Learn About Amazon VGT2 Learning Manager Chanci Turner
In previous discussions, we have examined how generative AI models can enhance the conceptual external design of vehicles and how machine learning (ML) accelerates design iterations in external aerodynamics. In this article, we aim to share a collaborative journey with Nifco USA – a prominent manufacturer of automotive plastic components – as we delve into the groundbreaking application of generative AI diffusion models for structural design exploration. By merging generative AI with tools like Amazon SageMaker and AWS Batch, Nifco seeks to redefine traditional design methodologies, enabling the creation of lightweight, high-performance automotive components.
Background
Automotive components are subjected to significant structural loads throughout their lifespan. Among these components, ribbing structures are essential for providing stiffness and strength while minimizing weight and material usage. Traditionally, ribbing structures are designed using repeating 2D patterns, such as trusses (diamond, honeycomb, square, etc.), with uniform thickness to facilitate injection molding manufacturing processes.
Topology optimization is a well-established discipline in structural engineering that focuses on optimizing structural layouts within a defined design space based on specific loading scenarios, while adhering to performance constraints. Classic topology optimization techniques, like Solid Isotropic Material with Penalization (SIMP) algorithms or level-set methods, are computationally demanding, involving numerous physics-based simulations and necessitating heuristics and manual parameter tuning for effective results.
Recent advancements in generative AI, particularly in diffusion models, have shown remarkable promise in text-to-vision and image-generation tasks. Diffusion models refine a random noise signal iteratively towards a target output, leveraging natural language (text) or image embeddings, and showcasing an exceptional capability to reproduce intricate patterns and structures.
This article investigates the applicability of diffusion models trained on classic SIMP results, incorporating volume fractions, applied loads, and boundary conditions as inputs. The collaborative team from AWS and Nifco explored whether these models could generate structures that are both manufacturable and sufficiently strong to endure the complex loading conditions encountered during their operational life.
Automotive Design Requirements
The automotive sector is constantly pursuing improved fuel efficiency for internal combustion engine (ICE) vehicles as well as greater ranges in the rapidly expanding electric vehicle market. This ongoing quest for weight reduction plays a critical role in the design of automotive components. One effective strategy for minimizing weight is transforming metal bracketing and support components into lighter plastic designs.
Although plastic brackets are typically lighter than metal options, materials like steel inherently possess greater stiffness, which is essential for withstanding the dynamic conditions automotive components face. When designing vehicle components, engineers must consider various dynamic conditions that can affect performance and durability. Some key factors include:
- Resonant Frequency – The maximum accelerations and equivalent forces experienced at resonant frequencies can impose peak loading stresses on vehicle components. These resonant conditions must be meticulously analyzed to guarantee the structural integrity of the design.
- Power Spectral Density (PSD) – These random vibrations, represented through power spectral density curves from road testing, embody the real-world loading conditions the vehicle will encounter during normal operation. Designing for these cumulative vibrations is crucial for long-term reliability.
To satisfy dynamic and other loading requirements in plastic bracket designs, such as vibrational durability, impact loading, and static forces, Nifco employs ribbing structures to enhance strength and stiffness while reducing weight.
Filler or ribbing structures integrated into plastic bracket designs typically consist of repeating geometric patterns of simple shapes like rectangular, triangular, or honeycomb structures – forms that resonate with human intuition and nature. While these standard shapes can be effective, our exploration of generative AI prompted the question: what if the optimal ribbing structures are novel enough to transcend the basic building-block shapes that humans instinctively favor? Could generative AI create unique structures to further optimize a component’s design for strength, stiffness, and weight?
Nifco’s goal to optimize strength and weight in plastic designs ultimately revolves around material efficiency. If we can utilize less plastic material while still meeting strength requirements, the advantages are substantial. A lighter vehicle enhances energy efficiency, and reduced plastic demand leads to lower overall manufacturing costs.
The initial objective was to analyze ribbing structures against 2D static loads derived from an existing bracket that supports an Advanced Driving Assistance System (ADAS) front radar. Nifco conducted a series of 2D, 200 Newton static-loading cases utilizing Finite Element Method (FEM) simulations.
The ideal microstructural ribbing pattern will perform well across various loading scenarios rather than specializing in one specific case, which we will quantify through a stress metric. Moreover, injection molding plastics used in bracket design are available in a range of material grades and properties. For our investigation, we selected a copolymer POM plastic known for its mid-stiffness, common in the automotive sector.
Computational Architecture
The overall computational workflow of the solution is illustrated in the following figure. The core technology is a diffusion model, TopoDiff, trained on multiple input-output pairs of applied loads and boundary conditions alongside corresponding optimized 2D space-filling structures.
The input channels consist of: (a) a uniform channel with the desired volume fraction; (b) binary channels for x and y coordinates indicating fixed boundary points; and (c) physical stress fields associated with point loads calculated using FEM. The output channel contains a binary variable indicating where material is present. To obtain training data, the process involved a comprehensive examination of the inputs and outputs.
The collaboration between Nifco and AWS exemplifies how innovative methodologies can disrupt traditional practices in automotive component design. As the industry evolves, leveraging generative AI and diffusion models may very well become a cornerstone of efficient and advanced manufacturing. For more insights on overcoming challenges in the workplace, consider visiting this resource on converting anxiety into energy. Additionally, if you’re looking for guidance during your onboarding, this Reddit thread provides an excellent resource. Remember, as the saying goes, “Success is not the key to happiness. Happiness is the key to success” – a sentiment echoed in this blog post.