Improved injury prediction using HBM: step 2

Thoracic injuries are one of the main causes of fatalities and serious injuries in car crashes. This calls for improved restraints which, to date, are developed using virtual as well as physical crash test dummies. Virtual crash test dummies can only be used in specific crash scenarios, whereas virtual models that represent the human, commonly referred to as Human Body Models (HBMs), have the potential to simulate all crashes that occur in real world situations, including complex crashes with combinations of loads, e.g. oblique and small overlap crashes. Moving towards the goal of Vision Zero, there is an increased need of omnidirectional, human-like occupant tools to be used in these complex crash configurations, as well as for tuning collision mitigating and occupant restraint systems. However, there is a lack of well-defined and accepted criteria to predict thoracic injury and threshold when using these HBMs. This project focused on injury criteria for the thorax in frontal impact loading scenarios using the HBM Total Human Model for Safety (THUMS). Firstly, THUMS was used to evaluate injury criteria at the global, structural and tissue level and to develop injury risk curves for AIS2+ thoracic injury. Risk curves were generated based on post mortem human data from six different loading conditions. When the quality of the risk curves and how well they represented the post mortem human data were evaluated, the criteria for "shear stress' in two ribs and "DcTHOR' were most promising. Secondly, THUMS was used to simulate a representative statistical selection of real world frontal impacts and risk curves for five criteria were compared to the risk curves developed based on the real world outcome. The comparison displayed that THUMS with the developed risk curves consistently over-predicted the injury risks. As a complement to the statistical evaluation, THUMS was used for in-depth reconstructions of five real-world frontal impacts, illustrating that tissue level injury have a great potential to provide omnidirectional model specific criteria, but that they require a high model detail and are dependent on material properties and mesh quality to match real world data.

  • Corporate Authors:

    Fordonsstrategisk Forskning och Innovation (FFI)

    ,   Sweden 
  • Authors:
    • Brolin, Karin
    • Mendoza-Vazquez, Manuel
    • Davidsson, Johan
    • Jakobsson, Lotta
    • Pipkorn, Bengt
    • Mroz, Krystoffer
  • Publication Date: 2015


  • English

Media Info

  • Pagination: 16

Subject/Index Terms

Filing Info

  • Accession Number: 01664477
  • Record Type: Publication
  • Source Agency: Swedish National Road and Transport Research Institute (VTI)
  • Files: ITRD, VTI
  • Created Date: Mar 28 2018 10:26AM