お問合せ
Unlock the Future of Vehicle Crash Optimization with Reduced Order Modeling (ROM)
Advanced Reduced Order Modeling for More Effective Structural Crash Optimization
Are you an engineer looking to revolutionize vehicle design and crash simulation? Discover how Reduced Order Modeling (ROM), specifically through ESI's advanced ADMORE technology, can transform your workflow. During our ESI LIVE 2023 virtual conference, my colleague Gavin England, PMM crash and safety here at ESI Group and I explored innovative methods that enable faster, cost-effective, and more accurate optimization of vehicle structures. This wasn't just theory—it is a real-world application with Renault, showcasing the power of ROM in actual crash simulation scenarios.
Our partnership with Renault utilized ADMORE's cutting-edge sPGD and ReCUR methods to optimize vehicle-side member reinforcement, dramatically cutting down simulation time and costs. These methods not only streamline the process but also open up new avenues for non-specialist engineers to participate in high-level simulations, democratizing the design process.
But let’s start with clarifying some fundamental questions before we move forward.
What is the Reduced Order Modeling Approach?
Reduced Order Modeling is a mathematical technique that simplifies complex simulations by reducing the number of variables and equations needed to describe a system. Instead of solving full-scale, high-fidelity models, ROM uses a smaller set of basis functions to approximate the system's behavior. This reduction in complexity makes it possible to perform simulations more quickly and with less computational power, without significantly compromising accuracy.
What is the Role of Reduced Order Models in Product Development?
Reduced Order Models (ROMs) play a crucial role in product development and the Design of experiments (DoE) by enabling rapid prototyping and iterative design processes. They allow engineers to quickly assess the impact of design changes, material selections, and other variables on the final product's performance. This capability is particularly valuable in industries like automotive engineering, where safety standards are stringent, and time-to-market is critical.
What is the Value of Using ROM in Engineering?
The primary advantage of using ROM is its ability to significantly reduce computational costs and time. Traditional high-fidelity models require extensive computational resources, often running for many hours on high-performance processors. Reduced Order Models can achieve similar levels of accuracy with much less computational effort, making it an invaluable tool in scenarios where multiple design iterations are necessary, such as crash simulations and vehicle design optimization.
Is Proper Orthogonal Decomposition the Only Method for Reduced Order Models?
No, Proper Orthogonal Decomposition is not the only ROM method. In their presentation, Gavin and Fatima highlighted two other main methods:
- sPGD (Sparse Proper Generalized Decomposition) | This method treats parameters as additional coordinates and creates a reduced representation using a combination of functions that depend separately on space, time, and parameters. It's particularly effective in capturing the system's behavior across a wide range of conditions.
- ReCUR (Regression CUR) |This method involves decomposing a matrix into selected rows and columns, along with a smaller matrix that links these components. ReCUR uses regression techniques, including machine learning methods like random forests, to predict the system's behavior for unseen parameter combinations.
How do Reduced Order Modeling, Machine Learning, and Hybrid Twin Relate to one Another?
Model Order Reduction, Machine Learning, Artificial Intelligence (AI), and the Hybrid Twin approach are interconnected concepts in the field of advanced computational modeling and simulation.
ROM provides a simplified yet accurate representation of complex system models, which is computationally efficient. Machine learning and hybrid AI enhance Reduced Order Models by enabling rapid physical insights, model predictions and early decision-making. Hybrid Twins leverage these technologies to create dynamic digital replicas of physical systems, allowing for real-time monitoring, optimization, and predictive analysis. Together, these technologies enable more effective and efficient system design, analysis, and operation.
What Are Practical Examples of Reduced Order Models?
There are plenty of use cases across industries. One very particular application is in the field of vehicle crash optimization, where Reduced Order Models can be applied to simulate the structural integrity and safety performance of various vehicle components. For instance, our partnership with Renault involved using ROM to optimize the reinforcement of vehicle-side members. By varying the thickness of different structural zones, we could efficiently explore the design space and identify optimal configurations that met safety and performance targets.
Why Download Our Technical Paper on “Advanced reduced order modelling for more effective structural crash optimization”?
Are you ready to dive deeper into the world of Reduced Order Modeling and its applications in crash optimization? We have outlined the full project in a technical paper for you to understand the scope and value and make sense of project results. Here are the 3 reasons why you should download this technical paper:
- In-depth Technical Insights: Understand the mechanics behind reduced-order modeling and its applications in vehicle crash optimization.
- Real-world Application: See how we partnered with Renault to tackle real challenges in crash simulation.
- Cost and Time Efficiency: Learn how to achieve optimal safety and performance with fewer resources.
Download the full technical paper now and discover how you can leverage ADMORE and ROM to optimize your vehicle designs efficiently and cost-effectively. Join us in pushing the boundaries of what's possible in automotive engineering!
Author
Dr. Fatima Daim
Team Leader, Computational Science at ESI Group
Dr. Fatima DAIM joined ESI Group in 2011 where she is currently a Team Leader in Research and Innovation Department. With her experience in applied mathematics and advanced simulation, she has contributed to the development of model order reduction technologies as well as in the introduction of techniques in data science and artificial intelligence. She is involved in number of academic and customer projects. Her work aims to demonstrate how reduced order modeling and machine learning can bring value to crash and manufacturing simulation.
Category: Automotive, Digital Transformation