Accelerating NVH optimization of eMBS system models through efficient parallelization of transient runup simulations
摘要
The modern, ever shorter development process of electrified drivetrains requires methods that enable a sufficiently fast acoustic evaluation of design decisions. The dynamic and acoustic behavior (NVH) forms a crucial aspect for the consumer’s purchase decision. Since experimental and empiric optimization approaches are highly inefficient, simulation methods such as the elastic multi-body simulation (eMBS) are used to sufficiently simulated the NVH-behavior of the drivetrain by performing transient run-up simulations. However, these simulations are highly computationally intensive, leading to computational times of multiple days or weeks, thus making those optimization approaches inefficient during iterative phases of the development process. Therefore, there is a need to reduce the overall simulation time for eMBS-based evaluation of design decisions.
This publication proposes a method on how to parallelize transient runup simulations to significantly decrease the needed simulation time for eMBS simulations. In this process, a partition of the time domain is achieved by analytically calculating the kinematic states of individual components at discrete time steps. As a result, a multitude of runup partitions can be computed in parallel, leading to a drastic reduction in simulation time. This method is demonstrated by examining an electrified powertrain. The application of the method shows a reduction in simulation time for a transient drivetrain runup of over 90%. The validation is carried out by comparing the parallelized computational results to a conventional transient simulation of the electrified drivetrain. Based on the proposed method the engineer is capable to efficiently use detailed eMBS system models within highly iterative phases of a development process for the evaluation of design decisions and therefore address NVH-optimizations more sufficiently.