In the face of the rapid growth in the scale and complexity of multidisciplinary systems, being able to develop reliable systems under ever-faster changing and more individual market requirements is becoming more and more challenging. The Model-Based Systems Engineering (MBSE) approach has already been researched heavily, and started to be introduced for the management of complexity, maintaining consistency, and reducing development costs and the time-to-market. However, a major drawback of the current MBSE methodologies is the lack of capability to integrate with domain-specific simulation models to investigate design concepts in the early phases of the development process. In order to address this issue, we propose a holistic system modeling approach that allows system engineers to link descriptive system models with domain-specific simulation models. In this paper, the Systems Modeling Language (SysML) is used as the standard architecture modeling language. A system modeling approach in SysML based on the system’s functional architecture for system design and validation is defined. The approach was developed to integrate domain-specific models into the system model using a SysML modeler with the capability of running and reusing simulation tasks via the coupling of external tools, which helps to bridge the existing gap between models on the system level and detail level. The feasibility of the proposed approach will be evaluated based on the case study of a wind turbine (WT) system. The study shows that our approach has the potential to enable the consistent, parameter-based interlinkage of domain-specific models based on always-up-to-date data, and to assist engineers in making design decisions to meet the system requirements accurately and rapidly in different engineering fields.

For the assessment of the Noise, Vibration and Harshness (NVH) behavior of complex systems in early phases of the development process, validated modelling methods are available that allow the prediction of structure-borne and airborne noise at system level. However, due to large model sizes, the identification of weak spots in the vibroacoustic transfer behavior and the derivation of optimization measures are highly complex, time-consuming, and mostly not practical. In the field of experimental NVH analysis, transfer path analysis (TPA) has been established as target-oriented methods for identifying vibroacoustic anomalies at system level. In this contribution, TPA methods from the field of measurement technology are selected and applied to numerical NVH system models, aiming for high accuracy and low computational effort in post-processing. The applications of the methods are shown and discussed using an elastic multi-body simulation model of a tractor drivetrain with a transient run-up of the vehicle speed as an example. The classical direct-force and component-based blocked-force TPA methods selected and adapted for this study allow for efficient calculation of the sound contributions of numerical models. At the same time, they overcome typical challenges of experimental TPA, such as exact force determination or consideration of rotational degrees of freedom. In addition, the comparison of the two methods shows that the path contributions are in general different for classical and component-based TPA and only under specific conditions the same. Both numerical TPA methods allow for the identification of weak spots in the NVH behavior in an efficient and target-oriented way.

Modal testing is used to experimentally determine the dynamic behavior of mechanical structures. The planning of the positions for exciting the structure, the so-called driving points, is essential for efficient experimental modal testing. Driving points are identified either by an initial assumption of possible excitation points and the experimental evaluation of their quality, or with the help of numerical models and indicators for optimal driving point selection. However, for practical applications, several driving points are usually required to excite all modes in the frequency range of interest, which is not covered in state-of-the-art indicators for driving point selection. This paper therefore presents a method for driving point identification with consideration of multiple driving points. Additionally, a criterion was developed that considers the excitability of modes, risks of double hits, and the excitation orientation simultaneously in order to increase the accuracy of the driving point identification. This criterion enables the evaluation of the excitation quality of each surface node and any mode combination using an automated selection method for determining the optimum set of driving points. The presented method assumes that natural frequencies and eigenvectors from preliminary numerical models are available for planning. By applying the method on application examples, the reduced effort required for a full modal analysis is demonstrated.