Recent advances in interactive physics simulation, originally developed for computer graphics, games, and virtual environments, have demonstrated that approximate solvers can capture gross system behavior in real time. These include rigid-body dynamics engines [1], particle-spring and constraintbased soft-body solvers[2], and specialized hybrid platforms. Out of such approaches (e.g., form-finding via dynamic relaxation or constraint-based simulation) have already proven useful for exploring large design spaces quickly for airbag volume simulation for engineering purpose [3].Automotive industry today uses game engines for realistic visualization [4] and training drive assistant algorithms in virtual environment [5] and synthetic data generated by them to train AI models for various applications, including self-driving technology [6]. Yet, the engineering community has not yet systematically evaluated these interactive physics methods for engineering simulations such as crash and impact scenarios. There is an opportunity to investigate whether game engine-based physics simulation can provide calibrated approximations for crash related metrics (force, acceleration, displacement) and bridge the gap between costly high-fidelity FEM and simplified analytical models.Example methodological categories of physics simulation for game development that can be pursuit in this thesis are: * Rigid-body dynamics approaches, where objects are assumed undeformable but connected via joints, constraints, and contact models. (Commonly used in the gaming industry, PhysX, Havok etc.)
Soft-body dynamics approaches, where objects are represented by particle-spring networks, constraint-based solvers, or simplified continuum models to capture deformation. (commonly used in video games to simulate clothes and hair, applicable within BeamNG, FLEXICX, etc)
Aim and PurposeThe aim of the thesis is to evaluate and demonstrate the applicability of game engine-based physics simulations for different crash scenarios in engineering. The purpose is not to replace FEM or physical testing, but to investigate whether these alternative methods can deliver sufficiently accurate and repeatable results for approximate studies, DOE, or surrogate model training.Expected Results
A structured comparison of rigid-body, soft-body, and specialized interactive physics approaches in the context of a generic crash simulation and occupant safety problems.
Implementation and testing of one or more representative case studies (e.g., impact of a rigid headform against a protective barrier or curtain).
Evaluation of accuracy, stability, and computational cost against reference data from FEM or physical testing.
Identification of strengths, limitations, and suitable application areas for these methods in engineering.
Guidelines for future work on how game engine physics simulations can complement established FEM and testing workflows.
Qualifications
Basic knowledge of mechanics, dynamics.
Familiarity with CAD modeling and handling of geometric data.
Strong programming skills in at least one language (e.g., Python, C#, C++, or MATLAB).
Previous exposure to simulation methods such as FEM, MBD, or multibody software., parametric modeling tools (e.g., Grasshopper/Rhino) or any scripting experience within 3D environments, game engines (e.g., Unity, Unreal, CryEngine) or physics middleware (PhysX, Bullet, Havok).
