The present thesis investigates into the topic of local heat treatments of aluminum alloys. In the following this means, that the annealing step is done only partially. Therefore, the non-homogeneous temperature distribution induces thermal stresses which have to be taken into account, if yielding of the material has to be avoided. For the numerical modeling of the whole process, all single steps are combined into a process chain framework. In addition, an optimizer is included in order to fully automatize the chain and enable the possibility of optimization. The structure of the thesis is as follows:
In the beginning, fundamental literature research of the three most important topics — thermal stress analysis, aluminum alloys, and optimization strategies — is given. Subsequent, an introduction to the theoretical topics of the thermal-elastic stress calculation and energy
equation is given while arguing that the finite volume method (FVM) is compatible to the well-known finite element method (FEM). During that, the coupled stress-energy system is discussed, and three possibilities of numerical implementations are given. Besides, numerical stabilization aspects are argued. After that, the most critical numerical boundary conditions are briefly described. Special ones are derived elaborately. To check the correct derivations and implementation of the numerical model, three different validation cases are analyzed and debated.
Furthermore, two discriminate residual calculations will be discussed, and a clear statement of the relaxation methods is presented.
Afterward, the whole modeling approach for the material is discussed profoundly while all input parameters and numerical models — that are used in the modeling approach — are given. Based on a clear statement of the problem which occur during local heat treatment, the single modeling steps are delineated. Thus, the Scheil-Gulliver approach is explained whereat the estimation of the radius distribution of the phases, which occur after solidification, is described.
Subsequently, a basic description of the kinetic modeling approach is given while all numerical models, inclusive all phase and matrix properties, are announced by using a wide range of literature resources. To complete the theoretical part, three different approaches for the estimation of the yield strength in 3D are discussed, and the advantages/disadvantages are pointed out.
Furthermore, a class of different kind of optimization algorithms is discussed and presented. To get familiar with some methods, simple object functions are used to describe and demonstrate the procedure of these algorithms. During the discussion, the advantages and
disadvantages of unique algorithms are indicated. Subsequently, the three different topics are combined in a newly developed process chain framework while various examples are examined intensively. The first investigation is related to the local heat treatment without material calculation while the challenge of the non-uniform heat-up is discussed. Afterward, two different cooling methods are investigated namely self-quenching and water-quenching while the thermal stresses will be analyzed and presented. Thereafter, the material calculation is included to predict the yield strength of the material after the local heat treatment and subsequent artificial aging in 3D by using the new developed Random-Cell-Poisson method. In the end, an optimization procedure is given to find the best process parameters for different local heat treatments, cooling methods, and artificial aging parameters. For that purpose, a genetic algorithm is used to detect the global maximum in the parameter space. Furthermore, the flexibility of the developed framework is presented by investigating a steel welding example.
In the end, the new scientific findings are summarized, and an outlook is given.