Multi-disciplinary design of a vibratory rammer with surrogate models and solution spaces
Please login to view abstract download link
A solution space–based approach to systems design has proven effective in determining admissible ranges of uncertain component properties. On these ranges, performance requirements are met [1]. An industry-relevant application of the proposed approach is the design of a vibratory rammer for soil compaction, previously investigated in [2], where hand–arm vibration reduction and machine maneuverability were addressed using simplified models of the machine and machine–soil interaction. That work enabled efficient requirement cascading; however, soil behavior is very simplified, and compaction performance was not considered. This work presents an extension that integrates a high-fidelity three-dimensional multi-body simulation model of a vibratory rammer with an advanced nonlinear soil constitutive model, thereby enabling physics-based evaluation of compaction performance, in addition to hand–arm vibration and maneuverability. The soil and the machine model are validated with acceleration data collected from sensors during the vibratory rammer operation. A design of experiments is conducted over the combined domains of machine design variables, operational parameters, and soil parameters. High-fidelity simulation results are used to train surrogate models, which facilitate efficient exploration of high-dimensional design variable spaces. With the surrogate models, solution spaces are optimized. Requirements for two types of essential design variables are generated: First, point-based target values for design variables without uncertainty, such as the operation frequency and the crank amplitude, and second, design variables with uncertainty, such as the rubber mount stiffness or mass distribution.
