Multi-layer Graph Theory Utilisation for Improving Traceability and Knowledge Management in Early Design Stages
Fecha
2017-01-01
Título de la revista
ISSN de la revista
Título del volumen
Editor
Elsevier
Resumen
Decision making processes in design often challenges designers to prioritise specifications and variables in order to develop solutions that are closer to the product's requirement goals. Concerning to support their decisions, different tools and methods are used by engineers and designers allowing to reduce uncertainty in design. Nevertheless, many of these decision support systems are focused in late design stages, such as detailed design and manufacturing design, even if the possibility to influence a new product is higher in early stages. The issues regarding to those situations are often associated to design processes related to multi-physics design, where the modification of geometric-related variables might affect the performance of the solution, and the analysis of tracking the influence of the modifications might generate reprocessing and loses of time, specially when those relations are tricky and are not easily identifiable by analysing equations and a manual analysis of requirements must be performed. This article is centred in proposing a traceability model for early design stages based in graph theory. The proposal supports the information generated in design, from the input requirements (linguistic field) up to mathematical modelling and variables definition (real numbers field). This information is arranged into different layers, allowing a multilevel approach in terms of information management. The model also features a novel solution for weighting vertex in graph model, featuring a model that balances the direction of improvement, the importance and flexibility of any specification and how its behaviour will affect the design variables associated to it. The goal of the proposed model is to offer to designers, since the conceptual design stage, a method that can show automatically the level of correlation between any pair variables and specifications by the use of information trees and featuring chains that can connect them whether there is or not a connection via equations. © 2017 The Authors.