Conference Agenda

A pair of Helmholtz coils consist of two identical circular coils aligned along their main axis. Following a specific sizing, it can generate a magnetic field mostly uniform along one direction between the two coils. When control over the three dimensions is desired, three pairs of such coils are required. They are oriented and scaled to prevent collisions between the pairs in space. This inevitably reduces the relative size of the region with a homogeneous magnetic field, degrading the performance and the integration of the whole system. The proposed work aims to improve this 200-year old device with the help of computational design, to find a better arrangement of the coils in innovative shapes. A topology optimization framework is developed, investigating the optimal distribution of coils in space that results in a large and balanced region with a uniform magnetic field, while still improving integration.

In this paper, a method for Topology Optimization (TO) of a Synchronous Reluctance motor is proposed using Deep Reinforcement Learning. Due to the need for simulating a large number of finite element models in a traditional topology optimization task, incorporating a study involving a different problem formulation (such as a varying design domain) can be an overwhelming task. A neural network-based agent trained using reinforcement learning formulation is able to extend the knowledge from one TO design problem to other similar TO tasks. The applicability of such learning is performed using a sequence based TO environment.

This paper presents an artificial neural network (ANN) approach for design optimization of electromagnetic planar coil. Two parallel approaches are implemented. First one is with radial basis function ANN used for objective function problem interpolation. Second one employs the ANN as an optimization problem parameters preselection. ANN interpolation model that employs, in a natural and effective way, an inversion algorithm providing a solution of the electromagnetic device design problem. Further development of this optimization model can propose an efficient general solution to the electromagnetic device design problem from the same complexity class. The interpolation ANN model and optimization ANN control are both applied to the magnetic planar high frequency coil design. The results obtained shows the effectiveness of the proposed optimization method.

We describe a wireless power transfer (WPT) system for linear drives based on inductive resonant coupling. The system is capable

of continuously transmitting power from the transmitter to the receiver coil, independent from its movement or position. The paper

proposes a solution and presents its rigorous analysis by means of 2-D and 3-D FEM simulations. 2-D and 3-D FEM simulations

are combined in order to compute the parameters of the coupling coils. The obtained solutions are further optimized in Matlab by

connecting an optimization algorithm of Linear Regression with a circuit model. A prototype of one optimized solution has been

built and tested. The presented comparison of the measured and simulated results reveal a high level of their agreement.

This digest presents an investigation on the optimal positioning of magnetic shunts in power transformers. Magnetic shunts are made of grain-oriented magnetic silicon steel laminations and are used to avoid hotspots in metallic regions inside the power transformer. The magnetodynamic equations are solved in terms of the electric vector potential and the magnetic scalar potential. Thermal effects are calculated using heat transfer coefficients. This digest proposes and solves an optimization problem where the objective is to position several shunts in the tank wall. There are many design variables, temperature, and geometrical constraints. Since the FE model to be solved is 3D, the computing resources must be efficiently used . A multi-threading method was developed and numerically tested where the 3D FE model, temperature calculation, and the Particle Swarm Optimizer are executed in a multicore server in a parallel scheme.

We propose an innovative technique for dealing with optimal shape design problems that exploits the flexibility of the Virtual Element Method in handling meshes with general polygonal and polyhedral elements. The shape synthesis of a magnetic pole is considered as the case study.

The high torque density of electrical motors is critical for many applications. This paper presents an optimization of the switched reluctance motor (SRM) geometry aiming to maximize the torque density. The optimal geometry of the SRM is obtained by utilizing the direct optimization approach, where the multi-objective genetic algorithm is used along with the finite element (FE) model of the motor. To reduce the computational complexity of the optimization, half of the motor was simulated in 2D. The optimization of the geometrical parameters of the SRM helped to achieve a higher torque density compared to the initial design. The static torque of the motor was improved by more than 10%, while the volume of the motor was reduced by more than 10%.

This paper presents the optimization of the design of an electrostatic micromotor using an ellipsoid method. The optimization of the micromotor, modelled with the Finite Element Method, would require a lot more solutions of the system and considerably more computing power to get to an optimum design. To solve this problem, this work uses the adjoint-state method to obtain the sensitivity information of the design and compare the results with the central finite difference method. This methodology gets new design with better torque characteristics with more precision without compromising the execution time by using the numerical modeling of the problem.

Index Terms— Adjoint method, Finite Elements, Micromotor, Sensitivity analysis.

While many left ventricular assist devices (LVADs) have been developed since the 1970s to treat adult patients with heart failure, there are limited LVAD options to treat pediatric patients with heart failure. To reduce blood trauma and improve device longevity, we are developing a set of passive magnetic levitation (maglev) bearings. These maglev bearings are made of two axially polarized permanent magnet (PM) rings oriented for radial stability. To identify the optimal PM ring geometry, we characterized the forces generated by PM rings of varied sizes in silico and validated the results experimentally. This validated numerical model can facilitate our design process for the bearing system of the pediatric LVAD.

The deviation of the built topology from the optimized one arising from manufacture tolerances can result in a degradation of the performance of the optimized solution for microscale magnetic devices. Therefore, Robust Optimization (RO) considering manufacturing tolerances has become a significant concern in Topology Optimization (TO) problem. To eliminate the extremely high computational burden and the checkboard pattern problem of existing RO methodology, a novel methodology based on robust min-cut for TO is firstly proposed. The proposed method finds the optimal direction through solving a robust min-cut problem. The numerical result of a magnetic actuator has validated the proposed method.

This paper offers a Topology Optimization (TO) method to replace conventional parameter optimization, as a common method for electrical machine’s or electromagnetic device’s design optimization. In most of the final designs gained from TO, there are challenges for the manufacturing process using the nowadays manufacturing technologies. This work investigates a two-level approach. In the first level a global optimization method, called Enhanced Binary Genetic Algorithm, is applied to the design domain. After optimization with this meta-heuristic method, some holes inside the structure appear. Morphological Reconstruction will improve this drawback in the second level, which is one of the image processing methods to reduce the manufacturing complexities with a minor influence on the technical characteristics. The goal is to find a lightweight structure, while preserving the force of electromagnetic device in a reasonable range. The proposed method is applied to an E-core actuator as a case study. After using the proposed hybrid optimization method, some air cells appear around the actuator’s coil, while extending the electric current carrying copper area to that air cells, the magnetic force considerably improves. Results show that it is possible to extend this work to different active parts of the electrical machines.

Linear induction motors (LIMs) are presently considered as an interesting and promising solution for future transportation systems, motion control applications, linear automation systems, etc. Contrary to their rotational counterparts, simulation based design and optimization of LIMs is still an open research subject due to their specific geometrical- and electromagnetic features. This paper presents 1-D analytical- and 2-D/3-D Finite-Element-Method-based numerical modeling and simulation of LIMs. To demonstrate different levels of accuracy associated with different levels of modeling complexity, a testing LIM has been developed, manufactured, and measured. A special attention in this study is paid to LIM-specific geometrical- and electromagnetic features and their influence on the LIM characteristics.

In this paper are presented the main steps for developing an analytical methodology to design Transverse Flux Permanent Magnet Synchronous Generators (TF-PMSG’s) for its use in small direct-drive wind turbines. First, to attain this aim, it is proposed the full magnetic circuit representation of the TF-PMSG. Next, the nodal analysis technique is used to find an algebraic expression to solve the airgap flux. The proposed analytical methodology for designing TF-PMSG’s has been programmed in Matlab and also it is well suited for analyzing their performance over a full range of operation speeds and load conditions. Finally, the methodology has been exhaustively validated through 3D Finite Element Analysis (FEA).

This paper proposes a new multi-objective cascade optimization approach to investigate the feasibility of a 12-stator/10-rotor-pole variable flux reluctance machine (VFRM) for heavy-duty applications requiring high-torque generation. The developed algorithm consists of five cascaded parts: initial sizing, optimization in radial, and tangential directions, hysteresis analysis, and thermal design. The first step, so-called initial sizing, scales a reference VFRM to reach the desired torque specification. The solution space of the optimization in both radial and tangential directions is determined using the scaled geometry. The radial parameters of the VFRM are optimized for the maximum torque density and efficiency. Then, the tangential parameters are optimized using the fixed radial parameters for the minimum torque ripple. Later, the optimized VFRM is analyzed considering the vector hysteresis behavior of the soft-magnetic material to verify the torque generation of the designed VFRM. Lastly, the thermal aspects of the VFRM are analyzed for the selected current density aiming at the final design with the cooling. A 2-D nonlinear finite element model is employed to calculate the objective function. In addition, the evolutionary genetic algorithm is used in both optimization stages. The proposed design strategy decreases the solution space of the optimization and increases the chance of reaching the global optimum design.

A key design problem for photo-electron guns is minimizing the electric field strength on the electrode surface to avoid field emission. Isogeometric analysis (IGA) allows to compute accurate approximations of the field by using non-uniform rational B-splines (NURBS) to describe both the computational domain and the numerical solution. The control points of these NURBS offer an intuitive set of degrees of freedom, and make it possible to freeform optimize the shape of the electrode efficiently. Several beam parameters are considered in the optimization to ensure proper gun performance. The results of an IGA-based shape optimization for a planned high-voltage upgrade of the photogun teststand Photo-CATCH at TU Darmstadt are presented.

Results of this study demonstrate an increase of 40% of the capacitance when the genetic algorithm optimization technique is applied. A non-conventional approach constructing the geometry of the capacitor's fingers with Bezier's curves was used for molding each finger as if they were dough evolving as a live organism trying to adapt to their environment, which demands the best capacitance to survive in it. A comparative study by numerical simulation between Conventional-Interdigital-Capacitor, Interdigital-Capacitor-With-More-Fingers (without perimeter variation) and the Evolutive-Capacitor were carried out to demonstrate the effectiveness of the proposed technique. Fabrication and characterization of the evolutive capacitor is also presented.