Effectiveness assessment of VRU protection systems

DI Peter Wimmer
Lead Researcher for occupant and VRU safety
8-Vehicle-Safety
Vulnerable road user (VRU) protection is one of the main research topics in vehicle safety. To assess the potential of different types of VRU protection systems for reducing injury severity under real-world conditions, VIRTUAL VEHICLE, in cooperation with BMW, has developed a unique simulation method. It provides detailed, vehicle-specific results across a broad range of critical situations while still being fast enough to be used in a standard vehicle development process.

When designing the VRU protection system configuration in a new car, one main question arises: What is the best system or combination of systems to minimize VRU injury severity under realworld conditions? Only limited testing is possible in this case, due to the large number of possibilities and the need to provide results already in early product development phases. Simulation can overcome
these shortcomings and deliver results on time. However, setting up an adequate simulation method and providing suitable solutions poses enormous challenges.

Significant challenges

To compare the effectiveness of different types of protection systems (active, passive, and integrated), a continuous tool chain covering all aspects from ordinary driving to in-crash is required. This is the first challenge when attempting to assess effectiveness. Many situations have to be considered to be close to real-world effectiveness. These situations should cover the whole range of different accident constellations. This ensures that the system performance is checked under both unusual and common conditions. Stochastic generation of critical situations based on real-world accidents addresses this challenge.
Simulating thousands of situations means the method has to be fast enough to provide results within a reasonable time. One approach would be to skip the slow simulation of the in-crash phase and replace it with general injury risk curves. However, this approach is unable to provide results for specific passive and integrated protection systems. Finding a way to reduce calculation time while still obtaining systemspecific results is the third challenge.

Key factor 1: Continuous tool chain

The first main element is an automated, continuous tool chain, which covers the whole accident situation from ordinary driving to in-crash. This is accomplished by using co-simulation to couple models of all relevant elements (vehicle, driver, sensors, safety systems, and environment) and a method for seamless exchange of tools for different accident phases during simulation (e.g. driving dynamics during precrash and finite element method (FEM) during in-crash phase to simulate the vehicle’s mechanical behaviour).

Key factor 2: Stochastic generation of critical situations

The number of documented real-world accidents is limited and will not cover the whole range of possible accident situations. Therefore, a method to create an arbitrary number of critical situations is needed. At the beginning, we start with real-world accident data and use this basis to generate new situations via a stochastic approach. This method ensures that the overall distributions of the generated accident parameters are similar to the distributions in the accident databases.

Key factor 3: Reduction of calculation time

Key factor 1 is needed to obtain detailed, vehicle-specific results. This approach includes complex FEM models which need hours to days for one run. These models are the bottleneck for the overall simulation and are too slow to allow for a large number of simulations. Therefore, an alternative is needed to obtain equally detailed results within an acceptable time. The problem has been solved by replacing the FEM models with non-physical black-box models. The great advantage of the black-box models is that, once they have been trained, they are very fast and provide a good approximation of the FEM models. These surrogate models need less than a minute to calculate the 10,000 cases used in the application example, which is described below.

Application example

As an example, the method is used to determine the effectiveness of three protection system configurations: an autonomous emergency braking (AEB) pedestrian system, an active bonnet (the bonnet pops up in the case of a vehicle-pedestrian collision to increase the deformation zone during impact), and a configuration that integrates both systems. The use-case is defined by 10,000 critical scenarios of pedestrians crossing the road from both sides which lead to an accident if there is no intervention by the driver or the AEB system. The AEB system reduces the accident rate by about 50%. In the remaining accidents, the speed reduction due to AEB intervention has more impact on reducing head injury severity than the active hood. Another benefit of the AEB system is that it also reduces leg injury, whereas the active hood only helps to reduce the head injury. Fig. 4 shows an overview of the results of this example.

Summary

The method developed by VIRTUAL VEHICLE allows for an effectiveness assessment of active, passive and integrated VRU protection systems under real-world conditions when using identical injury criteria. It includes a continuous tool chain for simulating an arbitrary number of accident scenarios from ordinary driving to in-crash without user interaction. It is sensitive to specific vehicle geometry, mechanical properties and the protection system configuration. The method is also easy to adapt, due to the modular co-simulation approach. It is perfectly suited to finding an optimal configuration of VRU protection systems, thereby making the roads safer for the most vulnerable traffic participants as well.



CONTACT US


I want to know more about VIRTUAL VEHICLE.

I agree that VIRTUAL VEHICLE will use the data I have provided for their own information and marketing activities relating to products, services and events. This declaration of consent can be revoked at any time by me.

By pressing the "Send" - button you agree to the Data Protection Notice.