1 |
Robust and stable discrete adjoint solver development for shape optimisation of incompressible flows with industrial applicationsWang, Yang January 2017 (has links)
This thesis investigates stabilisation of the SIMPLE-family discretisations for incompressible flow and their discrete adjoint counterparts. The SIMPLE method is presented from typical \prediction-correction" point of view, but also using a pressure Schur complement approach, which leads to a wider class of schemes. A novel semicoupled implicit solver with velocity coupling is proposed to improve stability. Skewness correction methods are applied to enhance solver accuracy on non-orthogonal grids. An algebraic multi grid linear solver from the HYPRE library is linked to flow and discrete adjoint solvers to further stabilise the computation and improve the convergence rate. With the improved implementation, both of flow and discrete adjoint solvers can be applied to a wide range of 2D and 3D test cases. Results show that the semi-coupled implicit solver is more robust compared to the standard SIMPLE solver. A shape optimisation of a S-bend air flow duct from a VW Golf vehicle is studied using a CAD-based parametrisation for two Reynolds numbers. The optimised shapes and their flows are analysed to con rm the physical nature of the improvement. A first application of the new stabilised discrete adjoint method to a reverse osmosis (RO) membrane channel flow is presented. A CFD model of the RO membrane process with a membrane boundary condition is added. Two objective functions, pressure drop and permeate flux, are evaluated for various spacer geometries such as open channel, cavity, submerged and zigzag spacer arrangements. The flow and the surface sensitivity of these two objective functions is computed and analysed for these geometries. An optimisation with a node-base parametrisation approach is carried out for the zigzag con guration channel flow in order to reduce the pressure drop. Results indicate that the pressure loss can be reduced by 24% with a slight reduction in permeate flux by 0.43%.
|
2 |
Gradient-Based Optimization of Highly Flexible Aeroelastic StructuresMcDonnell, Taylor G. 21 April 2023 (has links) (PDF)
Design optimization is a method that can be used to automate the design process to obtain better results. When applied to aeroelastic structures, design optimization often leads to the creation of highly flexible aeroelastic structures. There are, however, a number of conventional design procedures that must be modified when dealing with highly flexible aeroelastic structures. First, the deformed geometry must be the baseline for weight, structural, and stability analyses. Second, potential couplings between aeroelasticity and rigid-body dynamics must be considered. Third, dynamic analyses must be modified to handle large nonlinear displacements. These modifications to the conventional design process significantly increase the difficulty of developing an optimization framework appropriate for highly flexible aeroelastic structures. As a result, when designing these structures, often either gradient-free optimization is performed (which limits the optimization to relatively few design variables) or optimization is simply omitted from the design process. Both options significantly decrease the design exploration capabilities of a designer compared to a scenario in which gradient-based optimization is used. This dissertation therefore presents various contributions that allow gradient-based optimization to be more easily used to optimize highly flexible aeroelastic structures. One of our primary motivations for developing these capabilities is to accurately capture the design constraints of solar-regenerative high-altitude long-endurance (SR-HALE) aircraft. In this dissertation, we therefore present a SR-HALE aircraft optimization framework which accounts for the peculiarities of structurally flexible aircraft while remaining suitable for use with gradient-based optimization. These aircraft tend to be extremely large and light, which often leads to significant amounts of structural flexibility. Using this optimization framework, we design an aircraft that is capable of flying year-round at \SI{35}{\degree} latitude at \SI{18}{\kilo\meter} above sea level. We subject this aircraft to a number of constraints including energy capture, energy storage, material failure, local buckling, stall, static stability, and dynamic stability constraints. Critically, these constraints were designed to accurately model the actual design requirements of SR-HALE aircraft, rather than to provide a rough approximation of them. To demonstrate the design exploration capabilities of this framework, we also performed several parameters sweeps to determine optimal design sensitivities to altitude, latitude, battery specific energy, solar efficiency, avionics and payload power requirements, and minimum design velocity. Through this optimization framework, we demonstrate both the potential of gradient-based optimization applied to highly flexible aeroelastic structures and the challenges associated with it. One challenge associated with the gradient-based optimization of highly flexible aeroelastic structures, is the ability to accurately, efficiently, and reliably model the large deflections of these structures in gradient-based optimization frameworks. To enable large-scale optimization involving structural models with large deflections to be performed more easily, we present a finite-element implementation of geometrically exact beam theory which is designed specifically for gradient-based optimization. A key feature of this implementation of geometrically exact beam theory is its compatibility with forward and reverse-mode automatic differentiation, which allows accurate design sensitivities to be obtained with minimal development effort. Another key feature is its native support for unsteady adjoint sensitivity analysis, which allows design sensitivities to be obtained efficiently from time-marching simulations. Other features are also presented that build upon previous implementations of geometrically exact beam theory, including a singularity-free rotation parameterization based on Wiener-Milenkovi\'c parameters, an implementation of stiffness-proportional structural damping using a discretized form of the compatibility equations, and a reformulation of the equations of motion for geometrically exact beam theory from a fully implicit index-1 differential algebraic equation to a semi-explicit index-1 differential algebraic equation. Several examples are presented which verify the utility and validity of each of these features. Another challenge associated with the gradient-based optimization of highly flexible aeroelastic structures is the ability to reliably track and constrain individual dynamic stability modes across the design iterations of an optimization framework. To facilitate the development of mode-specific dynamic stability constraints in gradient-based optimization frameworks we develop a mode tracking method that uses an adaptive step size in order to maintain an arbitrarily high degree of confidence in mode correlations. This mode tracking method is then applied to track the modes of a linear two-dimensional aeroelastic system and a nonlinear three-dimensional aeroelastic system as velocity is increased. When used in a gradient-based optimization framework, this mode tracking method has the potential to allow continuous dynamic stability constraints to be constructed without constraint aggregation. It also has the potential to allow the stability and shape of specific modes to be constrained independently. Finally, to facilitate the development and use of highly flexible aeroelastic systems for use in gradient-based optimization frameworks, we introduce a general methodology for coupling aerodynamic and structural models together to form modular monolithic aeroelastic systems. We also propose efficient methods for computing the Jacobians of these coupled systems without significantly increasing the amount of time necessary to construct these systems. For completeness we also discuss how to ensure that the resulting system of equations constitutes a set of first-order index-1 differential algebraic equations. We then derive direct and adjoint sensitivities for these systems which are compatible with automatic differentiation so that derivatives for gradient-based optimization can be obtained with minimal development effort.
|
3 |
Qualification des simulations numériques par adaptation anisotropique de maillages / Qualification of numerical simulations by anisotropic mesh adaptationNguyen-Dinh, Maxime 19 March 2014 (has links)
La simulation numérique est largement utilisée pour évaluer les performances aérodynamiques des aéronefs ainsi qu'en optimisation de forme. Ainsi l'objectif de ces simulations est souvent le calcul de fonctions aérodynamiques. L'objet de cette thèse est d'étudier des méthodes d'adaptation de maillages basées sur la dérivée totale de ces fonctions par rapport aux coordonnées du maillage (notée dJ/dX). Celle-ci pouvant être calculée par la méthode adjointe discrète. La première partie de cette étude concerne l'application de méthodes d'adaptation de maillages appliquées à des écoulements de fluides parfaits. Le senseur qui détecte les zones de maillage à raffiner s'appuie sur la norme de cette dérivée pour adapter des maillages pour le calcul d'une fonction J. La seconde partie du travail est la construction et l'étude de critères plus fiables basés sur dJ/dX pour d'une part adapter des maillages et d'autre part estimer si un maillage est bien adapté ou non pour le calcul de la fonction J. De plus une méthode de remaillage plus efficace basée sur une EDP elliptique est aussi présentée. Cette nouvelle méthode est appliquée pour des écoulements bidimensionnels de fluides parfaits ainsi que pour un écoulement décrit par les équations RANS. La dernière partie de l'étude est consacrée à l'application de la méthode proposée à des cas tridimensionnels d'écoulement RANS sur des géométries d'intérêt industriel. / Numerical simulation is widely used for the assessment of aircraft aerodynamic performances and shape optimizations. Hence the objective of these simulations is often to compute aerodynamic outputs. The purpose of this thesis is to study mesh adaptation methods based on the total derivative of the outputs with respect to mesh coordinates (denoted dJ/dX). This derivative can be computed using the discrete adjoint method. The first part of this study is about the application of mesh adaptation methods applied for Eulerian flows. The mesh locations to refine are detected using a sensor based on the norm of the derivative dJ/dX. This study confirmed that this derivative is relevant in order to adapt a mesh for the computation of the output J. The second part of this work is the construction and the study of reliable criteria based on dJ/dX for both mesh adaptation and the quality assessment of a given mesh for the computation of the output J. Moreover a more efficient remeshing method based on an elliptic PDE is presented too. This new method is applied for both two-dimensional Eulerian flows and a flow described by the RANS equations. The last part of the study is devoted to the application of the proposed method to three-dimensional RANS flows on geometries of industrial interest.
|
4 |
Airfoil Optimization for Unsteady Flows with Application to High-lift Noise ReductionRumpfkeil, Markus Peer 26 February 2009 (has links)
The use of steady-state aerodynamic optimization methods in the computational
fluid dynamic (CFD) community is fairly well
established. In particular, the use of adjoint methods has proven to be very
beneficial because their cost is independent of the number of design variables.
The application of numerical optimization to airframe-generated noise, however, has not received as much attention, but with the significant
quieting of modern engines, airframe noise now competes with engine noise.
Optimal control techniques for unsteady flows are needed in order to be able to reduce airframe-generated noise.
In this thesis, a general framework is formulated to calculate the gradient of a cost function in a nonlinear unsteady flow environment
via the discrete adjoint method. The unsteady optimization algorithm developed in this work
utilizes a Newton-Krylov approach since the gradient-based optimizer uses the quasi-Newton method BFGS, Newton's method is applied to the
nonlinear flow problem, GMRES is used to solve the resulting linear problem inexactly, and last but not least the linear adjoint problem
is solved using Bi-CGSTAB. The flow is governed by the unsteady two-dimensional
compressible Navier-Stokes equations in conjunction with a one-equation turbulence model, which are discretized using
structured grids and a finite difference approach. The effectiveness of the unsteady optimization algorithm is demonstrated
by applying it to several problems of interest including shocktubes,
pulses in converging-diverging nozzles, rotating cylinders, transonic buffeting, and an unsteady trailing-edge flow.
In order to address radiated far-field noise, an acoustic wave propagation program based on
the Ffowcs Williams and Hawkings (FW-H) formulation is implemented and validated. The general framework is then used
to derive the adjoint equations for a novel hybrid URANS/FW-H optimization algorithm
in order to be able to optimize the shape of airfoils based on their calculated far-field pressure fluctuations.
Validation and application results for this novel hybrid URANS/FW-H optimization algorithm show that it is possible
to optimize the shape of an airfoil in an unsteady flow environment to minimize
its radiated far-field noise while maintaining good aerodynamic performance.
|
5 |
Airfoil Optimization for Unsteady Flows with Application to High-lift Noise ReductionRumpfkeil, Markus Peer 26 February 2009 (has links)
The use of steady-state aerodynamic optimization methods in the computational
fluid dynamic (CFD) community is fairly well
established. In particular, the use of adjoint methods has proven to be very
beneficial because their cost is independent of the number of design variables.
The application of numerical optimization to airframe-generated noise, however, has not received as much attention, but with the significant
quieting of modern engines, airframe noise now competes with engine noise.
Optimal control techniques for unsteady flows are needed in order to be able to reduce airframe-generated noise.
In this thesis, a general framework is formulated to calculate the gradient of a cost function in a nonlinear unsteady flow environment
via the discrete adjoint method. The unsteady optimization algorithm developed in this work
utilizes a Newton-Krylov approach since the gradient-based optimizer uses the quasi-Newton method BFGS, Newton's method is applied to the
nonlinear flow problem, GMRES is used to solve the resulting linear problem inexactly, and last but not least the linear adjoint problem
is solved using Bi-CGSTAB. The flow is governed by the unsteady two-dimensional
compressible Navier-Stokes equations in conjunction with a one-equation turbulence model, which are discretized using
structured grids and a finite difference approach. The effectiveness of the unsteady optimization algorithm is demonstrated
by applying it to several problems of interest including shocktubes,
pulses in converging-diverging nozzles, rotating cylinders, transonic buffeting, and an unsteady trailing-edge flow.
In order to address radiated far-field noise, an acoustic wave propagation program based on
the Ffowcs Williams and Hawkings (FW-H) formulation is implemented and validated. The general framework is then used
to derive the adjoint equations for a novel hybrid URANS/FW-H optimization algorithm
in order to be able to optimize the shape of airfoils based on their calculated far-field pressure fluctuations.
Validation and application results for this novel hybrid URANS/FW-H optimization algorithm show that it is possible
to optimize the shape of an airfoil in an unsteady flow environment to minimize
its radiated far-field noise while maintaining good aerodynamic performance.
|
6 |
Optimal Control Of Numerical Dissipation In Modified KFVS (m-KFVS) Using Discrete Adjoint MethodAnil, N 05 1900 (has links)
The kinetic schemes, also known as Boltzmann schemes are based on the moment-method-strategy, where an upwind scheme is first developed at the Boltzmann level and after taking suitable moments we arrive at an upwind scheme for the governing Euler or Navier-Stokes equations. The Kinetic Flux Vector Splitting (KFVS)scheme, which belongs to the family of kinetic schemes is being extensively used to compute inviscid as well as viscous flows around many complex configurations of practical interest over the past two decades. To resolve many flow features accurately, like suction peak, minimising the loss in stagnation pressure, shocks, slipstreams, triple points, vortex sheets, shock-shock interaction, mixing layers, flow separation in viscous flows require an accurate and low dissipative numerical scheme. The first order KFVS method even though is very robust suffers from the problem of having much more numerical diffusion than required, resulting in very badly smearing of the above features. However, numerical dissipation can be reduced considerably by using higher order kinetic schemes. But they require more points in the stencil and hence consume more computational time and memory. The second order schemes require flux or slope limiters in the neighbourhood of discontinuities to avoid spurious and physically meaningless wiggles or oscillations in pressure, temperature or density. The limiters generally restrict the residue fall in second order schemes while in first order schemes residue falls up to machine zero. Further, pressure and density contours or streamlines are much smoother for first order accurate schemes than second order accurate schemes. A question naturally arises about the possibility of constructing first order upwind schemes which retain almost all advantages mentioned above while at the same time crisply capture the flow features.
In the present work, an attempt has been made to address the above issues by developing yet another kinetic scheme, known as the low dissipative modified KFVS (m-KFVS) method based on modified CIR (MCIR) splitting with molecular velocity dependent dissipation control function. Different choices for the dissipation control function are presented. A detailed mathematical analysis and the underlying physical arguments behind these choices are presented. The expressions for the m-KFVS fluxes are derived. For one of the choices, the expressions for the split fluxes are similar to the usual first order KFVS method. The mathematical properties of 1D m-KFVS fluxes and the eigenvalues of the corresponding flux Jacobians are studied numerically. The analysis of numerical dissipation is carried out both at Boltzmann and Euler levels. The expression for stability criterion is derived. In order to be consistent with the interior scheme, modified solid wall and outer boundary conditions are derived by extending the MCIR idea to boundaries.
The cell-centred finite volume method based on m-KFVS is applied to several standard test cases for 1D, 2D and 3D inviscid flows. In the case of subsonic flows, the m-KFVS method produces much less numerical entropy compared to first order KFVS method and the results are comparable to second order accurate q-KFVS method. In transonic and supersonic flows, m-KFVS generates much less numerical dissipation compared to first order KFVS and even less compared to q-KFVS method. Further, the m-KFVS method captures the discontinuities more sharply with contours being smooth and near second order accuracy has been achieved in smooth regions, by still using first order stencil. Therefore, the numerical dissipation generated by m-KFVS is considerably reduced by suitably choosing the dissipation control variables. The Euler code based on m-KFVS method almost takes the same amount of computational time as that of KFVS method.
Although, the formal accuracy is of order one, the m-KFVS method resolves the flow features much more accurately compared to first order KFVS method but the numerical dissipation generated by m-KFVS method may not be minimal. Hence, the dissipation control vector is in general not optimal. If we can find the optimal dissipation control vector then we will be able to achieve the minimal dissipation. In the present work, the above objective is attained by posing the minimisation of numerical dissipation in m-KFVS method as an optimal control problem. Here, the control variables are the dissipation control vector. The discrete form of the cost function, which is to be minimised is considered as the sum of the squares of change in entropy at all cells in the computational domain. The number of control variables is equal to the total number of cells or finite volumes in the computational domain, as each cell has only one dissipation control variable.
In the present work, the minimum value of cost function is obtained by using gradient based optimisation method. The sensitivity gradients of the cost function with respect to the control variables are obtained using discrete adjoint approach. The discrete adjoint equations for the optimisation problem of minimising the numerical dissipation in m-KFVS method applied to 2D and 3D Euler equations are derived. The method of steepest descent is used to update the control variables. The automatic differentiation tool Tapenade has been used to ease the development of adjoint codes.
The m-KFVS code combined with discrete adjoint code is applied to several standard test cases for inviscid flows. The test cases considered are, low Mach number flows past NACA 0012 airfoil and two element Williams airfoil, transonic and supersonic flows past NACA 0012 airfoil and finally, transonic flow past Onera M6 wing. Numerical results have shown that the m-KFVS-adjoint method produces even less numerical dissipation compared to m-KFVS method and hence results in more accurate solution. The m-KFVS-adjoint code takes more computational time compared to m-KFVS code.
The present work demonstrates that it is possible to achieve near second order accuracy by formally first order accurate m-KFVS scheme while retaining advantages of first order accurate methods.
|
7 |
Adaptive Netzverfeinerung in der Formoptimierung mit der Methode der Diskreten AdjungiertenGünnel, Andreas 15 April 2010 (has links) (PDF)
Formoptimierung bezeichnet die Bestimmung der Geometrischen Gestalt eines Gebietes auf dem eine partielle Differentialgleichung (PDE) wirkt, sodass bestimmte gegebene Zielgrößen, welche von der Lösung der PDE abhängen, Extrema annehmen. Bei der Diskret Adjungierten Methode wird der Gradient einer Zielgröße bezüglich einer beliebigen Anzahl von Formparametern mit Hilfe der Lösung einer adjungierten Gleichung der diskretisierten PDE effizient ermittelt. Dieser Gradient wird dann in Verfahren der numerischen Optimierung verwendet um eine optimale Lösung zu suchen.
Da sowohl die Zielgröße als auch der Gradient für die diskretisierte PDE ermittelt werden, sind beide zunächst vom verwendeten Netz abhängig. Bei groben Netzen sind sogar Unstetigkeiten der diskreten Zielfunktion zu erwarten, wenn bei Änderungen der Formparameter sich das Netz unstetig ändert (z.B. Änderung Anzahl Knoten, Umschalten der Konnektivität). Mit zunehmender Feinheit der Netze verschwinden jedoch diese Unstetigkeiten aufgrund der Konvergenz der Diskretisierung.
Da im Zuge der Formoptimierung Zielgröße und Gradient für eine Vielzahl von Iterierten der Lösung bestimmt werden müssen, ist man bestrebt die Kosten einer einzelnen Auswertung möglichst gering zu halten, z.B. indem man mit nur moderat feinen oder adaptiv verfeinerten Netzen arbeitet.
Aufgabe dieser Diplomarbeit ist es zu untersuchen, ob mit gängigen Methoden adaptiv verfeinerte Netze hinreichende Genauigkeit der Auswertung von Zielgröße und Gradient erlauben und ob eventuell Anpassungen der Optimierungsstrategie an die adaptive Vernetzung notwendig sind. Für die Untersuchungen sind geeignete Modellprobleme aus der Festigkeitslehre zu wählen und zu untersuchen. / Shape optimization describes the determination of the geometric shape of a domain with a partial differential equation (PDE) with the purpose that a specific given performance function is minimized, its values depending on the solution of the PDE. The Discrete Adjoint Method can be used to evaluate the gradient of a performance function with respect to an arbitrary number of shape parameters by solving an adjoint equation of the discretized PDE. This gradient is used in the numerical optimization algorithm to search for the optimal solution.
As both function value and gradient are computed for the discretized PDE, they both fundamentally depend on the discretization. In using the coarse meshes, discontinuities in the discretized objective function can be expected if the changes in the shape parameters cause discontinuous changes in the mesh (e.g. change in the number of nodes, switching of connectivity). Due to the convergence of the discretization these discontinuities vanish with increasing fineness of the mesh.
In the course of shape optimization, function value and gradient require evaluation for a large number of iterations of the solution, therefore minimizing the costs of a single computation is desirable (e.g. using moderately or adaptively refined meshes).
Overall, the task of the diploma thesis is to investigate if adaptively refined meshes with established methods offer sufficient accuracy of the objective value and gradient, and if the optimization strategy requires readjustment to the adaptive mesh design. For the investigation, applicable model problems from the science of the strength of materials will be chosen and studied.
|
8 |
Contribution à une méthode de raffinement de maillage basée sur le vecteur adjoint pour le calcul de fonctions aérodynamiques / Contribution to a mesh refinement method based on the adjoint vector for the computation of aerodynamic outputsBourasseau, Sébastien 14 December 2015 (has links)
L’adaptation de maillage est un outil puissant pour l’obtention de simulations aérodynamiques précises à coût limité. Dans le cas particulier des simulations visant au calcul de fonctions aérodynamiques (efforts, moments, rendements...), plusieurs méthodes dites de raffinement ciblé (ou, en anglais, « goal-oriented ») basées sur le vecteur adjoint de la fonction d’intérêt ont été proposées. L’objectif de la thèse est l’extension d’une méthode de ce type basée sur la dérivée totale dJ/dX de la grandeur aérodynamique d’intérêt, J, par rapport aux coordonnées du maillage volumique, X. Les trois méthodes usuelles de calcul de gradient discret – la méthode de différentiation directe, la méthode adjointe-"paramètres" et la méthode adjointe-"maillage" évaluant dJ/dX – ont tout d’abord été étudiées et codées dans le logiciel elsA de l’ONERA pour des maillages non-structurés, pour des écoulements compressibles de fluide parfait et des écoulements laminaires. La seconde étape du travail a consisté à créer un senseur local θ basé sur dJ/dX qui identifie les zones du maillage volumique où la position des nœuds a une forte incidence sur l’évaluation de la fonction J. Ce senseur sert d’indicateur pour l’adaptation de différents maillages, pour différents régimes d’écoulement (subsonique, transsonique, supersonique), pour des configurations d’aérodynamique interne (aube et tuyère) et externe (profil d’aile). La méthode proposée est comparée à une méthode de raffinement ciblée très populaire (Venditti et Darmofal, 2001) et à une méthode de raffinement basée sur les caractéristiques de l’écoulement (ou, en anglais, « feature-based ») ; elle conduit à des résultats très satisfaisants. / Mesh adaptation is a powerful tool to obtain accurate aerodynamic simulations with limited cost. In the specific case of computation of aerodynamic functions (forces, moments, efficiency ...), goal-oriented methods based on the adjoint vector have been proposed. The aim of the thesis is the extension of a method of this type based on the total derivative dJ/dX of the aerodynamic output of interest, J, with respect to the volume mesh coordinates, X. The three common methods for calculating discrete gradient – the direct differentiation method, the parameter-adjoint method and mesh-adjoint method evaluating dJ/dX – have been studied first and coded in the elsA ONERA software for unstructured grids, for compressible inviscid and laminar flows. The second part of this work was has been to define a local sensor θ based on dJ/dX in order to identify zones where the volume mesh nodes position has a strong impact on the evaluation of the function J. This sensor is the selected indicator for different mesh adaptations for different flow regimes (subsonic, transonic, supersonic) for internal (blade and nozzle) and external (wing profile) aerodynamic configurations. The proposed method is compared to a well-known goal-oriented method (Darmofal and Venditti, 2001) and to a feature-based method ; it leads to very consistent results. very consistent results.
|
9 |
Modélisation et identification de paramètres pour les empreintes des faisceaux de haute énergie. / Modelling and parameter identification for energy beam footprintsBashtova, Kateryna 05 December 2016 (has links)
Le progrès technologique nécessite des techniques de plus en plus sophistiquées et précises de traitement de matériaux. Nous étudions le traitement de matériaux par faisceaux de haute énergie : un jet d’eau abrasif, une sonde ionique focalisée, un laser. L’évolution de la surface du matériau sous l’action du faisceau de haute énergie est modélisée par une EDP. Cette équation contient l’ensemble des coefficients inconnus - les paramètres de calibration de mo- dèle. Les paramètres inconnus peuvent être calibrés par minimisation de la fonction coût, c’est-à-dire, la fonction qui décrit la différence entre le résultat de la modélisation et les données expérimentales. Comme la surface modélisée est une solution du problème d’EDP, cela rentre dans le cadre de l’optimisation sous contrainte d’EDP. L’identification a été rendue bien posée par la régularisation du type Tikhonov. Le gradient de la fonction coût a été obtenu en utilisant les deux méthodes : l’approche adjointe et la différen- ciation automatique. Une fois la fonction coût et son gradient obtenus, nous avons utilisé un minimiseur L-BFGS pour réaliser la minimisation.Le problème de la non-unicité de la solution a été résolu pour le problème de traitement par le jet d’eau abrasif. Des effets secondaires ne sont pas inclus dans le modèle. Leur impact sur le procédé de calibration a été évité. Ensuite, le procédé de calibration a été validé pour les données synthétiques et expérimentales. Enfin, nous avons proposé un critère pour distinguer facilement entre le régime thermique et non- thermique d’ablation par laser. / The technological progress demands more and more sophisticated and precise techniques of the treatment of materials. We study the machining of the material with the high energy beams: the abrasive waterjet, the focused ion beam and the laser. Although the physics governing the energy beam interaction with material is very different for different application, we can use the same approach to the mathematical modeling of these processes.The evolution of the material surface under the energy beam impact is modeled by PDE equation. This equation contains a set of unknown parameters - the calibration parameters of the model. The unknown parameters can be identified by minimization of the cost function, i.e., function that describes the differ- ence between the result of modeling and the corresponding experimental data. As the modeled surface is a solution of the PDE problem, this minimization is an example of PDE-constrained optimization problem. The identification problem was regularized using Tikhonov regularization. The gradient of the cost function was obtained both by using the variational approach and by means of the automatic differentiation. Once the cost function and its gradient calculated, the minimization was performed using L-BFGS minimizer.For the abrasive waterjet application the problem of non-uniqueness of numerical solution is solved. The impact of the secondary effects non included into the model is avoided as well. The calibration procedure is validated on both synthetic and experimental data.For the laser application, we presented a simple criterion that allows to distinguish between the thermal and non-thermal laser ablation regimes.
|
10 |
Adaptive Netzverfeinerung in der Formoptimierung mit der Methode der Diskreten AdjungiertenGünnel, Andreas 22 January 2010 (has links)
Formoptimierung bezeichnet die Bestimmung der Geometrischen Gestalt eines Gebietes auf dem eine partielle Differentialgleichung (PDE) wirkt, sodass bestimmte gegebene Zielgrößen, welche von der Lösung der PDE abhängen, Extrema annehmen. Bei der Diskret Adjungierten Methode wird der Gradient einer Zielgröße bezüglich einer beliebigen Anzahl von Formparametern mit Hilfe der Lösung einer adjungierten Gleichung der diskretisierten PDE effizient ermittelt. Dieser Gradient wird dann in Verfahren der numerischen Optimierung verwendet um eine optimale Lösung zu suchen.
Da sowohl die Zielgröße als auch der Gradient für die diskretisierte PDE ermittelt werden, sind beide zunächst vom verwendeten Netz abhängig. Bei groben Netzen sind sogar Unstetigkeiten der diskreten Zielfunktion zu erwarten, wenn bei Änderungen der Formparameter sich das Netz unstetig ändert (z.B. Änderung Anzahl Knoten, Umschalten der Konnektivität). Mit zunehmender Feinheit der Netze verschwinden jedoch diese Unstetigkeiten aufgrund der Konvergenz der Diskretisierung.
Da im Zuge der Formoptimierung Zielgröße und Gradient für eine Vielzahl von Iterierten der Lösung bestimmt werden müssen, ist man bestrebt die Kosten einer einzelnen Auswertung möglichst gering zu halten, z.B. indem man mit nur moderat feinen oder adaptiv verfeinerten Netzen arbeitet.
Aufgabe dieser Diplomarbeit ist es zu untersuchen, ob mit gängigen Methoden adaptiv verfeinerte Netze hinreichende Genauigkeit der Auswertung von Zielgröße und Gradient erlauben und ob eventuell Anpassungen der Optimierungsstrategie an die adaptive Vernetzung notwendig sind. Für die Untersuchungen sind geeignete Modellprobleme aus der Festigkeitslehre zu wählen und zu untersuchen. / Shape optimization describes the determination of the geometric shape of a domain with a partial differential equation (PDE) with the purpose that a specific given performance function is minimized, its values depending on the solution of the PDE. The Discrete Adjoint Method can be used to evaluate the gradient of a performance function with respect to an arbitrary number of shape parameters by solving an adjoint equation of the discretized PDE. This gradient is used in the numerical optimization algorithm to search for the optimal solution.
As both function value and gradient are computed for the discretized PDE, they both fundamentally depend on the discretization. In using the coarse meshes, discontinuities in the discretized objective function can be expected if the changes in the shape parameters cause discontinuous changes in the mesh (e.g. change in the number of nodes, switching of connectivity). Due to the convergence of the discretization these discontinuities vanish with increasing fineness of the mesh.
In the course of shape optimization, function value and gradient require evaluation for a large number of iterations of the solution, therefore minimizing the costs of a single computation is desirable (e.g. using moderately or adaptively refined meshes).
Overall, the task of the diploma thesis is to investigate if adaptively refined meshes with established methods offer sufficient accuracy of the objective value and gradient, and if the optimization strategy requires readjustment to the adaptive mesh design. For the investigation, applicable model problems from the science of the strength of materials will be chosen and studied.
|
Page generated in 0.0545 seconds