, it is used as fast forward solver for a neural network approach to the inverse problem of reconstructing the shape of the aged material area in a plate. Finally, results of reconstruction, based on 500 case database simulations of a yoke and plate geometry, are presented, indicating the good quality of the reconstruction.
Compel-the International Journal for Computation and Mathematics in Electrical and Electronic Engineering | 2010
Christian Magele; Michael Jaindl; Alice Köstinger; Werner Renhart; Bogdan Cranganu-Cretu; Jasmin Smajic
Purpose – The purpose of this paper is to extend a (μ/ρ, λ) evolution strategy to perform remarkably more globally and to detect as many solutions as possible close to the Pareto optimal front.Design/methodology/approach – A C‐link cluster algorithm is used to group the parameter configurations of the current population into more or less independent clusters. Following this procedure, recombination (a classical operator of evolutionary strategies) is modified. Recombination within a cluster is performed with a higher probability than recombination of individuals coming from detached clusters.Findings – It is shown that this new method ends up virtually always in the global solution of a multi‐modal test function. When applied to a real‐world application, several solutions very close to the front of Pareto optimal solutions are detected.Research limitations/implications – Stochastic optimization strategies need a very large number of function calls to exhibit their ability to reach very good local if not t...
IEEE Transactions on Magnetics | 2008
Valentin Ionita; Bogdan Cranganu-Cretu
The local correlation between the macroscopic numerical simulations and the microscopic experimental behavior of soft magnetic materials (Fe-Si sheets) is investigated. The 2-D and 3-D magnetic field computation involves finite edge-elements and a nonlinear model of the material. The experimental investigation uses the microscopic magneto-optical imaging for thin samples of magnetic material.