Search results

Manifold‐mapping (MM) is an efficient surrogate‐based optimization technique aimed at the acceleration of very time‐consuming design problems. In this paper we present two new variants of the original algorithm that make it applicable to a broader range of optimization scenarios.

The first variant is useful when the optimization constraints are expressed by means of functions that are very expensive to compute. The second variant endows the original scheme with a trust‐region strategy and the result is a much more robust algorithm.

Two practical optimization problems from electromagnetics eventually show that the proposed variants perform efficiently.

The original MM algorithm is extended with two new variants. Therefore, the MM approach is applicable to a much larger set of design situations.

A flux‐difference splitting based on the polynomial character of the flux vectors is applied to steady Euler equations, discretized with a vertex‐centred finite volume method. In first order accurate form, a discrete set of equations is obtained which is both conservative and positive. Due to the positivity, the set of equations can be solved by collective relaxation methods in multigrid form. A full multigrid method based on successive relaxation, full weighting, bilinear interpolation and W‐cycle is used. Second order accuracy is obtained by the Chakravarthy‐Osher flux‐extrapolation technique, using the Roe‐Chakravarthy minmod limiter. In second order form, direct relaxation of the discrete equations is no longer possible due to the loss of positivity. A defect‐correction is used in order to solve the second order system.

Optimisation in electromagnetics, based on finite element models, is often very time‐consuming. In this paper, we present the space‐mapping (SM) technique which aims at speeding up such procedures by exploiting auxiliary models that are less accurate but much cheaper to compute.

The key element in this technique is the SM function. Its purpose is to relate the two models. The SM function, combined with the low accuracy model, makes a surrogate model that can be optimised more efficiently.

By two examples we show that the SM technique is effective. Further we show how the choice of the low accuracy model can influence the acceleration process. On one hand, taking into account more essential features of the problem helps speeding up the whole procedure. On the other hand, extremely simple auxiliary models can already yield a significant acceleration.

Obtaining the low accuracy model is not always straightforward. Some research could be done in this direction. The SM technique can also be applied iteratively, i.e. the auxiliary model is optimised aided by a coarser one. Thus, the generation of hierarchies of models seems to be a promising venue for the SM technique.

Optimisation in electromagnetics, based on finite element models, is often very time‐consuming. The results given show that the SM technique is effective for speeding up such procedures.

The purpose of this work is to investigate the similarity requirements for the application of multifidelity modeling (MFM) for the prediction of airfoil dynamic stall using computational fluid dynamics (CFD) simulations.

Dynamic stall is modeled using the unsteady Reynolds-averaged Navier–Stokes equations and Menter's shear stress transport turbulence model. Multifidelity models are created by varying the spatial and temporal discretizations. The effectiveness of the MFM method depends on the similarity between the high- (HF) and low-fidelity (LF) models. Their similarity is tested by computing the prediction error with respect to the HF model evaluations. The proposed approach is demonstrated on three airfoil shapes under deep dynamic stall at a Mach number 0.1 and Reynolds number 135,000.

The results show that varying the trust-region (TR) radius (λ) significantly affects the prediction accuracy of the MFM. The HF and LF simulation models hold similarity within small (λ ≤ 0.12) to medium (0.12 ≤ λ ≤ 0.23) TR radii producing a prediction error less than 5%, whereas for large TR radii (0.23 ≤ λ ≤ 0.41), the similarity is strongly affected by the time discretization and minimally by the spatial discretization.

The findings of this work present new knowledge for the construction of accurate MFMs for dynamic stall performance prediction using LF model spatial- and temporal discretization setup and the TR radius size. The approach used in this work is general and can be used for other unsteady applications involving CFD-based MFM and optimization.

An optimized multigrid method (NSFLEX‐MG) for the NSFLEX‐code (Navier‐Stokes solver using characteristic flux extrapolation) of MBB (Messerschmitt Bolkow Blohm GmbH) is described. The method is based on a correction scheme and implicit relaxation procedures and is applied to two‐dimensional test cases. The principal feature of the flow solver is a Godunov‐type averaging procedure based on the eigenvalues analysis of the Euler equations by means of which the inviscid fluxes are evaluated at the finite volume faces. Viscous fluxes are centrally differenced at each cell face. The performance of NSFLEX‐MG is demonstrated for a large range of Mach numbers for compressible inviscid and viscous (laminar and turbulent) flows over a RAE‐2822 airfoil and over a NACA‐0012 airfoil.

This work aims to present a multilevel optimization strategy based on manifold‐mapping combined with multiquadric interpolation for the coarse model construction.

In the proposed approach the coarse model is obtained by interpolating the fine model using multiquadrics in a small number of points. As the algorithm iterates the response surface model is improved by enriching the set of interpolation points.

This approach allows to accurately solve the TEAM Workshop Problem 25 using as little as 33 finite element simulations. Furthermore, it allows a robust sizing optimization of a cylindrical voice‐coil actuator with seven design variables.

Further analysis is required to gain a better understanding of the role that the initial coarse model accuracy plays in the convergence of the algorithm. The proposed model allows to carry out such analysis by varying the number of points included in the initial response surface model. The effect of the trust‐region stabilization in the presence of manifolds of equivalent solutions is also a topic of further investigations.

Unlike the closely related space‐mapping algorithm, the manifold‐mapping algorithm is guaranteed to converge to a fine model optimal solution. By combining it with multiquadric response surface models, its applicability is extended to problems for which other kinds of coarse model such as lumped parameter approximations for instance are tedious or impossible to construct.

This paper aims to introduce an original application of the corrected response surface method (CRSM) in the context of the optimal design of a permanent magnet synchronous machine used as an integrated starter generator. This method makes it possible to carry out this design in a very efficient manner, in comparison with conventional optimization approaches.

The search for optimal conditions is achieved by the joint use of two multi-physics models of the machine to be optimized. The former models most finely the physical functioning of the machine; it is called “fine model”. The second model describes the same physical phenomena as the fine model but must be much quicker to evaluate. Thus, to minimize its evaluation time, it is necessary to simplify it considerably. It is called “coarse model”. The lightness of the coarse model allows it to be used intensively by conventional optimization algorithms. On the other hand, the fine reference model makes it possible to recalibrate the results obtained from the coarse model at any instant, and mainly at the end of each classical optimization. The difference in definition between fine and coarse models implies that these two models do not give the same output values for the same input configuration. The approach described in this study proposes to correct the values of the coarse model outputs by constructing an adjustment (correcting) response surface. This gives the name to this method. It then becomes possible to have the entire load of the optimization carried over to the coarse model adjusted by the addition of this correction response surface.

The application of this method shows satisfactory results, in particular in comparison with those obtained with a traditional optimization approach based on a single (fine) model. It thus appears that the approach by CRSM makes it possible to converge much more quickly toward the optimal configurations. Also, the use of response surfaces for optimization makes it possible to capitalize the modeling data, thus making it possible to reuse them, if necessary, for subsequent optimal design studies. Numerous tests show that this approach is relatively robust to the variations of many important functioning parameters.

The CRSM technique is an indirect multi-model optimization method. This paper presents the application of this relatively undeveloped optimization approach, combining the features and benefits of (Indirect) efficient global optimization techniques and (multi-model) space mapping methods.

In the past decade, very effective techniques for the solution of the drift‐diffusion equations have been developed. This has enabled the simulation of a large variety of semiconductor devices in a robust and efficient way. However, the rapid development and miniaturization of devices necessitate the use of extended physical models in order to still provide an accurate description of device performance. In order to guarantee a similar degree of robustness and efficiency for these extended models, new algorithms have to be developed. In this paper we will present a number of advanced numerical techniques which have been developed for this purpose.

Considers the extension of a new class of higher‐order compact methods to nonuniform grids and examines the effect of pollution that arises with differencing the associated metric coefficients. Numerical studies for the standard model convection diffusion equation in 1D and 2D are carried out to validate the convergence behaviour and demonstrate the high‐order accuracy.

Development of techniques for expedited design optimization of complex and numerically expensive electromagnetic (EM) simulation models of antenna structures validated both numerically and experimentally. The paper aims to discuss these issues.

The optimization task is performed using a technique that combines gradient search with adjoint sensitivities, trust region framework, as well as EM simulation models with various levels of fidelity (coarse, medium and fine). Adaptive procedure for switching between the models of increasing accuracy in the course of the optimization process is implemented. Numerical and experimental case studies are provided to validate correctness of the design approach.

Appropriate combination of suitable design optimization algorithm embedded in a trust region framework, as well as model selection techniques, allows for considerable reduction of the antenna optimization cost compared to conventional methods.

The study demonstrates feasibility of EM-simulation-driven design optimization of antennas at low computational cost. The presented techniques reach beyond the common design approaches based on direct optimization of EM models using conventional gradient-based or derivative-free methods, particularly in terms of reliability and reduction of the computational costs of the design processes.

Simulation-driven design optimization of contemporary antenna structures is very challenging when high-fidelity EM simulations are utilized for performance utilization of structure at hand. The proposed variable-fidelity optimization technique with adjoint sensitivity and trust regions permits rapid optimization of numerically demanding antenna designs (here, dielectric resonator antenna and compact monopole), which cannot be achieved when conventional methods are of use. The design cost of proposed strategy is up to 60 percent lower than direct optimization exploiting adjoint sensitivities. Experimental validation of the results is also provided.