Die casting Unlocking the potential of mega-casting with AI-supported design

Large-scale casting has arrived in the automotive industry. However, the complex design requirements are pushing traditional development methods to their limits. Altair has now developed a solution for the structural design of mega-cast parts.

In the course of the transformation of the automotive industry, not only vehicle technologies but also manufacturing methods have evolved. An example of this progress are megacasting structures. Large casting components replace many sheet metal bending parts, which are typically used for body-in-whites. They can lower manufacturing costs and, thanks to additional design freedom and advantageous material properties, unlock new lightweight construction potential – especially in e-mobility.

Proven methods combined with machine learning

In practice, the implementation of large cast parts requires special machines and high precision to maintain manufacturing tolerances and ensure material quality and crash safety. With these complex design requirements for megacasting structures, traditional development methods reach their limits. Altair has developed a solution for the structural design of large cast parts using a generative multidisciplinary method. It combines proven methods such as topology optimization and response surface-based methods (Response Surface Method/RSM) with machine learning.

According to Altair, this combination makes it possible to derive optimal load paths for various loading conditions, taking into account different requirements such as casting conditions, as well as noise, vibration, and harshness (NVH), and crash safety. The connection of advanced CAE methods with innovative casting processes not only allows for optimal shaping and structural analysis but also revolutionizes the efficiency and quality of the manufactured components.

A holistic optimization process for AI-supported generative development

In the first phase of generative design, according to the announcement, the objective is to determine the optimal load paths for a cast construction, dividing the available space into shell structure and topology areas. The free-size optimization focuses on varying the different thicknesses of the shell elements. In topology optimization, on the other hand, the focus is on the efficient placement of the material.

With over 6 million elements, tools like Altair OptiStruct must be able to process a large number of design variables as well as consider various load cases from areas such as statics, driving dynamics, NVH (Noise, Vibration, Harshness), and crash.

The challenge is to determine a load path that considers diverse requirements such as crash loads. According to the announcement, crash tests, which include large deformations and contacts, play a decisive role and require special optimization methods due to the nonlinearities.

Gallery

To integrate crash and NVH load cases into the multi-model optimization problem, pragmatic linearization techniques are necessary, explains Altair. For the rear-impact load case, a linearization is applied that simulates deformation modes of the dynamic case. Similarly, the pole side impact is linearized by transferring measured contact forces to linear auxiliary load cases. From a combination of topology and free-form optimization, various design proposals then emerge, adapted to the requirements of different load cases such as NVH, crash, and driving dynamics.

Detecting desired component behavior with AI

According to the announcement, the RSM-based optimization approach in the second project phase is particularly valuable in crash optimization through mathematical models, which are used instead of finite element solvers and are intended to help represent complex crash requirements. To accelerate the process, an assessment is made by clustering and classifying the results. A surrogate model-based optimization process uses prediction models that are trained on a subset of DOE (Design of Experiments) points.

The casting process itself can also be optimized based on RSM (Response Surface Methodology), considering the design of the gating system. The number and configuration of gates influence the filling time and the material behavior during the casting process. The optimal configuration minimizes material waste while meeting structural requirements.

Furthermore, the gating design is adapted to optimize the filling processes in terms of structural and casting requirements. Rib direction and thickness affect the filling, while RSM and clustering optimize thickness distributions for crash requirements. The casting simulation is linked to these design variables, leading to a weight-optimal design according to Altair. The result of the two-phase design approach surpasses the classical design in most quantified requirements. Particularly in terms of weight, the superiority of the new design is evident with a reduction of 25 kilograms compared to the reference design.

The presented process enables the development of cast parts that meet various structural requirements of the automotive industry. The multidisciplinary optimization connects structural requirements, castability, and crash performance and offers efficiency advantages in terms of weight and crash safety compared to classical methods. Moreover, the approach allows for more efficient development, as the evaluation of component behavior becomes accessible to a broader user base. This, according to Altair, opens up immense design freedom in megacasting manufacturing. The result is high accuracy and valuable data analyses, without additional cost. (jup)