A Novel Framework for Modeling Reconfigurable Dynamic Systems

Abstract

Modular and reconfigurable design has gained significant attention in manufacturing systems, robotics, space, and automotive industries, aiming to reduce costs and enhance performance. While modal analysis traditionally provides the relationship between dynamic response and frequency, its applicability to reconfigurable systems is limited particulary for high frequency range. To tackle this limitation, this research proposes a novel approach for modeling reconfigurable dynamic systems, leveraging recent advancements in sub-structuring, impedance, and admittance techniques. The objective is to comprehensively understand, develop, and demonstrate the general theories of modeling reconfigurable systems, with a focus on applications in automotive engineering. In contemporary automotive engineering, achieving superior ride comfort and minimizing noise and vibration levels is paramount. This is especially critical for electric vehicles (EVs), where predicting road noise presents unique challenges due to the added weight of battery packs. This thesis introduces a Generalized technique to incorporate Frequency-Based Substructuring. This framework facilitates the development of mathematical models for multi-DoFs reconfigurable dynamical systems, allowing for a detailed analysis of vibrational responses within the automotive context. The GRCFBS approach enables the prediction and evaluation of the overall system response by analyzing the combined receptance matrix or FRFs of an assembly. This is achieved through a generalized mathematical algorithm that combines the subsystem components' Frequency Response Functions (FRFs). The approach systematically employs a reduced-order model by focusing on specific measurement and excitation points with only limited number of degrees of freedom (DoFs) at either connection region between subsystems or the internal region within individual subsystems, while still accounting for the essential translational and rotational DoFs required for modeling subsystems interaction. Subsequently, based on the excitation forces and the availability of the FRF (receptance) matrix, we can predict the overall system response using superposition and linear approximation. By systematically decomposing the system into its constituent subsystems and analyzing their independent responses, this method aids in system identification and understanding the interaction between subsystems. This technique not only helps in reducing model size, but also provides valuable insights into the transfer mechanisms of vibrations and noise throughout subsystem interfaces and connection points, contributing to noise and vibration mitigation in the automotive sector. Moreover, this methodology develops efficient reduced-order models, significantly reducing the need for time-consuming and expensive simulations, especially during the initial phases of development. By incorporating data from individual subsystem measurements, we can predict the overall system response based on focusing on a limited number of critical points regardless of complexity of the geometries, even when certain measurement points are inaccessible using traditional modal testing methods. Additionally, this methodology can potentially develop hybrid models that combine experimental and numerical data from various subsystems. Nonetheless, modeling of the interactions among the subsystems presents a significant challenge, which this study is also focused on. A variety of case studies illustrate the practical applications of this method, with further theoretical developments enhancing the reliability of the research findings.