Structured Methods for the Design and Optimization of Modular Embedded System Platforms: Axiomatic design, design structure matrices and optimization methods in architectural system design

Abstract

Interest towards electric mobility is on the increase due to global climate targets and the development of electric vehicle (EV) technologies. While EVs were not a satisfactory alternative to internal combustion engine (ICE) vehicles before, most problems have already been solved. Consequently, design and implementation of large-scale charging infrastructure is the next step in supporting the growing fleet of EVs around the world. Such infrastructure requires complex systems composed of mechanics, electronics and software, and suitable design methods are therefore needed. Methods which solve complex design problems generally deal with multiple-criteria decision making (MCDM).

In this thesis, the architectural design of modular embedded system platforms is studied. The thesis consists of three parts. In the first part, the concepts of design, architecture and modularity are introduced and a selection of existing design methods is shown. Based on a comparison between these methods, axiomatic design and design structure system are selected for further analysis. Subsequently, the theoretical bases for axiomatic design and design structure system are introduced and an integrated design flow is shown. Novel methods for design parameter utilization and modularity cost analysis are also developed. In the second part, the common properties and functionalities of public EV charging devices are introduced. In the third part, an exemplary EV charging platform architecture is designed using the introduced methods. Five different architectural options are evaluated and a single one is selected based on developed optimization indicators. Finally, an overview of the effects, risks and advantages of structured design methods and modularity is given.

Architectures can be categorized in two ways: Based on the level of modularity and based on production paradigm. There is also a mapping between these two views: Modular architectures are most suitable for mass customization and mass individualization, whereas integral architectures are most suitable for mass production. While many design methods for modular architectures exist, axiomatic design and design structure system represent strong contendants. In general, using axiomatic design tends to result in modular functional architectures, but it does not impose restrictions on physical modularity. Therefore, physical design must be done using another method such as design structure system. These methods and the novel cluster utilization and cost-analysis methods are also used successfully in the thesis to design and optimize an exemplary EV charging system platform. While structured design methods and modularity offer many advantages, implementing them in an R&D organization also poses some risks and implementation must therefore be carefully planned in advance.