These slides were presented at the NAFEMS World Congress 2025, held in Salzburg, Austria from May 19–22, 2025.

Abstract

The growth of computer-aided engineering (CAE) tools and simulation data offers new opportunities for engineering teams to develop new products faster. However, challenges persist due to fragmented workflows, siloed simulation data management, and inefficient manual metadata processes. Engineers often spend up to 30% of their time managing data instead of making critical design decisions, impacting time-to-market, operational costs, and the ability to deliver optimal solutions. Despite its many advantages, the use of CAE in the product development process presents several challenges. One major hurdle is the high computational power and specialized software required for advanced simulations. Additionally, accurately modeling real-world conditions in a virtual environment is complex and may not always capture all the nuances of physical behavior, leading to discrepancies between simulation results and real-world performance. CAE also relies heavily on accurate material data and boundary conditions; any errors or assumptions in these inputs can lead to inaccurate predictions. Lastly, while CAE speeds up development by reducing the need for physical prototypes, it may still require validation through physical testing, which can introduce delays and costs. These challenges highlight the need for continuous improvement in CAE technologies and the expertise of engineers to fully harness its potential. Artificial intelligence (AI) is emerging as a transformative tool in addressing the challenges encountered in running CAE simulations for various applications. AI-powered surrogate models can efficiently approximate complex CAE simulations, enabling rapid evaluations of design alternatives and scenario analyses. Moreover, AI techniques, such as deep learning, facilitate data-driven approaches for enhancing simulation accuracy, predicting aerodynamic behaviors, and identifying critical flow phenomena. Additionally, AI-driven optimization algorithms can efficiently search vast design spaces to identify optimal configurations and performance-enhancing parameters for defense systems. By leveraging AI technologies, researchers and engineers can overcome computational bottlenecks, expedite design iterations, and unlock new insights into aerodynamic phenomena, thereby advancing the development of next-generation products with improved performance, efficiency, and mission effectiveness. Graph neural networks (GNNs) architectures have emerged as a powerful approach for handling complex data structures and are finding increasing utility in CAE simulations. By representing the computational domain as a graph, with nodes corresponding to grid points and edges denoting connections, GNNs can capture intricate relationships and dependencies inherent in complex physical phenomena. This representation allows GNNs to effectively model spatial interactions, boundary conditions, and turbulence effects, leading to more accurate and efficient simulations. GNNs offer the ability to learn from large-scale datasets, enabling data-driven approaches for optimizing simulation parameters, predicting flow behavior, and identifying critical flow features. In this session, we will talk about how a GNN architecture is deployed for modeling the complex behavior of bipolar plates of PEM (Proton Exchange Membrane) fuel cells. Modeling fuel cells involves complex FEA and CFD methods. Geometry preparation for the FEA process is human intensive and solving the FEA simulation takes a minimum of 48 hours on 100s of CPUs. The deformed geometry from the FEA simulation is processed into a CFD model for the flow prediction. Using a surrogate model approach we will demonstrate how we can predict the structural deformation of the geometry starting from the CAD model. This presentation explores a framework to streamline multidisciplinary simulation workflows by integrating digital thread concepts and AI-driven methodologies. By unifying historical modeling and simulation data, automating metadata capture, and leveraging AI for optimization, this approach significantly enhances collaboration, decision-making, and productivity. The framework addresses key engineering challenges by introducing centralized data structures, automated processes, and AI-assisted modeling. A digital thread captures the complete lifecycle of simulation workflows, providing traceable and actionable insights across teams and disciplines. This ensures that data is not only accessible but also actionable, enabling engineers and decision-makers to make informed choices that accelerate development and improve product outcomes. Key topics of the presentation include: Engineering Problem: The complexities of managing multidisciplinary CAE workflows due to disconnected tools, fragmented datasets, and manual processes. These challenges are particularly critical in industries such as automotive and aerospace, where thermal, structural, and aerodynamic simulations must converge seamlessly for optimal product development. Methods: A detailed explanation of how a centralized simulation data management system, such as a hierarchical framework (project > folder > study > job), can organize and share simulation files and metadata effectively. Coupled with AI models, these systems enable predictive insights, simulation acceleration, and iterative design optimization. Results: Case studies from automotive and aerospace industries (such as General Motors Motorsports, Boom Supersonic, Denso Manufacturing, Hankook Tire, and more) will demonstrate the impact of this approach. Examples include reducing simulation runtimes by up to 50%, improving simulation throughput, and enhancing collaboration across global engineering teams. These results underscore the potential of AI and digital thread technology to address bottlenecks, improve resource utilization, and foster innovation. Business Outcomes: The implementation of these methodologies has led to business impact, such as faster product development, reduced operational costs, enhanced sustainability, and higher-quality designs. By enabling real-time collaboration and decision-making, these solutions empower teams to meet customer demands more effectively. The session will feature modern modeling and simulation applications, including a demonstration of AI-enhanced workflows and digital thread capabilities. Attendees will learn how these tools can be leveraged to modernize their simulation processes, enabling faster innovation cycles and delivering superior engineering solutions. Through this presentation, participants will gain actionable insights into modernizing CAE workflows and integrating advanced technologies to remain competitive in a rapidly evolving engineering landscape. This approach bridges the gap between traditional engineering methodologies and future-forward practices, equipping teams with the tools needed to navigate increasingly complex design challenges with confidence.