Unveiling Variety-induced Complexity in Vehicle Development: Towards Complexity Management Using Change Prediction

Abstract

The dimension of product variety can turn complicated product development into an extremely complex task. Meanwhile, studies have shown that the development effort grows super-linearly with increased complexity. Differentiated products and speed of change can however be vital to meet the customer on her terms. The term variety-induced complexity was invented to describe the costs brought on by differentiation and numerous publications have been dedicated to study ing them. Perhaps even more studies have been dedicated to the strategies used to keep them to a minimum. One of the most prolific strategies is design re-use, deriving product variants from a common platform. It is a strategy to allow product variety and manage the complexity. Variety is still omnipresent, a fact companies need to deal with. Complexity costs during product development are often obscure and relate to the time design engineers spend on non-value adding activities. Mean while, the development task undergoes a small revolution. A ’good’ design in an individual product is one that satisfy requirements. In a platform, it is also one with a positive contribu tion to the platform’s complexity costs over time. Complexity creates complexity, and if not managed properly it creates an inertia that threaten the long-term success of product platforms. This thesis studies an established automotive company and identifies that decision are often made in the disfavour of the product platform. The organisation, culture and information struc tures are in many ways biased to the benefit of individual products. We propose a framework that places the configurable design in the centre of decision-making. The aim is to give design engineers means to model and optimise the design and configuration space of a product platform to minimise complexity costs while meeting the needs of a range of customers. In early design phases, the functional bandwidth the system presents to the customer is de fined. By modelling the functions and the different ways to solve them, the system’s design bandwidth is defined. Adding knowledge about how different solutions interact allows us to eval uate how coupled the design is. That is, by simulating changes to the design, and analysing how they propagate, the complexity and robustness of individual system variants can be assessed. This framework was applied to external vehicle Rear-view Mirrors (RVMs). By modelling an existing design, we showed that knowledge can be generated to support a selection of the ar chitectural options and configurations with the least negative effects on complexity costs and robustness. That is, some solutions tend to increase the risks in the system and should be har monised against their value. The framework also appeared promising to efficiently model and compare many configurations during design space exploration and to identify variants likely to be affected by a re-design.