174 A.S. Mohammed and A. Pavic Fig. 21.4 Schematic overview of the results comparison approaches 21.4.2 Discussion Section 21.4.1 provided comparison between the measured and simulated vibration responses in terms of vibration magnitudes up to 90% probability of non-exceedance and the maximum 1-second RMS. Hence, they can be used to describe the performance of the simulated models. Simulated vibration responses for all non-interactive models are found to significantly overestimate the measured vibration responses in most of the cases, especially for the resonant responses (Fig. 21.5 and Table 21.2). Less conservative vibration responses are generally obtained when any of the interactive models are used. This is in line with previous findings that neglecting the interaction between walking people and the supporting structure can result in overestimation of vibration responses [4, 15]. Interestingly, none of the utilised HSI models of walking people performed particularly better than the others. However, the models presented by Silva [8] and Alonso [6] appear to perform slightly better than the others in most of the simulation cases (Fig. 21.5 and Table 21.2). It is worth mentioning that the simulated vibration responses using any of the HSI models has less difference from the measured responses in the non-resonant cases (when pacing frequency is 2.05 Hz) than in the resonant case (1.90 Hz). This could indicate sensitivity to damping or amplitude-dependent parameters of the HSI models of walking humans. Another possible source of error is the variability of walking human model parameters between different people (intersubject variability). This could be better addressed by repeating the test and simulations for a larger number of people. The utilised measured walking forces model appears to have a non-negligible effect on the accuracy of the simulated vibration responses (Fig. 21.5 and Table 21.2). Therefore, despite the improved accuracy of the simulated responses, when any of the HSI models of walking humans is used, it can be even more improved when a more realistic walking forces model is used (the measured forces). This emphasises the importance of utilising a narrow-band near-periodic walking force model instead of the traditional periodic walking force model. Figure 21.8 shows an example of measured and simulated vibration responses when a measured force and an interactive model are used.
RkJQdWJsaXNoZXIy MTMzNzEzMQ==