Managing dependencies in mechatronic design: a case study on dependency management between mechanical design and system design

Abstract

In this paper, we have investigated the role of dependencies in the design process of mechatronic products. Since explicit modeling of dependencies is largely considered unnecessary today, current languages do not support dependency modeling due to lack of sufficiently expressive language constructs. However, this paper argues that modeling dependencies is important in managing the overall design process. The paper highlights dependencies between two important viewpoints: system design and mechanical design. We have looked closely at how mechanical design (supported by CAD tools) establishes a backbone for the overall design concept. Mechanical design cannot be isolated from other design activities, and the mismanagement of dependencies there leads to problems in other domains too. To illustrate the process, the paper presents an example of modeling dependencies between system hierarchy in OMG SysML™ and the CAD assembly in Solid Edge for a mechatronic design example. The paper presents two different approaches to capturing dependencies--using a general purpose modeling language such as SysML and using a domain specific modeling language (DSML). We argue for using a DSML instead of a general purpose language and provide a DSML called the dependency modeling language (DML). An example DML model for a two degree of freedom robot use case is discussed. The paper also illustrates the complete process of capturing dependencies in a general purpose modeling language like SysML, which served as a good exercise on how to fetch data from a CAD tool and how to represent dependencies inside a significantly different modeling language. Lessons learned from doing this were applied to the construction of DML. Our aim for the future is to reduce the human effort required to build dependency models. Machine learning techniques and automated model transformations are valuable techniques to support this cause.