Join our welcoming International Design Community.
Complete this application form to join now.
Author: Holley, Vincent
Supervisor: Yannou, Bernard; Jankovic, Marija
Institution: Laboratoire Génie Industriel, École Centrale des Arts et Manufactures, École Centrale Paris
Multi-physics systems design, including the design of mechatronics systems, involvings designers in different disciplines (e.g., mechanics, electronics, physics of sensors, etc.), particularly design for systems intended for operation in severe conditions (withstanding shocks, vibrations, high temperatures, and high pressures in limited dimensions), raises many of the challenging issues in the design of complex systems. Consequently, highly integrated products are characterized by multiple functional flows passing through common components. Very high performance requirements from the different designers may over-constrain architectural modules, as well as connections, and the performance of some functions. The integration of multi-physics functions within products of limited size that operate in severe conditions results in an intense interaction between design parameters and expected functionality. As soon as a design parameter is changed, the performance of several functions may be impacted. This is due to a high degree of performance optimization and the fact that several functions are part of the functional flow stemmingfrom a single component. In addition, some disciplines may be more constrained than others, depending upon given performance challenges and the concept architecture being considered. Hereafter, we refer to architectural modules, connections, and disciplines as constrainable objects. Today, with no prediction tool for locating the aspects that are likely to be highly constrained, consequences may be dramatic. For instance, project management for systems in the oil industry is often responsible for unacceptable additions to project overhead costs and project timelines for a project that may simply fail in the end.
In our study, we propose to semantically enrich conventional representation models of product complexity. We use a design structure matrix (DSM) to represent admissible architecture connections and dependency configurations, a domain mapping matrix (DMM) to link functions and architecture, and quality function deployment (QFD), in a non-conventional way, in order to propagate the designers’ aims for performance of the components more than the traditional “voice of the customer”. We enrich DSM representations with a physical connection typology, allowing a range of choices at an early design stage. For a given connection, information regarding the nature of likely design difficulties is incorporated into a data model. We enrich DMM representations with functional flow sequencing along the architectural modules. We adapt the QFD method to capture the voice of the""engineering disciplines involved in the project; this ontological enrichment of design data makes it easier to envision and manage design challenges for multi-physics systems. Seven design assessment cards are proposed to the design team as meaningful tools used to converge from a set of potential architectural configurations towards a single architecture. This convergence process is driven by the necessity of avoiding highly constrained constrainable objects, achieved by balancing and spreading the design constraints throughout the system. The seven assessment cards are organized into two major design quality vectors: the ambition vector and the difficulty vector. The ambition vector indicates degrees of freedom in exploration of the architecture design space. The difficulty vector offers heuristic information on the nature and levels of the difficulties in meeting performance targets. The resulting method, which we call the multi-physics design scorecard (MPDS), was applied to the design of a power electronics controller (PEC), a regulator board involving three sectors: mechanics, electronics, and packaging. Data gathering and implementation of the MPDS method took the design team just one day. The method immediately generated improved architectures, guaranteeing at the same time a more robust further design process.