Comparison of anthropomorphic test dummies with a pediatric cadaver restrained by a three-point belt in frontal sled tests

Validation data for assessing dummy child biofidelity are limited, especially with regard to whole-body kinematics. Therefore, the goal of this study was to assess the kinematic biofidelity of current child dummies relative to results obtained from analysis of a child cadaver sled test. The baseline data were obtained from a previously unpublished test performed with a 13-year old pediatric cadaver restrained by a three-point belt. The cadaver test conditions were reconstructed using two dummies with anthropometry closest to that of the cadaver, the HIII 10-year old and HIII 5 female dummies. Due to anthropometric and age-equivalent differences between the dummies and the child cadaver, geometric scaling was performed on the signals based on the seated height and material properties. Kinematic evaluations of head, hip, and knee trajectories were obtained from film analysis. Accelerations of the head, shoulder and lap belt loads were measured and compared among the dummy and child cadaver data. While this study shows that the HIII 10-year old, scaled HIII 5th female and scaled pediatric cadaver reasonably agree for the shoulder belt force, the resultant head acceleration, and the maximum head excursion, differences in kinematics were identified between the dummies and the cadaver. Some of these differences in dummy kinematics were attributed to nonbiofidelic motion of the rigid thoracic spine with extensive bending at the cervical and thoracic spine junction. In addition to new cadaver data, the study provides insight into the applicability of geometric scaling for dummy evaluation and suggestions for improved dummy biofidelity. INTRODUCTION The design and evaluation of current child restraint systems relies heavily on the biofidelity of current anthropometric test devices. Given the paucity of biomechanical data available for both development and validation of child dummies, child response targets have largely been achieved through geometric and material property scaling of adult responses. Dimensional scaling, however, involves a number of assumptions in terms of geometric similarity and loading of homologous structures that may not be justified with the differences in regional dimensions and mass distributions between children and adults (e.g., the child head comprises a disproportionate share of the overall body mass relative to the adult head). Furthermore, the scaling of response for material property from adults to children are usually limited to one tissue type (e.g., bone) within a body region whereas the actual response typically involves the composite response effects of a large number of soft and hard tissue types. Therefore, it is essential that dummies are evaluated under whole body loading conditions similar to the test environment in which they will be used to design restraint systems. While reconstructions of crashes involving children provides valuable injury data (c.f., Ash et al., 2009), the lack of information regarding initial occupant, restraint, and vehicle conditions introduces large uncertainties into reproducing these events in the laboratory for the purpose of dummy validation. For the validation of occupant kinematics, pediatric cadaver tests arguably provide the best condition for evaluation of dummy seat belt and booster seat restraint of children (Kallieris, 1976) despite some differences in muscular effects between dummies and cadavers. In the study presented here, data from a