The objective of this project was to establish a process to perform reliability-based design optimization for improved robustness and understanding of variability inside door slam durability performance. The existing analysis process assesses only the nominal door design. The updated process was automated to include a large-scale DOE which defines the range of durability performance. Then a large-scale Monte Carlo simulation is run to provide a probabilistic distribution of the predicted durability performance.
Many variables are included in the DOE based on manufacturing build tolerance, material & weld properties, and material thickness variation due stamping/thinning. 300 finite element model runs are done to complete the DOE and define the fatigue performance domain. From these full FEA model runs, Kriging surfaces are created to quickly provide estimated performance for 4096 Monte Carlo simulation data points. These data are used to define the variability distribution of the fatigue life performance.
A fatigue life distribution is automatically generated for any spot weld or finite element (fatigue crack initiation point) calculated to have 25 lives or less. Customer usage over the life of the vehicle represents 1 life. Our CAE target traditionally has been 1.6 life based on Weibull distribution, Generally, before this automated process was put in place, our engineering best practices would require our team to investigate any location with less than 5 lives. Individual manual investigation was time prohibitive to pursue issues with predicted durability above 4 lives. Traditionally this proved to drive robust results; this process with 1.6 life target has been in place for 25+ years. With this new automated process, we expanded our search to include any locations with predicted durability up to 25 lives to increase the robustness of the designs.