Towards Efficient Evaluation of Impacts of Geometrical Variation on Perceived Quality in Early Phases
In a perfect world, manufacturing would lead to components without geometrical variation. This would mean that every component produced in a series would have the exact same dimensions and, when observed by a trained eye, look exactly identical. However, the world is not perfect and neither are manufacturing processes.
When developing new products, this fact must be considered since a product, in most cases, consists of several components that are assembled together. The geometrical component variation may impact the function of the product, the ability to assemble the product and the aesthetics of the product. Geometrical variation may thereby affect the customer’s perception of the overall quality of the product. That is the reason why the appearance of the relationships between visible components, also defined as split-lines, in many cases is of exceptional importance in the automotive industry. This overall quality as perceived through the human senses is here referred to as Perceived Quality (PQ).
This thesis is concerned with methods for simulating and evaluating PQ in early phases of the product development process. Early simulation and evaluation of PQ will minimize the number of corrections in late phases, saving both time and money. The potential for savings is of substantial importance as projects in the automotive industry are conducted with ever decreasing budgets and with ever tighter schedules. However, in these early phases, the maturity of the data (i.e. geometrical models) used for simulation activities is low, which has earlier restricted the possibly to perform certain types of simulations.
The thesis includes a proposed method for how to perform non-rigid variation simulation in early phases when data maturity is low. It solves the issue of data immaturity by introducing so-called reference data linked to previous projects. Further, a framework is presented for managing and supporting evaluation of PQ during the development process based on the maturity of the data available. Some activities in the framework have been widely used in the automotive industry and some are based on recent research within the area.
An interview study is reported that investigated on what vehicle areas, non-rigid variation simulation was to be recommended before rigid variation simulation, in order to further support the work with PQ. A number of critical areas were identified along with a number of factors important when performing non-rigid variation in early phases. Further the need for non-rigid variation simulation was addressed by the respondents.
Finally, a user-study directed towards the visualization activity when evaluating PQ has been performed. Here, the ability to identify geometrical deviations on rendered images was investigated. The results showed that there is a significance difference in how certain relevant deviations (here meaning deviations normal to the surface extent) on non-rigid components are perceived, compared to how the corresponding deviations are perceived when performing rigid simulation.
visualization and evaluation