Classifiers to Augment the CDC System to Distinguish the Role of Structure in a Frontal Impact Taxonomy 2012-01-0575

The purpose of the study was to distinguish the role of vehicle structure in frontal impacts in published coded National Automotive Sampling System (NASS-CDS) data. The criteria used: Collision Deformation Classification (CDC) coding rules, crush profile locator data and the projected location of longitudinal structural members in models of vehicle class sizes used by NASS-CDS. Two classifiers were developed to augment the CDC system. The Coincidence classifier indicates the relationship between the quadrant of the clock face the crash vector originates in and the aspect of the end plane the center of damage is located. It has three values: Linear (12 o'clock impacts) Consistent and Variant ("oblique" Principal Directions of Force or PDOFs). The second classifier indicates the number of longitudinal members engaged: 0, 1 or 2.

NASS-CDS data for sample years 2005 to 2009 was filtered for occupants involved in impacts with the highest ranked speed change assigned to the front-end plane. The set excluded impacts assigned a PDOF greater ± 50 degrees and small over-lap (SOI) type impacts. A set of 12,165 raw records (4,655,471 weighted) for occupants in wide type impacts that extending from the bumper to the belt line or above was collected and subdivided into linear and oblique categories.

1,910 raw records involved Consistent impacts; 75 (3.9%) occupants (16,481 weighted or 3.6%) were in impacts involving 0 rails. 164 (8.6%) or 40,189 weighted (8.7%) involved 1 rail and 1,671 (87.5%) records or 405,022 weighted (87.7%) involved 2 rails. 160 raw occupant records involved Variant impacts, 2 (1.2%) or 73 occupants weighted (≻1%) involved 0 rails. 80 (50%) or 20,175 weighted (43.9%) involved 1 rail and 78 (48.8%) of the raw records (25,682 weighted or 56%) involved 2 rails.

To evaluate the theory of the Coincidence classifier, a matrix of impact configurations were submitted to NHTSA for simulation. Impacts predicting head motion beyond the perimeter of the steering wheel were deemed oblique. Twenty-one simulations with vectors of 10, 20 and 50 degrees resulted in eight predicted occupant kinematics consistent with an "oblique" impact.