Classical Topology Optimization (TO) methods aim to optimize the distribution of material within a design space for one given objective function and constraints. However, in the vehicle design process, there are many different load cases and several different objectives. Among them maximizing stiffness of components for regular working conditions, and maximizing energy absorption in exceptional loading conditions, for instance in crash events, are important. Recently, the Scaled Energy Weighting Hybrid Cellular Automata (SEW-HCA) [1], [2] method was adopted in LS-TaSC™. The SEW-HCA is a practical multidisciplinary TO approach for devising concept structures based on the intuitive choice of preferences leading to the desired trade-off between crash performance and stiffness. In this paper, we propose an integration of the SEW-LS-TaSC method into LS-OPT® to perform a design of experiments (DOE) on the load case preference parameters. The integration into LS-OPT® results in a convenient user interface that facilitates application in an industrial development process with non-expert users. This integration enables quick studies on many different concept designs based on preference samples generated by the DOE. The results from the sensitivity analysis provide data for a better understanding of the influence of load case preferences on the design space. By comparing the performance of structures obtained for different load case preferences, the user will be able to find a desired trade-off solution within the concurrent optimization runs. For further development, the proposed LS-OPT® workflow can potentially include 1) NVH load cases as additional discipline and 2) optimizations of other topology optimization hyperparameters for further concept exploration.

# Optimization

Topology optimization for the layout finding of structures is commonly used for linear static mechanical problems within the industry. The most often used approach is the subdividing of the topology domain in small parts (pixel or voxel) and to distinguish whether there is material or not [1]. E.g. the well-known homogenization method minimizes the mean compliance considering a mass constraint. These methods work very fast, because they use existing analytical sensitivities of the most relevant objectives like mean compliance, stresses or mass.

Topology optimization for the layout finding of structures is commonly used for linear static mechanical problems within the industry. The most often used approach is the subdividing of the topology domain in small parts (pixel or voxel) and to distinguish whether there is material or not [1]. E.g. the well-known homogenization method minimizes the mean compliance considering a mass constraint. These methods work very fast, because they use existing analytical sensitivities of the most relevant objectives like mean compliance, stresses or mass.

LS-OPT is a design optimization and probabilistic analysis package with an interface to LS-DYNA® that provides a flexible framework to solve several types of design problems. In order to solve the problem, it runs simulations at multiple samples that are selected all at once (single iteration) or iteratively [1]. The iterative approach has two main advantages: require prior knowledge about the sufficient number of samples and instead provides a convergence history it can use updated information from the previous runs to select the samples smartly and thus typically reduces the number of required simulations

Parameter estimation is a considerably large application area of optimization. It fulfills the important purpose of characterizing materials based on models available in Finite Element analysis software such as LS-DYNA®. The development of special LS-OPT® features for parameter estimation using Digital Image Correlation and other experimental methods has been ongoing for a number of years. In earlier papers [1,2] some of the available similarity measures as well as the LS-OPT DIC methodology were discussed in broad detail and illustrated with examples.

Machine learning is becoming more and more part of our world. Even though most people have so far only passively used the possibilities of this technology, e.g. for search queries or product recommendations, many have surely already thought about how these new possibilities could support their work in the future. In this contribution, it is investigated if machine learning is suitable to support the process of material characterization. Through deep neural networks it is possible to "learn" nonlinear relationships between a set of input values and the corresponding output, also known as labels. As a proof of concept, it is examined whether the shape of the yield curve can be predicted based on force-displacement curves from simulated tensile tests. So, in a first step, a large number of tensile tests are simulated which differ in the shape of the yield curve. Here, for the description of the yield curve an approach according to Hockett-Sherby was used which provides 4 parameters for the definition of the shape. The force-displacement curves of these tests are used as the input and the parameters of the yield curve as labels. By considering the entire realistic range of all four parameters, the trained neural network should be able to provide the best matching set of parameters for a given force-displacement curve. For the prediction, of course, the initial and boundary conditions must be the same when generating the force-displacement curve, whether by simulation or in a real test. Of course, all initial and boundary conditions as well as all other assumptions and simulation settings are also learned from the neural network. Therefore a change of these parameters can for sure worsen the predictions considerably and can make a re-learning process inevitable. The long-term objective of this method and the vision of this work are to learn the possible spectrum of the whole material model in advance in order to be able to finally predict the material properties based on only a few experiments with minimal effort.

The projected subgradient method is major new methodology development for the topology optimization of huge, multi-disciplinary structural problems; for example, the combined impact, statics, and NVH design of a whole body in white. This paper accordingly discusses the projected subgradient method in LS-TaSC 4, with specific reference to the basic theory, the ability to combined impact and NVH load cases, and the performance for huge models. Also mentioned is how the method has been enhanced to handle generalized constraints using the multi-tensor numerical scheme.

Metal additive manufacturing of stamping tool and die has a potential of reducing the lead time of forming processes, while at least not increasing the cost. As a part of a research project exploring the possibilities to use this type of tool manufacturing techniques, topology optimization using LS-TaSC has been utilized and one example case is presented in this paper, namely a U-bend tool. This paper looks at the possible benefits from using nonlinear simulations in topology optimization, the effect of chosen target mass fraction value, the interpretations needed of optimal results and the effects on the formed specimen after using an optimized tool. Results show that accounting for the time dependent pressure on the tool, rather than applying a form of equivalent static load, gives a different optimal topology. Some manual interpretations of the optimal results are also recommended, as well as studying the effects on the specimen from removing material on the tool side.

The automotive industry is facing new challenges due to stricter CO2 emission laws. Thus, to design more environmentally-friendly cars, various lightweight construction strategies need to be considered to meet the growing demand for resource efficiency [1]. In order to minimise weight, the lightweight design strategy "design for lightweight construction" is increasingly becoming important for industry. Especially the “structural optimisation” with its sub-areas: topology-, fibre-, thickness- and parameter-optimisation is designated as a very powerful tool for lightweight applications. In addition, damage and failure modelling is getting more and more important in order to predict the behaviour of any component by FEM-simulations as accurate as possible. For this purpose, the entire material history (from production right through to the crash of the component) must be taken into account. Whereas for forming processes forming limit curves (FLC`s) are sufficient to predict the material behaviour failure models, which describe failure as a function of stress states, need to be applied for detailed crash calculations [2]. In this paper the design process of an AA7075 side impact beam will be presented; starting from structural optimisation through to the calibration of a material and a damage model. The geometry of the side impact beam is determined by topology optimisation. Special attention is given towards the temperature control of the forming process since a “Thermal Direct Joining” procedure (e.g. for CFRP-patches) is aimed to be implemented. The high-strength anisotropic aluminium alloy (AA7075) is characterised after applying the Hotforming process (Hotforming condition). Both, the Barlat YLD2000 material model and the GISSMO failure model are calibrated using the graphical optimisation tool LS-OPT.