Visualising the Effect of Geometrical Variation in Assembled Products. Predicting Visual Quality Appearance

Doctoral thesis, 2005

Today, when an industrial design concept is evaluated in an aesthetic manner, little or no consideration is given to geometrical variation, i.e. the concept is evaluated with nominal geometry. If geometrical variation were considered as early on as in the concept phase, when industrial design concepts are evaluated, the possibility of discovering aspects that can influence the visual quality appearance would be enhanced. By using non-nominal models during the design process, important geometric aspects can be stressed and the need for physical prototypes and test series can be reduced. In the automotive industry, the relationships between different visible parts of the interior and exterior are critical for the overall visual quality appearance of the product. These relations should have a tight and parallel appearance. This thesis presents methods and a tool for predicting how geometrical variation affects visual quality appearance. The adopted approach has been to combine computer-aided tolerance analysis with high qualitative computer graphics in order to visualise the effect of geometrical variation. The two main relations - gap and flush of a split line - are adjusted in accordance to calculated geometrical deviation. Furthermore, the ability to judge distance in virtual environments is also investigated. If virtual non-nominal models are to be used for evaluation of variation, it is essential to have knowledge about the difference between the estimated and real actual distances. Otherwise, decisions are made based on the wrong facts, and this contributes to increased uncertainly. An initial study showed that there is a significant difference in perception between physical and virtual representations. A second study investigated how ten different parameters influenced the ability to judge distance. Significant main effects and interactions were identified for the two relations, gap and flush. Finally, all methods and the tool have been implemented in industrial projects in order to evaluate whether developed methods and tools actually work as intended. Qualitative analysis of collected data showed that proposed methods and tools gained high acceptance from both industrial designers and engineering designers. Although many aspects still remain to be investigated further, the final conclusion for this project is that tremendous progress has been achieved towards the aim of predicting visual quality appearance in an accurate and reliable manner.