108 M. R. Seyedi et al. Fig. 11.1 Different platforms to conduct the rollover accident: (a) rollover sled test [3]; (b) Modified dolly rollover for heavy vehicles [4]; (c) Standard dolly test Fig. 11.2 The sensitivity analysis process inertial properties, the center of the gravity (COG), and overall dimensions were extracted. Then, the data were used to build the dynamic model of the bus in PC-Crash software. The details about the theory of dynamic model and its accuracy and reliability can be found in several studies [5–10]. Next, to validate the model, the results from PC-Crash included overall motion, a number of quarter turns (Nq), roll distance (Rd), and angular rates were compared with similar experimental results. Figure 11.2 shows the steps that have been followed to perform the sensitivity analysis using the simulation results. The input parameters with proper range and uniform distribution were selected to create the design space using Latin Hypercube Sampling (LHS). Among various available surrogate models [11–13], Kriging method was used to build the response surfaces for all outputs. Finally, after checking the accuracy of the model the sensitivity of selected rollover outcomes to the input parameters was evaluated by computing the standardized regression correlation (SRC). By comparing the results of SA given by these methods, this study provides an insight into the relationship between initial conditions of dolly rollover test and rollover outcomes.
RkJQdWJsaXNoZXIy MTMzNzEzMQ==