In a traffic accident, it is not only the impact itself that determines possible injuries. The crash is often preceded by a highly dynamic situation: the vehicle brakes abruptly, occupants try to brace themselves or react reflexively to the acceleration. These movements change the posture and speed of the limbs and thus significantly influence the risk of injury at the moment of impact.
This raises a key question for accident research: How can these movements be simulated realistically?
This is where mathematical models come into play. Today, Finite Element Human Body Models (FE-HBM) are primarily used in numerical crash simulations. These high-resolution models describe the human body with hundreds of thousands of elements and enable detailed biomechanical analysis of stresses, tissue deformations, and possible injuries.
A prominent example is the THUMS® model (Total Human Model for Safety), which was originally developed by Toyota Motor Corporation and is now widely used in accident research. Such models provide very precise information about injury risks, but they have one disadvantage: the simulations are extremely computationally intensive and therefore only allow calculations over very short periods of time (i.e., tenths of a second or less).
The phase immediately before the actual impact, known as the Pre-Crash Phase, poses a particular challenge. It can last more than a second and involves active movements of the human body. A complete simulation of this phase using a finite element model would be extremely complex. Although theoretically possible, it would not be practical for the intended purpose.
Two Modeling Approaches – Two Strengths
Multibody models (Multibody System Digital Human Models, MKS-DHM) offer an alternative. In these models, the human body is described as a system of rigid segments with a few hundred degrees of freedom, which are connected to each other via joints.
The major advantage of this approach lies in its efficiency: movements can be calculated very quickly, allowing dynamic scenarios – such as braking maneuvers or reactions to vehicle acceleration – to be simulated accurately over a period of seconds.
However, multibody models also have limitations. Since the body segments are assumed to be rigid, biomechanical effects such as tissue deformation are not taken into account. This means that statements about injury risks are only possible to a limited extent.
From a mathematical perspective, both modeling approaches complement each other perfectly:
- Multi-Body Models are ideal for describing movements.
- Finite Element Models enable detailed analysis of stresses and injuries.
The challenge is to connect both worlds.
EMMA as a Bridge Between Movement and Crash Simulation
The EMMA (Ergo-dynamic Moving Manikin) simulation environment was developed at Fraunhofer ITWM for this purpose. EMMA is based on a multi-body model of the human body and allows efficient calculation of movements under various boundary conditions, such as vehicle acceleration or braking maneuvers.
In the »EMMA4Drive« project, EMMA was expanded to include an interface for finite element simulations. The aim is to transfer motion data from the multi-body model directly to a detailed FE Human Model.
The human model integrated into EMMA is based on the THUMS® model of a 50th percentile male. This refers to a statistical reference figure – specifically, an average male in terms of height, weight, proportions, etc.
This allows the postures and movements calculated in EMMA to be transferred to the FE model and then reused in a crash simulation. A practical advantage of this approach is that it greatly simplifies the otherwise very time-consuming positioning of the FE Model. Normally, such models must be brought into the desired body posture before a simulation through lengthy positioning calculations or manual adjustments. By coupling with EMMA, this position can be derived directly from a calculated movement.
New Project »EMMA2HBM«: Transferring Movements Between Two Model Worlds
One component of our work is the »EMMA2HBM« project, which investigates how movements from a multi-body human model can be transferred to a high-resolution finite element human model.
The focus is on comparing two digital representations of the same body: the THUMS® Finite Element Model of a 50th percentile male and the Multi-Body Model of the EMMA simulation software, which was developed based on the same reference model. While the finite element model takes deformable structures and tissue deformations into account, the multibody model describes the body as a system of rigid segments connected by idealized rotational joints.
These different mathematical modeling approaches mean that even with identical body postures, deviations can occur – for example, in the position of individual limbs or in the centers of mass of body regions. The first step of the project is therefore to systematically compare selected body poses and quantify possible modeling differences.
Based on this, it is investigated whether movements from the multi-body model can be transferred to the Finite Element Model. To do this, the FE Model is first stabilized under gravity and then excited via time-dependent force vectors calculated from the multi-body simulation.
If this approach proves successful, finite element human models could in future not only be positioned statically, but also controlled by calculated movement sequences. This would allow the kinematic efficiency of Multi-Body Models to be combined with the biomechanical detail of Finite Element Model – a promising step towards more realistic simulations of human movement and stress.
The »EMMA2HBM« project is a study conducted by the Fraunhofer network »Numerical Simulation of Products and Processes«, which we are carrying out in collaboration with Fraunhofer EMI (Fraunhofer Institute for High-Speed Dynamics, Ernst Mach Institute, EMI). The network works across institutes on tasks related to the development and improvement of simulation methods.
Find out more about EMMA on our website at: www.itwm.fraunhofer.de/emma
Connection to the THUMS® model for crash simulations: The THUMS®-based human model for evaluating new vehicle concepts and driver assistance systems. © Fraunhofer ITWM