AI/ML: Real-Time Geometry-to-Field Prediction with Geometric Deep Learning

24 March 2025, Online (webex)

Geometric Deep Learning is the current state-of-the-art for training AI models on simulation data. It enables Neural Networks to understand geometric and spatial information and directly work with file formats like VTK.

This course will teach participants how to create Deep Learning models from multiple simulation files and to make instantaneous simulation predictions for new geometries. Research examples are:

• Prediction of airfoil pressure distribution
• Multi-Objective Fluid-Acoustic Shape Optimization
• Fast predictions of stress distribution

Learning Objectives

• Understand the core principles of geometric deep learning (manifolds, graphs, mesh representations).
• Formulate CFD and FEA problems as graph-structured data.
• Build and train a simple graph neural network (GNN) for surrogate modeling or mesh refinement.
• Evaluate GNN performance on machine learning and simulation benchmarks.

Hands-On Exercise
Participants will implement and train a GNN surrogate model for a simple flow and a simple stress-analysis problem, then compare accuracy and runtime against a baseline.

Course Contents

• A high-level overview of Geometric Deep Learning

• Introduction to geometric deep learning and graph neural networks
• How geometric deep learning works with simulation data
• Example Application for FEA (Predicting Brake Disc Temperature from geometry)
• Example Application for CFD (Predicting turbulent kinetic energy from geometry)
• Overview of possible applications for engineering simulation

• Introducing the mathematical graph

• The core concepts of the mathematical graph

• How graphs appear in nature
• Nodes, node features, and labels
• Edges, edge features and topology
• Positional/structural encodings
• Global graph features
• Graph variants (heterogenous, temporal, structural)
• Weight and directionality
• How graphs describe natural phenomena (Example graphs)

• From simple graphs to describing simulation data

• Grid graphs
• Arbitrary graphs
• Describing complex geometry with graphs
• Representing simulation data as a graph
• Encoding boundary conditions
• Graph neural networks and geometric deep learning

• Comparing graph neural networks to conventional neural network types
• GNN layer properties

• Computational efficiency
• Size independence
• Localization
• Specifying importances
• Inductive generalization

• GNN layer types
• Message passing
• Learning outcomes (Tasks)

• Node prediction
• Edge prediction
• Graph-level prediction

• Training and validation

• Training metrics; Avoiding leakage
• Physics-aware metrics
• Physics-based loss terms (PINN)
• Applying GNNs to simulation data (practical example)

• Converting simulation data to a graph (Feature scaling, nondimensionalization)
• Setting up a GNN
• The training/validation loop
• Testing model validity

T​rainer

M​ax Kassera (yasAI)
Max Kassera studied mechanical engineering with a minor in economics at the University of Kaiserslautern-Landau, where he first applied machine learning and artificial intelligence to turbocharger design in 2017. After graduating, he was awarded two German government grants to develop AI software for mechanical engineering, which led to the incorporation of yasAI in 2022. With yasAI, Max began training engineers in applying AI to simulation projects with a focus on simulations and fluid mechanics.

Requierements

• Familiarity with FEA/CFD and meshes
• Prior machine learning knowledge is not required but beneficial
• Programming experience is not required

Duration & Format

Organisation

Duration
9​:00 am - 5:00 pm (MEZ, UTC+1, Berlin)
Login phase from 30 min before course starts.

Language
English

Course Fee
Non NAFEMS members: 990 Euro / person*
NAFEMS member: 700 Euro / person*
Included in the fees are digital course notes and a certificate.
* plus VAT if applicable.
Please note - unpaid registrations will be cancelled one week prior to the event start date unless previous contact has been made with our Accounts Department or the course organiser. If not, then our cancellation policy will be enforced. If required, the cancellation policy can be viewed on the event page on our website.

NAFEMS membership fees (company)
A standard NAFEMS site membership costs 1,365 Euro per year, an academic site and entry membership costs 855 Euro per year.

Cancellation Policy

Course cancellation by NAFEMS
If not enough participants we keep the right to cancel the course one week before. The course can be canceled also in case of disease of the speakers or force majeure. In these cases the course fees will be returned.

Organisation / Contact
NAFEMS
e-mail: roger.oswald@nafems.org

Accreditation Policy

The course is agreed and under control of NAFEMS Education and Training Working Group (ETWG).