This content is not included in your SAE MOBILUS subscription, or you are not logged in.
Similarity between Damaging Events Using Pseudo Damage Density
ISSN: 1946-3995, e-ISSN: 1946-4002
Published November 10, 2020 by SAE International in United States
Citation: Altmann, C. and Ferris, J., "Similarity between Damaging Events Using Pseudo Damage Density," SAE Int. J. Passeng. Cars - Mech. Syst. 13(3):2020, https://doi.org/10.4271/06-13-03-0016.
Load-time histories can be used to predict vehicle durability by calculating the pseudo damage (PD) through one or more load paths for a vehicle. When the dynamics of each load path are taken into account, a PD density (damage per distance traveled) can be expressed for each load path for any given road input to a vehicle. When damage is expressed as a PD density for a segment of road, separable damaging events can be identified using the PD density in all load paths of interest for a vehicle. However, it would be beneficial if events with similar damage characteristics can be identified and grouped together to provide an additional level of durability information. The objective of this work is to develop a similarity test for identifying the similarity/dissimilarity between multiple damaging events using the damage characteristics in multiple load paths. The damage characteristics for events are defined using the distribution of PD density samples for all known load paths. The similarity test developed is paired with existing clustering algorithms to identify groups of damaging events based on user-provided information. A proof of concept is provided using 25 damaging events identified from a Federal Highway Administration Long-Term Pavement Performance program test site. For each damaging event, PD density samples are known for two load paths (spring and damper for a Golden Quarter-Car). To better understand the extension of the analysis to customer usage, the 25 events are broken into two sets: durability test events and customer usage events. Based on the selection of a clustering algorithm and a rejection criterion, two examples of clusters are achieved. The first example provides the best matching durability test event to a customer usage event, and the second example provides the best grouping of damaging events, independent of the damaging event original definition (e.g., durability test event or customer usage event).