Algorithmic Product Design

Graph Neural Network-Powered Finite Element Analysis of Semicrystalline Polymers

In finite element simulations (FEA) for semicrystalline polymers, reducing computational time and energy use is essential for both efficiency and sustainability. Traditional simulations, which rely on iterative optimization, are energy-intensive, contributing to higher carbon footprints and operational costs. As demand for high-fidelity models grows, so does the environmental impact of these simulations.

A Graph Neural Network (GNN)-based approach offers a sustainable solution by drastically reducing computation time. Once trained on simulation data, GNNs predict material parameters in seconds, eliminating resource-heavy iterative processes. This not only accelerates research and innovation but also cuts energy consumption, supporting lower carbon emissions and more sustainable digital manufacturing practices.

Enhancing Metal 3D Printing with Optimized Infill Patterns and Shrinkage Control

In metal 3D printing, the way material is structured internally plays a crucial role in the quality and accuracy of the final part. Our process uses metal paste deposition, which closely resembles the widely known fused filament fabrication (FFF) used for printing thermoplastics. As a result, many of the same infill patterns developed for plastics are often applied to metal printing. However, because metal components must undergo a sintering process, where the material shrinks as it solidifies, these patterns may behave differently, affecting the precision and integrity of the finished part.

To address this challenge, we combine advanced simulations with experimental testing to analyze how different infill patterns perform under the unique conditions of metal printing. By capturing 3D surface data at the bead level, we gain insights into the impact of these patterns on shrinkage and identify new designs better suited to metal paste deposition. The goal is to create optimized infill structures that reduce shrinkage, enhance dimensional accuracy, and improve the overall quality of printed parts.

Thermal Insights: Merging Simulation and Real-Time Data for 3D Printing Precision

For FDM temperature plays a critical role in the success of 3D printing, influencing the flow of material through the extruder and the adhesion between layers. To ensure optimal print quality, it is essential to accurately capture and understand the thermal behavior throughout the process.

Our approach integrates real-time data from thermal cameras with simulations, allowing for precise calibration of thermal behavior such as temperature evolution and cooling rates. This bead-level thermal analysis provides deep insights, helping us fine-tune parameters in real time to achieve consistent, high-quality results.

Infill Pattern Generation

Generative Design of Engineering Materials

Framework Overview of scalable AI-based Generative Design of Engineering Materials

AI-based generative design in the context of engineering materials is an innovative approach that harnesses the power of artificial intelligence to revolutionize how materials are developed, optimized, and utilized in various engineering applications.

Our approach seeks to harness the combined insights of additive manufacturing and machine learning, drawing extensively from accumulated expertise and comprehensive specifications of materials and existing product patterns. We aim to evolve these established patterns, utilizing them as a foundation for generating innovative designs that hold the potential for future scalability. Currently, we employ advanced machine learning techniques, specifically Convolutional Neural Networks (CNN) and Convolutional Variational Autoencoders (CVAE). These are instrumental in enhancing our pattern recognition capabilities through sophisticated interpolation methods derived from the encoded representations of the patterns. This integration of cutting-edge technologies not only improves the efficiency and effectiveness of our design process but also paves the way for groundbreaking advancements in product development.

Page updated

Report abuse